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} -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{]}} -fdump-go-spec=@var{file}
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 -fplan9-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 -fobjc-std=objc1 @gol
216 -freplace-objc-classes @gol
219 -Wassign-intercept @gol
220 -Wno-protocol -Wselector @gol
221 -Wstrict-selector-match @gol
222 -Wundeclared-selector}
224 @item Language Independent Options
225 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
226 @gccoptlist{-fmessage-length=@var{n} @gol
227 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
228 -fdiagnostics-show-option}
230 @item Warning Options
231 @xref{Warning Options,,Options to Request or Suppress Warnings}.
232 @gccoptlist{-fsyntax-only fmax-errors=@var{n} -pedantic @gol
233 -pedantic-errors @gol
234 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
235 -Wno-attributes -Wno-builtin-macro-redefined @gol
236 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
237 -Wchar-subscripts -Wclobbered -Wcomment @gol
238 -Wconversion -Wcoverage-mismatch -Wcpp -Wno-deprecated @gol
239 -Wno-deprecated-declarations -Wdisabled-optimization @gol
240 -Wno-div-by-zero -Wdouble-promotion -Wempty-body -Wenum-compare @gol
241 -Wno-endif-labels -Werror -Werror=* @gol
242 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
243 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
244 -Wformat-security -Wformat-y2k @gol
245 -Wframe-larger-than=@var{len} -Wjump-misses-init -Wignored-qualifiers @gol
246 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
247 -Winit-self -Winline @gol
248 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
249 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
250 -Wlogical-op -Wlong-long @gol
251 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
252 -Wmissing-format-attribute -Wmissing-include-dirs @gol
254 -Wno-multichar -Wnonnull -Wno-overflow @gol
255 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
256 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
257 -Wpointer-arith -Wno-pointer-to-int-cast @gol
258 -Wredundant-decls @gol
259 -Wreturn-type -Wsequence-point -Wshadow @gol
260 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
261 -Wstrict-aliasing -Wstrict-aliasing=n @gol
262 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
263 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]} @gol
264 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
265 -Wsystem-headers -Wtrampolines -Wtrigraphs -Wtype-limits -Wundef @gol
266 -Wuninitialized -Wunknown-pragmas -Wno-pragmas @gol
267 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
268 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value @gol
269 -Wunused-variable -Wunused-but-set-parameter -Wunused-but-set-variable @gol
270 -Wvariadic-macros -Wvla -Wvolatile-register-var -Wwrite-strings}
272 @item C and Objective-C-only Warning Options
273 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
274 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
275 -Wold-style-declaration -Wold-style-definition @gol
276 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
277 -Wdeclaration-after-statement -Wpointer-sign}
279 @item Debugging Options
280 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
281 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
282 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
283 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
284 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
285 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
286 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
287 -fdump-statistics @gol
289 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
293 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
299 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
300 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
301 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
302 -fdump-tree-nrv -fdump-tree-vect @gol
303 -fdump-tree-sink @gol
304 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
305 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
306 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
307 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
308 -ftree-vectorizer-verbose=@var{n} @gol
309 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
310 -fdump-final-insns=@var{file} @gol
311 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
312 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
313 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
314 -fenable-icf-debug @gol
315 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
316 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
317 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
318 -fstack-usage -ftest-coverage -ftime-report -fvar-tracking @gol
319 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
320 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
321 -ggdb -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
322 -gvms -gxcoff -gxcoff+ @gol
323 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
324 -fdebug-prefix-map=@var{old}=@var{new} @gol
325 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
326 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
327 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
328 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
329 -print-prog-name=@var{program} -print-search-dirs -Q @gol
330 -print-sysroot -print-sysroot-headers-suffix @gol
331 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
333 @item Optimization Options
334 @xref{Optimize Options,,Options that Control Optimization}.
335 @gccoptlist{-falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
336 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
337 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
338 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
339 -fcheck-data-deps -fcombine-stack-adjustments -fconserve-stack @gol
340 -fcprop-registers -fcrossjumping @gol
341 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules @gol
342 -fcx-limited-range @gol
343 -fdata-sections -fdce -fdce @gol
344 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
345 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffast-math @gol
346 -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
347 -fforward-propagate -ffp-contract=@var{style} -ffunction-sections @gol
348 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
349 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
350 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
351 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg @gol
352 -fipa-pta -fipa-profile -fipa-pure-const -fipa-reference @gol
353 -fipa-struct-reorg -fira-algorithm=@var{algorithm} @gol
354 -fira-region=@var{region} @gol
355 -fira-loop-pressure -fno-ira-share-save-slots @gol
356 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
357 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
358 -floop-block -floop-flatten -floop-interchange -floop-strip-mine @gol
359 -floop-parallelize-all -flto -flto-compression-level
360 -flto-partition=@var{alg} -flto-report -fmerge-all-constants @gol
361 -fmerge-constants -fmodulo-sched -fmodulo-sched-allow-regmoves @gol
362 -fmove-loop-invariants fmudflap -fmudflapir -fmudflapth -fno-branch-count-reg @gol
363 -fno-default-inline @gol
364 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
365 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
366 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
367 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
368 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
369 -fpartial-inlining -fpeel-loops -fpredictive-commoning @gol
370 -fprefetch-loop-arrays @gol
371 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
372 -fprofile-generate=@var{path} @gol
373 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
374 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
375 -freorder-blocks-and-partition -freorder-functions @gol
376 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
377 -frounding-math -fsched2-use-superblocks -fsched-pressure @gol
378 -fsched-spec-load -fsched-spec-load-dangerous @gol
379 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
380 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
381 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
382 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
383 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
384 -fselective-scheduling -fselective-scheduling2 @gol
385 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
386 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
387 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
388 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
390 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
391 -ftree-copyrename -ftree-dce -ftree-dominator-opts -ftree-dse @gol
392 -ftree-forwprop -ftree-fre -ftree-loop-if-convert @gol
393 -ftree-loop-if-convert-memory-writes -ftree-loop-im @gol
394 -ftree-phiprop -ftree-loop-distribution -ftree-loop-distribute-patterns @gol
395 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
396 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
397 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
398 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
399 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
400 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
401 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
402 -fwhole-program -fwpa -fuse-linker-plugin @gol
403 --param @var{name}=@var{value}
404 -O -O0 -O1 -O2 -O3 -Os -Ofast}
406 @item Preprocessor Options
407 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
408 @gccoptlist{-A@var{question}=@var{answer} @gol
409 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
410 -C -dD -dI -dM -dN @gol
411 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
412 -idirafter @var{dir} @gol
413 -include @var{file} -imacros @var{file} @gol
414 -iprefix @var{file} -iwithprefix @var{dir} @gol
415 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
416 -imultilib @var{dir} -isysroot @var{dir} @gol
417 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
418 -P -fworking-directory -remap @gol
419 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
420 -Xpreprocessor @var{option}}
422 @item Assembler Option
423 @xref{Assembler Options,,Passing Options to the Assembler}.
424 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
427 @xref{Link Options,,Options for Linking}.
428 @gccoptlist{@var{object-file-name} -l@var{library} @gol
429 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
430 -s -static -static-libgcc -static-libstdc++ -shared @gol
431 -shared-libgcc -symbolic @gol
432 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
435 @item Directory Options
436 @xref{Directory Options,,Options for Directory Search}.
437 @gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir}}
438 -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-
441 @item Machine Dependent Options
442 @xref{Submodel Options,,Hardware Models and Configurations}.
443 @c This list is ordered alphanumerically by subsection name.
444 @c Try and put the significant identifier (CPU or system) first,
445 @c so users have a clue at guessing where the ones they want will be.
448 @gccoptlist{-EB -EL @gol
449 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
450 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
453 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
454 -mabi=@var{name} @gol
455 -mapcs-stack-check -mno-apcs-stack-check @gol
456 -mapcs-float -mno-apcs-float @gol
457 -mapcs-reentrant -mno-apcs-reentrant @gol
458 -msched-prolog -mno-sched-prolog @gol
459 -mlittle-endian -mbig-endian -mwords-little-endian @gol
460 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
461 -mfp16-format=@var{name}
462 -mthumb-interwork -mno-thumb-interwork @gol
463 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
464 -mstructure-size-boundary=@var{n} @gol
465 -mabort-on-noreturn @gol
466 -mlong-calls -mno-long-calls @gol
467 -msingle-pic-base -mno-single-pic-base @gol
468 -mpic-register=@var{reg} @gol
469 -mnop-fun-dllimport @gol
470 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
471 -mpoke-function-name @gol
473 -mtpcs-frame -mtpcs-leaf-frame @gol
474 -mcaller-super-interworking -mcallee-super-interworking @gol
476 -mword-relocations @gol
477 -mfix-cortex-m3-ldrd}
480 @gccoptlist{-mmcu=@var{mcu} -mno-interrupts @gol
481 -mcall-prologues -mtiny-stack -mint8}
483 @emph{Blackfin Options}
484 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
485 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
486 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
487 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
488 -mno-id-shared-library -mshared-library-id=@var{n} @gol
489 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
490 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
491 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
495 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
496 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
497 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
498 -mstack-align -mdata-align -mconst-align @gol
499 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
500 -melf -maout -melinux -mlinux -sim -sim2 @gol
501 -mmul-bug-workaround -mno-mul-bug-workaround}
504 @gccoptlist{-mmac -mpush-args}
506 @emph{Darwin Options}
507 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
508 -arch_only -bind_at_load -bundle -bundle_loader @gol
509 -client_name -compatibility_version -current_version @gol
511 -dependency-file -dylib_file -dylinker_install_name @gol
512 -dynamic -dynamiclib -exported_symbols_list @gol
513 -filelist -flat_namespace -force_cpusubtype_ALL @gol
514 -force_flat_namespace -headerpad_max_install_names @gol
516 -image_base -init -install_name -keep_private_externs @gol
517 -multi_module -multiply_defined -multiply_defined_unused @gol
518 -noall_load -no_dead_strip_inits_and_terms @gol
519 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
520 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
521 -private_bundle -read_only_relocs -sectalign @gol
522 -sectobjectsymbols -whyload -seg1addr @gol
523 -sectcreate -sectobjectsymbols -sectorder @gol
524 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
525 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
526 -segprot -segs_read_only_addr -segs_read_write_addr @gol
527 -single_module -static -sub_library -sub_umbrella @gol
528 -twolevel_namespace -umbrella -undefined @gol
529 -unexported_symbols_list -weak_reference_mismatches @gol
530 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
531 -mkernel -mone-byte-bool}
533 @emph{DEC Alpha Options}
534 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
535 -mieee -mieee-with-inexact -mieee-conformant @gol
536 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
537 -mtrap-precision=@var{mode} -mbuild-constants @gol
538 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
539 -mbwx -mmax -mfix -mcix @gol
540 -mfloat-vax -mfloat-ieee @gol
541 -mexplicit-relocs -msmall-data -mlarge-data @gol
542 -msmall-text -mlarge-text @gol
543 -mmemory-latency=@var{time}}
545 @emph{DEC Alpha/VMS Options}
546 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
549 @gccoptlist{-msmall-model -mno-lsim}
552 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
553 -mhard-float -msoft-float @gol
554 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
555 -mdouble -mno-double @gol
556 -mmedia -mno-media -mmuladd -mno-muladd @gol
557 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
558 -mlinked-fp -mlong-calls -malign-labels @gol
559 -mlibrary-pic -macc-4 -macc-8 @gol
560 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
561 -moptimize-membar -mno-optimize-membar @gol
562 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
563 -mvliw-branch -mno-vliw-branch @gol
564 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
565 -mno-nested-cond-exec -mtomcat-stats @gol
569 @emph{GNU/Linux Options}
570 @gccoptlist{-mglibc -muclibc -mbionic -mandroid @gol
571 -tno-android-cc -tno-android-ld}
573 @emph{H8/300 Options}
574 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
577 @gccoptlist{-march=@var{architecture-type} @gol
578 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
579 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
580 -mfixed-range=@var{register-range} @gol
581 -mjump-in-delay -mlinker-opt -mlong-calls @gol
582 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
583 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
584 -mno-jump-in-delay -mno-long-load-store @gol
585 -mno-portable-runtime -mno-soft-float @gol
586 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
587 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
588 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
589 -munix=@var{unix-std} -nolibdld -static -threads}
591 @emph{i386 and x86-64 Options}
592 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
593 -mfpmath=@var{unit} @gol
594 -masm=@var{dialect} -mno-fancy-math-387 @gol
595 -mno-fp-ret-in-387 -msoft-float @gol
596 -mno-wide-multiply -mrtd -malign-double @gol
597 -mpreferred-stack-boundary=@var{num}
598 -mincoming-stack-boundary=@var{num} @gol
599 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip -mvzeroupper @gol
600 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
601 -maes -mpclmul -mfsgsbase -mrdrnd -mf16c -mfused-madd @gol
602 -msse4a -m3dnow -mpopcnt -mabm -mbmi -mtbm -mfma4 -mxop -mlwp @gol
603 -mthreads -mno-align-stringops -minline-all-stringops @gol
604 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
605 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
606 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
607 -mveclibabi=@var{type} -mvect8-ret-in-mem @gol
608 -mpc32 -mpc64 -mpc80 -mstackrealign @gol
609 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
610 -mcmodel=@var{code-model} -mabi=@var{name} @gol
611 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
612 -msse2avx -mfentry -m8bit-idiv}
615 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
616 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
617 -mconstant-gp -mauto-pic -mfused-madd @gol
618 -minline-float-divide-min-latency @gol
619 -minline-float-divide-max-throughput @gol
620 -mno-inline-float-divide @gol
621 -minline-int-divide-min-latency @gol
622 -minline-int-divide-max-throughput @gol
623 -mno-inline-int-divide @gol
624 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
625 -mno-inline-sqrt @gol
626 -mdwarf2-asm -mearly-stop-bits @gol
627 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
628 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
629 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
630 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
631 -msched-spec-ldc -msched-spec-control-ldc @gol
632 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
633 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
634 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
635 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
637 @emph{IA-64/VMS Options}
638 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
641 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
642 -msign-extend-enabled -muser-enabled}
644 @emph{M32R/D Options}
645 @gccoptlist{-m32r2 -m32rx -m32r @gol
647 -malign-loops -mno-align-loops @gol
648 -missue-rate=@var{number} @gol
649 -mbranch-cost=@var{number} @gol
650 -mmodel=@var{code-size-model-type} @gol
651 -msdata=@var{sdata-type} @gol
652 -mno-flush-func -mflush-func=@var{name} @gol
653 -mno-flush-trap -mflush-trap=@var{number} @gol
657 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
659 @emph{M680x0 Options}
660 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
661 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
662 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
663 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
664 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
665 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
666 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
667 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
670 @emph{M68hc1x Options}
671 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
672 -mauto-incdec -minmax -mlong-calls -mshort @gol
673 -msoft-reg-count=@var{count}}
676 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
677 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
678 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
679 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
680 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
683 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
684 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
685 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
686 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
689 @emph{MicroBlaze Options}
690 @gccoptlist{-msoft-float -mhard-float -msmall-divides -mcpu=@var{cpu} @gol
691 -mmemcpy -mxl-soft-mul -mxl-soft-div -mxl-barrel-shift @gol
692 -mxl-pattern-compare -mxl-stack-check -mxl-gp-opt -mno-clearbss @gol
693 -mxl-multiply-high -mxl-float-convert -mxl-float-sqrt @gol
694 -mxl-mode-@var{app-model}}
697 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
698 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
699 -mips64 -mips64r2 @gol
700 -mips16 -mno-mips16 -mflip-mips16 @gol
701 -minterlink-mips16 -mno-interlink-mips16 @gol
702 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
703 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
704 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
705 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
706 -mfpu=@var{fpu-type} @gol
707 -msmartmips -mno-smartmips @gol
708 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
709 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
710 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
711 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
712 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
713 -membedded-data -mno-embedded-data @gol
714 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
715 -mcode-readable=@var{setting} @gol
716 -msplit-addresses -mno-split-addresses @gol
717 -mexplicit-relocs -mno-explicit-relocs @gol
718 -mcheck-zero-division -mno-check-zero-division @gol
719 -mdivide-traps -mdivide-breaks @gol
720 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
721 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
722 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
723 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
724 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
725 -mflush-func=@var{func} -mno-flush-func @gol
726 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
727 -mfp-exceptions -mno-fp-exceptions @gol
728 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
729 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
732 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
733 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
734 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
735 -mno-base-addresses -msingle-exit -mno-single-exit}
737 @emph{MN10300 Options}
738 @gccoptlist{-mmult-bug -mno-mult-bug @gol
739 -mno-am33 -mam33 -mam33-2 -mam34 @gol
740 -mtune=@var{cpu-type} @gol
741 -mreturn-pointer-on-d0 @gol
744 @emph{PDP-11 Options}
745 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
746 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
747 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
748 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
749 -mbranch-expensive -mbranch-cheap @gol
750 -munix-asm -mdec-asm}
752 @emph{picoChip Options}
753 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N} @gol
754 -msymbol-as-address -mno-inefficient-warnings}
756 @emph{PowerPC Options}
757 See RS/6000 and PowerPC Options.
759 @emph{RS/6000 and PowerPC Options}
760 @gccoptlist{-mcpu=@var{cpu-type} @gol
761 -mtune=@var{cpu-type} @gol
762 -mcmodel=@var{code-model} @gol
763 -mpower -mno-power -mpower2 -mno-power2 @gol
764 -mpowerpc -mpowerpc64 -mno-powerpc @gol
765 -maltivec -mno-altivec @gol
766 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
767 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
768 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
769 -mfprnd -mno-fprnd @gol
770 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
771 -mnew-mnemonics -mold-mnemonics @gol
772 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
773 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
774 -malign-power -malign-natural @gol
775 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
776 -msingle-float -mdouble-float -msimple-fpu @gol
777 -mstring -mno-string -mupdate -mno-update @gol
778 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
779 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
780 -mstrict-align -mno-strict-align -mrelocatable @gol
781 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
782 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
783 -mdynamic-no-pic -maltivec -mswdiv -msingle-pic-base @gol
784 -mprioritize-restricted-insns=@var{priority} @gol
785 -msched-costly-dep=@var{dependence_type} @gol
786 -minsert-sched-nops=@var{scheme} @gol
787 -mcall-sysv -mcall-netbsd @gol
788 -maix-struct-return -msvr4-struct-return @gol
789 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
790 -mblock-move-inline-limit=@var{num} @gol
791 -misel -mno-isel @gol
792 -misel=yes -misel=no @gol
794 -mspe=yes -mspe=no @gol
796 -mgen-cell-microcode -mwarn-cell-microcode @gol
797 -mvrsave -mno-vrsave @gol
798 -mmulhw -mno-mulhw @gol
799 -mdlmzb -mno-dlmzb @gol
800 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
801 -mprototype -mno-prototype @gol
802 -msim -mmvme -mads -myellowknife -memb -msdata @gol
803 -msdata=@var{opt} -mvxworks -G @var{num} -pthread @gol
804 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision
805 -mno-recip-precision @gol
806 -mveclibabi=@var{type} -mfriz -mno-friz}
809 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
811 -mbig-endian-data -mlittle-endian-data @gol
814 -mas100-syntax -mno-as100-syntax@gol
816 -mmax-constant-size=@gol
818 -msave-acc-in-interrupts}
820 @emph{S/390 and zSeries Options}
821 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
822 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
823 -mlong-double-64 -mlong-double-128 @gol
824 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
825 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
826 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
827 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
828 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
831 @gccoptlist{-meb -mel @gol
835 -mscore5 -mscore5u -mscore7 -mscore7d}
838 @gccoptlist{-m1 -m2 -m2e @gol
839 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
841 -m4-nofpu -m4-single-only -m4-single -m4 @gol
842 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
843 -m5-64media -m5-64media-nofpu @gol
844 -m5-32media -m5-32media-nofpu @gol
845 -m5-compact -m5-compact-nofpu @gol
846 -mb -ml -mdalign -mrelax @gol
847 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
848 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
849 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
850 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
851 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
852 -maccumulate-outgoing-args -minvalid-symbols}
854 @emph{Solaris 2 Options}
855 @gccoptlist{-mimpure-text -mno-impure-text @gol
856 -threads -pthreads -pthread}
859 @gccoptlist{-mcpu=@var{cpu-type} @gol
860 -mtune=@var{cpu-type} @gol
861 -mcmodel=@var{code-model} @gol
862 -m32 -m64 -mapp-regs -mno-app-regs @gol
863 -mfaster-structs -mno-faster-structs @gol
864 -mfpu -mno-fpu -mhard-float -msoft-float @gol
865 -mhard-quad-float -msoft-quad-float @gol
867 -mstack-bias -mno-stack-bias @gol
868 -munaligned-doubles -mno-unaligned-doubles @gol
869 -mv8plus -mno-v8plus -mvis -mno-vis}
872 @gccoptlist{-mwarn-reloc -merror-reloc @gol
873 -msafe-dma -munsafe-dma @gol
875 -msmall-mem -mlarge-mem -mstdmain @gol
876 -mfixed-range=@var{register-range} @gol
878 -maddress-space-conversion -mno-address-space-conversion @gol
879 -mcache-size=@var{cache-size} @gol
880 -matomic-updates -mno-atomic-updates}
882 @emph{System V Options}
883 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
886 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
887 -mprolog-function -mno-prolog-function -mspace @gol
888 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
889 -mapp-regs -mno-app-regs @gol
890 -mdisable-callt -mno-disable-callt @gol
898 @gccoptlist{-mg -mgnu -munix}
900 @emph{VxWorks Options}
901 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
902 -Xbind-lazy -Xbind-now}
904 @emph{x86-64 Options}
905 See i386 and x86-64 Options.
907 @emph{i386 and x86-64 Windows Options}
908 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
909 -mnop-fun-dllimport -mthread @gol
910 -municode -mwin32 -mwindows -fno-set-stack-executable}
912 @emph{Xstormy16 Options}
915 @emph{Xtensa Options}
916 @gccoptlist{-mconst16 -mno-const16 @gol
917 -mfused-madd -mno-fused-madd @gol
919 -mserialize-volatile -mno-serialize-volatile @gol
920 -mtext-section-literals -mno-text-section-literals @gol
921 -mtarget-align -mno-target-align @gol
922 -mlongcalls -mno-longcalls}
924 @emph{zSeries Options}
925 See S/390 and zSeries Options.
927 @item Code Generation Options
928 @xref{Code Gen Options,,Options for Code Generation Conventions}.
929 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
930 -ffixed-@var{reg} -fexceptions @gol
931 -fnon-call-exceptions -funwind-tables @gol
932 -fasynchronous-unwind-tables @gol
933 -finhibit-size-directive -finstrument-functions @gol
934 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
935 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
936 -fno-common -fno-ident @gol
937 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
938 -fno-jump-tables @gol
939 -frecord-gcc-switches @gol
940 -freg-struct-return -fshort-enums @gol
941 -fshort-double -fshort-wchar @gol
942 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
943 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
944 -fno-stack-limit -fsplit-stack @gol
945 -fleading-underscore -ftls-model=@var{model} @gol
946 -ftrapv -fwrapv -fbounds-check @gol
947 -fvisibility -fstrict-volatile-bitfields}
951 * Overall Options:: Controlling the kind of output:
952 an executable, object files, assembler files,
953 or preprocessed source.
954 * C Dialect Options:: Controlling the variant of C language compiled.
955 * C++ Dialect Options:: Variations on C++.
956 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
958 * Language Independent Options:: Controlling how diagnostics should be
960 * Warning Options:: How picky should the compiler be?
961 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
962 * Optimize Options:: How much optimization?
963 * Preprocessor Options:: Controlling header files and macro definitions.
964 Also, getting dependency information for Make.
965 * Assembler Options:: Passing options to the assembler.
966 * Link Options:: Specifying libraries and so on.
967 * Directory Options:: Where to find header files and libraries.
968 Where to find the compiler executable files.
969 * Spec Files:: How to pass switches to sub-processes.
970 * Target Options:: Running a cross-compiler, or an old version of GCC.
973 @node Overall Options
974 @section Options Controlling the Kind of Output
976 Compilation can involve up to four stages: preprocessing, compilation
977 proper, assembly and linking, always in that order. GCC is capable of
978 preprocessing and compiling several files either into several
979 assembler input files, or into one assembler input file; then each
980 assembler input file produces an object file, and linking combines all
981 the object files (those newly compiled, and those specified as input)
982 into an executable file.
984 @cindex file name suffix
985 For any given input file, the file name suffix determines what kind of
990 C source code which must be preprocessed.
993 C source code which should not be preprocessed.
996 C++ source code which should not be preprocessed.
999 Objective-C source code. Note that you must link with the @file{libobjc}
1000 library to make an Objective-C program work.
1003 Objective-C source code which should not be preprocessed.
1007 Objective-C++ source code. Note that you must link with the @file{libobjc}
1008 library to make an Objective-C++ program work. Note that @samp{.M} refers
1009 to a literal capital M@.
1011 @item @var{file}.mii
1012 Objective-C++ source code which should not be preprocessed.
1015 C, C++, Objective-C or Objective-C++ header file to be turned into a
1016 precompiled header (default), or C, C++ header file to be turned into an
1017 Ada spec (via the @option{-fdump-ada-spec} switch).
1020 @itemx @var{file}.cp
1021 @itemx @var{file}.cxx
1022 @itemx @var{file}.cpp
1023 @itemx @var{file}.CPP
1024 @itemx @var{file}.c++
1026 C++ source code which must be preprocessed. Note that in @samp{.cxx},
1027 the last two letters must both be literally @samp{x}. Likewise,
1028 @samp{.C} refers to a literal capital C@.
1032 Objective-C++ source code which must be preprocessed.
1034 @item @var{file}.mii
1035 Objective-C++ source code which should not be preprocessed.
1039 @itemx @var{file}.hp
1040 @itemx @var{file}.hxx
1041 @itemx @var{file}.hpp
1042 @itemx @var{file}.HPP
1043 @itemx @var{file}.h++
1044 @itemx @var{file}.tcc
1045 C++ header file to be turned into a precompiled header or Ada spec.
1048 @itemx @var{file}.for
1049 @itemx @var{file}.ftn
1050 Fixed form Fortran source code which should not be preprocessed.
1053 @itemx @var{file}.FOR
1054 @itemx @var{file}.fpp
1055 @itemx @var{file}.FPP
1056 @itemx @var{file}.FTN
1057 Fixed form Fortran source code which must be preprocessed (with the traditional
1060 @item @var{file}.f90
1061 @itemx @var{file}.f95
1062 @itemx @var{file}.f03
1063 @itemx @var{file}.f08
1064 Free form Fortran source code which should not be preprocessed.
1066 @item @var{file}.F90
1067 @itemx @var{file}.F95
1068 @itemx @var{file}.F03
1069 @itemx @var{file}.F08
1070 Free form Fortran source code which must be preprocessed (with the
1071 traditional preprocessor).
1076 @c FIXME: Descriptions of Java file types.
1082 @item @var{file}.ads
1083 Ada source code file which contains a library unit declaration (a
1084 declaration of a package, subprogram, or generic, or a generic
1085 instantiation), or a library unit renaming declaration (a package,
1086 generic, or subprogram renaming declaration). Such files are also
1089 @item @var{file}.adb
1090 Ada source code file containing a library unit body (a subprogram or
1091 package body). Such files are also called @dfn{bodies}.
1093 @c GCC also knows about some suffixes for languages not yet included:
1104 @itemx @var{file}.sx
1105 Assembler code which must be preprocessed.
1108 An object file to be fed straight into linking.
1109 Any file name with no recognized suffix is treated this way.
1113 You can specify the input language explicitly with the @option{-x} option:
1116 @item -x @var{language}
1117 Specify explicitly the @var{language} for the following input files
1118 (rather than letting the compiler choose a default based on the file
1119 name suffix). This option applies to all following input files until
1120 the next @option{-x} option. Possible values for @var{language} are:
1122 c c-header cpp-output
1123 c++ c++-header c++-cpp-output
1124 objective-c objective-c-header objective-c-cpp-output
1125 objective-c++ objective-c++-header objective-c++-cpp-output
1126 assembler assembler-with-cpp
1128 f77 f77-cpp-input f95 f95-cpp-input
1134 Turn off any specification of a language, so that subsequent files are
1135 handled according to their file name suffixes (as they are if @option{-x}
1136 has not been used at all).
1138 @item -pass-exit-codes
1139 @opindex pass-exit-codes
1140 Normally the @command{gcc} program will exit with the code of 1 if any
1141 phase of the compiler returns a non-success return code. If you specify
1142 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1143 numerically highest error produced by any phase that returned an error
1144 indication. The C, C++, and Fortran frontends return 4, if an internal
1145 compiler error is encountered.
1148 If you only want some of the stages of compilation, you can use
1149 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1150 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1151 @command{gcc} is to stop. Note that some combinations (for example,
1152 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1157 Compile or assemble the source files, but do not link. The linking
1158 stage simply is not done. The ultimate output is in the form of an
1159 object file for each source file.
1161 By default, the object file name for a source file is made by replacing
1162 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1164 Unrecognized input files, not requiring compilation or assembly, are
1169 Stop after the stage of compilation proper; do not assemble. The output
1170 is in the form of an assembler code file for each non-assembler input
1173 By default, the assembler file name for a source file is made by
1174 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1176 Input files that don't require compilation are ignored.
1180 Stop after the preprocessing stage; do not run the compiler proper. The
1181 output is in the form of preprocessed source code, which is sent to the
1184 Input files which don't require preprocessing are ignored.
1186 @cindex output file option
1189 Place output in file @var{file}. This applies regardless to whatever
1190 sort of output is being produced, whether it be an executable file,
1191 an object file, an assembler file or preprocessed C code.
1193 If @option{-o} is not specified, the default is to put an executable
1194 file in @file{a.out}, the object file for
1195 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1196 assembler file in @file{@var{source}.s}, a precompiled header file in
1197 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1202 Print (on standard error output) the commands executed to run the stages
1203 of compilation. Also print the version number of the compiler driver
1204 program and of the preprocessor and the compiler proper.
1208 Like @option{-v} except the commands are not executed and arguments
1209 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1210 This is useful for shell scripts to capture the driver-generated command lines.
1214 Use pipes rather than temporary files for communication between the
1215 various stages of compilation. This fails to work on some systems where
1216 the assembler is unable to read from a pipe; but the GNU assembler has
1221 Print (on the standard output) a description of the command line options
1222 understood by @command{gcc}. If the @option{-v} option is also specified
1223 then @option{--help} will also be passed on to the various processes
1224 invoked by @command{gcc}, so that they can display the command line options
1225 they accept. If the @option{-Wextra} option has also been specified
1226 (prior to the @option{--help} option), then command line options which
1227 have no documentation associated with them will also be displayed.
1230 @opindex target-help
1231 Print (on the standard output) a description of target-specific command
1232 line options for each tool. For some targets extra target-specific
1233 information may also be printed.
1235 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1236 Print (on the standard output) a description of the command line
1237 options understood by the compiler that fit into all specified classes
1238 and qualifiers. These are the supported classes:
1241 @item @samp{optimizers}
1242 This will display all of the optimization options supported by the
1245 @item @samp{warnings}
1246 This will display all of the options controlling warning messages
1247 produced by the compiler.
1250 This will display target-specific options. Unlike the
1251 @option{--target-help} option however, target-specific options of the
1252 linker and assembler will not be displayed. This is because those
1253 tools do not currently support the extended @option{--help=} syntax.
1256 This will display the values recognized by the @option{--param}
1259 @item @var{language}
1260 This will display the options supported for @var{language}, where
1261 @var{language} is the name of one of the languages supported in this
1265 This will display the options that are common to all languages.
1268 These are the supported qualifiers:
1271 @item @samp{undocumented}
1272 Display only those options which are undocumented.
1275 Display options which take an argument that appears after an equal
1276 sign in the same continuous piece of text, such as:
1277 @samp{--help=target}.
1279 @item @samp{separate}
1280 Display options which take an argument that appears as a separate word
1281 following the original option, such as: @samp{-o output-file}.
1284 Thus for example to display all the undocumented target-specific
1285 switches supported by the compiler the following can be used:
1288 --help=target,undocumented
1291 The sense of a qualifier can be inverted by prefixing it with the
1292 @samp{^} character, so for example to display all binary warning
1293 options (i.e., ones that are either on or off and that do not take an
1294 argument), which have a description the following can be used:
1297 --help=warnings,^joined,^undocumented
1300 The argument to @option{--help=} should not consist solely of inverted
1303 Combining several classes is possible, although this usually
1304 restricts the output by so much that there is nothing to display. One
1305 case where it does work however is when one of the classes is
1306 @var{target}. So for example to display all the target-specific
1307 optimization options the following can be used:
1310 --help=target,optimizers
1313 The @option{--help=} option can be repeated on the command line. Each
1314 successive use will display its requested class of options, skipping
1315 those that have already been displayed.
1317 If the @option{-Q} option appears on the command line before the
1318 @option{--help=} option, then the descriptive text displayed by
1319 @option{--help=} is changed. Instead of describing the displayed
1320 options, an indication is given as to whether the option is enabled,
1321 disabled or set to a specific value (assuming that the compiler
1322 knows this at the point where the @option{--help=} option is used).
1324 Here is a truncated example from the ARM port of @command{gcc}:
1327 % gcc -Q -mabi=2 --help=target -c
1328 The following options are target specific:
1330 -mabort-on-noreturn [disabled]
1334 The output is sensitive to the effects of previous command line
1335 options, so for example it is possible to find out which optimizations
1336 are enabled at @option{-O2} by using:
1339 -Q -O2 --help=optimizers
1342 Alternatively you can discover which binary optimizations are enabled
1343 by @option{-O3} by using:
1346 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1347 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1348 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1351 @item -no-canonical-prefixes
1352 @opindex no-canonical-prefixes
1353 Do not expand any symbolic links, resolve references to @samp{/../}
1354 or @samp{/./}, or make the path absolute when generating a relative
1359 Display the version number and copyrights of the invoked GCC@.
1363 Invoke all subcommands under a wrapper program. It takes a single
1364 comma separated list as an argument, which will be used to invoke
1368 gcc -c t.c -wrapper gdb,--args
1371 This will invoke all subprograms of gcc under "gdb --args",
1372 thus cc1 invocation will be "gdb --args cc1 ...".
1374 @item -fplugin=@var{name}.so
1375 Load the plugin code in file @var{name}.so, assumed to be a
1376 shared object to be dlopen'd by the compiler. The base name of
1377 the shared object file is used to identify the plugin for the
1378 purposes of argument parsing (See
1379 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1380 Each plugin should define the callback functions specified in the
1383 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1384 Define an argument called @var{key} with a value of @var{value}
1385 for the plugin called @var{name}.
1387 @item -fdump-ada-spec@r{[}-slim@r{]}
1388 For C and C++ source and include files, generate corresponding Ada
1389 specs. @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1390 GNAT User's Guide}, which provides detailed documentation on this feature.
1392 @item -fdump-go-spec=@var{file}
1393 For input files in any language, generate corresponding Go
1394 declarations in @var{file}. This generates Go @code{const},
1395 @code{type}, @code{var}, and @code{func} declarations which may be a
1396 useful way to start writing a Go interface to code written in some
1399 @include @value{srcdir}/../libiberty/at-file.texi
1403 @section Compiling C++ Programs
1405 @cindex suffixes for C++ source
1406 @cindex C++ source file suffixes
1407 C++ source files conventionally use one of the suffixes @samp{.C},
1408 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1409 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1410 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1411 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1412 files with these names and compiles them as C++ programs even if you
1413 call the compiler the same way as for compiling C programs (usually
1414 with the name @command{gcc}).
1418 However, the use of @command{gcc} does not add the C++ library.
1419 @command{g++} is a program that calls GCC and treats @samp{.c},
1420 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1421 files unless @option{-x} is used, and automatically specifies linking
1422 against the C++ library. This program is also useful when
1423 precompiling a C header file with a @samp{.h} extension for use in C++
1424 compilations. On many systems, @command{g++} is also installed with
1425 the name @command{c++}.
1427 @cindex invoking @command{g++}
1428 When you compile C++ programs, you may specify many of the same
1429 command-line options that you use for compiling programs in any
1430 language; or command-line options meaningful for C and related
1431 languages; or options that are meaningful only for C++ programs.
1432 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1433 explanations of options for languages related to C@.
1434 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1435 explanations of options that are meaningful only for C++ programs.
1437 @node C Dialect Options
1438 @section Options Controlling C Dialect
1439 @cindex dialect options
1440 @cindex language dialect options
1441 @cindex options, dialect
1443 The following options control the dialect of C (or languages derived
1444 from C, such as C++, Objective-C and Objective-C++) that the compiler
1448 @cindex ANSI support
1452 In C mode, this is equivalent to @samp{-std=c90}. In C++ mode, it is
1453 equivalent to @samp{-std=c++98}.
1455 This turns off certain features of GCC that are incompatible with ISO
1456 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1457 such as the @code{asm} and @code{typeof} keywords, and
1458 predefined macros such as @code{unix} and @code{vax} that identify the
1459 type of system you are using. It also enables the undesirable and
1460 rarely used ISO trigraph feature. For the C compiler,
1461 it disables recognition of C++ style @samp{//} comments as well as
1462 the @code{inline} keyword.
1464 The alternate keywords @code{__asm__}, @code{__extension__},
1465 @code{__inline__} and @code{__typeof__} continue to work despite
1466 @option{-ansi}. You would not want to use them in an ISO C program, of
1467 course, but it is useful to put them in header files that might be included
1468 in compilations done with @option{-ansi}. Alternate predefined macros
1469 such as @code{__unix__} and @code{__vax__} are also available, with or
1470 without @option{-ansi}.
1472 The @option{-ansi} option does not cause non-ISO programs to be
1473 rejected gratuitously. For that, @option{-pedantic} is required in
1474 addition to @option{-ansi}. @xref{Warning Options}.
1476 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1477 option is used. Some header files may notice this macro and refrain
1478 from declaring certain functions or defining certain macros that the
1479 ISO standard doesn't call for; this is to avoid interfering with any
1480 programs that might use these names for other things.
1482 Functions that would normally be built in but do not have semantics
1483 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1484 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1485 built-in functions provided by GCC}, for details of the functions
1490 Determine the language standard. @xref{Standards,,Language Standards
1491 Supported by GCC}, for details of these standard versions. This option
1492 is currently only supported when compiling C or C++.
1494 The compiler can accept several base standards, such as @samp{c90} or
1495 @samp{c++98}, and GNU dialects of those standards, such as
1496 @samp{gnu90} or @samp{gnu++98}. By specifying a base standard, the
1497 compiler will accept all programs following that standard and those
1498 using GNU extensions that do not contradict it. For example,
1499 @samp{-std=c90} turns off certain features of GCC that are
1500 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1501 keywords, but not other GNU extensions that do not have a meaning in
1502 ISO C90, such as omitting the middle term of a @code{?:}
1503 expression. On the other hand, by specifying a GNU dialect of a
1504 standard, all features the compiler support are enabled, even when
1505 those features change the meaning of the base standard and some
1506 strict-conforming programs may be rejected. The particular standard
1507 is used by @option{-pedantic} to identify which features are GNU
1508 extensions given that version of the standard. For example
1509 @samp{-std=gnu90 -pedantic} would warn about C++ style @samp{//}
1510 comments, while @samp{-std=gnu99 -pedantic} would not.
1512 A value for this option must be provided; possible values are
1518 Support all ISO C90 programs (certain GNU extensions that conflict
1519 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1521 @item iso9899:199409
1522 ISO C90 as modified in amendment 1.
1528 ISO C99. Note that this standard is not yet fully supported; see
1529 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1530 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1533 ISO C1X, the draft of the next revision of the ISO C standard.
1534 Support is limited and experimental and features enabled by this
1535 option may be changed or removed if changed in or removed from the
1540 GNU dialect of ISO C90 (including some C99 features). This
1541 is the default for C code.
1545 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1546 this will become the default. The name @samp{gnu9x} is deprecated.
1549 GNU dialect of ISO C1X. Support is limited and experimental and
1550 features enabled by this option may be changed or removed if changed
1551 in or removed from the standard draft.
1554 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1558 GNU dialect of @option{-std=c++98}. This is the default for
1562 The working draft of the upcoming ISO C++0x standard. This option
1563 enables experimental features that are likely to be included in
1564 C++0x. The working draft is constantly changing, and any feature that is
1565 enabled by this flag may be removed from future versions of GCC if it is
1566 not part of the C++0x standard.
1569 GNU dialect of @option{-std=c++0x}. This option enables
1570 experimental features that may be removed in future versions of GCC.
1573 @item -fgnu89-inline
1574 @opindex fgnu89-inline
1575 The option @option{-fgnu89-inline} tells GCC to use the traditional
1576 GNU semantics for @code{inline} functions when in C99 mode.
1577 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1578 is accepted and ignored by GCC versions 4.1.3 up to but not including
1579 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1580 C99 mode. Using this option is roughly equivalent to adding the
1581 @code{gnu_inline} function attribute to all inline functions
1582 (@pxref{Function Attributes}).
1584 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1585 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1586 specifies the default behavior). This option was first supported in
1587 GCC 4.3. This option is not supported in @option{-std=c90} or
1588 @option{-std=gnu90} mode.
1590 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1591 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1592 in effect for @code{inline} functions. @xref{Common Predefined
1593 Macros,,,cpp,The C Preprocessor}.
1595 @item -aux-info @var{filename}
1597 Output to the given filename prototyped declarations for all functions
1598 declared and/or defined in a translation unit, including those in header
1599 files. This option is silently ignored in any language other than C@.
1601 Besides declarations, the file indicates, in comments, the origin of
1602 each declaration (source file and line), whether the declaration was
1603 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1604 @samp{O} for old, respectively, in the first character after the line
1605 number and the colon), and whether it came from a declaration or a
1606 definition (@samp{C} or @samp{F}, respectively, in the following
1607 character). In the case of function definitions, a K&R-style list of
1608 arguments followed by their declarations is also provided, inside
1609 comments, after the declaration.
1613 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1614 keyword, so that code can use these words as identifiers. You can use
1615 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1616 instead. @option{-ansi} implies @option{-fno-asm}.
1618 In C++, this switch only affects the @code{typeof} keyword, since
1619 @code{asm} and @code{inline} are standard keywords. You may want to
1620 use the @option{-fno-gnu-keywords} flag instead, which has the same
1621 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1622 switch only affects the @code{asm} and @code{typeof} keywords, since
1623 @code{inline} is a standard keyword in ISO C99.
1626 @itemx -fno-builtin-@var{function}
1627 @opindex fno-builtin
1628 @cindex built-in functions
1629 Don't recognize built-in functions that do not begin with
1630 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1631 functions provided by GCC}, for details of the functions affected,
1632 including those which are not built-in functions when @option{-ansi} or
1633 @option{-std} options for strict ISO C conformance are used because they
1634 do not have an ISO standard meaning.
1636 GCC normally generates special code to handle certain built-in functions
1637 more efficiently; for instance, calls to @code{alloca} may become single
1638 instructions that adjust the stack directly, and calls to @code{memcpy}
1639 may become inline copy loops. The resulting code is often both smaller
1640 and faster, but since the function calls no longer appear as such, you
1641 cannot set a breakpoint on those calls, nor can you change the behavior
1642 of the functions by linking with a different library. In addition,
1643 when a function is recognized as a built-in function, GCC may use
1644 information about that function to warn about problems with calls to
1645 that function, or to generate more efficient code, even if the
1646 resulting code still contains calls to that function. For example,
1647 warnings are given with @option{-Wformat} for bad calls to
1648 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1649 known not to modify global memory.
1651 With the @option{-fno-builtin-@var{function}} option
1652 only the built-in function @var{function} is
1653 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1654 function is named that is not built-in in this version of GCC, this
1655 option is ignored. There is no corresponding
1656 @option{-fbuiltin-@var{function}} option; if you wish to enable
1657 built-in functions selectively when using @option{-fno-builtin} or
1658 @option{-ffreestanding}, you may define macros such as:
1661 #define abs(n) __builtin_abs ((n))
1662 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1667 @cindex hosted environment
1669 Assert that compilation takes place in a hosted environment. This implies
1670 @option{-fbuiltin}. A hosted environment is one in which the
1671 entire standard library is available, and in which @code{main} has a return
1672 type of @code{int}. Examples are nearly everything except a kernel.
1673 This is equivalent to @option{-fno-freestanding}.
1675 @item -ffreestanding
1676 @opindex ffreestanding
1677 @cindex hosted environment
1679 Assert that compilation takes place in a freestanding environment. This
1680 implies @option{-fno-builtin}. A freestanding environment
1681 is one in which the standard library may not exist, and program startup may
1682 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1683 This is equivalent to @option{-fno-hosted}.
1685 @xref{Standards,,Language Standards Supported by GCC}, for details of
1686 freestanding and hosted environments.
1690 @cindex OpenMP parallel
1691 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1692 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1693 compiler generates parallel code according to the OpenMP Application
1694 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1695 implies @option{-pthread}, and thus is only supported on targets that
1696 have support for @option{-pthread}.
1698 @item -fms-extensions
1699 @opindex fms-extensions
1700 Accept some non-standard constructs used in Microsoft header files.
1702 It allows for c++ that member-names in structures can be similiar
1703 to previous types declarations.
1712 Some cases of unnamed fields in structures and unions are only
1713 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1714 fields within structs/unions}, for details.
1716 @item -fplan9-extensions
1717 Accept some non-standard constructs used in Plan 9 code.
1719 This enables @option{-fms-extensions}, permits passing pointers to
1720 structures with anonymous fields to functions which expect pointers to
1721 elements of the type of the field, and permits referring to anonymous
1722 fields declared using a typedef. @xref{Unnamed Fields,,Unnamed
1723 struct/union fields within structs/unions}, for details. This is only
1724 supported for C, not C++.
1728 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1729 options for strict ISO C conformance) implies @option{-trigraphs}.
1731 @item -no-integrated-cpp
1732 @opindex no-integrated-cpp
1733 Performs a compilation in two passes: preprocessing and compiling. This
1734 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1735 @option{-B} option. The user supplied compilation step can then add in
1736 an additional preprocessing step after normal preprocessing but before
1737 compiling. The default is to use the integrated cpp (internal cpp)
1739 The semantics of this option will change if "cc1", "cc1plus", and
1740 "cc1obj" are merged.
1742 @cindex traditional C language
1743 @cindex C language, traditional
1745 @itemx -traditional-cpp
1746 @opindex traditional-cpp
1747 @opindex traditional
1748 Formerly, these options caused GCC to attempt to emulate a pre-standard
1749 C compiler. They are now only supported with the @option{-E} switch.
1750 The preprocessor continues to support a pre-standard mode. See the GNU
1751 CPP manual for details.
1753 @item -fcond-mismatch
1754 @opindex fcond-mismatch
1755 Allow conditional expressions with mismatched types in the second and
1756 third arguments. The value of such an expression is void. This option
1757 is not supported for C++.
1759 @item -flax-vector-conversions
1760 @opindex flax-vector-conversions
1761 Allow implicit conversions between vectors with differing numbers of
1762 elements and/or incompatible element types. This option should not be
1765 @item -funsigned-char
1766 @opindex funsigned-char
1767 Let the type @code{char} be unsigned, like @code{unsigned char}.
1769 Each kind of machine has a default for what @code{char} should
1770 be. It is either like @code{unsigned char} by default or like
1771 @code{signed char} by default.
1773 Ideally, a portable program should always use @code{signed char} or
1774 @code{unsigned char} when it depends on the signedness of an object.
1775 But many programs have been written to use plain @code{char} and
1776 expect it to be signed, or expect it to be unsigned, depending on the
1777 machines they were written for. This option, and its inverse, let you
1778 make such a program work with the opposite default.
1780 The type @code{char} is always a distinct type from each of
1781 @code{signed char} or @code{unsigned char}, even though its behavior
1782 is always just like one of those two.
1785 @opindex fsigned-char
1786 Let the type @code{char} be signed, like @code{signed char}.
1788 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1789 the negative form of @option{-funsigned-char}. Likewise, the option
1790 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1792 @item -fsigned-bitfields
1793 @itemx -funsigned-bitfields
1794 @itemx -fno-signed-bitfields
1795 @itemx -fno-unsigned-bitfields
1796 @opindex fsigned-bitfields
1797 @opindex funsigned-bitfields
1798 @opindex fno-signed-bitfields
1799 @opindex fno-unsigned-bitfields
1800 These options control whether a bit-field is signed or unsigned, when the
1801 declaration does not use either @code{signed} or @code{unsigned}. By
1802 default, such a bit-field is signed, because this is consistent: the
1803 basic integer types such as @code{int} are signed types.
1806 @node C++ Dialect Options
1807 @section Options Controlling C++ Dialect
1809 @cindex compiler options, C++
1810 @cindex C++ options, command line
1811 @cindex options, C++
1812 This section describes the command-line options that are only meaningful
1813 for C++ programs; but you can also use most of the GNU compiler options
1814 regardless of what language your program is in. For example, you
1815 might compile a file @code{firstClass.C} like this:
1818 g++ -g -frepo -O -c firstClass.C
1822 In this example, only @option{-frepo} is an option meant
1823 only for C++ programs; you can use the other options with any
1824 language supported by GCC@.
1826 Here is a list of options that are @emph{only} for compiling C++ programs:
1830 @item -fabi-version=@var{n}
1831 @opindex fabi-version
1832 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1833 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1834 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1835 the version that conforms most closely to the C++ ABI specification.
1836 Therefore, the ABI obtained using version 0 will change as ABI bugs
1839 The default is version 2.
1841 Version 3 corrects an error in mangling a constant address as a
1844 Version 4 implements a standard mangling for vector types.
1846 See also @option{-Wabi}.
1848 @item -fno-access-control
1849 @opindex fno-access-control
1850 Turn off all access checking. This switch is mainly useful for working
1851 around bugs in the access control code.
1855 Check that the pointer returned by @code{operator new} is non-null
1856 before attempting to modify the storage allocated. This check is
1857 normally unnecessary because the C++ standard specifies that
1858 @code{operator new} will only return @code{0} if it is declared
1859 @samp{throw()}, in which case the compiler will always check the
1860 return value even without this option. In all other cases, when
1861 @code{operator new} has a non-empty exception specification, memory
1862 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1863 @samp{new (nothrow)}.
1865 @item -fconserve-space
1866 @opindex fconserve-space
1867 Put uninitialized or runtime-initialized global variables into the
1868 common segment, as C does. This saves space in the executable at the
1869 cost of not diagnosing duplicate definitions. If you compile with this
1870 flag and your program mysteriously crashes after @code{main()} has
1871 completed, you may have an object that is being destroyed twice because
1872 two definitions were merged.
1874 This option is no longer useful on most targets, now that support has
1875 been added for putting variables into BSS without making them common.
1877 @item -fno-deduce-init-list
1878 @opindex fno-deduce-init-list
1879 Disable deduction of a template type parameter as
1880 std::initializer_list from a brace-enclosed initializer list, i.e.
1883 template <class T> auto forward(T t) -> decltype (realfn (t))
1890 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1894 This option is present because this deduction is an extension to the
1895 current specification in the C++0x working draft, and there was
1896 some concern about potential overload resolution problems.
1898 @item -ffriend-injection
1899 @opindex ffriend-injection
1900 Inject friend functions into the enclosing namespace, so that they are
1901 visible outside the scope of the class in which they are declared.
1902 Friend functions were documented to work this way in the old Annotated
1903 C++ Reference Manual, and versions of G++ before 4.1 always worked
1904 that way. However, in ISO C++ a friend function which is not declared
1905 in an enclosing scope can only be found using argument dependent
1906 lookup. This option causes friends to be injected as they were in
1909 This option is for compatibility, and may be removed in a future
1912 @item -fno-elide-constructors
1913 @opindex fno-elide-constructors
1914 The C++ standard allows an implementation to omit creating a temporary
1915 which is only used to initialize another object of the same type.
1916 Specifying this option disables that optimization, and forces G++ to
1917 call the copy constructor in all cases.
1919 @item -fno-enforce-eh-specs
1920 @opindex fno-enforce-eh-specs
1921 Don't generate code to check for violation of exception specifications
1922 at runtime. This option violates the C++ standard, but may be useful
1923 for reducing code size in production builds, much like defining
1924 @samp{NDEBUG}. This does not give user code permission to throw
1925 exceptions in violation of the exception specifications; the compiler
1926 will still optimize based on the specifications, so throwing an
1927 unexpected exception will result in undefined behavior.
1930 @itemx -fno-for-scope
1932 @opindex fno-for-scope
1933 If @option{-ffor-scope} is specified, the scope of variables declared in
1934 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1935 as specified by the C++ standard.
1936 If @option{-fno-for-scope} is specified, the scope of variables declared in
1937 a @i{for-init-statement} extends to the end of the enclosing scope,
1938 as was the case in old versions of G++, and other (traditional)
1939 implementations of C++.
1941 The default if neither flag is given to follow the standard,
1942 but to allow and give a warning for old-style code that would
1943 otherwise be invalid, or have different behavior.
1945 @item -fno-gnu-keywords
1946 @opindex fno-gnu-keywords
1947 Do not recognize @code{typeof} as a keyword, so that code can use this
1948 word as an identifier. You can use the keyword @code{__typeof__} instead.
1949 @option{-ansi} implies @option{-fno-gnu-keywords}.
1951 @item -fno-implicit-templates
1952 @opindex fno-implicit-templates
1953 Never emit code for non-inline templates which are instantiated
1954 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1955 @xref{Template Instantiation}, for more information.
1957 @item -fno-implicit-inline-templates
1958 @opindex fno-implicit-inline-templates
1959 Don't emit code for implicit instantiations of inline templates, either.
1960 The default is to handle inlines differently so that compiles with and
1961 without optimization will need the same set of explicit instantiations.
1963 @item -fno-implement-inlines
1964 @opindex fno-implement-inlines
1965 To save space, do not emit out-of-line copies of inline functions
1966 controlled by @samp{#pragma implementation}. This will cause linker
1967 errors if these functions are not inlined everywhere they are called.
1969 @item -fms-extensions
1970 @opindex fms-extensions
1971 Disable pedantic warnings about constructs used in MFC, such as implicit
1972 int and getting a pointer to member function via non-standard syntax.
1974 @item -fno-nonansi-builtins
1975 @opindex fno-nonansi-builtins
1976 Disable built-in declarations of functions that are not mandated by
1977 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1978 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1981 @opindex fnothrow-opt
1982 Treat a @code{throw()} exception specification as though it were a
1983 @code{noexcept} specification to reduce or eliminate the text size
1984 overhead relative to a function with no exception specification. If
1985 the function has local variables of types with non-trivial
1986 destructors, the exception specification will actually make the
1987 function smaller because the EH cleanups for those variables can be
1988 optimized away. The semantic effect is that an exception thrown out of
1989 a function with such an exception specification will result in a call
1990 to @code{terminate} rather than @code{unexpected}.
1992 @item -fno-operator-names
1993 @opindex fno-operator-names
1994 Do not treat the operator name keywords @code{and}, @code{bitand},
1995 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1996 synonyms as keywords.
1998 @item -fno-optional-diags
1999 @opindex fno-optional-diags
2000 Disable diagnostics that the standard says a compiler does not need to
2001 issue. Currently, the only such diagnostic issued by G++ is the one for
2002 a name having multiple meanings within a class.
2005 @opindex fpermissive
2006 Downgrade some diagnostics about nonconformant code from errors to
2007 warnings. Thus, using @option{-fpermissive} will allow some
2008 nonconforming code to compile.
2010 @item -fno-pretty-templates
2011 @opindex fno-pretty-templates
2012 When an error message refers to a specialization of a function
2013 template, the compiler will normally print the signature of the
2014 template followed by the template arguments and any typedefs or
2015 typenames in the signature (e.g. @code{void f(T) [with T = int]}
2016 rather than @code{void f(int)}) so that it's clear which template is
2017 involved. When an error message refers to a specialization of a class
2018 template, the compiler will omit any template arguments which match
2019 the default template arguments for that template. If either of these
2020 behaviors make it harder to understand the error message rather than
2021 easier, using @option{-fno-pretty-templates} will disable them.
2025 Enable automatic template instantiation at link time. This option also
2026 implies @option{-fno-implicit-templates}. @xref{Template
2027 Instantiation}, for more information.
2031 Disable generation of information about every class with virtual
2032 functions for use by the C++ runtime type identification features
2033 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
2034 of the language, you can save some space by using this flag. Note that
2035 exception handling uses the same information, but it will generate it as
2036 needed. The @samp{dynamic_cast} operator can still be used for casts that
2037 do not require runtime type information, i.e.@: casts to @code{void *} or to
2038 unambiguous base classes.
2042 Emit statistics about front-end processing at the end of the compilation.
2043 This information is generally only useful to the G++ development team.
2045 @item -fstrict-enums
2046 @opindex fstrict-enums
2047 Allow the compiler to optimize using the assumption that a value of
2048 enumeration type can only be one of the values of the enumeration (as
2049 defined in the C++ standard; basically, a value which can be
2050 represented in the minimum number of bits needed to represent all the
2051 enumerators). This assumption may not be valid if the program uses a
2052 cast to convert an arbitrary integer value to the enumeration type.
2054 @item -ftemplate-depth=@var{n}
2055 @opindex ftemplate-depth
2056 Set the maximum instantiation depth for template classes to @var{n}.
2057 A limit on the template instantiation depth is needed to detect
2058 endless recursions during template class instantiation. ANSI/ISO C++
2059 conforming programs must not rely on a maximum depth greater than 17
2060 (changed to 1024 in C++0x).
2062 @item -fno-threadsafe-statics
2063 @opindex fno-threadsafe-statics
2064 Do not emit the extra code to use the routines specified in the C++
2065 ABI for thread-safe initialization of local statics. You can use this
2066 option to reduce code size slightly in code that doesn't need to be
2069 @item -fuse-cxa-atexit
2070 @opindex fuse-cxa-atexit
2071 Register destructors for objects with static storage duration with the
2072 @code{__cxa_atexit} function rather than the @code{atexit} function.
2073 This option is required for fully standards-compliant handling of static
2074 destructors, but will only work if your C library supports
2075 @code{__cxa_atexit}.
2077 @item -fno-use-cxa-get-exception-ptr
2078 @opindex fno-use-cxa-get-exception-ptr
2079 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2080 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2081 if the runtime routine is not available.
2083 @item -fvisibility-inlines-hidden
2084 @opindex fvisibility-inlines-hidden
2085 This switch declares that the user does not attempt to compare
2086 pointers to inline methods where the addresses of the two functions
2087 were taken in different shared objects.
2089 The effect of this is that GCC may, effectively, mark inline methods with
2090 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2091 appear in the export table of a DSO and do not require a PLT indirection
2092 when used within the DSO@. Enabling this option can have a dramatic effect
2093 on load and link times of a DSO as it massively reduces the size of the
2094 dynamic export table when the library makes heavy use of templates.
2096 The behavior of this switch is not quite the same as marking the
2097 methods as hidden directly, because it does not affect static variables
2098 local to the function or cause the compiler to deduce that
2099 the function is defined in only one shared object.
2101 You may mark a method as having a visibility explicitly to negate the
2102 effect of the switch for that method. For example, if you do want to
2103 compare pointers to a particular inline method, you might mark it as
2104 having default visibility. Marking the enclosing class with explicit
2105 visibility will have no effect.
2107 Explicitly instantiated inline methods are unaffected by this option
2108 as their linkage might otherwise cross a shared library boundary.
2109 @xref{Template Instantiation}.
2111 @item -fvisibility-ms-compat
2112 @opindex fvisibility-ms-compat
2113 This flag attempts to use visibility settings to make GCC's C++
2114 linkage model compatible with that of Microsoft Visual Studio.
2116 The flag makes these changes to GCC's linkage model:
2120 It sets the default visibility to @code{hidden}, like
2121 @option{-fvisibility=hidden}.
2124 Types, but not their members, are not hidden by default.
2127 The One Definition Rule is relaxed for types without explicit
2128 visibility specifications which are defined in more than one different
2129 shared object: those declarations are permitted if they would have
2130 been permitted when this option was not used.
2133 In new code it is better to use @option{-fvisibility=hidden} and
2134 export those classes which are intended to be externally visible.
2135 Unfortunately it is possible for code to rely, perhaps accidentally,
2136 on the Visual Studio behavior.
2138 Among the consequences of these changes are that static data members
2139 of the same type with the same name but defined in different shared
2140 objects will be different, so changing one will not change the other;
2141 and that pointers to function members defined in different shared
2142 objects may not compare equal. When this flag is given, it is a
2143 violation of the ODR to define types with the same name differently.
2147 Do not use weak symbol support, even if it is provided by the linker.
2148 By default, G++ will use weak symbols if they are available. This
2149 option exists only for testing, and should not be used by end-users;
2150 it will result in inferior code and has no benefits. This option may
2151 be removed in a future release of G++.
2155 Do not search for header files in the standard directories specific to
2156 C++, but do still search the other standard directories. (This option
2157 is used when building the C++ library.)
2160 In addition, these optimization, warning, and code generation options
2161 have meanings only for C++ programs:
2164 @item -fno-default-inline
2165 @opindex fno-default-inline
2166 Do not assume @samp{inline} for functions defined inside a class scope.
2167 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2168 functions will have linkage like inline functions; they just won't be
2171 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2174 Warn when G++ generates code that is probably not compatible with the
2175 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2176 all such cases, there are probably some cases that are not warned about,
2177 even though G++ is generating incompatible code. There may also be
2178 cases where warnings are emitted even though the code that is generated
2181 You should rewrite your code to avoid these warnings if you are
2182 concerned about the fact that code generated by G++ may not be binary
2183 compatible with code generated by other compilers.
2185 The known incompatibilities in @option{-fabi-version=2} (the default) include:
2190 A template with a non-type template parameter of reference type is
2191 mangled incorrectly:
2194 template <int &> struct S @{@};
2198 This is fixed in @option{-fabi-version=3}.
2201 SIMD vector types declared using @code{__attribute ((vector_size))} are
2202 mangled in a non-standard way that does not allow for overloading of
2203 functions taking vectors of different sizes.
2205 The mangling is changed in @option{-fabi-version=4}.
2208 The known incompatibilities in @option{-fabi-version=1} include:
2213 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2214 pack data into the same byte as a base class. For example:
2217 struct A @{ virtual void f(); int f1 : 1; @};
2218 struct B : public A @{ int f2 : 1; @};
2222 In this case, G++ will place @code{B::f2} into the same byte
2223 as@code{A::f1}; other compilers will not. You can avoid this problem
2224 by explicitly padding @code{A} so that its size is a multiple of the
2225 byte size on your platform; that will cause G++ and other compilers to
2226 layout @code{B} identically.
2229 Incorrect handling of tail-padding for virtual bases. G++ does not use
2230 tail padding when laying out virtual bases. For example:
2233 struct A @{ virtual void f(); char c1; @};
2234 struct B @{ B(); char c2; @};
2235 struct C : public A, public virtual B @{@};
2239 In this case, G++ will not place @code{B} into the tail-padding for
2240 @code{A}; other compilers will. You can avoid this problem by
2241 explicitly padding @code{A} so that its size is a multiple of its
2242 alignment (ignoring virtual base classes); that will cause G++ and other
2243 compilers to layout @code{C} identically.
2246 Incorrect handling of bit-fields with declared widths greater than that
2247 of their underlying types, when the bit-fields appear in a union. For
2251 union U @{ int i : 4096; @};
2255 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2256 union too small by the number of bits in an @code{int}.
2259 Empty classes can be placed at incorrect offsets. For example:
2269 struct C : public B, public A @{@};
2273 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2274 it should be placed at offset zero. G++ mistakenly believes that the
2275 @code{A} data member of @code{B} is already at offset zero.
2278 Names of template functions whose types involve @code{typename} or
2279 template template parameters can be mangled incorrectly.
2282 template <typename Q>
2283 void f(typename Q::X) @{@}
2285 template <template <typename> class Q>
2286 void f(typename Q<int>::X) @{@}
2290 Instantiations of these templates may be mangled incorrectly.
2294 It also warns psABI related changes. The known psABI changes at this
2300 For SYSV/x86-64, when passing union with long double, it is changed to
2301 pass in memory as specified in psABI. For example:
2311 @code{union U} will always be passed in memory.
2315 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2316 @opindex Wctor-dtor-privacy
2317 @opindex Wno-ctor-dtor-privacy
2318 Warn when a class seems unusable because all the constructors or
2319 destructors in that class are private, and it has neither friends nor
2320 public static member functions.
2322 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
2324 @opindex Wno-noexcept
2325 Warn when a noexcept-expression evaluates to false because of a call
2326 to a function that does not have a non-throwing exception
2327 specification (i.e. @samp{throw()} or @samp{noexcept}) but is known by
2328 the compiler to never throw an exception.
2330 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2331 @opindex Wnon-virtual-dtor
2332 @opindex Wno-non-virtual-dtor
2333 Warn when a class has virtual functions and accessible non-virtual
2334 destructor, in which case it would be possible but unsafe to delete
2335 an instance of a derived class through a pointer to the base class.
2336 This warning is also enabled if -Weffc++ is specified.
2338 @item -Wreorder @r{(C++ and Objective-C++ only)}
2340 @opindex Wno-reorder
2341 @cindex reordering, warning
2342 @cindex warning for reordering of member initializers
2343 Warn when the order of member initializers given in the code does not
2344 match the order in which they must be executed. For instance:
2350 A(): j (0), i (1) @{ @}
2354 The compiler will rearrange the member initializers for @samp{i}
2355 and @samp{j} to match the declaration order of the members, emitting
2356 a warning to that effect. This warning is enabled by @option{-Wall}.
2359 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2362 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2365 Warn about violations of the following style guidelines from Scott Meyers'
2366 @cite{Effective C++} book:
2370 Item 11: Define a copy constructor and an assignment operator for classes
2371 with dynamically allocated memory.
2374 Item 12: Prefer initialization to assignment in constructors.
2377 Item 14: Make destructors virtual in base classes.
2380 Item 15: Have @code{operator=} return a reference to @code{*this}.
2383 Item 23: Don't try to return a reference when you must return an object.
2387 Also warn about violations of the following style guidelines from
2388 Scott Meyers' @cite{More Effective C++} book:
2392 Item 6: Distinguish between prefix and postfix forms of increment and
2393 decrement operators.
2396 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2400 When selecting this option, be aware that the standard library
2401 headers do not obey all of these guidelines; use @samp{grep -v}
2402 to filter out those warnings.
2404 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2405 @opindex Wstrict-null-sentinel
2406 @opindex Wno-strict-null-sentinel
2407 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2408 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2409 to @code{__null}. Although it is a null pointer constant not a null pointer,
2410 it is guaranteed to be of the same size as a pointer. But this use is
2411 not portable across different compilers.
2413 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2414 @opindex Wno-non-template-friend
2415 @opindex Wnon-template-friend
2416 Disable warnings when non-templatized friend functions are declared
2417 within a template. Since the advent of explicit template specification
2418 support in G++, if the name of the friend is an unqualified-id (i.e.,
2419 @samp{friend foo(int)}), the C++ language specification demands that the
2420 friend declare or define an ordinary, nontemplate function. (Section
2421 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2422 could be interpreted as a particular specialization of a templatized
2423 function. Because this non-conforming behavior is no longer the default
2424 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2425 check existing code for potential trouble spots and is on by default.
2426 This new compiler behavior can be turned off with
2427 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2428 but disables the helpful warning.
2430 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2431 @opindex Wold-style-cast
2432 @opindex Wno-old-style-cast
2433 Warn if an old-style (C-style) cast to a non-void type is used within
2434 a C++ program. The new-style casts (@samp{dynamic_cast},
2435 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2436 less vulnerable to unintended effects and much easier to search for.
2438 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2439 @opindex Woverloaded-virtual
2440 @opindex Wno-overloaded-virtual
2441 @cindex overloaded virtual function, warning
2442 @cindex warning for overloaded virtual function
2443 Warn when a function declaration hides virtual functions from a
2444 base class. For example, in:
2451 struct B: public A @{
2456 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2464 will fail to compile.
2466 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2467 @opindex Wno-pmf-conversions
2468 @opindex Wpmf-conversions
2469 Disable the diagnostic for converting a bound pointer to member function
2472 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2473 @opindex Wsign-promo
2474 @opindex Wno-sign-promo
2475 Warn when overload resolution chooses a promotion from unsigned or
2476 enumerated type to a signed type, over a conversion to an unsigned type of
2477 the same size. Previous versions of G++ would try to preserve
2478 unsignedness, but the standard mandates the current behavior.
2483 A& operator = (int);
2493 In this example, G++ will synthesize a default @samp{A& operator =
2494 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2497 @node Objective-C and Objective-C++ Dialect Options
2498 @section Options Controlling Objective-C and Objective-C++ Dialects
2500 @cindex compiler options, Objective-C and Objective-C++
2501 @cindex Objective-C and Objective-C++ options, command line
2502 @cindex options, Objective-C and Objective-C++
2503 (NOTE: This manual does not describe the Objective-C and Objective-C++
2504 languages themselves. See @xref{Standards,,Language Standards
2505 Supported by GCC}, for references.)
2507 This section describes the command-line options that are only meaningful
2508 for Objective-C and Objective-C++ programs, but you can also use most of
2509 the language-independent GNU compiler options.
2510 For example, you might compile a file @code{some_class.m} like this:
2513 gcc -g -fgnu-runtime -O -c some_class.m
2517 In this example, @option{-fgnu-runtime} is an option meant only for
2518 Objective-C and Objective-C++ programs; you can use the other options with
2519 any language supported by GCC@.
2521 Note that since Objective-C is an extension of the C language, Objective-C
2522 compilations may also use options specific to the C front-end (e.g.,
2523 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2524 C++-specific options (e.g., @option{-Wabi}).
2526 Here is a list of options that are @emph{only} for compiling Objective-C
2527 and Objective-C++ programs:
2530 @item -fconstant-string-class=@var{class-name}
2531 @opindex fconstant-string-class
2532 Use @var{class-name} as the name of the class to instantiate for each
2533 literal string specified with the syntax @code{@@"@dots{}"}. The default
2534 class name is @code{NXConstantString} if the GNU runtime is being used, and
2535 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2536 @option{-fconstant-cfstrings} option, if also present, will override the
2537 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2538 to be laid out as constant CoreFoundation strings.
2541 @opindex fgnu-runtime
2542 Generate object code compatible with the standard GNU Objective-C
2543 runtime. This is the default for most types of systems.
2545 @item -fnext-runtime
2546 @opindex fnext-runtime
2547 Generate output compatible with the NeXT runtime. This is the default
2548 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2549 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2552 @item -fno-nil-receivers
2553 @opindex fno-nil-receivers
2554 Assume that all Objective-C message dispatches (@code{[receiver
2555 message:arg]}) in this translation unit ensure that the receiver is
2556 not @code{nil}. This allows for more efficient entry points in the
2557 runtime to be used. Currently, this option is only available in
2558 conjunction with the NeXT runtime on Mac OS X 10.3 and later.
2560 @item -fobjc-call-cxx-cdtors
2561 @opindex fobjc-call-cxx-cdtors
2562 For each Objective-C class, check if any of its instance variables is a
2563 C++ object with a non-trivial default constructor. If so, synthesize a
2564 special @code{- (id) .cxx_construct} instance method that will run
2565 non-trivial default constructors on any such instance variables, in order,
2566 and then return @code{self}. Similarly, check if any instance variable
2567 is a C++ object with a non-trivial destructor, and if so, synthesize a
2568 special @code{- (void) .cxx_destruct} method that will run
2569 all such default destructors, in reverse order.
2571 The @code{- (id) .cxx_construct} and @code{- (void) .cxx_destruct}
2572 methods thusly generated will only operate on instance variables
2573 declared in the current Objective-C class, and not those inherited
2574 from superclasses. It is the responsibility of the Objective-C
2575 runtime to invoke all such methods in an object's inheritance
2576 hierarchy. The @code{- (id) .cxx_construct} methods will be invoked
2577 by the runtime immediately after a new object instance is allocated;
2578 the @code{- (void) .cxx_destruct} methods will be invoked immediately
2579 before the runtime deallocates an object instance.
2581 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2582 support for invoking the @code{- (id) .cxx_construct} and
2583 @code{- (void) .cxx_destruct} methods.
2585 @item -fobjc-direct-dispatch
2586 @opindex fobjc-direct-dispatch
2587 Allow fast jumps to the message dispatcher. On Darwin this is
2588 accomplished via the comm page.
2590 @item -fobjc-exceptions
2591 @opindex fobjc-exceptions
2592 Enable syntactic support for structured exception handling in
2593 Objective-C, similar to what is offered by C++ and Java. This option
2594 is required to use the Objective-C keywords @code{@@try},
2595 @code{@@throw}, @code{@@catch}, @code{@@finally} and
2596 @code{@@synchronized}. This option is available with both the GNU
2597 runtime and the NeXT runtime (but not available in conjunction with
2598 the NeXT runtime on Mac OS X 10.2 and earlier).
2602 Enable garbage collection (GC) in Objective-C and Objective-C++
2603 programs. This option is only available with the NeXT runtime; the
2604 GNU runtime has a different garbage collection implementation that
2605 does not require special compiler flags.
2607 @item -fobjc-std=objc1
2609 Conform to the language syntax of Objective-C 1.0, the language
2610 recognized by GCC 4.0. This only affects the Objective-C additions to
2611 the C/C++ language; it does not affect conformance to C/C++ standards,
2612 which is controlled by the separate C/C++ dialect option flags. When
2613 this option is used with the Objective-C or Objective-C++ compiler,
2614 any Objective-C syntax that is not recognized by GCC 4.0 is rejected.
2615 This is useful if you need to make sure that your Objective-C code can
2616 be compiled with older versions of GCC.
2618 @item -freplace-objc-classes
2619 @opindex freplace-objc-classes
2620 Emit a special marker instructing @command{ld(1)} not to statically link in
2621 the resulting object file, and allow @command{dyld(1)} to load it in at
2622 run time instead. This is used in conjunction with the Fix-and-Continue
2623 debugging mode, where the object file in question may be recompiled and
2624 dynamically reloaded in the course of program execution, without the need
2625 to restart the program itself. Currently, Fix-and-Continue functionality
2626 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2631 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2632 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2633 compile time) with static class references that get initialized at load time,
2634 which improves run-time performance. Specifying the @option{-fzero-link} flag
2635 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2636 to be retained. This is useful in Zero-Link debugging mode, since it allows
2637 for individual class implementations to be modified during program execution.
2638 The GNU runtime currently always retains calls to @code{objc_get_class("@dots{}")}
2639 regardless of command line options.
2643 Dump interface declarations for all classes seen in the source file to a
2644 file named @file{@var{sourcename}.decl}.
2646 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2647 @opindex Wassign-intercept
2648 @opindex Wno-assign-intercept
2649 Warn whenever an Objective-C assignment is being intercepted by the
2652 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2653 @opindex Wno-protocol
2655 If a class is declared to implement a protocol, a warning is issued for
2656 every method in the protocol that is not implemented by the class. The
2657 default behavior is to issue a warning for every method not explicitly
2658 implemented in the class, even if a method implementation is inherited
2659 from the superclass. If you use the @option{-Wno-protocol} option, then
2660 methods inherited from the superclass are considered to be implemented,
2661 and no warning is issued for them.
2663 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2665 @opindex Wno-selector
2666 Warn if multiple methods of different types for the same selector are
2667 found during compilation. The check is performed on the list of methods
2668 in the final stage of compilation. Additionally, a check is performed
2669 for each selector appearing in a @code{@@selector(@dots{})}
2670 expression, and a corresponding method for that selector has been found
2671 during compilation. Because these checks scan the method table only at
2672 the end of compilation, these warnings are not produced if the final
2673 stage of compilation is not reached, for example because an error is
2674 found during compilation, or because the @option{-fsyntax-only} option is
2677 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2678 @opindex Wstrict-selector-match
2679 @opindex Wno-strict-selector-match
2680 Warn if multiple methods with differing argument and/or return types are
2681 found for a given selector when attempting to send a message using this
2682 selector to a receiver of type @code{id} or @code{Class}. When this flag
2683 is off (which is the default behavior), the compiler will omit such warnings
2684 if any differences found are confined to types which share the same size
2687 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2688 @opindex Wundeclared-selector
2689 @opindex Wno-undeclared-selector
2690 Warn if a @code{@@selector(@dots{})} expression referring to an
2691 undeclared selector is found. A selector is considered undeclared if no
2692 method with that name has been declared before the
2693 @code{@@selector(@dots{})} expression, either explicitly in an
2694 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2695 an @code{@@implementation} section. This option always performs its
2696 checks as soon as a @code{@@selector(@dots{})} expression is found,
2697 while @option{-Wselector} only performs its checks in the final stage of
2698 compilation. This also enforces the coding style convention
2699 that methods and selectors must be declared before being used.
2701 @item -print-objc-runtime-info
2702 @opindex print-objc-runtime-info
2703 Generate C header describing the largest structure that is passed by
2708 @node Language Independent Options
2709 @section Options to Control Diagnostic Messages Formatting
2710 @cindex options to control diagnostics formatting
2711 @cindex diagnostic messages
2712 @cindex message formatting
2714 Traditionally, diagnostic messages have been formatted irrespective of
2715 the output device's aspect (e.g.@: its width, @dots{}). The options described
2716 below can be used to control the diagnostic messages formatting
2717 algorithm, e.g.@: how many characters per line, how often source location
2718 information should be reported. Right now, only the C++ front end can
2719 honor these options. However it is expected, in the near future, that
2720 the remaining front ends would be able to digest them correctly.
2723 @item -fmessage-length=@var{n}
2724 @opindex fmessage-length
2725 Try to format error messages so that they fit on lines of about @var{n}
2726 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2727 the front ends supported by GCC@. If @var{n} is zero, then no
2728 line-wrapping will be done; each error message will appear on a single
2731 @opindex fdiagnostics-show-location
2732 @item -fdiagnostics-show-location=once
2733 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2734 reporter to emit @emph{once} source location information; that is, in
2735 case the message is too long to fit on a single physical line and has to
2736 be wrapped, the source location won't be emitted (as prefix) again,
2737 over and over, in subsequent continuation lines. This is the default
2740 @item -fdiagnostics-show-location=every-line
2741 Only meaningful in line-wrapping mode. Instructs the diagnostic
2742 messages reporter to emit the same source location information (as
2743 prefix) for physical lines that result from the process of breaking
2744 a message which is too long to fit on a single line.
2746 @item -fdiagnostics-show-option
2747 @opindex fdiagnostics-show-option
2748 This option instructs the diagnostic machinery to add text to each
2749 diagnostic emitted, which indicates which command line option directly
2750 controls that diagnostic, when such an option is known to the
2751 diagnostic machinery.
2753 @item -Wcoverage-mismatch
2754 @opindex Wcoverage-mismatch
2755 Warn if feedback profiles do not match when using the
2756 @option{-fprofile-use} option.
2757 If a source file was changed between @option{-fprofile-gen} and
2758 @option{-fprofile-use}, the files with the profile feedback can fail
2759 to match the source file and GCC can not use the profile feedback
2760 information. By default, this warning is enabled and is treated as an
2761 error. @option{-Wno-coverage-mismatch} can be used to disable the
2762 warning or @option{-Wno-error=coverage-mismatch} can be used to
2763 disable the error. Disable the error for this warning can result in
2764 poorly optimized code, so disabling the error is useful only in the
2765 case of very minor changes such as bug fixes to an existing code-base.
2766 Completely disabling the warning is not recommended.
2770 @node Warning Options
2771 @section Options to Request or Suppress Warnings
2772 @cindex options to control warnings
2773 @cindex warning messages
2774 @cindex messages, warning
2775 @cindex suppressing warnings
2777 Warnings are diagnostic messages that report constructions which
2778 are not inherently erroneous but which are risky or suggest there
2779 may have been an error.
2781 The following language-independent options do not enable specific
2782 warnings but control the kinds of diagnostics produced by GCC.
2785 @cindex syntax checking
2787 @opindex fsyntax-only
2788 Check the code for syntax errors, but don't do anything beyond that.
2790 @item -fmax-errors=@var{n}
2791 @opindex fmax-errors
2792 Limits the maximum number of error messages to @var{n}, at which point
2793 GCC bails out rather than attempting to continue processing the source
2794 code. If @var{n} is 0 (the default), there is no limit on the number
2795 of error messages produced. If @option{-Wfatal-errors} is also
2796 specified, then @option{-Wfatal-errors} takes precedence over this
2801 Inhibit all warning messages.
2806 Make all warnings into errors.
2811 Make the specified warning into an error. The specifier for a warning
2812 is appended, for example @option{-Werror=switch} turns the warnings
2813 controlled by @option{-Wswitch} into errors. This switch takes a
2814 negative form, to be used to negate @option{-Werror} for specific
2815 warnings, for example @option{-Wno-error=switch} makes
2816 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2817 is in effect. You can use the @option{-fdiagnostics-show-option}
2818 option to have each controllable warning amended with the option which
2819 controls it, to determine what to use with this option.
2821 Note that specifying @option{-Werror=}@var{foo} automatically implies
2822 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2825 @item -Wfatal-errors
2826 @opindex Wfatal-errors
2827 @opindex Wno-fatal-errors
2828 This option causes the compiler to abort compilation on the first error
2829 occurred rather than trying to keep going and printing further error
2834 You can request many specific warnings with options beginning
2835 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2836 implicit declarations. Each of these specific warning options also
2837 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2838 example, @option{-Wno-implicit}. This manual lists only one of the
2839 two forms, whichever is not the default. For further,
2840 language-specific options also refer to @ref{C++ Dialect Options} and
2841 @ref{Objective-C and Objective-C++ Dialect Options}.
2843 When an unrecognized warning option is requested (e.g.,
2844 @option{-Wunknown-warning}), GCC will emit a diagnostic stating
2845 that the option is not recognized. However, if the @option{-Wno-} form
2846 is used, the behavior is slightly different: No diagnostic will be
2847 produced for @option{-Wno-unknown-warning} unless other diagnostics
2848 are being produced. This allows the use of new @option{-Wno-} options
2849 with old compilers, but if something goes wrong, the compiler will
2850 warn that an unrecognized option was used.
2855 Issue all the warnings demanded by strict ISO C and ISO C++;
2856 reject all programs that use forbidden extensions, and some other
2857 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2858 version of the ISO C standard specified by any @option{-std} option used.
2860 Valid ISO C and ISO C++ programs should compile properly with or without
2861 this option (though a rare few will require @option{-ansi} or a
2862 @option{-std} option specifying the required version of ISO C)@. However,
2863 without this option, certain GNU extensions and traditional C and C++
2864 features are supported as well. With this option, they are rejected.
2866 @option{-pedantic} does not cause warning messages for use of the
2867 alternate keywords whose names begin and end with @samp{__}. Pedantic
2868 warnings are also disabled in the expression that follows
2869 @code{__extension__}. However, only system header files should use
2870 these escape routes; application programs should avoid them.
2871 @xref{Alternate Keywords}.
2873 Some users try to use @option{-pedantic} to check programs for strict ISO
2874 C conformance. They soon find that it does not do quite what they want:
2875 it finds some non-ISO practices, but not all---only those for which
2876 ISO C @emph{requires} a diagnostic, and some others for which
2877 diagnostics have been added.
2879 A feature to report any failure to conform to ISO C might be useful in
2880 some instances, but would require considerable additional work and would
2881 be quite different from @option{-pedantic}. We don't have plans to
2882 support such a feature in the near future.
2884 Where the standard specified with @option{-std} represents a GNU
2885 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
2886 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2887 extended dialect is based. Warnings from @option{-pedantic} are given
2888 where they are required by the base standard. (It would not make sense
2889 for such warnings to be given only for features not in the specified GNU
2890 C dialect, since by definition the GNU dialects of C include all
2891 features the compiler supports with the given option, and there would be
2892 nothing to warn about.)
2894 @item -pedantic-errors
2895 @opindex pedantic-errors
2896 Like @option{-pedantic}, except that errors are produced rather than
2902 This enables all the warnings about constructions that some users
2903 consider questionable, and that are easy to avoid (or modify to
2904 prevent the warning), even in conjunction with macros. This also
2905 enables some language-specific warnings described in @ref{C++ Dialect
2906 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2908 @option{-Wall} turns on the following warning flags:
2910 @gccoptlist{-Waddress @gol
2911 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2913 -Wchar-subscripts @gol
2914 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2915 -Wimplicit-int @r{(C and Objective-C only)} @gol
2916 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
2919 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2920 -Wmissing-braces @gol
2926 -Wsequence-point @gol
2927 -Wsign-compare @r{(only in C++)} @gol
2928 -Wstrict-aliasing @gol
2929 -Wstrict-overflow=1 @gol
2932 -Wuninitialized @gol
2933 -Wunknown-pragmas @gol
2934 -Wunused-function @gol
2937 -Wunused-variable @gol
2938 -Wvolatile-register-var @gol
2941 Note that some warning flags are not implied by @option{-Wall}. Some of
2942 them warn about constructions that users generally do not consider
2943 questionable, but which occasionally you might wish to check for;
2944 others warn about constructions that are necessary or hard to avoid in
2945 some cases, and there is no simple way to modify the code to suppress
2946 the warning. Some of them are enabled by @option{-Wextra} but many of
2947 them must be enabled individually.
2953 This enables some extra warning flags that are not enabled by
2954 @option{-Wall}. (This option used to be called @option{-W}. The older
2955 name is still supported, but the newer name is more descriptive.)
2957 @gccoptlist{-Wclobbered @gol
2959 -Wignored-qualifiers @gol
2960 -Wmissing-field-initializers @gol
2961 -Wmissing-parameter-type @r{(C only)} @gol
2962 -Wold-style-declaration @r{(C only)} @gol
2963 -Woverride-init @gol
2966 -Wuninitialized @gol
2967 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2968 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2971 The option @option{-Wextra} also prints warning messages for the
2977 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2978 @samp{>}, or @samp{>=}.
2981 (C++ only) An enumerator and a non-enumerator both appear in a
2982 conditional expression.
2985 (C++ only) Ambiguous virtual bases.
2988 (C++ only) Subscripting an array which has been declared @samp{register}.
2991 (C++ only) Taking the address of a variable which has been declared
2995 (C++ only) A base class is not initialized in a derived class' copy
3000 @item -Wchar-subscripts
3001 @opindex Wchar-subscripts
3002 @opindex Wno-char-subscripts
3003 Warn if an array subscript has type @code{char}. This is a common cause
3004 of error, as programmers often forget that this type is signed on some
3006 This warning is enabled by @option{-Wall}.
3010 @opindex Wno-comment
3011 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3012 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3013 This warning is enabled by @option{-Wall}.
3016 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3018 Suppress warning messages emitted by @code{#warning} directives.
3020 @item -Wdouble-promotion @r{(C, C++, Objective-C and Objective-C++ only)}
3021 @opindex Wdouble-promotion
3022 @opindex Wno-double-promotion
3023 Give a warning when a value of type @code{float} is implicitly
3024 promoted to @code{double}. CPUs with a 32-bit ``single-precision''
3025 floating-point unit implement @code{float} in hardware, but emulate
3026 @code{double} in software. On such a machine, doing computations
3027 using @code{double} values is much more expensive because of the
3028 overhead required for software emulation.
3030 It is easy to accidentally do computations with @code{double} because
3031 floating-point literals are implicitly of type @code{double}. For
3035 float area(float radius)
3037 return 3.14159 * radius * radius;
3041 the compiler will perform the entire computation with @code{double}
3042 because the floating-point literal is a @code{double}.
3047 @opindex ffreestanding
3048 @opindex fno-builtin
3049 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3050 the arguments supplied have types appropriate to the format string
3051 specified, and that the conversions specified in the format string make
3052 sense. This includes standard functions, and others specified by format
3053 attributes (@pxref{Function Attributes}), in the @code{printf},
3054 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3055 not in the C standard) families (or other target-specific families).
3056 Which functions are checked without format attributes having been
3057 specified depends on the standard version selected, and such checks of
3058 functions without the attribute specified are disabled by
3059 @option{-ffreestanding} or @option{-fno-builtin}.
3061 The formats are checked against the format features supported by GNU
3062 libc version 2.2. These include all ISO C90 and C99 features, as well
3063 as features from the Single Unix Specification and some BSD and GNU
3064 extensions. Other library implementations may not support all these
3065 features; GCC does not support warning about features that go beyond a
3066 particular library's limitations. However, if @option{-pedantic} is used
3067 with @option{-Wformat}, warnings will be given about format features not
3068 in the selected standard version (but not for @code{strfmon} formats,
3069 since those are not in any version of the C standard). @xref{C Dialect
3070 Options,,Options Controlling C Dialect}.
3072 Since @option{-Wformat} also checks for null format arguments for
3073 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3075 @option{-Wformat} is included in @option{-Wall}. For more control over some
3076 aspects of format checking, the options @option{-Wformat-y2k},
3077 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3078 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3079 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
3082 @opindex Wformat-y2k
3083 @opindex Wno-format-y2k
3084 If @option{-Wformat} is specified, also warn about @code{strftime}
3085 formats which may yield only a two-digit year.
3087 @item -Wno-format-contains-nul
3088 @opindex Wno-format-contains-nul
3089 @opindex Wformat-contains-nul
3090 If @option{-Wformat} is specified, do not warn about format strings that
3093 @item -Wno-format-extra-args
3094 @opindex Wno-format-extra-args
3095 @opindex Wformat-extra-args
3096 If @option{-Wformat} is specified, do not warn about excess arguments to a
3097 @code{printf} or @code{scanf} format function. The C standard specifies
3098 that such arguments are ignored.
3100 Where the unused arguments lie between used arguments that are
3101 specified with @samp{$} operand number specifications, normally
3102 warnings are still given, since the implementation could not know what
3103 type to pass to @code{va_arg} to skip the unused arguments. However,
3104 in the case of @code{scanf} formats, this option will suppress the
3105 warning if the unused arguments are all pointers, since the Single
3106 Unix Specification says that such unused arguments are allowed.
3108 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3109 @opindex Wno-format-zero-length
3110 @opindex Wformat-zero-length
3111 If @option{-Wformat} is specified, do not warn about zero-length formats.
3112 The C standard specifies that zero-length formats are allowed.
3114 @item -Wformat-nonliteral
3115 @opindex Wformat-nonliteral
3116 @opindex Wno-format-nonliteral
3117 If @option{-Wformat} is specified, also warn if the format string is not a
3118 string literal and so cannot be checked, unless the format function
3119 takes its format arguments as a @code{va_list}.
3121 @item -Wformat-security
3122 @opindex Wformat-security
3123 @opindex Wno-format-security
3124 If @option{-Wformat} is specified, also warn about uses of format
3125 functions that represent possible security problems. At present, this
3126 warns about calls to @code{printf} and @code{scanf} functions where the
3127 format string is not a string literal and there are no format arguments,
3128 as in @code{printf (foo);}. This may be a security hole if the format
3129 string came from untrusted input and contains @samp{%n}. (This is
3130 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3131 in future warnings may be added to @option{-Wformat-security} that are not
3132 included in @option{-Wformat-nonliteral}.)
3136 @opindex Wno-format=2
3137 Enable @option{-Wformat} plus format checks not included in
3138 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3139 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3141 @item -Wnonnull @r{(C and Objective-C only)}
3143 @opindex Wno-nonnull
3144 Warn about passing a null pointer for arguments marked as
3145 requiring a non-null value by the @code{nonnull} function attribute.
3147 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3148 can be disabled with the @option{-Wno-nonnull} option.
3150 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3152 @opindex Wno-init-self
3153 Warn about uninitialized variables which are initialized with themselves.
3154 Note this option can only be used with the @option{-Wuninitialized} option.
3156 For example, GCC will warn about @code{i} being uninitialized in the
3157 following snippet only when @option{-Winit-self} has been specified:
3168 @item -Wimplicit-int @r{(C and Objective-C only)}
3169 @opindex Wimplicit-int
3170 @opindex Wno-implicit-int
3171 Warn when a declaration does not specify a type.
3172 This warning is enabled by @option{-Wall}.
3174 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3175 @opindex Wimplicit-function-declaration
3176 @opindex Wno-implicit-function-declaration
3177 Give a warning whenever a function is used before being declared. In
3178 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3179 enabled by default and it is made into an error by
3180 @option{-pedantic-errors}. This warning is also enabled by
3183 @item -Wimplicit @r{(C and Objective-C only)}
3185 @opindex Wno-implicit
3186 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3187 This warning is enabled by @option{-Wall}.
3189 @item -Wignored-qualifiers @r{(C and C++ only)}
3190 @opindex Wignored-qualifiers
3191 @opindex Wno-ignored-qualifiers
3192 Warn if the return type of a function has a type qualifier
3193 such as @code{const}. For ISO C such a type qualifier has no effect,
3194 since the value returned by a function is not an lvalue.
3195 For C++, the warning is only emitted for scalar types or @code{void}.
3196 ISO C prohibits qualified @code{void} return types on function
3197 definitions, so such return types always receive a warning
3198 even without this option.
3200 This warning is also enabled by @option{-Wextra}.
3205 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3206 a function with external linkage, returning int, taking either zero
3207 arguments, two, or three arguments of appropriate types. This warning
3208 is enabled by default in C++ and is enabled by either @option{-Wall}
3209 or @option{-pedantic}.
3211 @item -Wmissing-braces
3212 @opindex Wmissing-braces
3213 @opindex Wno-missing-braces
3214 Warn if an aggregate or union initializer is not fully bracketed. In
3215 the following example, the initializer for @samp{a} is not fully
3216 bracketed, but that for @samp{b} is fully bracketed.
3219 int a[2][2] = @{ 0, 1, 2, 3 @};
3220 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3223 This warning is enabled by @option{-Wall}.
3225 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3226 @opindex Wmissing-include-dirs
3227 @opindex Wno-missing-include-dirs
3228 Warn if a user-supplied include directory does not exist.
3231 @opindex Wparentheses
3232 @opindex Wno-parentheses
3233 Warn if parentheses are omitted in certain contexts, such
3234 as when there is an assignment in a context where a truth value
3235 is expected, or when operators are nested whose precedence people
3236 often get confused about.
3238 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3239 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3240 interpretation from that of ordinary mathematical notation.
3242 Also warn about constructions where there may be confusion to which
3243 @code{if} statement an @code{else} branch belongs. Here is an example of
3258 In C/C++, every @code{else} branch belongs to the innermost possible
3259 @code{if} statement, which in this example is @code{if (b)}. This is
3260 often not what the programmer expected, as illustrated in the above
3261 example by indentation the programmer chose. When there is the
3262 potential for this confusion, GCC will issue a warning when this flag
3263 is specified. To eliminate the warning, add explicit braces around
3264 the innermost @code{if} statement so there is no way the @code{else}
3265 could belong to the enclosing @code{if}. The resulting code would
3282 Also warn for dangerous uses of the
3283 ?: with omitted middle operand GNU extension. When the condition
3284 in the ?: operator is a boolean expression the omitted value will
3285 be always 1. Often the user expects it to be a value computed
3286 inside the conditional expression instead.
3288 This warning is enabled by @option{-Wall}.
3290 @item -Wsequence-point
3291 @opindex Wsequence-point
3292 @opindex Wno-sequence-point
3293 Warn about code that may have undefined semantics because of violations
3294 of sequence point rules in the C and C++ standards.
3296 The C and C++ standards defines the order in which expressions in a C/C++
3297 program are evaluated in terms of @dfn{sequence points}, which represent
3298 a partial ordering between the execution of parts of the program: those
3299 executed before the sequence point, and those executed after it. These
3300 occur after the evaluation of a full expression (one which is not part
3301 of a larger expression), after the evaluation of the first operand of a
3302 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3303 function is called (but after the evaluation of its arguments and the
3304 expression denoting the called function), and in certain other places.
3305 Other than as expressed by the sequence point rules, the order of
3306 evaluation of subexpressions of an expression is not specified. All
3307 these rules describe only a partial order rather than a total order,
3308 since, for example, if two functions are called within one expression
3309 with no sequence point between them, the order in which the functions
3310 are called is not specified. However, the standards committee have
3311 ruled that function calls do not overlap.
3313 It is not specified when between sequence points modifications to the
3314 values of objects take effect. Programs whose behavior depends on this
3315 have undefined behavior; the C and C++ standards specify that ``Between
3316 the previous and next sequence point an object shall have its stored
3317 value modified at most once by the evaluation of an expression.
3318 Furthermore, the prior value shall be read only to determine the value
3319 to be stored.''. If a program breaks these rules, the results on any
3320 particular implementation are entirely unpredictable.
3322 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3323 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3324 diagnosed by this option, and it may give an occasional false positive
3325 result, but in general it has been found fairly effective at detecting
3326 this sort of problem in programs.
3328 The standard is worded confusingly, therefore there is some debate
3329 over the precise meaning of the sequence point rules in subtle cases.
3330 Links to discussions of the problem, including proposed formal
3331 definitions, may be found on the GCC readings page, at
3332 @uref{http://gcc.gnu.org/@/readings.html}.
3334 This warning is enabled by @option{-Wall} for C and C++.
3337 @opindex Wreturn-type
3338 @opindex Wno-return-type
3339 Warn whenever a function is defined with a return-type that defaults
3340 to @code{int}. Also warn about any @code{return} statement with no
3341 return-value in a function whose return-type is not @code{void}
3342 (falling off the end of the function body is considered returning
3343 without a value), and about a @code{return} statement with an
3344 expression in a function whose return-type is @code{void}.
3346 For C++, a function without return type always produces a diagnostic
3347 message, even when @option{-Wno-return-type} is specified. The only
3348 exceptions are @samp{main} and functions defined in system headers.
3350 This warning is enabled by @option{-Wall}.
3355 Warn whenever a @code{switch} statement has an index of enumerated type
3356 and lacks a @code{case} for one or more of the named codes of that
3357 enumeration. (The presence of a @code{default} label prevents this
3358 warning.) @code{case} labels outside the enumeration range also
3359 provoke warnings when this option is used (even if there is a
3360 @code{default} label).
3361 This warning is enabled by @option{-Wall}.
3363 @item -Wswitch-default
3364 @opindex Wswitch-default
3365 @opindex Wno-switch-default
3366 Warn whenever a @code{switch} statement does not have a @code{default}
3370 @opindex Wswitch-enum
3371 @opindex Wno-switch-enum
3372 Warn whenever a @code{switch} statement has an index of enumerated type
3373 and lacks a @code{case} for one or more of the named codes of that
3374 enumeration. @code{case} labels outside the enumeration range also
3375 provoke warnings when this option is used. The only difference
3376 between @option{-Wswitch} and this option is that this option gives a
3377 warning about an omitted enumeration code even if there is a
3378 @code{default} label.
3380 @item -Wsync-nand @r{(C and C++ only)}
3382 @opindex Wno-sync-nand
3383 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3384 built-in functions are used. These functions changed semantics in GCC 4.4.
3388 @opindex Wno-trigraphs
3389 Warn if any trigraphs are encountered that might change the meaning of
3390 the program (trigraphs within comments are not warned about).
3391 This warning is enabled by @option{-Wall}.
3393 @item -Wunused-but-set-parameter
3394 @opindex Wunused-but-set-parameter
3395 @opindex Wno-unused-but-set-parameter
3396 Warn whenever a function parameter is assigned to, but otherwise unused
3397 (aside from its declaration).
3399 To suppress this warning use the @samp{unused} attribute
3400 (@pxref{Variable Attributes}).
3402 This warning is also enabled by @option{-Wunused} together with
3405 @item -Wunused-but-set-variable
3406 @opindex Wunused-but-set-variable
3407 @opindex Wno-unused-but-set-variable
3408 Warn whenever a local variable is assigned to, but otherwise unused
3409 (aside from its declaration).
3410 This warning is enabled by @option{-Wall}.
3412 To suppress this warning use the @samp{unused} attribute
3413 (@pxref{Variable Attributes}).
3415 This warning is also enabled by @option{-Wunused}, which is enabled
3418 @item -Wunused-function
3419 @opindex Wunused-function
3420 @opindex Wno-unused-function
3421 Warn whenever a static function is declared but not defined or a
3422 non-inline static function is unused.
3423 This warning is enabled by @option{-Wall}.
3425 @item -Wunused-label
3426 @opindex Wunused-label
3427 @opindex Wno-unused-label
3428 Warn whenever a label is declared but not used.
3429 This warning is enabled by @option{-Wall}.
3431 To suppress this warning use the @samp{unused} attribute
3432 (@pxref{Variable Attributes}).
3434 @item -Wunused-parameter
3435 @opindex Wunused-parameter
3436 @opindex Wno-unused-parameter
3437 Warn whenever a function parameter is unused aside from its declaration.
3439 To suppress this warning use the @samp{unused} attribute
3440 (@pxref{Variable Attributes}).
3442 @item -Wno-unused-result
3443 @opindex Wunused-result
3444 @opindex Wno-unused-result
3445 Do not warn if a caller of a function marked with attribute
3446 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3447 its return value. The default is @option{-Wunused-result}.
3449 @item -Wunused-variable
3450 @opindex Wunused-variable
3451 @opindex Wno-unused-variable
3452 Warn whenever a local variable or non-constant static variable is unused
3453 aside from its declaration.
3454 This warning is enabled by @option{-Wall}.
3456 To suppress this warning use the @samp{unused} attribute
3457 (@pxref{Variable Attributes}).
3459 @item -Wunused-value
3460 @opindex Wunused-value
3461 @opindex Wno-unused-value
3462 Warn whenever a statement computes a result that is explicitly not
3463 used. To suppress this warning cast the unused expression to
3464 @samp{void}. This includes an expression-statement or the left-hand
3465 side of a comma expression that contains no side effects. For example,
3466 an expression such as @samp{x[i,j]} will cause a warning, while
3467 @samp{x[(void)i,j]} will not.
3469 This warning is enabled by @option{-Wall}.
3474 All the above @option{-Wunused} options combined.
3476 In order to get a warning about an unused function parameter, you must
3477 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3478 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3480 @item -Wuninitialized
3481 @opindex Wuninitialized
3482 @opindex Wno-uninitialized
3483 Warn if an automatic variable is used without first being initialized
3484 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3485 warn if a non-static reference or non-static @samp{const} member
3486 appears in a class without constructors.
3488 If you want to warn about code which uses the uninitialized value of the
3489 variable in its own initializer, use the @option{-Winit-self} option.
3491 These warnings occur for individual uninitialized or clobbered
3492 elements of structure, union or array variables as well as for
3493 variables which are uninitialized or clobbered as a whole. They do
3494 not occur for variables or elements declared @code{volatile}. Because
3495 these warnings depend on optimization, the exact variables or elements
3496 for which there are warnings will depend on the precise optimization
3497 options and version of GCC used.
3499 Note that there may be no warning about a variable that is used only
3500 to compute a value that itself is never used, because such
3501 computations may be deleted by data flow analysis before the warnings
3504 These warnings are made optional because GCC is not smart
3505 enough to see all the reasons why the code might be correct
3506 despite appearing to have an error. Here is one example of how
3527 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3528 always initialized, but GCC doesn't know this. Here is
3529 another common case:
3534 if (change_y) save_y = y, y = new_y;
3536 if (change_y) y = save_y;
3541 This has no bug because @code{save_y} is used only if it is set.
3543 @cindex @code{longjmp} warnings
3544 This option also warns when a non-volatile automatic variable might be
3545 changed by a call to @code{longjmp}. These warnings as well are possible
3546 only in optimizing compilation.
3548 The compiler sees only the calls to @code{setjmp}. It cannot know
3549 where @code{longjmp} will be called; in fact, a signal handler could
3550 call it at any point in the code. As a result, you may get a warning
3551 even when there is in fact no problem because @code{longjmp} cannot
3552 in fact be called at the place which would cause a problem.
3554 Some spurious warnings can be avoided if you declare all the functions
3555 you use that never return as @code{noreturn}. @xref{Function
3558 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3560 @item -Wunknown-pragmas
3561 @opindex Wunknown-pragmas
3562 @opindex Wno-unknown-pragmas
3563 @cindex warning for unknown pragmas
3564 @cindex unknown pragmas, warning
3565 @cindex pragmas, warning of unknown
3566 Warn when a #pragma directive is encountered which is not understood by
3567 GCC@. If this command line option is used, warnings will even be issued
3568 for unknown pragmas in system header files. This is not the case if
3569 the warnings were only enabled by the @option{-Wall} command line option.
3572 @opindex Wno-pragmas
3574 Do not warn about misuses of pragmas, such as incorrect parameters,
3575 invalid syntax, or conflicts between pragmas. See also
3576 @samp{-Wunknown-pragmas}.
3578 @item -Wstrict-aliasing
3579 @opindex Wstrict-aliasing
3580 @opindex Wno-strict-aliasing
3581 This option is only active when @option{-fstrict-aliasing} is active.
3582 It warns about code which might break the strict aliasing rules that the
3583 compiler is using for optimization. The warning does not catch all
3584 cases, but does attempt to catch the more common pitfalls. It is
3585 included in @option{-Wall}.
3586 It is equivalent to @option{-Wstrict-aliasing=3}
3588 @item -Wstrict-aliasing=n
3589 @opindex Wstrict-aliasing=n
3590 @opindex Wno-strict-aliasing=n
3591 This option is only active when @option{-fstrict-aliasing} is active.
3592 It warns about code which might break the strict aliasing rules that the
3593 compiler is using for optimization.
3594 Higher levels correspond to higher accuracy (fewer false positives).
3595 Higher levels also correspond to more effort, similar to the way -O works.
3596 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3599 Level 1: Most aggressive, quick, least accurate.
3600 Possibly useful when higher levels
3601 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3602 false negatives. However, it has many false positives.
3603 Warns for all pointer conversions between possibly incompatible types,
3604 even if never dereferenced. Runs in the frontend only.
3606 Level 2: Aggressive, quick, not too precise.
3607 May still have many false positives (not as many as level 1 though),
3608 and few false negatives (but possibly more than level 1).
3609 Unlike level 1, it only warns when an address is taken. Warns about
3610 incomplete types. Runs in the frontend only.
3612 Level 3 (default for @option{-Wstrict-aliasing}):
3613 Should have very few false positives and few false
3614 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3615 Takes care of the common pun+dereference pattern in the frontend:
3616 @code{*(int*)&some_float}.
3617 If optimization is enabled, it also runs in the backend, where it deals
3618 with multiple statement cases using flow-sensitive points-to information.
3619 Only warns when the converted pointer is dereferenced.
3620 Does not warn about incomplete types.
3622 @item -Wstrict-overflow
3623 @itemx -Wstrict-overflow=@var{n}
3624 @opindex Wstrict-overflow
3625 @opindex Wno-strict-overflow
3626 This option is only active when @option{-fstrict-overflow} is active.
3627 It warns about cases where the compiler optimizes based on the
3628 assumption that signed overflow does not occur. Note that it does not
3629 warn about all cases where the code might overflow: it only warns
3630 about cases where the compiler implements some optimization. Thus
3631 this warning depends on the optimization level.
3633 An optimization which assumes that signed overflow does not occur is
3634 perfectly safe if the values of the variables involved are such that
3635 overflow never does, in fact, occur. Therefore this warning can
3636 easily give a false positive: a warning about code which is not
3637 actually a problem. To help focus on important issues, several
3638 warning levels are defined. No warnings are issued for the use of
3639 undefined signed overflow when estimating how many iterations a loop
3640 will require, in particular when determining whether a loop will be
3644 @item -Wstrict-overflow=1
3645 Warn about cases which are both questionable and easy to avoid. For
3646 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3647 compiler will simplify this to @code{1}. This level of
3648 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3649 are not, and must be explicitly requested.
3651 @item -Wstrict-overflow=2
3652 Also warn about other cases where a comparison is simplified to a
3653 constant. For example: @code{abs (x) >= 0}. This can only be
3654 simplified when @option{-fstrict-overflow} is in effect, because
3655 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3656 zero. @option{-Wstrict-overflow} (with no level) is the same as
3657 @option{-Wstrict-overflow=2}.
3659 @item -Wstrict-overflow=3
3660 Also warn about other cases where a comparison is simplified. For
3661 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3663 @item -Wstrict-overflow=4
3664 Also warn about other simplifications not covered by the above cases.
3665 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3667 @item -Wstrict-overflow=5
3668 Also warn about cases where the compiler reduces the magnitude of a
3669 constant involved in a comparison. For example: @code{x + 2 > y} will
3670 be simplified to @code{x + 1 >= y}. This is reported only at the
3671 highest warning level because this simplification applies to many
3672 comparisons, so this warning level will give a very large number of
3676 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]}
3677 @opindex Wsuggest-attribute=
3678 @opindex Wno-suggest-attribute=
3679 Warn for cases where adding an attribute may be beneficial. The
3680 attributes currently supported are listed below.
3683 @item -Wsuggest-attribute=pure
3684 @itemx -Wsuggest-attribute=const
3685 @itemx -Wsuggest-attribute=noreturn
3686 @opindex Wsuggest-attribute=pure
3687 @opindex Wno-suggest-attribute=pure
3688 @opindex Wsuggest-attribute=const
3689 @opindex Wno-suggest-attribute=const
3690 @opindex Wsuggest-attribute=noreturn
3691 @opindex Wno-suggest-attribute=noreturn
3693 Warn about functions which might be candidates for attributes
3694 @code{pure}, @code{const} or @code{noreturn}. The compiler only warns for
3695 functions visible in other compilation units or (in the case of @code{pure} and
3696 @code{const}) if it cannot prove that the function returns normally. A function
3697 returns normally if it doesn't contain an infinite loop nor returns abnormally
3698 by throwing, calling @code{abort()} or trapping. This analysis requires option
3699 @option{-fipa-pure-const}, which is enabled by default at @option{-O} and
3700 higher. Higher optimization levels improve the accuracy of the analysis.
3703 @item -Warray-bounds
3704 @opindex Wno-array-bounds
3705 @opindex Warray-bounds
3706 This option is only active when @option{-ftree-vrp} is active
3707 (default for @option{-O2} and above). It warns about subscripts to arrays
3708 that are always out of bounds. This warning is enabled by @option{-Wall}.
3710 @item -Wno-div-by-zero
3711 @opindex Wno-div-by-zero
3712 @opindex Wdiv-by-zero
3713 Do not warn about compile-time integer division by zero. Floating point
3714 division by zero is not warned about, as it can be a legitimate way of
3715 obtaining infinities and NaNs.
3717 @item -Wsystem-headers
3718 @opindex Wsystem-headers
3719 @opindex Wno-system-headers
3720 @cindex warnings from system headers
3721 @cindex system headers, warnings from
3722 Print warning messages for constructs found in system header files.
3723 Warnings from system headers are normally suppressed, on the assumption
3724 that they usually do not indicate real problems and would only make the
3725 compiler output harder to read. Using this command line option tells
3726 GCC to emit warnings from system headers as if they occurred in user
3727 code. However, note that using @option{-Wall} in conjunction with this
3728 option will @emph{not} warn about unknown pragmas in system
3729 headers---for that, @option{-Wunknown-pragmas} must also be used.
3732 @opindex Wtrampolines
3733 @opindex Wno-trampolines
3734 Warn about trampolines generated for pointers to nested functions.
3736 A trampoline is a small piece of data or code that is created at run
3737 time on the stack when the address of a nested function is taken, and
3738 is used to call the nested function indirectly. For some targets, it
3739 is made up of data only and thus requires no special treatment. But,
3740 for most targets, it is made up of code and thus requires the stack
3741 to be made executable in order for the program to work properly.
3744 @opindex Wfloat-equal
3745 @opindex Wno-float-equal
3746 Warn if floating point values are used in equality comparisons.
3748 The idea behind this is that sometimes it is convenient (for the
3749 programmer) to consider floating-point values as approximations to
3750 infinitely precise real numbers. If you are doing this, then you need
3751 to compute (by analyzing the code, or in some other way) the maximum or
3752 likely maximum error that the computation introduces, and allow for it
3753 when performing comparisons (and when producing output, but that's a
3754 different problem). In particular, instead of testing for equality, you
3755 would check to see whether the two values have ranges that overlap; and
3756 this is done with the relational operators, so equality comparisons are
3759 @item -Wtraditional @r{(C and Objective-C only)}
3760 @opindex Wtraditional
3761 @opindex Wno-traditional
3762 Warn about certain constructs that behave differently in traditional and
3763 ISO C@. Also warn about ISO C constructs that have no traditional C
3764 equivalent, and/or problematic constructs which should be avoided.
3768 Macro parameters that appear within string literals in the macro body.
3769 In traditional C macro replacement takes place within string literals,
3770 but does not in ISO C@.
3773 In traditional C, some preprocessor directives did not exist.
3774 Traditional preprocessors would only consider a line to be a directive
3775 if the @samp{#} appeared in column 1 on the line. Therefore
3776 @option{-Wtraditional} warns about directives that traditional C
3777 understands but would ignore because the @samp{#} does not appear as the
3778 first character on the line. It also suggests you hide directives like
3779 @samp{#pragma} not understood by traditional C by indenting them. Some
3780 traditional implementations would not recognize @samp{#elif}, so it
3781 suggests avoiding it altogether.
3784 A function-like macro that appears without arguments.
3787 The unary plus operator.
3790 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3791 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3792 constants.) Note, these suffixes appear in macros defined in the system
3793 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3794 Use of these macros in user code might normally lead to spurious
3795 warnings, however GCC's integrated preprocessor has enough context to
3796 avoid warning in these cases.
3799 A function declared external in one block and then used after the end of
3803 A @code{switch} statement has an operand of type @code{long}.
3806 A non-@code{static} function declaration follows a @code{static} one.
3807 This construct is not accepted by some traditional C compilers.
3810 The ISO type of an integer constant has a different width or
3811 signedness from its traditional type. This warning is only issued if
3812 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3813 typically represent bit patterns, are not warned about.
3816 Usage of ISO string concatenation is detected.
3819 Initialization of automatic aggregates.
3822 Identifier conflicts with labels. Traditional C lacks a separate
3823 namespace for labels.
3826 Initialization of unions. If the initializer is zero, the warning is
3827 omitted. This is done under the assumption that the zero initializer in
3828 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3829 initializer warnings and relies on default initialization to zero in the
3833 Conversions by prototypes between fixed/floating point values and vice
3834 versa. The absence of these prototypes when compiling with traditional
3835 C would cause serious problems. This is a subset of the possible
3836 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3839 Use of ISO C style function definitions. This warning intentionally is
3840 @emph{not} issued for prototype declarations or variadic functions
3841 because these ISO C features will appear in your code when using
3842 libiberty's traditional C compatibility macros, @code{PARAMS} and
3843 @code{VPARAMS}. This warning is also bypassed for nested functions
3844 because that feature is already a GCC extension and thus not relevant to
3845 traditional C compatibility.
3848 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3849 @opindex Wtraditional-conversion
3850 @opindex Wno-traditional-conversion
3851 Warn if a prototype causes a type conversion that is different from what
3852 would happen to the same argument in the absence of a prototype. This
3853 includes conversions of fixed point to floating and vice versa, and
3854 conversions changing the width or signedness of a fixed point argument
3855 except when the same as the default promotion.
3857 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3858 @opindex Wdeclaration-after-statement
3859 @opindex Wno-declaration-after-statement
3860 Warn when a declaration is found after a statement in a block. This
3861 construct, known from C++, was introduced with ISO C99 and is by default
3862 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3863 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3868 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3870 @item -Wno-endif-labels
3871 @opindex Wno-endif-labels
3872 @opindex Wendif-labels
3873 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3878 Warn whenever a local variable or type declaration shadows another variable,
3879 parameter, type, or class member (in C++), or whenever a built-in function
3880 is shadowed. Note that in C++, the compiler will not warn if a local variable
3881 shadows a struct/class/enum, but will warn if it shadows an explicit typedef.
3883 @item -Wlarger-than=@var{len}
3884 @opindex Wlarger-than=@var{len}
3885 @opindex Wlarger-than-@var{len}
3886 Warn whenever an object of larger than @var{len} bytes is defined.
3888 @item -Wframe-larger-than=@var{len}
3889 @opindex Wframe-larger-than
3890 Warn if the size of a function frame is larger than @var{len} bytes.
3891 The computation done to determine the stack frame size is approximate
3892 and not conservative.
3893 The actual requirements may be somewhat greater than @var{len}
3894 even if you do not get a warning. In addition, any space allocated
3895 via @code{alloca}, variable-length arrays, or related constructs
3896 is not included by the compiler when determining
3897 whether or not to issue a warning.
3899 @item -Wunsafe-loop-optimizations
3900 @opindex Wunsafe-loop-optimizations
3901 @opindex Wno-unsafe-loop-optimizations
3902 Warn if the loop cannot be optimized because the compiler could not
3903 assume anything on the bounds of the loop indices. With
3904 @option{-funsafe-loop-optimizations} warn if the compiler made
3907 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3908 @opindex Wno-pedantic-ms-format
3909 @opindex Wpedantic-ms-format
3910 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3911 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3912 depending on the MS runtime, when you are using the options @option{-Wformat}
3913 and @option{-pedantic} without gnu-extensions.
3915 @item -Wpointer-arith
3916 @opindex Wpointer-arith
3917 @opindex Wno-pointer-arith
3918 Warn about anything that depends on the ``size of'' a function type or
3919 of @code{void}. GNU C assigns these types a size of 1, for
3920 convenience in calculations with @code{void *} pointers and pointers
3921 to functions. In C++, warn also when an arithmetic operation involves
3922 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3925 @opindex Wtype-limits
3926 @opindex Wno-type-limits
3927 Warn if a comparison is always true or always false due to the limited
3928 range of the data type, but do not warn for constant expressions. For
3929 example, warn if an unsigned variable is compared against zero with
3930 @samp{<} or @samp{>=}. This warning is also enabled by
3933 @item -Wbad-function-cast @r{(C and Objective-C only)}
3934 @opindex Wbad-function-cast
3935 @opindex Wno-bad-function-cast
3936 Warn whenever a function call is cast to a non-matching type.
3937 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3939 @item -Wc++-compat @r{(C and Objective-C only)}
3940 Warn about ISO C constructs that are outside of the common subset of
3941 ISO C and ISO C++, e.g.@: request for implicit conversion from
3942 @code{void *} to a pointer to non-@code{void} type.
3944 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3945 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3946 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3947 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3951 @opindex Wno-cast-qual
3952 Warn whenever a pointer is cast so as to remove a type qualifier from
3953 the target type. For example, warn if a @code{const char *} is cast
3954 to an ordinary @code{char *}.
3956 Also warn when making a cast which introduces a type qualifier in an
3957 unsafe way. For example, casting @code{char **} to @code{const char **}
3958 is unsafe, as in this example:
3961 /* p is char ** value. */
3962 const char **q = (const char **) p;
3963 /* Assignment of readonly string to const char * is OK. */
3965 /* Now char** pointer points to read-only memory. */
3970 @opindex Wcast-align
3971 @opindex Wno-cast-align
3972 Warn whenever a pointer is cast such that the required alignment of the
3973 target is increased. For example, warn if a @code{char *} is cast to
3974 an @code{int *} on machines where integers can only be accessed at
3975 two- or four-byte boundaries.
3977 @item -Wwrite-strings
3978 @opindex Wwrite-strings
3979 @opindex Wno-write-strings
3980 When compiling C, give string constants the type @code{const
3981 char[@var{length}]} so that copying the address of one into a
3982 non-@code{const} @code{char *} pointer will get a warning. These
3983 warnings will help you find at compile time code that can try to write
3984 into a string constant, but only if you have been very careful about
3985 using @code{const} in declarations and prototypes. Otherwise, it will
3986 just be a nuisance. This is why we did not make @option{-Wall} request
3989 When compiling C++, warn about the deprecated conversion from string
3990 literals to @code{char *}. This warning is enabled by default for C++
3995 @opindex Wno-clobbered
3996 Warn for variables that might be changed by @samp{longjmp} or
3997 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
4000 @opindex Wconversion
4001 @opindex Wno-conversion
4002 Warn for implicit conversions that may alter a value. This includes
4003 conversions between real and integer, like @code{abs (x)} when
4004 @code{x} is @code{double}; conversions between signed and unsigned,
4005 like @code{unsigned ui = -1}; and conversions to smaller types, like
4006 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
4007 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
4008 changed by the conversion like in @code{abs (2.0)}. Warnings about
4009 conversions between signed and unsigned integers can be disabled by
4010 using @option{-Wno-sign-conversion}.
4012 For C++, also warn for confusing overload resolution for user-defined
4013 conversions; and conversions that will never use a type conversion
4014 operator: conversions to @code{void}, the same type, a base class or a
4015 reference to them. Warnings about conversions between signed and
4016 unsigned integers are disabled by default in C++ unless
4017 @option{-Wsign-conversion} is explicitly enabled.
4019 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
4020 @opindex Wconversion-null
4021 @opindex Wno-conversion-null
4022 Do not warn for conversions between @code{NULL} and non-pointer
4023 types. @option{-Wconversion-null} is enabled by default.
4026 @opindex Wempty-body
4027 @opindex Wno-empty-body
4028 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
4029 while} statement. This warning is also enabled by @option{-Wextra}.
4031 @item -Wenum-compare
4032 @opindex Wenum-compare
4033 @opindex Wno-enum-compare
4034 Warn about a comparison between values of different enum types. In C++
4035 this warning is enabled by default. In C this warning is enabled by
4038 @item -Wjump-misses-init @r{(C, Objective-C only)}
4039 @opindex Wjump-misses-init
4040 @opindex Wno-jump-misses-init
4041 Warn if a @code{goto} statement or a @code{switch} statement jumps
4042 forward across the initialization of a variable, or jumps backward to a
4043 label after the variable has been initialized. This only warns about
4044 variables which are initialized when they are declared. This warning is
4045 only supported for C and Objective C; in C++ this sort of branch is an
4048 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
4049 can be disabled with the @option{-Wno-jump-misses-init} option.
4051 @item -Wsign-compare
4052 @opindex Wsign-compare
4053 @opindex Wno-sign-compare
4054 @cindex warning for comparison of signed and unsigned values
4055 @cindex comparison of signed and unsigned values, warning
4056 @cindex signed and unsigned values, comparison warning
4057 Warn when a comparison between signed and unsigned values could produce
4058 an incorrect result when the signed value is converted to unsigned.
4059 This warning is also enabled by @option{-Wextra}; to get the other warnings
4060 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
4062 @item -Wsign-conversion
4063 @opindex Wsign-conversion
4064 @opindex Wno-sign-conversion
4065 Warn for implicit conversions that may change the sign of an integer
4066 value, like assigning a signed integer expression to an unsigned
4067 integer variable. An explicit cast silences the warning. In C, this
4068 option is enabled also by @option{-Wconversion}.
4072 @opindex Wno-address
4073 Warn about suspicious uses of memory addresses. These include using
4074 the address of a function in a conditional expression, such as
4075 @code{void func(void); if (func)}, and comparisons against the memory
4076 address of a string literal, such as @code{if (x == "abc")}. Such
4077 uses typically indicate a programmer error: the address of a function
4078 always evaluates to true, so their use in a conditional usually
4079 indicate that the programmer forgot the parentheses in a function
4080 call; and comparisons against string literals result in unspecified
4081 behavior and are not portable in C, so they usually indicate that the
4082 programmer intended to use @code{strcmp}. This warning is enabled by
4086 @opindex Wlogical-op
4087 @opindex Wno-logical-op
4088 Warn about suspicious uses of logical operators in expressions.
4089 This includes using logical operators in contexts where a
4090 bit-wise operator is likely to be expected.
4092 @item -Waggregate-return
4093 @opindex Waggregate-return
4094 @opindex Wno-aggregate-return
4095 Warn if any functions that return structures or unions are defined or
4096 called. (In languages where you can return an array, this also elicits
4099 @item -Wno-attributes
4100 @opindex Wno-attributes
4101 @opindex Wattributes
4102 Do not warn if an unexpected @code{__attribute__} is used, such as
4103 unrecognized attributes, function attributes applied to variables,
4104 etc. This will not stop errors for incorrect use of supported
4107 @item -Wno-builtin-macro-redefined
4108 @opindex Wno-builtin-macro-redefined
4109 @opindex Wbuiltin-macro-redefined
4110 Do not warn if certain built-in macros are redefined. This suppresses
4111 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4112 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4114 @item -Wstrict-prototypes @r{(C and Objective-C only)}
4115 @opindex Wstrict-prototypes
4116 @opindex Wno-strict-prototypes
4117 Warn if a function is declared or defined without specifying the
4118 argument types. (An old-style function definition is permitted without
4119 a warning if preceded by a declaration which specifies the argument
4122 @item -Wold-style-declaration @r{(C and Objective-C only)}
4123 @opindex Wold-style-declaration
4124 @opindex Wno-old-style-declaration
4125 Warn for obsolescent usages, according to the C Standard, in a
4126 declaration. For example, warn if storage-class specifiers like
4127 @code{static} are not the first things in a declaration. This warning
4128 is also enabled by @option{-Wextra}.
4130 @item -Wold-style-definition @r{(C and Objective-C only)}
4131 @opindex Wold-style-definition
4132 @opindex Wno-old-style-definition
4133 Warn if an old-style function definition is used. A warning is given
4134 even if there is a previous prototype.
4136 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
4137 @opindex Wmissing-parameter-type
4138 @opindex Wno-missing-parameter-type
4139 A function parameter is declared without a type specifier in K&R-style
4146 This warning is also enabled by @option{-Wextra}.
4148 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4149 @opindex Wmissing-prototypes
4150 @opindex Wno-missing-prototypes
4151 Warn if a global function is defined without a previous prototype
4152 declaration. This warning is issued even if the definition itself
4153 provides a prototype. The aim is to detect global functions that fail
4154 to be declared in header files.
4156 @item -Wmissing-declarations
4157 @opindex Wmissing-declarations
4158 @opindex Wno-missing-declarations
4159 Warn if a global function is defined without a previous declaration.
4160 Do so even if the definition itself provides a prototype.
4161 Use this option to detect global functions that are not declared in
4162 header files. In C++, no warnings are issued for function templates,
4163 or for inline functions, or for functions in anonymous namespaces.
4165 @item -Wmissing-field-initializers
4166 @opindex Wmissing-field-initializers
4167 @opindex Wno-missing-field-initializers
4171 Warn if a structure's initializer has some fields missing. For
4172 example, the following code would cause such a warning, because
4173 @code{x.h} is implicitly zero:
4176 struct s @{ int f, g, h; @};
4177 struct s x = @{ 3, 4 @};
4180 This option does not warn about designated initializers, so the following
4181 modification would not trigger a warning:
4184 struct s @{ int f, g, h; @};
4185 struct s x = @{ .f = 3, .g = 4 @};
4188 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4189 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4191 @item -Wmissing-format-attribute
4192 @opindex Wmissing-format-attribute
4193 @opindex Wno-missing-format-attribute
4196 Warn about function pointers which might be candidates for @code{format}
4197 attributes. Note these are only possible candidates, not absolute ones.
4198 GCC will guess that function pointers with @code{format} attributes that
4199 are used in assignment, initialization, parameter passing or return
4200 statements should have a corresponding @code{format} attribute in the
4201 resulting type. I.e.@: the left-hand side of the assignment or
4202 initialization, the type of the parameter variable, or the return type
4203 of the containing function respectively should also have a @code{format}
4204 attribute to avoid the warning.
4206 GCC will also warn about function definitions which might be
4207 candidates for @code{format} attributes. Again, these are only
4208 possible candidates. GCC will guess that @code{format} attributes
4209 might be appropriate for any function that calls a function like
4210 @code{vprintf} or @code{vscanf}, but this might not always be the
4211 case, and some functions for which @code{format} attributes are
4212 appropriate may not be detected.
4214 @item -Wno-multichar
4215 @opindex Wno-multichar
4217 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4218 Usually they indicate a typo in the user's code, as they have
4219 implementation-defined values, and should not be used in portable code.
4221 @item -Wnormalized=<none|id|nfc|nfkc>
4222 @opindex Wnormalized=
4225 @cindex character set, input normalization
4226 In ISO C and ISO C++, two identifiers are different if they are
4227 different sequences of characters. However, sometimes when characters
4228 outside the basic ASCII character set are used, you can have two
4229 different character sequences that look the same. To avoid confusion,
4230 the ISO 10646 standard sets out some @dfn{normalization rules} which
4231 when applied ensure that two sequences that look the same are turned into
4232 the same sequence. GCC can warn you if you are using identifiers which
4233 have not been normalized; this option controls that warning.
4235 There are four levels of warning that GCC supports. The default is
4236 @option{-Wnormalized=nfc}, which warns about any identifier which is
4237 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4238 recommended form for most uses.
4240 Unfortunately, there are some characters which ISO C and ISO C++ allow
4241 in identifiers that when turned into NFC aren't allowable as
4242 identifiers. That is, there's no way to use these symbols in portable
4243 ISO C or C++ and have all your identifiers in NFC@.
4244 @option{-Wnormalized=id} suppresses the warning for these characters.
4245 It is hoped that future versions of the standards involved will correct
4246 this, which is why this option is not the default.
4248 You can switch the warning off for all characters by writing
4249 @option{-Wnormalized=none}. You would only want to do this if you
4250 were using some other normalization scheme (like ``D''), because
4251 otherwise you can easily create bugs that are literally impossible to see.
4253 Some characters in ISO 10646 have distinct meanings but look identical
4254 in some fonts or display methodologies, especially once formatting has
4255 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4256 LETTER N'', will display just like a regular @code{n} which has been
4257 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4258 normalization scheme to convert all these into a standard form as
4259 well, and GCC will warn if your code is not in NFKC if you use
4260 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4261 about every identifier that contains the letter O because it might be
4262 confused with the digit 0, and so is not the default, but may be
4263 useful as a local coding convention if the programming environment is
4264 unable to be fixed to display these characters distinctly.
4266 @item -Wno-deprecated
4267 @opindex Wno-deprecated
4268 @opindex Wdeprecated
4269 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4271 @item -Wno-deprecated-declarations
4272 @opindex Wno-deprecated-declarations
4273 @opindex Wdeprecated-declarations
4274 Do not warn about uses of functions (@pxref{Function Attributes}),
4275 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4276 Attributes}) marked as deprecated by using the @code{deprecated}
4280 @opindex Wno-overflow
4282 Do not warn about compile-time overflow in constant expressions.
4284 @item -Woverride-init @r{(C and Objective-C only)}
4285 @opindex Woverride-init
4286 @opindex Wno-override-init
4290 Warn if an initialized field without side effects is overridden when
4291 using designated initializers (@pxref{Designated Inits, , Designated
4294 This warning is included in @option{-Wextra}. To get other
4295 @option{-Wextra} warnings without this one, use @samp{-Wextra
4296 -Wno-override-init}.
4301 Warn if a structure is given the packed attribute, but the packed
4302 attribute has no effect on the layout or size of the structure.
4303 Such structures may be mis-aligned for little benefit. For
4304 instance, in this code, the variable @code{f.x} in @code{struct bar}
4305 will be misaligned even though @code{struct bar} does not itself
4306 have the packed attribute:
4313 @} __attribute__((packed));
4321 @item -Wpacked-bitfield-compat
4322 @opindex Wpacked-bitfield-compat
4323 @opindex Wno-packed-bitfield-compat
4324 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4325 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4326 the change can lead to differences in the structure layout. GCC
4327 informs you when the offset of such a field has changed in GCC 4.4.
4328 For example there is no longer a 4-bit padding between field @code{a}
4329 and @code{b} in this structure:
4336 @} __attribute__ ((packed));
4339 This warning is enabled by default. Use
4340 @option{-Wno-packed-bitfield-compat} to disable this warning.
4345 Warn if padding is included in a structure, either to align an element
4346 of the structure or to align the whole structure. Sometimes when this
4347 happens it is possible to rearrange the fields of the structure to
4348 reduce the padding and so make the structure smaller.
4350 @item -Wredundant-decls
4351 @opindex Wredundant-decls
4352 @opindex Wno-redundant-decls
4353 Warn if anything is declared more than once in the same scope, even in
4354 cases where multiple declaration is valid and changes nothing.
4356 @item -Wnested-externs @r{(C and Objective-C only)}
4357 @opindex Wnested-externs
4358 @opindex Wno-nested-externs
4359 Warn if an @code{extern} declaration is encountered within a function.
4364 Warn if a function can not be inlined and it was declared as inline.
4365 Even with this option, the compiler will not warn about failures to
4366 inline functions declared in system headers.
4368 The compiler uses a variety of heuristics to determine whether or not
4369 to inline a function. For example, the compiler takes into account
4370 the size of the function being inlined and the amount of inlining
4371 that has already been done in the current function. Therefore,
4372 seemingly insignificant changes in the source program can cause the
4373 warnings produced by @option{-Winline} to appear or disappear.
4375 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4376 @opindex Wno-invalid-offsetof
4377 @opindex Winvalid-offsetof
4378 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4379 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4380 to a non-POD type is undefined. In existing C++ implementations,
4381 however, @samp{offsetof} typically gives meaningful results even when
4382 applied to certain kinds of non-POD types. (Such as a simple
4383 @samp{struct} that fails to be a POD type only by virtue of having a
4384 constructor.) This flag is for users who are aware that they are
4385 writing nonportable code and who have deliberately chosen to ignore the
4388 The restrictions on @samp{offsetof} may be relaxed in a future version
4389 of the C++ standard.
4391 @item -Wno-int-to-pointer-cast
4392 @opindex Wno-int-to-pointer-cast
4393 @opindex Wint-to-pointer-cast
4394 Suppress warnings from casts to pointer type of an integer of a
4395 different size. In C++, casting to a pointer type of smaller size is
4396 an error. @option{Wint-to-pointer-cast} is enabled by default.
4399 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4400 @opindex Wno-pointer-to-int-cast
4401 @opindex Wpointer-to-int-cast
4402 Suppress warnings from casts from a pointer to an integer type of a
4406 @opindex Winvalid-pch
4407 @opindex Wno-invalid-pch
4408 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4409 the search path but can't be used.
4413 @opindex Wno-long-long
4414 Warn if @samp{long long} type is used. This is enabled by either
4415 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4416 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4418 @item -Wvariadic-macros
4419 @opindex Wvariadic-macros
4420 @opindex Wno-variadic-macros
4421 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4422 alternate syntax when in pedantic ISO C99 mode. This is default.
4423 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4428 Warn if variable length array is used in the code.
4429 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4430 the variable length array.
4432 @item -Wvolatile-register-var
4433 @opindex Wvolatile-register-var
4434 @opindex Wno-volatile-register-var
4435 Warn if a register variable is declared volatile. The volatile
4436 modifier does not inhibit all optimizations that may eliminate reads
4437 and/or writes to register variables. This warning is enabled by
4440 @item -Wdisabled-optimization
4441 @opindex Wdisabled-optimization
4442 @opindex Wno-disabled-optimization
4443 Warn if a requested optimization pass is disabled. This warning does
4444 not generally indicate that there is anything wrong with your code; it
4445 merely indicates that GCC's optimizers were unable to handle the code
4446 effectively. Often, the problem is that your code is too big or too
4447 complex; GCC will refuse to optimize programs when the optimization
4448 itself is likely to take inordinate amounts of time.
4450 @item -Wpointer-sign @r{(C and Objective-C only)}
4451 @opindex Wpointer-sign
4452 @opindex Wno-pointer-sign
4453 Warn for pointer argument passing or assignment with different signedness.
4454 This option is only supported for C and Objective-C@. It is implied by
4455 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4456 @option{-Wno-pointer-sign}.
4458 @item -Wstack-protector
4459 @opindex Wstack-protector
4460 @opindex Wno-stack-protector
4461 This option is only active when @option{-fstack-protector} is active. It
4462 warns about functions that will not be protected against stack smashing.
4465 @opindex Wno-mudflap
4466 Suppress warnings about constructs that cannot be instrumented by
4469 @item -Woverlength-strings
4470 @opindex Woverlength-strings
4471 @opindex Wno-overlength-strings
4472 Warn about string constants which are longer than the ``minimum
4473 maximum'' length specified in the C standard. Modern compilers
4474 generally allow string constants which are much longer than the
4475 standard's minimum limit, but very portable programs should avoid
4476 using longer strings.
4478 The limit applies @emph{after} string constant concatenation, and does
4479 not count the trailing NUL@. In C90, the limit was 509 characters; in
4480 C99, it was raised to 4095. C++98 does not specify a normative
4481 minimum maximum, so we do not diagnose overlength strings in C++@.
4483 This option is implied by @option{-pedantic}, and can be disabled with
4484 @option{-Wno-overlength-strings}.
4486 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4487 @opindex Wunsuffixed-float-constants
4489 GCC will issue a warning for any floating constant that does not have
4490 a suffix. When used together with @option{-Wsystem-headers} it will
4491 warn about such constants in system header files. This can be useful
4492 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4493 from the decimal floating-point extension to C99.
4496 @node Debugging Options
4497 @section Options for Debugging Your Program or GCC
4498 @cindex options, debugging
4499 @cindex debugging information options
4501 GCC has various special options that are used for debugging
4502 either your program or GCC:
4507 Produce debugging information in the operating system's native format
4508 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4511 On most systems that use stabs format, @option{-g} enables use of extra
4512 debugging information that only GDB can use; this extra information
4513 makes debugging work better in GDB but will probably make other debuggers
4515 refuse to read the program. If you want to control for certain whether
4516 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4517 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4519 GCC allows you to use @option{-g} with
4520 @option{-O}. The shortcuts taken by optimized code may occasionally
4521 produce surprising results: some variables you declared may not exist
4522 at all; flow of control may briefly move where you did not expect it;
4523 some statements may not be executed because they compute constant
4524 results or their values were already at hand; some statements may
4525 execute in different places because they were moved out of loops.
4527 Nevertheless it proves possible to debug optimized output. This makes
4528 it reasonable to use the optimizer for programs that might have bugs.
4530 The following options are useful when GCC is generated with the
4531 capability for more than one debugging format.
4535 Produce debugging information for use by GDB@. This means to use the
4536 most expressive format available (DWARF 2, stabs, or the native format
4537 if neither of those are supported), including GDB extensions if at all
4542 Produce debugging information in stabs format (if that is supported),
4543 without GDB extensions. This is the format used by DBX on most BSD
4544 systems. On MIPS, Alpha and System V Release 4 systems this option
4545 produces stabs debugging output which is not understood by DBX or SDB@.
4546 On System V Release 4 systems this option requires the GNU assembler.
4548 @item -feliminate-unused-debug-symbols
4549 @opindex feliminate-unused-debug-symbols
4550 Produce debugging information in stabs format (if that is supported),
4551 for only symbols that are actually used.
4553 @item -femit-class-debug-always
4554 Instead of emitting debugging information for a C++ class in only one
4555 object file, emit it in all object files using the class. This option
4556 should be used only with debuggers that are unable to handle the way GCC
4557 normally emits debugging information for classes because using this
4558 option will increase the size of debugging information by as much as a
4563 Produce debugging information in stabs format (if that is supported),
4564 using GNU extensions understood only by the GNU debugger (GDB)@. The
4565 use of these extensions is likely to make other debuggers crash or
4566 refuse to read the program.
4570 Produce debugging information in COFF format (if that is supported).
4571 This is the format used by SDB on most System V systems prior to
4576 Produce debugging information in XCOFF format (if that is supported).
4577 This is the format used by the DBX debugger on IBM RS/6000 systems.
4581 Produce debugging information in XCOFF format (if that is supported),
4582 using GNU extensions understood only by the GNU debugger (GDB)@. The
4583 use of these extensions is likely to make other debuggers crash or
4584 refuse to read the program, and may cause assemblers other than the GNU
4585 assembler (GAS) to fail with an error.
4587 @item -gdwarf-@var{version}
4588 @opindex gdwarf-@var{version}
4589 Produce debugging information in DWARF format (if that is
4590 supported). This is the format used by DBX on IRIX 6. The value
4591 of @var{version} may be either 2, 3 or 4; the default version is 2.
4593 Note that with DWARF version 2 some ports require, and will always
4594 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4596 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4597 for maximum benefit.
4599 @item -gstrict-dwarf
4600 @opindex gstrict-dwarf
4601 Disallow using extensions of later DWARF standard version than selected
4602 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4603 DWARF extensions from later standard versions is allowed.
4605 @item -gno-strict-dwarf
4606 @opindex gno-strict-dwarf
4607 Allow using extensions of later DWARF standard version than selected with
4608 @option{-gdwarf-@var{version}}.
4612 Produce debugging information in VMS debug format (if that is
4613 supported). This is the format used by DEBUG on VMS systems.
4616 @itemx -ggdb@var{level}
4617 @itemx -gstabs@var{level}
4618 @itemx -gcoff@var{level}
4619 @itemx -gxcoff@var{level}
4620 @itemx -gvms@var{level}
4621 Request debugging information and also use @var{level} to specify how
4622 much information. The default level is 2.
4624 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4627 Level 1 produces minimal information, enough for making backtraces in
4628 parts of the program that you don't plan to debug. This includes
4629 descriptions of functions and external variables, but no information
4630 about local variables and no line numbers.
4632 Level 3 includes extra information, such as all the macro definitions
4633 present in the program. Some debuggers support macro expansion when
4634 you use @option{-g3}.
4636 @option{-gdwarf-2} does not accept a concatenated debug level, because
4637 GCC used to support an option @option{-gdwarf} that meant to generate
4638 debug information in version 1 of the DWARF format (which is very
4639 different from version 2), and it would have been too confusing. That
4640 debug format is long obsolete, but the option cannot be changed now.
4641 Instead use an additional @option{-g@var{level}} option to change the
4642 debug level for DWARF.
4646 Turn off generation of debug info, if leaving out this option would have
4647 generated it, or turn it on at level 2 otherwise. The position of this
4648 argument in the command line does not matter, it takes effect after all
4649 other options are processed, and it does so only once, no matter how
4650 many times it is given. This is mainly intended to be used with
4651 @option{-fcompare-debug}.
4653 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4654 @opindex fdump-final-insns
4655 Dump the final internal representation (RTL) to @var{file}. If the
4656 optional argument is omitted (or if @var{file} is @code{.}), the name
4657 of the dump file will be determined by appending @code{.gkd} to the
4658 compilation output file name.
4660 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4661 @opindex fcompare-debug
4662 @opindex fno-compare-debug
4663 If no error occurs during compilation, run the compiler a second time,
4664 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4665 passed to the second compilation. Dump the final internal
4666 representation in both compilations, and print an error if they differ.
4668 If the equal sign is omitted, the default @option{-gtoggle} is used.
4670 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4671 and nonzero, implicitly enables @option{-fcompare-debug}. If
4672 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4673 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4676 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4677 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4678 of the final representation and the second compilation, preventing even
4679 @env{GCC_COMPARE_DEBUG} from taking effect.
4681 To verify full coverage during @option{-fcompare-debug} testing, set
4682 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4683 which GCC will reject as an invalid option in any actual compilation
4684 (rather than preprocessing, assembly or linking). To get just a
4685 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4686 not overridden} will do.
4688 @item -fcompare-debug-second
4689 @opindex fcompare-debug-second
4690 This option is implicitly passed to the compiler for the second
4691 compilation requested by @option{-fcompare-debug}, along with options to
4692 silence warnings, and omitting other options that would cause
4693 side-effect compiler outputs to files or to the standard output. Dump
4694 files and preserved temporary files are renamed so as to contain the
4695 @code{.gk} additional extension during the second compilation, to avoid
4696 overwriting those generated by the first.
4698 When this option is passed to the compiler driver, it causes the
4699 @emph{first} compilation to be skipped, which makes it useful for little
4700 other than debugging the compiler proper.
4702 @item -feliminate-dwarf2-dups
4703 @opindex feliminate-dwarf2-dups
4704 Compress DWARF2 debugging information by eliminating duplicated
4705 information about each symbol. This option only makes sense when
4706 generating DWARF2 debugging information with @option{-gdwarf-2}.
4708 @item -femit-struct-debug-baseonly
4709 Emit debug information for struct-like types
4710 only when the base name of the compilation source file
4711 matches the base name of file in which the struct was defined.
4713 This option substantially reduces the size of debugging information,
4714 but at significant potential loss in type information to the debugger.
4715 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4716 See @option{-femit-struct-debug-detailed} for more detailed control.
4718 This option works only with DWARF 2.
4720 @item -femit-struct-debug-reduced
4721 Emit debug information for struct-like types
4722 only when the base name of the compilation source file
4723 matches the base name of file in which the type was defined,
4724 unless the struct is a template or defined in a system header.
4726 This option significantly reduces the size of debugging information,
4727 with some potential loss in type information to the debugger.
4728 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4729 See @option{-femit-struct-debug-detailed} for more detailed control.
4731 This option works only with DWARF 2.
4733 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4734 Specify the struct-like types
4735 for which the compiler will generate debug information.
4736 The intent is to reduce duplicate struct debug information
4737 between different object files within the same program.
4739 This option is a detailed version of
4740 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4741 which will serve for most needs.
4743 A specification has the syntax@*
4744 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4746 The optional first word limits the specification to
4747 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4748 A struct type is used directly when it is the type of a variable, member.
4749 Indirect uses arise through pointers to structs.
4750 That is, when use of an incomplete struct would be legal, the use is indirect.
4752 @samp{struct one direct; struct two * indirect;}.
4754 The optional second word limits the specification to
4755 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4756 Generic structs are a bit complicated to explain.
4757 For C++, these are non-explicit specializations of template classes,
4758 or non-template classes within the above.
4759 Other programming languages have generics,
4760 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4762 The third word specifies the source files for those
4763 structs for which the compiler will emit debug information.
4764 The values @samp{none} and @samp{any} have the normal meaning.
4765 The value @samp{base} means that
4766 the base of name of the file in which the type declaration appears
4767 must match the base of the name of the main compilation file.
4768 In practice, this means that
4769 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4770 but types declared in other header will not.
4771 The value @samp{sys} means those types satisfying @samp{base}
4772 or declared in system or compiler headers.
4774 You may need to experiment to determine the best settings for your application.
4776 The default is @samp{-femit-struct-debug-detailed=all}.
4778 This option works only with DWARF 2.
4780 @item -fenable-icf-debug
4781 @opindex fenable-icf-debug
4782 Generate additional debug information to support identical code folding (ICF).
4783 This option only works with DWARF version 2 or higher.
4785 @item -fno-merge-debug-strings
4786 @opindex fmerge-debug-strings
4787 @opindex fno-merge-debug-strings
4788 Direct the linker to not merge together strings in the debugging
4789 information which are identical in different object files. Merging is
4790 not supported by all assemblers or linkers. Merging decreases the size
4791 of the debug information in the output file at the cost of increasing
4792 link processing time. Merging is enabled by default.
4794 @item -fdebug-prefix-map=@var{old}=@var{new}
4795 @opindex fdebug-prefix-map
4796 When compiling files in directory @file{@var{old}}, record debugging
4797 information describing them as in @file{@var{new}} instead.
4799 @item -fno-dwarf2-cfi-asm
4800 @opindex fdwarf2-cfi-asm
4801 @opindex fno-dwarf2-cfi-asm
4802 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4803 instead of using GAS @code{.cfi_*} directives.
4805 @cindex @command{prof}
4808 Generate extra code to write profile information suitable for the
4809 analysis program @command{prof}. You must use this option when compiling
4810 the source files you want data about, and you must also use it when
4813 @cindex @command{gprof}
4816 Generate extra code to write profile information suitable for the
4817 analysis program @command{gprof}. You must use this option when compiling
4818 the source files you want data about, and you must also use it when
4823 Makes the compiler print out each function name as it is compiled, and
4824 print some statistics about each pass when it finishes.
4827 @opindex ftime-report
4828 Makes the compiler print some statistics about the time consumed by each
4829 pass when it finishes.
4832 @opindex fmem-report
4833 Makes the compiler print some statistics about permanent memory
4834 allocation when it finishes.
4836 @item -fpre-ipa-mem-report
4837 @opindex fpre-ipa-mem-report
4838 @item -fpost-ipa-mem-report
4839 @opindex fpost-ipa-mem-report
4840 Makes the compiler print some statistics about permanent memory
4841 allocation before or after interprocedural optimization.
4844 @opindex fstack-usage
4845 Makes the compiler output stack usage information for the program, on a
4846 per-function basis. The filename for the dump is made by appending
4847 @file{.su} to the AUXNAME. AUXNAME is generated from the name of
4848 the output file, if explicitly specified and it is not an executable,
4849 otherwise it is the basename of the source file. An entry is made up
4854 The name of the function.
4858 One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
4861 The qualifier @code{static} means that the function manipulates the stack
4862 statically: a fixed number of bytes are allocated for the frame on function
4863 entry and released on function exit; no stack adjustments are otherwise made
4864 in the function. The second field is this fixed number of bytes.
4866 The qualifier @code{dynamic} means that the function manipulates the stack
4867 dynamically: in addition to the static allocation described above, stack
4868 adjustments are made in the body of the function, for example to push/pop
4869 arguments around function calls. If the qualifier @code{bounded} is also
4870 present, the amount of these adjustments is bounded at compile-time and
4871 the second field is an upper bound of the total amount of stack used by
4872 the function. If it is not present, the amount of these adjustments is
4873 not bounded at compile-time and the second field only represents the
4876 @item -fprofile-arcs
4877 @opindex fprofile-arcs
4878 Add code so that program flow @dfn{arcs} are instrumented. During
4879 execution the program records how many times each branch and call is
4880 executed and how many times it is taken or returns. When the compiled
4881 program exits it saves this data to a file called
4882 @file{@var{auxname}.gcda} for each source file. The data may be used for
4883 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4884 test coverage analysis (@option{-ftest-coverage}). Each object file's
4885 @var{auxname} is generated from the name of the output file, if
4886 explicitly specified and it is not the final executable, otherwise it is
4887 the basename of the source file. In both cases any suffix is removed
4888 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4889 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4890 @xref{Cross-profiling}.
4892 @cindex @command{gcov}
4896 This option is used to compile and link code instrumented for coverage
4897 analysis. The option is a synonym for @option{-fprofile-arcs}
4898 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4899 linking). See the documentation for those options for more details.
4904 Compile the source files with @option{-fprofile-arcs} plus optimization
4905 and code generation options. For test coverage analysis, use the
4906 additional @option{-ftest-coverage} option. You do not need to profile
4907 every source file in a program.
4910 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4911 (the latter implies the former).
4914 Run the program on a representative workload to generate the arc profile
4915 information. This may be repeated any number of times. You can run
4916 concurrent instances of your program, and provided that the file system
4917 supports locking, the data files will be correctly updated. Also
4918 @code{fork} calls are detected and correctly handled (double counting
4922 For profile-directed optimizations, compile the source files again with
4923 the same optimization and code generation options plus
4924 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4925 Control Optimization}).
4928 For test coverage analysis, use @command{gcov} to produce human readable
4929 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4930 @command{gcov} documentation for further information.
4934 With @option{-fprofile-arcs}, for each function of your program GCC
4935 creates a program flow graph, then finds a spanning tree for the graph.
4936 Only arcs that are not on the spanning tree have to be instrumented: the
4937 compiler adds code to count the number of times that these arcs are
4938 executed. When an arc is the only exit or only entrance to a block, the
4939 instrumentation code can be added to the block; otherwise, a new basic
4940 block must be created to hold the instrumentation code.
4943 @item -ftest-coverage
4944 @opindex ftest-coverage
4945 Produce a notes file that the @command{gcov} code-coverage utility
4946 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4947 show program coverage. Each source file's note file is called
4948 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4949 above for a description of @var{auxname} and instructions on how to
4950 generate test coverage data. Coverage data will match the source files
4951 more closely, if you do not optimize.
4953 @item -fdbg-cnt-list
4954 @opindex fdbg-cnt-list
4955 Print the name and the counter upperbound for all debug counters.
4957 @item -fdbg-cnt=@var{counter-value-list}
4959 Set the internal debug counter upperbound. @var{counter-value-list}
4960 is a comma-separated list of @var{name}:@var{value} pairs
4961 which sets the upperbound of each debug counter @var{name} to @var{value}.
4962 All debug counters have the initial upperbound of @var{UINT_MAX},
4963 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4964 e.g. With -fdbg-cnt=dce:10,tail_call:0
4965 dbg_cnt(dce) will return true only for first 10 invocations
4966 and dbg_cnt(tail_call) will return false always.
4968 @item -d@var{letters}
4969 @itemx -fdump-rtl-@var{pass}
4971 Says to make debugging dumps during compilation at times specified by
4972 @var{letters}. This is used for debugging the RTL-based passes of the
4973 compiler. The file names for most of the dumps are made by appending
4974 a pass number and a word to the @var{dumpname}, and the files are
4975 created in the directory of the output file. Note that the pass
4976 number is computed statically as passes get registered into the pass
4977 manager. Thus the numbering is not related to the dynamic order of
4978 execution of passes. In particular, a pass installed by a plugin
4979 could have a number over 200 even if it executed quite early.
4980 @var{dumpname} is generated from the name of the output file, if
4981 explicitly specified and it is not an executable, otherwise it is the
4982 basename of the source file. These switches may have different effects
4983 when @option{-E} is used for preprocessing.
4985 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4986 @option{-d} option @var{letters}. Here are the possible
4987 letters for use in @var{pass} and @var{letters}, and their meanings:
4991 @item -fdump-rtl-alignments
4992 @opindex fdump-rtl-alignments
4993 Dump after branch alignments have been computed.
4995 @item -fdump-rtl-asmcons
4996 @opindex fdump-rtl-asmcons
4997 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4999 @item -fdump-rtl-auto_inc_dec
5000 @opindex fdump-rtl-auto_inc_dec
5001 Dump after auto-inc-dec discovery. This pass is only run on
5002 architectures that have auto inc or auto dec instructions.
5004 @item -fdump-rtl-barriers
5005 @opindex fdump-rtl-barriers
5006 Dump after cleaning up the barrier instructions.
5008 @item -fdump-rtl-bbpart
5009 @opindex fdump-rtl-bbpart
5010 Dump after partitioning hot and cold basic blocks.
5012 @item -fdump-rtl-bbro
5013 @opindex fdump-rtl-bbro
5014 Dump after block reordering.
5016 @item -fdump-rtl-btl1
5017 @itemx -fdump-rtl-btl2
5018 @opindex fdump-rtl-btl2
5019 @opindex fdump-rtl-btl2
5020 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
5021 after the two branch
5022 target load optimization passes.
5024 @item -fdump-rtl-bypass
5025 @opindex fdump-rtl-bypass
5026 Dump after jump bypassing and control flow optimizations.
5028 @item -fdump-rtl-combine
5029 @opindex fdump-rtl-combine
5030 Dump after the RTL instruction combination pass.
5032 @item -fdump-rtl-compgotos
5033 @opindex fdump-rtl-compgotos
5034 Dump after duplicating the computed gotos.
5036 @item -fdump-rtl-ce1
5037 @itemx -fdump-rtl-ce2
5038 @itemx -fdump-rtl-ce3
5039 @opindex fdump-rtl-ce1
5040 @opindex fdump-rtl-ce2
5041 @opindex fdump-rtl-ce3
5042 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
5043 @option{-fdump-rtl-ce3} enable dumping after the three
5044 if conversion passes.
5046 @itemx -fdump-rtl-cprop_hardreg
5047 @opindex fdump-rtl-cprop_hardreg
5048 Dump after hard register copy propagation.
5050 @itemx -fdump-rtl-csa
5051 @opindex fdump-rtl-csa
5052 Dump after combining stack adjustments.
5054 @item -fdump-rtl-cse1
5055 @itemx -fdump-rtl-cse2
5056 @opindex fdump-rtl-cse1
5057 @opindex fdump-rtl-cse2
5058 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
5059 the two common sub-expression elimination passes.
5061 @itemx -fdump-rtl-dce
5062 @opindex fdump-rtl-dce
5063 Dump after the standalone dead code elimination passes.
5065 @itemx -fdump-rtl-dbr
5066 @opindex fdump-rtl-dbr
5067 Dump after delayed branch scheduling.
5069 @item -fdump-rtl-dce1
5070 @itemx -fdump-rtl-dce2
5071 @opindex fdump-rtl-dce1
5072 @opindex fdump-rtl-dce2
5073 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
5074 the two dead store elimination passes.
5077 @opindex fdump-rtl-eh
5078 Dump after finalization of EH handling code.
5080 @item -fdump-rtl-eh_ranges
5081 @opindex fdump-rtl-eh_ranges
5082 Dump after conversion of EH handling range regions.
5084 @item -fdump-rtl-expand
5085 @opindex fdump-rtl-expand
5086 Dump after RTL generation.
5088 @item -fdump-rtl-fwprop1
5089 @itemx -fdump-rtl-fwprop2
5090 @opindex fdump-rtl-fwprop1
5091 @opindex fdump-rtl-fwprop2
5092 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
5093 dumping after the two forward propagation passes.
5095 @item -fdump-rtl-gcse1
5096 @itemx -fdump-rtl-gcse2
5097 @opindex fdump-rtl-gcse1
5098 @opindex fdump-rtl-gcse2
5099 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
5100 after global common subexpression elimination.
5102 @item -fdump-rtl-init-regs
5103 @opindex fdump-rtl-init-regs
5104 Dump after the initialization of the registers.
5106 @item -fdump-rtl-initvals
5107 @opindex fdump-rtl-initvals
5108 Dump after the computation of the initial value sets.
5110 @itemx -fdump-rtl-into_cfglayout
5111 @opindex fdump-rtl-into_cfglayout
5112 Dump after converting to cfglayout mode.
5114 @item -fdump-rtl-ira
5115 @opindex fdump-rtl-ira
5116 Dump after iterated register allocation.
5118 @item -fdump-rtl-jump
5119 @opindex fdump-rtl-jump
5120 Dump after the second jump optimization.
5122 @item -fdump-rtl-loop2
5123 @opindex fdump-rtl-loop2
5124 @option{-fdump-rtl-loop2} enables dumping after the rtl
5125 loop optimization passes.
5127 @item -fdump-rtl-mach
5128 @opindex fdump-rtl-mach
5129 Dump after performing the machine dependent reorganization pass, if that
5132 @item -fdump-rtl-mode_sw
5133 @opindex fdump-rtl-mode_sw
5134 Dump after removing redundant mode switches.
5136 @item -fdump-rtl-rnreg
5137 @opindex fdump-rtl-rnreg
5138 Dump after register renumbering.
5140 @itemx -fdump-rtl-outof_cfglayout
5141 @opindex fdump-rtl-outof_cfglayout
5142 Dump after converting from cfglayout mode.
5144 @item -fdump-rtl-peephole2
5145 @opindex fdump-rtl-peephole2
5146 Dump after the peephole pass.
5148 @item -fdump-rtl-postreload
5149 @opindex fdump-rtl-postreload
5150 Dump after post-reload optimizations.
5152 @itemx -fdump-rtl-pro_and_epilogue
5153 @opindex fdump-rtl-pro_and_epilogue
5154 Dump after generating the function pro and epilogues.
5156 @item -fdump-rtl-regmove
5157 @opindex fdump-rtl-regmove
5158 Dump after the register move pass.
5160 @item -fdump-rtl-sched1
5161 @itemx -fdump-rtl-sched2
5162 @opindex fdump-rtl-sched1
5163 @opindex fdump-rtl-sched2
5164 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
5165 after the basic block scheduling passes.
5167 @item -fdump-rtl-see
5168 @opindex fdump-rtl-see
5169 Dump after sign extension elimination.
5171 @item -fdump-rtl-seqabstr
5172 @opindex fdump-rtl-seqabstr
5173 Dump after common sequence discovery.
5175 @item -fdump-rtl-shorten
5176 @opindex fdump-rtl-shorten
5177 Dump after shortening branches.
5179 @item -fdump-rtl-sibling
5180 @opindex fdump-rtl-sibling
5181 Dump after sibling call optimizations.
5183 @item -fdump-rtl-split1
5184 @itemx -fdump-rtl-split2
5185 @itemx -fdump-rtl-split3
5186 @itemx -fdump-rtl-split4
5187 @itemx -fdump-rtl-split5
5188 @opindex fdump-rtl-split1
5189 @opindex fdump-rtl-split2
5190 @opindex fdump-rtl-split3
5191 @opindex fdump-rtl-split4
5192 @opindex fdump-rtl-split5
5193 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5194 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5195 @option{-fdump-rtl-split5} enable dumping after five rounds of
5196 instruction splitting.
5198 @item -fdump-rtl-sms
5199 @opindex fdump-rtl-sms
5200 Dump after modulo scheduling. This pass is only run on some
5203 @item -fdump-rtl-stack
5204 @opindex fdump-rtl-stack
5205 Dump after conversion from GCC's "flat register file" registers to the
5206 x87's stack-like registers. This pass is only run on x86 variants.
5208 @item -fdump-rtl-subreg1
5209 @itemx -fdump-rtl-subreg2
5210 @opindex fdump-rtl-subreg1
5211 @opindex fdump-rtl-subreg2
5212 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5213 the two subreg expansion passes.
5215 @item -fdump-rtl-unshare
5216 @opindex fdump-rtl-unshare
5217 Dump after all rtl has been unshared.
5219 @item -fdump-rtl-vartrack
5220 @opindex fdump-rtl-vartrack
5221 Dump after variable tracking.
5223 @item -fdump-rtl-vregs
5224 @opindex fdump-rtl-vregs
5225 Dump after converting virtual registers to hard registers.
5227 @item -fdump-rtl-web
5228 @opindex fdump-rtl-web
5229 Dump after live range splitting.
5231 @item -fdump-rtl-regclass
5232 @itemx -fdump-rtl-subregs_of_mode_init
5233 @itemx -fdump-rtl-subregs_of_mode_finish
5234 @itemx -fdump-rtl-dfinit
5235 @itemx -fdump-rtl-dfinish
5236 @opindex fdump-rtl-regclass
5237 @opindex fdump-rtl-subregs_of_mode_init
5238 @opindex fdump-rtl-subregs_of_mode_finish
5239 @opindex fdump-rtl-dfinit
5240 @opindex fdump-rtl-dfinish
5241 These dumps are defined but always produce empty files.
5243 @item -fdump-rtl-all
5244 @opindex fdump-rtl-all
5245 Produce all the dumps listed above.
5249 Annotate the assembler output with miscellaneous debugging information.
5253 Dump all macro definitions, at the end of preprocessing, in addition to
5258 Produce a core dump whenever an error occurs.
5262 Print statistics on memory usage, at the end of the run, to
5267 Annotate the assembler output with a comment indicating which
5268 pattern and alternative was used. The length of each instruction is
5273 Dump the RTL in the assembler output as a comment before each instruction.
5274 Also turns on @option{-dp} annotation.
5278 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5279 dump a representation of the control flow graph suitable for viewing with VCG
5280 to @file{@var{file}.@var{pass}.vcg}.
5284 Just generate RTL for a function instead of compiling it. Usually used
5285 with @option{-fdump-rtl-expand}.
5289 @opindex fdump-noaddr
5290 When doing debugging dumps, suppress address output. This makes it more
5291 feasible to use diff on debugging dumps for compiler invocations with
5292 different compiler binaries and/or different
5293 text / bss / data / heap / stack / dso start locations.
5295 @item -fdump-unnumbered
5296 @opindex fdump-unnumbered
5297 When doing debugging dumps, suppress instruction numbers and address output.
5298 This makes it more feasible to use diff on debugging dumps for compiler
5299 invocations with different options, in particular with and without
5302 @item -fdump-unnumbered-links
5303 @opindex fdump-unnumbered-links
5304 When doing debugging dumps (see @option{-d} option above), suppress
5305 instruction numbers for the links to the previous and next instructions
5308 @item -fdump-translation-unit @r{(C++ only)}
5309 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5310 @opindex fdump-translation-unit
5311 Dump a representation of the tree structure for the entire translation
5312 unit to a file. The file name is made by appending @file{.tu} to the
5313 source file name, and the file is created in the same directory as the
5314 output file. If the @samp{-@var{options}} form is used, @var{options}
5315 controls the details of the dump as described for the
5316 @option{-fdump-tree} options.
5318 @item -fdump-class-hierarchy @r{(C++ only)}
5319 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5320 @opindex fdump-class-hierarchy
5321 Dump a representation of each class's hierarchy and virtual function
5322 table layout to a file. The file name is made by appending
5323 @file{.class} to the source file name, and the file is created in the
5324 same directory as the output file. If the @samp{-@var{options}} form
5325 is used, @var{options} controls the details of the dump as described
5326 for the @option{-fdump-tree} options.
5328 @item -fdump-ipa-@var{switch}
5330 Control the dumping at various stages of inter-procedural analysis
5331 language tree to a file. The file name is generated by appending a
5332 switch specific suffix to the source file name, and the file is created
5333 in the same directory as the output file. The following dumps are
5338 Enables all inter-procedural analysis dumps.
5341 Dumps information about call-graph optimization, unused function removal,
5342 and inlining decisions.
5345 Dump after function inlining.
5349 @item -fdump-statistics-@var{option}
5350 @opindex fdump-statistics
5351 Enable and control dumping of pass statistics in a separate file. The
5352 file name is generated by appending a suffix ending in
5353 @samp{.statistics} to the source file name, and the file is created in
5354 the same directory as the output file. If the @samp{-@var{option}}
5355 form is used, @samp{-stats} will cause counters to be summed over the
5356 whole compilation unit while @samp{-details} will dump every event as
5357 the passes generate them. The default with no option is to sum
5358 counters for each function compiled.
5360 @item -fdump-tree-@var{switch}
5361 @itemx -fdump-tree-@var{switch}-@var{options}
5363 Control the dumping at various stages of processing the intermediate
5364 language tree to a file. The file name is generated by appending a
5365 switch specific suffix to the source file name, and the file is
5366 created in the same directory as the output file. If the
5367 @samp{-@var{options}} form is used, @var{options} is a list of
5368 @samp{-} separated options that control the details of the dump. Not
5369 all options are applicable to all dumps, those which are not
5370 meaningful will be ignored. The following options are available
5374 Print the address of each node. Usually this is not meaningful as it
5375 changes according to the environment and source file. Its primary use
5376 is for tying up a dump file with a debug environment.
5378 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5379 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5380 use working backward from mangled names in the assembly file.
5382 Inhibit dumping of members of a scope or body of a function merely
5383 because that scope has been reached. Only dump such items when they
5384 are directly reachable by some other path. When dumping pretty-printed
5385 trees, this option inhibits dumping the bodies of control structures.
5387 Print a raw representation of the tree. By default, trees are
5388 pretty-printed into a C-like representation.
5390 Enable more detailed dumps (not honored by every dump option).
5392 Enable dumping various statistics about the pass (not honored by every dump
5395 Enable showing basic block boundaries (disabled in raw dumps).
5397 Enable showing virtual operands for every statement.
5399 Enable showing line numbers for statements.
5401 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5403 Enable showing the tree dump for each statement.
5405 Enable showing the EH region number holding each statement.
5407 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5408 and @option{lineno}.
5411 The following tree dumps are possible:
5415 @opindex fdump-tree-original
5416 Dump before any tree based optimization, to @file{@var{file}.original}.
5419 @opindex fdump-tree-optimized
5420 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5423 @opindex fdump-tree-gimple
5424 Dump each function before and after the gimplification pass to a file. The
5425 file name is made by appending @file{.gimple} to the source file name.
5428 @opindex fdump-tree-cfg
5429 Dump the control flow graph of each function to a file. The file name is
5430 made by appending @file{.cfg} to the source file name.
5433 @opindex fdump-tree-vcg
5434 Dump the control flow graph of each function to a file in VCG format. The
5435 file name is made by appending @file{.vcg} to the source file name. Note
5436 that if the file contains more than one function, the generated file cannot
5437 be used directly by VCG@. You will need to cut and paste each function's
5438 graph into its own separate file first.
5441 @opindex fdump-tree-ch
5442 Dump each function after copying loop headers. The file name is made by
5443 appending @file{.ch} to the source file name.
5446 @opindex fdump-tree-ssa
5447 Dump SSA related information to a file. The file name is made by appending
5448 @file{.ssa} to the source file name.
5451 @opindex fdump-tree-alias
5452 Dump aliasing information for each function. The file name is made by
5453 appending @file{.alias} to the source file name.
5456 @opindex fdump-tree-ccp
5457 Dump each function after CCP@. The file name is made by appending
5458 @file{.ccp} to the source file name.
5461 @opindex fdump-tree-storeccp
5462 Dump each function after STORE-CCP@. The file name is made by appending
5463 @file{.storeccp} to the source file name.
5466 @opindex fdump-tree-pre
5467 Dump trees after partial redundancy elimination. The file name is made
5468 by appending @file{.pre} to the source file name.
5471 @opindex fdump-tree-fre
5472 Dump trees after full redundancy elimination. The file name is made
5473 by appending @file{.fre} to the source file name.
5476 @opindex fdump-tree-copyprop
5477 Dump trees after copy propagation. The file name is made
5478 by appending @file{.copyprop} to the source file name.
5480 @item store_copyprop
5481 @opindex fdump-tree-store_copyprop
5482 Dump trees after store copy-propagation. The file name is made
5483 by appending @file{.store_copyprop} to the source file name.
5486 @opindex fdump-tree-dce
5487 Dump each function after dead code elimination. The file name is made by
5488 appending @file{.dce} to the source file name.
5491 @opindex fdump-tree-mudflap
5492 Dump each function after adding mudflap instrumentation. The file name is
5493 made by appending @file{.mudflap} to the source file name.
5496 @opindex fdump-tree-sra
5497 Dump each function after performing scalar replacement of aggregates. The
5498 file name is made by appending @file{.sra} to the source file name.
5501 @opindex fdump-tree-sink
5502 Dump each function after performing code sinking. The file name is made
5503 by appending @file{.sink} to the source file name.
5506 @opindex fdump-tree-dom
5507 Dump each function after applying dominator tree optimizations. The file
5508 name is made by appending @file{.dom} to the source file name.
5511 @opindex fdump-tree-dse
5512 Dump each function after applying dead store elimination. The file
5513 name is made by appending @file{.dse} to the source file name.
5516 @opindex fdump-tree-phiopt
5517 Dump each function after optimizing PHI nodes into straightline code. The file
5518 name is made by appending @file{.phiopt} to the source file name.
5521 @opindex fdump-tree-forwprop
5522 Dump each function after forward propagating single use variables. The file
5523 name is made by appending @file{.forwprop} to the source file name.
5526 @opindex fdump-tree-copyrename
5527 Dump each function after applying the copy rename optimization. The file
5528 name is made by appending @file{.copyrename} to the source file name.
5531 @opindex fdump-tree-nrv
5532 Dump each function after applying the named return value optimization on
5533 generic trees. The file name is made by appending @file{.nrv} to the source
5537 @opindex fdump-tree-vect
5538 Dump each function after applying vectorization of loops. The file name is
5539 made by appending @file{.vect} to the source file name.
5542 @opindex fdump-tree-slp
5543 Dump each function after applying vectorization of basic blocks. The file name
5544 is made by appending @file{.slp} to the source file name.
5547 @opindex fdump-tree-vrp
5548 Dump each function after Value Range Propagation (VRP). The file name
5549 is made by appending @file{.vrp} to the source file name.
5552 @opindex fdump-tree-all
5553 Enable all the available tree dumps with the flags provided in this option.
5556 @item -ftree-vectorizer-verbose=@var{n}
5557 @opindex ftree-vectorizer-verbose
5558 This option controls the amount of debugging output the vectorizer prints.
5559 This information is written to standard error, unless
5560 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5561 in which case it is output to the usual dump listing file, @file{.vect}.
5562 For @var{n}=0 no diagnostic information is reported.
5563 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5564 and the total number of loops that got vectorized.
5565 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5566 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5567 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5568 level that @option{-fdump-tree-vect-stats} uses.
5569 Higher verbosity levels mean either more information dumped for each
5570 reported loop, or same amount of information reported for more loops:
5571 if @var{n}=3, vectorizer cost model information is reported.
5572 If @var{n}=4, alignment related information is added to the reports.
5573 If @var{n}=5, data-references related information (e.g.@: memory dependences,
5574 memory access-patterns) is added to the reports.
5575 If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5576 that did not pass the first analysis phase (i.e., may not be countable, or
5577 may have complicated control-flow).
5578 If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5579 If @var{n}=8, SLP related information is added to the reports.
5580 For @var{n}=9, all the information the vectorizer generates during its
5581 analysis and transformation is reported. This is the same verbosity level
5582 that @option{-fdump-tree-vect-details} uses.
5584 @item -frandom-seed=@var{string}
5585 @opindex frandom-seed
5586 This option provides a seed that GCC uses when it would otherwise use
5587 random numbers. It is used to generate certain symbol names
5588 that have to be different in every compiled file. It is also used to
5589 place unique stamps in coverage data files and the object files that
5590 produce them. You can use the @option{-frandom-seed} option to produce
5591 reproducibly identical object files.
5593 The @var{string} should be different for every file you compile.
5595 @item -fsched-verbose=@var{n}
5596 @opindex fsched-verbose
5597 On targets that use instruction scheduling, this option controls the
5598 amount of debugging output the scheduler prints. This information is
5599 written to standard error, unless @option{-fdump-rtl-sched1} or
5600 @option{-fdump-rtl-sched2} is specified, in which case it is output
5601 to the usual dump listing file, @file{.sched1} or @file{.sched2}
5602 respectively. However for @var{n} greater than nine, the output is
5603 always printed to standard error.
5605 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5606 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5607 For @var{n} greater than one, it also output basic block probabilities,
5608 detailed ready list information and unit/insn info. For @var{n} greater
5609 than two, it includes RTL at abort point, control-flow and regions info.
5610 And for @var{n} over four, @option{-fsched-verbose} also includes
5614 @itemx -save-temps=cwd
5616 Store the usual ``temporary'' intermediate files permanently; place them
5617 in the current directory and name them based on the source file. Thus,
5618 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5619 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5620 preprocessed @file{foo.i} output file even though the compiler now
5621 normally uses an integrated preprocessor.
5623 When used in combination with the @option{-x} command line option,
5624 @option{-save-temps} is sensible enough to avoid over writing an
5625 input source file with the same extension as an intermediate file.
5626 The corresponding intermediate file may be obtained by renaming the
5627 source file before using @option{-save-temps}.
5629 If you invoke GCC in parallel, compiling several different source
5630 files that share a common base name in different subdirectories or the
5631 same source file compiled for multiple output destinations, it is
5632 likely that the different parallel compilers will interfere with each
5633 other, and overwrite the temporary files. For instance:
5636 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5637 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5640 may result in @file{foo.i} and @file{foo.o} being written to
5641 simultaneously by both compilers.
5643 @item -save-temps=obj
5644 @opindex save-temps=obj
5645 Store the usual ``temporary'' intermediate files permanently. If the
5646 @option{-o} option is used, the temporary files are based on the
5647 object file. If the @option{-o} option is not used, the
5648 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5653 gcc -save-temps=obj -c foo.c
5654 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5655 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5658 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5659 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5660 @file{dir2/yfoobar.o}.
5662 @item -time@r{[}=@var{file}@r{]}
5664 Report the CPU time taken by each subprocess in the compilation
5665 sequence. For C source files, this is the compiler proper and assembler
5666 (plus the linker if linking is done).
5668 Without the specification of an output file, the output looks like this:
5675 The first number on each line is the ``user time'', that is time spent
5676 executing the program itself. The second number is ``system time'',
5677 time spent executing operating system routines on behalf of the program.
5678 Both numbers are in seconds.
5680 With the specification of an output file, the output is appended to the
5681 named file, and it looks like this:
5684 0.12 0.01 cc1 @var{options}
5685 0.00 0.01 as @var{options}
5688 The ``user time'' and the ``system time'' are moved before the program
5689 name, and the options passed to the program are displayed, so that one
5690 can later tell what file was being compiled, and with which options.
5692 @item -fvar-tracking
5693 @opindex fvar-tracking
5694 Run variable tracking pass. It computes where variables are stored at each
5695 position in code. Better debugging information is then generated
5696 (if the debugging information format supports this information).
5698 It is enabled by default when compiling with optimization (@option{-Os},
5699 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5700 the debug info format supports it.
5702 @item -fvar-tracking-assignments
5703 @opindex fvar-tracking-assignments
5704 @opindex fno-var-tracking-assignments
5705 Annotate assignments to user variables early in the compilation and
5706 attempt to carry the annotations over throughout the compilation all the
5707 way to the end, in an attempt to improve debug information while
5708 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5710 It can be enabled even if var-tracking is disabled, in which case
5711 annotations will be created and maintained, but discarded at the end.
5713 @item -fvar-tracking-assignments-toggle
5714 @opindex fvar-tracking-assignments-toggle
5715 @opindex fno-var-tracking-assignments-toggle
5716 Toggle @option{-fvar-tracking-assignments}, in the same way that
5717 @option{-gtoggle} toggles @option{-g}.
5719 @item -print-file-name=@var{library}
5720 @opindex print-file-name
5721 Print the full absolute name of the library file @var{library} that
5722 would be used when linking---and don't do anything else. With this
5723 option, GCC does not compile or link anything; it just prints the
5726 @item -print-multi-directory
5727 @opindex print-multi-directory
5728 Print the directory name corresponding to the multilib selected by any
5729 other switches present in the command line. This directory is supposed
5730 to exist in @env{GCC_EXEC_PREFIX}.
5732 @item -print-multi-lib
5733 @opindex print-multi-lib
5734 Print the mapping from multilib directory names to compiler switches
5735 that enable them. The directory name is separated from the switches by
5736 @samp{;}, and each switch starts with an @samp{@@} instead of the
5737 @samp{-}, without spaces between multiple switches. This is supposed to
5738 ease shell-processing.
5740 @item -print-multi-os-directory
5741 @opindex print-multi-os-directory
5742 Print the path to OS libraries for the selected
5743 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5744 present in the @file{lib} subdirectory and no multilibs are used, this is
5745 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5746 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5747 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5748 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5750 @item -print-prog-name=@var{program}
5751 @opindex print-prog-name
5752 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5754 @item -print-libgcc-file-name
5755 @opindex print-libgcc-file-name
5756 Same as @option{-print-file-name=libgcc.a}.
5758 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5759 but you do want to link with @file{libgcc.a}. You can do
5762 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5765 @item -print-search-dirs
5766 @opindex print-search-dirs
5767 Print the name of the configured installation directory and a list of
5768 program and library directories @command{gcc} will search---and don't do anything else.
5770 This is useful when @command{gcc} prints the error message
5771 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5772 To resolve this you either need to put @file{cpp0} and the other compiler
5773 components where @command{gcc} expects to find them, or you can set the environment
5774 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5775 Don't forget the trailing @samp{/}.
5776 @xref{Environment Variables}.
5778 @item -print-sysroot
5779 @opindex print-sysroot
5780 Print the target sysroot directory that will be used during
5781 compilation. This is the target sysroot specified either at configure
5782 time or using the @option{--sysroot} option, possibly with an extra
5783 suffix that depends on compilation options. If no target sysroot is
5784 specified, the option prints nothing.
5786 @item -print-sysroot-headers-suffix
5787 @opindex print-sysroot-headers-suffix
5788 Print the suffix added to the target sysroot when searching for
5789 headers, or give an error if the compiler is not configured with such
5790 a suffix---and don't do anything else.
5793 @opindex dumpmachine
5794 Print the compiler's target machine (for example,
5795 @samp{i686-pc-linux-gnu})---and don't do anything else.
5798 @opindex dumpversion
5799 Print the compiler version (for example, @samp{3.0})---and don't do
5804 Print the compiler's built-in specs---and don't do anything else. (This
5805 is used when GCC itself is being built.) @xref{Spec Files}.
5807 @item -feliminate-unused-debug-types
5808 @opindex feliminate-unused-debug-types
5809 Normally, when producing DWARF2 output, GCC will emit debugging
5810 information for all types declared in a compilation
5811 unit, regardless of whether or not they are actually used
5812 in that compilation unit. Sometimes this is useful, such as
5813 if, in the debugger, you want to cast a value to a type that is
5814 not actually used in your program (but is declared). More often,
5815 however, this results in a significant amount of wasted space.
5816 With this option, GCC will avoid producing debug symbol output
5817 for types that are nowhere used in the source file being compiled.
5820 @node Optimize Options
5821 @section Options That Control Optimization
5822 @cindex optimize options
5823 @cindex options, optimization
5825 These options control various sorts of optimizations.
5827 Without any optimization option, the compiler's goal is to reduce the
5828 cost of compilation and to make debugging produce the expected
5829 results. Statements are independent: if you stop the program with a
5830 breakpoint between statements, you can then assign a new value to any
5831 variable or change the program counter to any other statement in the
5832 function and get exactly the results you would expect from the source
5835 Turning on optimization flags makes the compiler attempt to improve
5836 the performance and/or code size at the expense of compilation time
5837 and possibly the ability to debug the program.
5839 The compiler performs optimization based on the knowledge it has of the
5840 program. Compiling multiple files at once to a single output file mode allows
5841 the compiler to use information gained from all of the files when compiling
5844 Not all optimizations are controlled directly by a flag. Only
5845 optimizations that have a flag are listed in this section.
5847 Most optimizations are only enabled if an @option{-O} level is set on
5848 the command line. Otherwise they are disabled, even if individual
5849 optimization flags are specified.
5851 Depending on the target and how GCC was configured, a slightly different
5852 set of optimizations may be enabled at each @option{-O} level than
5853 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5854 to find out the exact set of optimizations that are enabled at each level.
5855 @xref{Overall Options}, for examples.
5862 Optimize. Optimizing compilation takes somewhat more time, and a lot
5863 more memory for a large function.
5865 With @option{-O}, the compiler tries to reduce code size and execution
5866 time, without performing any optimizations that take a great deal of
5869 @option{-O} turns on the following optimization flags:
5872 -fcprop-registers @gol
5875 -fdelayed-branch @gol
5877 -fguess-branch-probability @gol
5878 -fif-conversion2 @gol
5879 -fif-conversion @gol
5880 -fipa-pure-const @gol
5882 -fipa-reference @gol
5884 -fsplit-wide-types @gol
5886 -ftree-builtin-call-dce @gol
5889 -ftree-copyrename @gol
5891 -ftree-dominator-opts @gol
5893 -ftree-forwprop @gol
5901 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5902 where doing so does not interfere with debugging.
5906 Optimize even more. GCC performs nearly all supported optimizations
5907 that do not involve a space-speed tradeoff.
5908 As compared to @option{-O}, this option increases both compilation time
5909 and the performance of the generated code.
5911 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5912 also turns on the following optimization flags:
5913 @gccoptlist{-fthread-jumps @gol
5914 -falign-functions -falign-jumps @gol
5915 -falign-loops -falign-labels @gol
5918 -fcse-follow-jumps -fcse-skip-blocks @gol
5919 -fdelete-null-pointer-checks @gol
5920 -fexpensive-optimizations @gol
5921 -fgcse -fgcse-lm @gol
5922 -finline-small-functions @gol
5923 -findirect-inlining @gol
5925 -foptimize-sibling-calls @gol
5926 -fpartial-inlining @gol
5929 -freorder-blocks -freorder-functions @gol
5930 -frerun-cse-after-loop @gol
5931 -fsched-interblock -fsched-spec @gol
5932 -fschedule-insns -fschedule-insns2 @gol
5933 -fstrict-aliasing -fstrict-overflow @gol
5934 -ftree-switch-conversion @gol
5938 Please note the warning under @option{-fgcse} about
5939 invoking @option{-O2} on programs that use computed gotos.
5943 Optimize yet more. @option{-O3} turns on all optimizations specified
5944 by @option{-O2} and also turns on the @option{-finline-functions},
5945 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5946 @option{-fgcse-after-reload}, @option{-ftree-vectorize} and
5947 @option{-fipa-cp-clone} options.
5951 Reduce compilation time and make debugging produce the expected
5952 results. This is the default.
5956 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5957 do not typically increase code size. It also performs further
5958 optimizations designed to reduce code size.
5960 @option{-Os} disables the following optimization flags:
5961 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5962 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5963 -fprefetch-loop-arrays -ftree-vect-loop-version}
5967 Disregard strict standards compliance. @option{-Ofast} enables all
5968 @option{-O3} optimizations. It also enables optimizations that are not
5969 valid for all standard compliant programs.
5970 It turns on @option{-ffast-math}.
5972 If you use multiple @option{-O} options, with or without level numbers,
5973 the last such option is the one that is effective.
5976 Options of the form @option{-f@var{flag}} specify machine-independent
5977 flags. Most flags have both positive and negative forms; the negative
5978 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5979 below, only one of the forms is listed---the one you typically will
5980 use. You can figure out the other form by either removing @samp{no-}
5983 The following options control specific optimizations. They are either
5984 activated by @option{-O} options or are related to ones that are. You
5985 can use the following flags in the rare cases when ``fine-tuning'' of
5986 optimizations to be performed is desired.
5989 @item -fno-default-inline
5990 @opindex fno-default-inline
5991 Do not make member functions inline by default merely because they are
5992 defined inside the class scope (C++ only). Otherwise, when you specify
5993 @w{@option{-O}}, member functions defined inside class scope are compiled
5994 inline by default; i.e., you don't need to add @samp{inline} in front of
5995 the member function name.
5997 @item -fno-defer-pop
5998 @opindex fno-defer-pop
5999 Always pop the arguments to each function call as soon as that function
6000 returns. For machines which must pop arguments after a function call,
6001 the compiler normally lets arguments accumulate on the stack for several
6002 function calls and pops them all at once.
6004 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6006 @item -fforward-propagate
6007 @opindex fforward-propagate
6008 Perform a forward propagation pass on RTL@. The pass tries to combine two
6009 instructions and checks if the result can be simplified. If loop unrolling
6010 is active, two passes are performed and the second is scheduled after
6013 This option is enabled by default at optimization levels @option{-O},
6014 @option{-O2}, @option{-O3}, @option{-Os}.
6016 @item -ffp-contract=@var{style}
6017 @opindex ffp-contract
6018 @option{-ffp-contract=off} disables floating-point expression contraction.
6019 @option{-ffp-contract=fast} enables floating-point expression contraction
6020 such as forming of fused multiply-add operations if the target has
6021 native support for them.
6022 @option{-ffp-contract=on} enables floating-point expression contraction
6023 if allowed by the language standard. This is currently not implemented
6024 and treated equal to @option{-ffp-contract=off}.
6026 The default is @option{-ffp-contract=fast}.
6028 @item -fomit-frame-pointer
6029 @opindex fomit-frame-pointer
6030 Don't keep the frame pointer in a register for functions that
6031 don't need one. This avoids the instructions to save, set up and
6032 restore frame pointers; it also makes an extra register available
6033 in many functions. @strong{It also makes debugging impossible on
6036 On some machines, such as the VAX, this flag has no effect, because
6037 the standard calling sequence automatically handles the frame pointer
6038 and nothing is saved by pretending it doesn't exist. The
6039 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
6040 whether a target machine supports this flag. @xref{Registers,,Register
6041 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
6043 Starting with GCC version 4.6, the default setting (when not optimizing for
6044 size) for 32-bit Linux x86 and 32-bit Darwin x86 targets has been changed to
6045 @option{-fomit-frame-pointer}. The default can be reverted to
6046 @option{-fno-omit-frame-pointer} by configuring GCC with the
6047 @option{--enable-frame-pointer} configure option.
6049 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6051 @item -foptimize-sibling-calls
6052 @opindex foptimize-sibling-calls
6053 Optimize sibling and tail recursive calls.
6055 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6059 Don't pay attention to the @code{inline} keyword. Normally this option
6060 is used to keep the compiler from expanding any functions inline.
6061 Note that if you are not optimizing, no functions can be expanded inline.
6063 @item -finline-small-functions
6064 @opindex finline-small-functions
6065 Integrate functions into their callers when their body is smaller than expected
6066 function call code (so overall size of program gets smaller). The compiler
6067 heuristically decides which functions are simple enough to be worth integrating
6070 Enabled at level @option{-O2}.
6072 @item -findirect-inlining
6073 @opindex findirect-inlining
6074 Inline also indirect calls that are discovered to be known at compile
6075 time thanks to previous inlining. This option has any effect only
6076 when inlining itself is turned on by the @option{-finline-functions}
6077 or @option{-finline-small-functions} options.
6079 Enabled at level @option{-O2}.
6081 @item -finline-functions
6082 @opindex finline-functions
6083 Integrate all simple functions into their callers. The compiler
6084 heuristically decides which functions are simple enough to be worth
6085 integrating in this way.
6087 If all calls to a given function are integrated, and the function is
6088 declared @code{static}, then the function is normally not output as
6089 assembler code in its own right.
6091 Enabled at level @option{-O3}.
6093 @item -finline-functions-called-once
6094 @opindex finline-functions-called-once
6095 Consider all @code{static} functions called once for inlining into their
6096 caller even if they are not marked @code{inline}. If a call to a given
6097 function is integrated, then the function is not output as assembler code
6100 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
6102 @item -fearly-inlining
6103 @opindex fearly-inlining
6104 Inline functions marked by @code{always_inline} and functions whose body seems
6105 smaller than the function call overhead early before doing
6106 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
6107 makes profiling significantly cheaper and usually inlining faster on programs
6108 having large chains of nested wrapper functions.
6114 Perform interprocedural scalar replacement of aggregates, removal of
6115 unused parameters and replacement of parameters passed by reference
6116 by parameters passed by value.
6118 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6120 @item -finline-limit=@var{n}
6121 @opindex finline-limit
6122 By default, GCC limits the size of functions that can be inlined. This flag
6123 allows coarse control of this limit. @var{n} is the size of functions that
6124 can be inlined in number of pseudo instructions.
6126 Inlining is actually controlled by a number of parameters, which may be
6127 specified individually by using @option{--param @var{name}=@var{value}}.
6128 The @option{-finline-limit=@var{n}} option sets some of these parameters
6132 @item max-inline-insns-single
6133 is set to @var{n}/2.
6134 @item max-inline-insns-auto
6135 is set to @var{n}/2.
6138 See below for a documentation of the individual
6139 parameters controlling inlining and for the defaults of these parameters.
6141 @emph{Note:} there may be no value to @option{-finline-limit} that results
6142 in default behavior.
6144 @emph{Note:} pseudo instruction represents, in this particular context, an
6145 abstract measurement of function's size. In no way does it represent a count
6146 of assembly instructions and as such its exact meaning might change from one
6147 release to an another.
6149 @item -fkeep-inline-functions
6150 @opindex fkeep-inline-functions
6151 In C, emit @code{static} functions that are declared @code{inline}
6152 into the object file, even if the function has been inlined into all
6153 of its callers. This switch does not affect functions using the
6154 @code{extern inline} extension in GNU C90@. In C++, emit any and all
6155 inline functions into the object file.
6157 @item -fkeep-static-consts
6158 @opindex fkeep-static-consts
6159 Emit variables declared @code{static const} when optimization isn't turned
6160 on, even if the variables aren't referenced.
6162 GCC enables this option by default. If you want to force the compiler to
6163 check if the variable was referenced, regardless of whether or not
6164 optimization is turned on, use the @option{-fno-keep-static-consts} option.
6166 @item -fmerge-constants
6167 @opindex fmerge-constants
6168 Attempt to merge identical constants (string constants and floating point
6169 constants) across compilation units.
6171 This option is the default for optimized compilation if the assembler and
6172 linker support it. Use @option{-fno-merge-constants} to inhibit this
6175 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6177 @item -fmerge-all-constants
6178 @opindex fmerge-all-constants
6179 Attempt to merge identical constants and identical variables.
6181 This option implies @option{-fmerge-constants}. In addition to
6182 @option{-fmerge-constants} this considers e.g.@: even constant initialized
6183 arrays or initialized constant variables with integral or floating point
6184 types. Languages like C or C++ require each variable, including multiple
6185 instances of the same variable in recursive calls, to have distinct locations,
6186 so using this option will result in non-conforming
6189 @item -fmodulo-sched
6190 @opindex fmodulo-sched
6191 Perform swing modulo scheduling immediately before the first scheduling
6192 pass. This pass looks at innermost loops and reorders their
6193 instructions by overlapping different iterations.
6195 @item -fmodulo-sched-allow-regmoves
6196 @opindex fmodulo-sched-allow-regmoves
6197 Perform more aggressive SMS based modulo scheduling with register moves
6198 allowed. By setting this flag certain anti-dependences edges will be
6199 deleted which will trigger the generation of reg-moves based on the
6200 life-range analysis. This option is effective only with
6201 @option{-fmodulo-sched} enabled.
6203 @item -fno-branch-count-reg
6204 @opindex fno-branch-count-reg
6205 Do not use ``decrement and branch'' instructions on a count register,
6206 but instead generate a sequence of instructions that decrement a
6207 register, compare it against zero, then branch based upon the result.
6208 This option is only meaningful on architectures that support such
6209 instructions, which include x86, PowerPC, IA-64 and S/390.
6211 The default is @option{-fbranch-count-reg}.
6213 @item -fno-function-cse
6214 @opindex fno-function-cse
6215 Do not put function addresses in registers; make each instruction that
6216 calls a constant function contain the function's address explicitly.
6218 This option results in less efficient code, but some strange hacks
6219 that alter the assembler output may be confused by the optimizations
6220 performed when this option is not used.
6222 The default is @option{-ffunction-cse}
6224 @item -fno-zero-initialized-in-bss
6225 @opindex fno-zero-initialized-in-bss
6226 If the target supports a BSS section, GCC by default puts variables that
6227 are initialized to zero into BSS@. This can save space in the resulting
6230 This option turns off this behavior because some programs explicitly
6231 rely on variables going to the data section. E.g., so that the
6232 resulting executable can find the beginning of that section and/or make
6233 assumptions based on that.
6235 The default is @option{-fzero-initialized-in-bss}.
6237 @item -fmudflap -fmudflapth -fmudflapir
6241 @cindex bounds checking
6243 For front-ends that support it (C and C++), instrument all risky
6244 pointer/array dereferencing operations, some standard library
6245 string/heap functions, and some other associated constructs with
6246 range/validity tests. Modules so instrumented should be immune to
6247 buffer overflows, invalid heap use, and some other classes of C/C++
6248 programming errors. The instrumentation relies on a separate runtime
6249 library (@file{libmudflap}), which will be linked into a program if
6250 @option{-fmudflap} is given at link time. Run-time behavior of the
6251 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6252 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6255 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6256 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6257 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6258 instrumentation should ignore pointer reads. This produces less
6259 instrumentation (and therefore faster execution) and still provides
6260 some protection against outright memory corrupting writes, but allows
6261 erroneously read data to propagate within a program.
6263 @item -fthread-jumps
6264 @opindex fthread-jumps
6265 Perform optimizations where we check to see if a jump branches to a
6266 location where another comparison subsumed by the first is found. If
6267 so, the first branch is redirected to either the destination of the
6268 second branch or a point immediately following it, depending on whether
6269 the condition is known to be true or false.
6271 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6273 @item -fsplit-wide-types
6274 @opindex fsplit-wide-types
6275 When using a type that occupies multiple registers, such as @code{long
6276 long} on a 32-bit system, split the registers apart and allocate them
6277 independently. This normally generates better code for those types,
6278 but may make debugging more difficult.
6280 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6283 @item -fcse-follow-jumps
6284 @opindex fcse-follow-jumps
6285 In common subexpression elimination (CSE), scan through jump instructions
6286 when the target of the jump is not reached by any other path. For
6287 example, when CSE encounters an @code{if} statement with an
6288 @code{else} clause, CSE will follow the jump when the condition
6291 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6293 @item -fcse-skip-blocks
6294 @opindex fcse-skip-blocks
6295 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6296 follow jumps which conditionally skip over blocks. When CSE
6297 encounters a simple @code{if} statement with no else clause,
6298 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6299 body of the @code{if}.
6301 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6303 @item -frerun-cse-after-loop
6304 @opindex frerun-cse-after-loop
6305 Re-run common subexpression elimination after loop optimizations has been
6308 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6312 Perform a global common subexpression elimination pass.
6313 This pass also performs global constant and copy propagation.
6315 @emph{Note:} When compiling a program using computed gotos, a GCC
6316 extension, you may get better runtime performance if you disable
6317 the global common subexpression elimination pass by adding
6318 @option{-fno-gcse} to the command line.
6320 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6324 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6325 attempt to move loads which are only killed by stores into themselves. This
6326 allows a loop containing a load/store sequence to be changed to a load outside
6327 the loop, and a copy/store within the loop.
6329 Enabled by default when gcse is enabled.
6333 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6334 global common subexpression elimination. This pass will attempt to move
6335 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6336 loops containing a load/store sequence can be changed to a load before
6337 the loop and a store after the loop.
6339 Not enabled at any optimization level.
6343 When @option{-fgcse-las} is enabled, the global common subexpression
6344 elimination pass eliminates redundant loads that come after stores to the
6345 same memory location (both partial and full redundancies).
6347 Not enabled at any optimization level.
6349 @item -fgcse-after-reload
6350 @opindex fgcse-after-reload
6351 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6352 pass is performed after reload. The purpose of this pass is to cleanup
6355 @item -funsafe-loop-optimizations
6356 @opindex funsafe-loop-optimizations
6357 If given, the loop optimizer will assume that loop indices do not
6358 overflow, and that the loops with nontrivial exit condition are not
6359 infinite. This enables a wider range of loop optimizations even if
6360 the loop optimizer itself cannot prove that these assumptions are valid.
6361 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6362 if it finds this kind of loop.
6364 @item -fcrossjumping
6365 @opindex fcrossjumping
6366 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6367 resulting code may or may not perform better than without cross-jumping.
6369 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6371 @item -fauto-inc-dec
6372 @opindex fauto-inc-dec
6373 Combine increments or decrements of addresses with memory accesses.
6374 This pass is always skipped on architectures that do not have
6375 instructions to support this. Enabled by default at @option{-O} and
6376 higher on architectures that support this.
6380 Perform dead code elimination (DCE) on RTL@.
6381 Enabled by default at @option{-O} and higher.
6385 Perform dead store elimination (DSE) on RTL@.
6386 Enabled by default at @option{-O} and higher.
6388 @item -fif-conversion
6389 @opindex fif-conversion
6390 Attempt to transform conditional jumps into branch-less equivalents. This
6391 include use of conditional moves, min, max, set flags and abs instructions, and
6392 some tricks doable by standard arithmetics. The use of conditional execution
6393 on chips where it is available is controlled by @code{if-conversion2}.
6395 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6397 @item -fif-conversion2
6398 @opindex fif-conversion2
6399 Use conditional execution (where available) to transform conditional jumps into
6400 branch-less equivalents.
6402 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6404 @item -fdelete-null-pointer-checks
6405 @opindex fdelete-null-pointer-checks
6406 Assume that programs cannot safely dereference null pointers, and that
6407 no code or data element resides there. This enables simple constant
6408 folding optimizations at all optimization levels. In addition, other
6409 optimization passes in GCC use this flag to control global dataflow
6410 analyses that eliminate useless checks for null pointers; these assume
6411 that if a pointer is checked after it has already been dereferenced,
6414 Note however that in some environments this assumption is not true.
6415 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6416 for programs which depend on that behavior.
6418 Some targets, especially embedded ones, disable this option at all levels.
6419 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6420 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6421 are enabled independently at different optimization levels.
6423 @item -fexpensive-optimizations
6424 @opindex fexpensive-optimizations
6425 Perform a number of minor optimizations that are relatively expensive.
6427 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6429 @item -foptimize-register-move
6431 @opindex foptimize-register-move
6433 Attempt to reassign register numbers in move instructions and as
6434 operands of other simple instructions in order to maximize the amount of
6435 register tying. This is especially helpful on machines with two-operand
6438 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6441 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6443 @item -fira-algorithm=@var{algorithm}
6444 Use specified coloring algorithm for the integrated register
6445 allocator. The @var{algorithm} argument should be @code{priority} or
6446 @code{CB}. The first algorithm specifies Chow's priority coloring,
6447 the second one specifies Chaitin-Briggs coloring. The second
6448 algorithm can be unimplemented for some architectures. If it is
6449 implemented, it is the default because Chaitin-Briggs coloring as a
6450 rule generates a better code.
6452 @item -fira-region=@var{region}
6453 Use specified regions for the integrated register allocator. The
6454 @var{region} argument should be one of @code{all}, @code{mixed}, or
6455 @code{one}. The first value means using all loops as register
6456 allocation regions, the second value which is the default means using
6457 all loops except for loops with small register pressure as the
6458 regions, and third one means using all function as a single region.
6459 The first value can give best result for machines with small size and
6460 irregular register set, the third one results in faster and generates
6461 decent code and the smallest size code, and the default value usually
6462 give the best results in most cases and for most architectures.
6464 @item -fira-loop-pressure
6465 @opindex fira-loop-pressure
6466 Use IRA to evaluate register pressure in loops for decision to move
6467 loop invariants. Usage of this option usually results in generation
6468 of faster and smaller code on machines with big register files (>= 32
6469 registers) but it can slow compiler down.
6471 This option is enabled at level @option{-O3} for some targets.
6473 @item -fno-ira-share-save-slots
6474 @opindex fno-ira-share-save-slots
6475 Switch off sharing stack slots used for saving call used hard
6476 registers living through a call. Each hard register will get a
6477 separate stack slot and as a result function stack frame will be
6480 @item -fno-ira-share-spill-slots
6481 @opindex fno-ira-share-spill-slots
6482 Switch off sharing stack slots allocated for pseudo-registers. Each
6483 pseudo-register which did not get a hard register will get a separate
6484 stack slot and as a result function stack frame will be bigger.
6486 @item -fira-verbose=@var{n}
6487 @opindex fira-verbose
6488 Set up how verbose dump file for the integrated register allocator
6489 will be. Default value is 5. If the value is greater or equal to 10,
6490 the dump file will be stderr as if the value were @var{n} minus 10.
6492 @item -fdelayed-branch
6493 @opindex fdelayed-branch
6494 If supported for the target machine, attempt to reorder instructions
6495 to exploit instruction slots available after delayed branch
6498 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6500 @item -fschedule-insns
6501 @opindex fschedule-insns
6502 If supported for the target machine, attempt to reorder instructions to
6503 eliminate execution stalls due to required data being unavailable. This
6504 helps machines that have slow floating point or memory load instructions
6505 by allowing other instructions to be issued until the result of the load
6506 or floating point instruction is required.
6508 Enabled at levels @option{-O2}, @option{-O3}.
6510 @item -fschedule-insns2
6511 @opindex fschedule-insns2
6512 Similar to @option{-fschedule-insns}, but requests an additional pass of
6513 instruction scheduling after register allocation has been done. This is
6514 especially useful on machines with a relatively small number of
6515 registers and where memory load instructions take more than one cycle.
6517 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6519 @item -fno-sched-interblock
6520 @opindex fno-sched-interblock
6521 Don't schedule instructions across basic blocks. This is normally
6522 enabled by default when scheduling before register allocation, i.e.@:
6523 with @option{-fschedule-insns} or at @option{-O2} or higher.
6525 @item -fno-sched-spec
6526 @opindex fno-sched-spec
6527 Don't allow speculative motion of non-load instructions. This is normally
6528 enabled by default when scheduling before register allocation, i.e.@:
6529 with @option{-fschedule-insns} or at @option{-O2} or higher.
6531 @item -fsched-pressure
6532 @opindex fsched-pressure
6533 Enable register pressure sensitive insn scheduling before the register
6534 allocation. This only makes sense when scheduling before register
6535 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6536 @option{-O2} or higher. Usage of this option can improve the
6537 generated code and decrease its size by preventing register pressure
6538 increase above the number of available hard registers and as a
6539 consequence register spills in the register allocation.
6541 @item -fsched-spec-load
6542 @opindex fsched-spec-load
6543 Allow speculative motion of some load instructions. This only makes
6544 sense when scheduling before register allocation, i.e.@: with
6545 @option{-fschedule-insns} or at @option{-O2} or higher.
6547 @item -fsched-spec-load-dangerous
6548 @opindex fsched-spec-load-dangerous
6549 Allow speculative motion of more load instructions. This only makes
6550 sense when scheduling before register allocation, i.e.@: with
6551 @option{-fschedule-insns} or at @option{-O2} or higher.
6553 @item -fsched-stalled-insns
6554 @itemx -fsched-stalled-insns=@var{n}
6555 @opindex fsched-stalled-insns
6556 Define how many insns (if any) can be moved prematurely from the queue
6557 of stalled insns into the ready list, during the second scheduling pass.
6558 @option{-fno-sched-stalled-insns} means that no insns will be moved
6559 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6560 on how many queued insns can be moved prematurely.
6561 @option{-fsched-stalled-insns} without a value is equivalent to
6562 @option{-fsched-stalled-insns=1}.
6564 @item -fsched-stalled-insns-dep
6565 @itemx -fsched-stalled-insns-dep=@var{n}
6566 @opindex fsched-stalled-insns-dep
6567 Define how many insn groups (cycles) will be examined for a dependency
6568 on a stalled insn that is candidate for premature removal from the queue
6569 of stalled insns. This has an effect only during the second scheduling pass,
6570 and only if @option{-fsched-stalled-insns} is used.
6571 @option{-fno-sched-stalled-insns-dep} is equivalent to
6572 @option{-fsched-stalled-insns-dep=0}.
6573 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6574 @option{-fsched-stalled-insns-dep=1}.
6576 @item -fsched2-use-superblocks
6577 @opindex fsched2-use-superblocks
6578 When scheduling after register allocation, do use superblock scheduling
6579 algorithm. Superblock scheduling allows motion across basic block boundaries
6580 resulting on faster schedules. This option is experimental, as not all machine
6581 descriptions used by GCC model the CPU closely enough to avoid unreliable
6582 results from the algorithm.
6584 This only makes sense when scheduling after register allocation, i.e.@: with
6585 @option{-fschedule-insns2} or at @option{-O2} or higher.
6587 @item -fsched-group-heuristic
6588 @opindex fsched-group-heuristic
6589 Enable the group heuristic in the scheduler. This heuristic favors
6590 the instruction that belongs to a schedule group. This is enabled
6591 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6592 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6594 @item -fsched-critical-path-heuristic
6595 @opindex fsched-critical-path-heuristic
6596 Enable the critical-path heuristic in the scheduler. This heuristic favors
6597 instructions on the critical path. This is enabled by default when
6598 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6599 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6601 @item -fsched-spec-insn-heuristic
6602 @opindex fsched-spec-insn-heuristic
6603 Enable the speculative instruction heuristic in the scheduler. This
6604 heuristic favors speculative instructions with greater dependency weakness.
6605 This is enabled by default when scheduling is enabled, i.e.@:
6606 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6607 or at @option{-O2} or higher.
6609 @item -fsched-rank-heuristic
6610 @opindex fsched-rank-heuristic
6611 Enable the rank heuristic in the scheduler. This heuristic favors
6612 the instruction belonging to a basic block with greater size or frequency.
6613 This is enabled by default when scheduling is enabled, i.e.@:
6614 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6615 at @option{-O2} or higher.
6617 @item -fsched-last-insn-heuristic
6618 @opindex fsched-last-insn-heuristic
6619 Enable the last-instruction heuristic in the scheduler. This heuristic
6620 favors the instruction that is less dependent on the last instruction
6621 scheduled. This is enabled by default when scheduling is enabled,
6622 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6623 at @option{-O2} or higher.
6625 @item -fsched-dep-count-heuristic
6626 @opindex fsched-dep-count-heuristic
6627 Enable the dependent-count heuristic in the scheduler. This heuristic
6628 favors the instruction that has more instructions depending on it.
6629 This is enabled by default when scheduling is enabled, i.e.@:
6630 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6631 at @option{-O2} or higher.
6633 @item -freschedule-modulo-scheduled-loops
6634 @opindex freschedule-modulo-scheduled-loops
6635 The modulo scheduling comes before the traditional scheduling, if a loop
6636 was modulo scheduled we may want to prevent the later scheduling passes
6637 from changing its schedule, we use this option to control that.
6639 @item -fselective-scheduling
6640 @opindex fselective-scheduling
6641 Schedule instructions using selective scheduling algorithm. Selective
6642 scheduling runs instead of the first scheduler pass.
6644 @item -fselective-scheduling2
6645 @opindex fselective-scheduling2
6646 Schedule instructions using selective scheduling algorithm. Selective
6647 scheduling runs instead of the second scheduler pass.
6649 @item -fsel-sched-pipelining
6650 @opindex fsel-sched-pipelining
6651 Enable software pipelining of innermost loops during selective scheduling.
6652 This option has no effect until one of @option{-fselective-scheduling} or
6653 @option{-fselective-scheduling2} is turned on.
6655 @item -fsel-sched-pipelining-outer-loops
6656 @opindex fsel-sched-pipelining-outer-loops
6657 When pipelining loops during selective scheduling, also pipeline outer loops.
6658 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6660 @item -fcaller-saves
6661 @opindex fcaller-saves
6662 Enable values to be allocated in registers that will be clobbered by
6663 function calls, by emitting extra instructions to save and restore the
6664 registers around such calls. Such allocation is done only when it
6665 seems to result in better code than would otherwise be produced.
6667 This option is always enabled by default on certain machines, usually
6668 those which have no call-preserved registers to use instead.
6670 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6672 @item -fcombine-stack-adjustments
6673 @opindex fcombine-stack-adjustments
6674 Tracks stack adjustments (pushes and pops) and stack memory references
6675 and then tries to find ways to combine them.
6677 Enabled by default at @option{-O1} and higher.
6679 @item -fconserve-stack
6680 @opindex fconserve-stack
6681 Attempt to minimize stack usage. The compiler will attempt to use less
6682 stack space, even if that makes the program slower. This option
6683 implies setting the @option{large-stack-frame} parameter to 100
6684 and the @option{large-stack-frame-growth} parameter to 400.
6686 @item -ftree-reassoc
6687 @opindex ftree-reassoc
6688 Perform reassociation on trees. This flag is enabled by default
6689 at @option{-O} and higher.
6693 Perform partial redundancy elimination (PRE) on trees. This flag is
6694 enabled by default at @option{-O2} and @option{-O3}.
6696 @item -ftree-forwprop
6697 @opindex ftree-forwprop
6698 Perform forward propagation on trees. This flag is enabled by default
6699 at @option{-O} and higher.
6703 Perform full redundancy elimination (FRE) on trees. The difference
6704 between FRE and PRE is that FRE only considers expressions
6705 that are computed on all paths leading to the redundant computation.
6706 This analysis is faster than PRE, though it exposes fewer redundancies.
6707 This flag is enabled by default at @option{-O} and higher.
6709 @item -ftree-phiprop
6710 @opindex ftree-phiprop
6711 Perform hoisting of loads from conditional pointers on trees. This
6712 pass is enabled by default at @option{-O} and higher.
6714 @item -ftree-copy-prop
6715 @opindex ftree-copy-prop
6716 Perform copy propagation on trees. This pass eliminates unnecessary
6717 copy operations. This flag is enabled by default at @option{-O} and
6720 @item -fipa-pure-const
6721 @opindex fipa-pure-const
6722 Discover which functions are pure or constant.
6723 Enabled by default at @option{-O} and higher.
6725 @item -fipa-reference
6726 @opindex fipa-reference
6727 Discover which static variables do not escape cannot escape the
6729 Enabled by default at @option{-O} and higher.
6731 @item -fipa-struct-reorg
6732 @opindex fipa-struct-reorg
6733 Perform structure reorganization optimization, that change C-like structures
6734 layout in order to better utilize spatial locality. This transformation is
6735 affective for programs containing arrays of structures. Available in two
6736 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6737 or static (which uses built-in heuristics). It works only in whole program
6738 mode, so it requires @option{-fwhole-program} to be
6739 enabled. Structures considered @samp{cold} by this transformation are not
6740 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6742 With this flag, the program debug info reflects a new structure layout.
6746 Perform interprocedural pointer analysis and interprocedural modification
6747 and reference analysis. This option can cause excessive memory and
6748 compile-time usage on large compilation units. It is not enabled by
6749 default at any optimization level.
6752 @opindex fipa-profile
6753 Perform interprocedural profile propagation. The functions called only from
6754 cold functions are marked as cold. Also functions executed once (such as
6755 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
6756 functions and loop less parts of functions executed once are then optimized for
6758 Enabled by default at @option{-O} and higher.
6762 Perform interprocedural constant propagation.
6763 This optimization analyzes the program to determine when values passed
6764 to functions are constants and then optimizes accordingly.
6765 This optimization can substantially increase performance
6766 if the application has constants passed to functions.
6767 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6769 @item -fipa-cp-clone
6770 @opindex fipa-cp-clone
6771 Perform function cloning to make interprocedural constant propagation stronger.
6772 When enabled, interprocedural constant propagation will perform function cloning
6773 when externally visible function can be called with constant arguments.
6774 Because this optimization can create multiple copies of functions,
6775 it may significantly increase code size
6776 (see @option{--param ipcp-unit-growth=@var{value}}).
6777 This flag is enabled by default at @option{-O3}.
6779 @item -fipa-matrix-reorg
6780 @opindex fipa-matrix-reorg
6781 Perform matrix flattening and transposing.
6782 Matrix flattening tries to replace an @math{m}-dimensional matrix
6783 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6784 This reduces the level of indirection needed for accessing the elements
6785 of the matrix. The second optimization is matrix transposing that
6786 attempts to change the order of the matrix's dimensions in order to
6787 improve cache locality.
6788 Both optimizations need the @option{-fwhole-program} flag.
6789 Transposing is enabled only if profiling information is available.
6793 Perform forward store motion on trees. This flag is
6794 enabled by default at @option{-O} and higher.
6796 @item -ftree-bit-ccp
6797 @opindex ftree-bit-ccp
6798 Perform sparse conditional bit constant propagation on trees and propagate
6799 pointer alignment information.
6800 This pass only operates on local scalar variables and is enabled by default
6801 at @option{-O} and higher. It requires that @option{-ftree-ccp} is enabled.
6805 Perform sparse conditional constant propagation (CCP) on trees. This
6806 pass only operates on local scalar variables and is enabled by default
6807 at @option{-O} and higher.
6809 @item -ftree-switch-conversion
6810 Perform conversion of simple initializations in a switch to
6811 initializations from a scalar array. This flag is enabled by default
6812 at @option{-O2} and higher.
6816 Perform dead code elimination (DCE) on trees. This flag is enabled by
6817 default at @option{-O} and higher.
6819 @item -ftree-builtin-call-dce
6820 @opindex ftree-builtin-call-dce
6821 Perform conditional dead code elimination (DCE) for calls to builtin functions
6822 that may set @code{errno} but are otherwise side-effect free. This flag is
6823 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6826 @item -ftree-dominator-opts
6827 @opindex ftree-dominator-opts
6828 Perform a variety of simple scalar cleanups (constant/copy
6829 propagation, redundancy elimination, range propagation and expression
6830 simplification) based on a dominator tree traversal. This also
6831 performs jump threading (to reduce jumps to jumps). This flag is
6832 enabled by default at @option{-O} and higher.
6836 Perform dead store elimination (DSE) on trees. A dead store is a store into
6837 a memory location which will later be overwritten by another store without
6838 any intervening loads. In this case the earlier store can be deleted. This
6839 flag is enabled by default at @option{-O} and higher.
6843 Perform loop header copying on trees. This is beneficial since it increases
6844 effectiveness of code motion optimizations. It also saves one jump. This flag
6845 is enabled by default at @option{-O} and higher. It is not enabled
6846 for @option{-Os}, since it usually increases code size.
6848 @item -ftree-loop-optimize
6849 @opindex ftree-loop-optimize
6850 Perform loop optimizations on trees. This flag is enabled by default
6851 at @option{-O} and higher.
6853 @item -ftree-loop-linear
6854 @opindex ftree-loop-linear
6855 Perform linear loop transformations on tree. This flag can improve cache
6856 performance and allow further loop optimizations to take place.
6858 @item -floop-interchange
6859 @opindex floop-interchange
6860 Perform loop interchange transformations on loops. Interchanging two
6861 nested loops switches the inner and outer loops. For example, given a
6866 A(J, I) = A(J, I) * C
6870 loop interchange will transform the loop as if the user had written:
6874 A(J, I) = A(J, I) * C
6878 which can be beneficial when @code{N} is larger than the caches,
6879 because in Fortran, the elements of an array are stored in memory
6880 contiguously by column, and the original loop iterates over rows,
6881 potentially creating at each access a cache miss. This optimization
6882 applies to all the languages supported by GCC and is not limited to
6883 Fortran. To use this code transformation, GCC has to be configured
6884 with @option{--with-ppl} and @option{--with-cloog} to enable the
6885 Graphite loop transformation infrastructure.
6887 @item -floop-strip-mine
6888 @opindex floop-strip-mine
6889 Perform loop strip mining transformations on loops. Strip mining
6890 splits a loop into two nested loops. The outer loop has strides
6891 equal to the strip size and the inner loop has strides of the
6892 original loop within a strip. The strip length can be changed
6893 using the @option{loop-block-tile-size} parameter. For example,
6900 loop strip mining will transform the loop as if the user had written:
6903 DO I = II, min (II + 50, N)
6908 This optimization applies to all the languages supported by GCC and is
6909 not limited to Fortran. To use this code transformation, GCC has to
6910 be configured with @option{--with-ppl} and @option{--with-cloog} to
6911 enable the Graphite loop transformation infrastructure.
6914 @opindex floop-block
6915 Perform loop blocking transformations on loops. Blocking strip mines
6916 each loop in the loop nest such that the memory accesses of the
6917 element loops fit inside caches. The strip length can be changed
6918 using the @option{loop-block-tile-size} parameter. For example, given
6923 A(J, I) = B(I) + C(J)
6927 loop blocking will transform the loop as if the user had written:
6931 DO I = II, min (II + 50, N)
6932 DO J = JJ, min (JJ + 50, M)
6933 A(J, I) = B(I) + C(J)
6939 which can be beneficial when @code{M} is larger than the caches,
6940 because the innermost loop will iterate over a smaller amount of data
6941 that can be kept in the caches. This optimization applies to all the
6942 languages supported by GCC and is not limited to Fortran. To use this
6943 code transformation, GCC has to be configured with @option{--with-ppl}
6944 and @option{--with-cloog} to enable the Graphite loop transformation
6947 @item -fgraphite-identity
6948 @opindex fgraphite-identity
6949 Enable the identity transformation for graphite. For every SCoP we generate
6950 the polyhedral representation and transform it back to gimple. Using
6951 @option{-fgraphite-identity} we can check the costs or benefits of the
6952 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6953 are also performed by the code generator CLooG, like index splitting and
6954 dead code elimination in loops.
6956 @item -floop-flatten
6957 @opindex floop-flatten
6958 Removes the loop nesting structure: transforms the loop nest into a
6959 single loop. This transformation can be useful to vectorize all the
6960 levels of the loop nest.
6962 @item -floop-parallelize-all
6963 @opindex floop-parallelize-all
6964 Use the Graphite data dependence analysis to identify loops that can
6965 be parallelized. Parallelize all the loops that can be analyzed to
6966 not contain loop carried dependences without checking that it is
6967 profitable to parallelize the loops.
6969 @item -fcheck-data-deps
6970 @opindex fcheck-data-deps
6971 Compare the results of several data dependence analyzers. This option
6972 is used for debugging the data dependence analyzers.
6974 @item -ftree-loop-if-convert
6975 Attempt to transform conditional jumps in the innermost loops to
6976 branch-less equivalents. The intent is to remove control-flow from
6977 the innermost loops in order to improve the ability of the
6978 vectorization pass to handle these loops. This is enabled by default
6979 if vectorization is enabled.
6981 @item -ftree-loop-if-convert-stores
6982 Attempt to also if-convert conditional jumps containing memory writes.
6983 This transformation can be unsafe for multi-threaded programs as it
6984 transforms conditional memory writes into unconditional memory writes.
6987 for (i = 0; i < N; i++)
6991 would be transformed to
6993 for (i = 0; i < N; i++)
6994 A[i] = cond ? expr : A[i];
6996 potentially producing data races.
6998 @item -ftree-loop-distribution
6999 Perform loop distribution. This flag can improve cache performance on
7000 big loop bodies and allow further loop optimizations, like
7001 parallelization or vectorization, to take place. For example, the loop
7018 @item -ftree-loop-distribute-patterns
7019 Perform loop distribution of patterns that can be code generated with
7020 calls to a library. This flag is enabled by default at @option{-O3}.
7022 This pass distributes the initialization loops and generates a call to
7023 memset zero. For example, the loop
7039 and the initialization loop is transformed into a call to memset zero.
7041 @item -ftree-loop-im
7042 @opindex ftree-loop-im
7043 Perform loop invariant motion on trees. This pass moves only invariants that
7044 would be hard to handle at RTL level (function calls, operations that expand to
7045 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
7046 operands of conditions that are invariant out of the loop, so that we can use
7047 just trivial invariantness analysis in loop unswitching. The pass also includes
7050 @item -ftree-loop-ivcanon
7051 @opindex ftree-loop-ivcanon
7052 Create a canonical counter for number of iterations in the loop for that
7053 determining number of iterations requires complicated analysis. Later
7054 optimizations then may determine the number easily. Useful especially
7055 in connection with unrolling.
7059 Perform induction variable optimizations (strength reduction, induction
7060 variable merging and induction variable elimination) on trees.
7062 @item -ftree-parallelize-loops=n
7063 @opindex ftree-parallelize-loops
7064 Parallelize loops, i.e., split their iteration space to run in n threads.
7065 This is only possible for loops whose iterations are independent
7066 and can be arbitrarily reordered. The optimization is only
7067 profitable on multiprocessor machines, for loops that are CPU-intensive,
7068 rather than constrained e.g.@: by memory bandwidth. This option
7069 implies @option{-pthread}, and thus is only supported on targets
7070 that have support for @option{-pthread}.
7074 Perform function-local points-to analysis on trees. This flag is
7075 enabled by default at @option{-O} and higher.
7079 Perform scalar replacement of aggregates. This pass replaces structure
7080 references with scalars to prevent committing structures to memory too
7081 early. This flag is enabled by default at @option{-O} and higher.
7083 @item -ftree-copyrename
7084 @opindex ftree-copyrename
7085 Perform copy renaming on trees. This pass attempts to rename compiler
7086 temporaries to other variables at copy locations, usually resulting in
7087 variable names which more closely resemble the original variables. This flag
7088 is enabled by default at @option{-O} and higher.
7092 Perform temporary expression replacement during the SSA->normal phase. Single
7093 use/single def temporaries are replaced at their use location with their
7094 defining expression. This results in non-GIMPLE code, but gives the expanders
7095 much more complex trees to work on resulting in better RTL generation. This is
7096 enabled by default at @option{-O} and higher.
7098 @item -ftree-vectorize
7099 @opindex ftree-vectorize
7100 Perform loop vectorization on trees. This flag is enabled by default at
7103 @item -ftree-slp-vectorize
7104 @opindex ftree-slp-vectorize
7105 Perform basic block vectorization on trees. This flag is enabled by default at
7106 @option{-O3} and when @option{-ftree-vectorize} is enabled.
7108 @item -ftree-vect-loop-version
7109 @opindex ftree-vect-loop-version
7110 Perform loop versioning when doing loop vectorization on trees. When a loop
7111 appears to be vectorizable except that data alignment or data dependence cannot
7112 be determined at compile time then vectorized and non-vectorized versions of
7113 the loop are generated along with runtime checks for alignment or dependence
7114 to control which version is executed. This option is enabled by default
7115 except at level @option{-Os} where it is disabled.
7117 @item -fvect-cost-model
7118 @opindex fvect-cost-model
7119 Enable cost model for vectorization.
7123 Perform Value Range Propagation on trees. This is similar to the
7124 constant propagation pass, but instead of values, ranges of values are
7125 propagated. This allows the optimizers to remove unnecessary range
7126 checks like array bound checks and null pointer checks. This is
7127 enabled by default at @option{-O2} and higher. Null pointer check
7128 elimination is only done if @option{-fdelete-null-pointer-checks} is
7133 Perform tail duplication to enlarge superblock size. This transformation
7134 simplifies the control flow of the function allowing other optimizations to do
7137 @item -funroll-loops
7138 @opindex funroll-loops
7139 Unroll loops whose number of iterations can be determined at compile
7140 time or upon entry to the loop. @option{-funroll-loops} implies
7141 @option{-frerun-cse-after-loop}. This option makes code larger,
7142 and may or may not make it run faster.
7144 @item -funroll-all-loops
7145 @opindex funroll-all-loops
7146 Unroll all loops, even if their number of iterations is uncertain when
7147 the loop is entered. This usually makes programs run more slowly.
7148 @option{-funroll-all-loops} implies the same options as
7149 @option{-funroll-loops},
7151 @item -fsplit-ivs-in-unroller
7152 @opindex fsplit-ivs-in-unroller
7153 Enables expressing of values of induction variables in later iterations
7154 of the unrolled loop using the value in the first iteration. This breaks
7155 long dependency chains, thus improving efficiency of the scheduling passes.
7157 Combination of @option{-fweb} and CSE is often sufficient to obtain the
7158 same effect. However in cases the loop body is more complicated than
7159 a single basic block, this is not reliable. It also does not work at all
7160 on some of the architectures due to restrictions in the CSE pass.
7162 This optimization is enabled by default.
7164 @item -fvariable-expansion-in-unroller
7165 @opindex fvariable-expansion-in-unroller
7166 With this option, the compiler will create multiple copies of some
7167 local variables when unrolling a loop which can result in superior code.
7169 @item -fpartial-inlining
7170 @opindex fpartial-inlining
7171 Inline parts of functions. This option has any effect only
7172 when inlining itself is turned on by the @option{-finline-functions}
7173 or @option{-finline-small-functions} options.
7175 Enabled at level @option{-O2}.
7177 @item -fpredictive-commoning
7178 @opindex fpredictive-commoning
7179 Perform predictive commoning optimization, i.e., reusing computations
7180 (especially memory loads and stores) performed in previous
7181 iterations of loops.
7183 This option is enabled at level @option{-O3}.
7185 @item -fprefetch-loop-arrays
7186 @opindex fprefetch-loop-arrays
7187 If supported by the target machine, generate instructions to prefetch
7188 memory to improve the performance of loops that access large arrays.
7190 This option may generate better or worse code; results are highly
7191 dependent on the structure of loops within the source code.
7193 Disabled at level @option{-Os}.
7196 @itemx -fno-peephole2
7197 @opindex fno-peephole
7198 @opindex fno-peephole2
7199 Disable any machine-specific peephole optimizations. The difference
7200 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7201 are implemented in the compiler; some targets use one, some use the
7202 other, a few use both.
7204 @option{-fpeephole} is enabled by default.
7205 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7207 @item -fno-guess-branch-probability
7208 @opindex fno-guess-branch-probability
7209 Do not guess branch probabilities using heuristics.
7211 GCC will use heuristics to guess branch probabilities if they are
7212 not provided by profiling feedback (@option{-fprofile-arcs}). These
7213 heuristics are based on the control flow graph. If some branch probabilities
7214 are specified by @samp{__builtin_expect}, then the heuristics will be
7215 used to guess branch probabilities for the rest of the control flow graph,
7216 taking the @samp{__builtin_expect} info into account. The interactions
7217 between the heuristics and @samp{__builtin_expect} can be complex, and in
7218 some cases, it may be useful to disable the heuristics so that the effects
7219 of @samp{__builtin_expect} are easier to understand.
7221 The default is @option{-fguess-branch-probability} at levels
7222 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7224 @item -freorder-blocks
7225 @opindex freorder-blocks
7226 Reorder basic blocks in the compiled function in order to reduce number of
7227 taken branches and improve code locality.
7229 Enabled at levels @option{-O2}, @option{-O3}.
7231 @item -freorder-blocks-and-partition
7232 @opindex freorder-blocks-and-partition
7233 In addition to reordering basic blocks in the compiled function, in order
7234 to reduce number of taken branches, partitions hot and cold basic blocks
7235 into separate sections of the assembly and .o files, to improve
7236 paging and cache locality performance.
7238 This optimization is automatically turned off in the presence of
7239 exception handling, for linkonce sections, for functions with a user-defined
7240 section attribute and on any architecture that does not support named
7243 @item -freorder-functions
7244 @opindex freorder-functions
7245 Reorder functions in the object file in order to
7246 improve code locality. This is implemented by using special
7247 subsections @code{.text.hot} for most frequently executed functions and
7248 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
7249 the linker so object file format must support named sections and linker must
7250 place them in a reasonable way.
7252 Also profile feedback must be available in to make this option effective. See
7253 @option{-fprofile-arcs} for details.
7255 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7257 @item -fstrict-aliasing
7258 @opindex fstrict-aliasing
7259 Allow the compiler to assume the strictest aliasing rules applicable to
7260 the language being compiled. For C (and C++), this activates
7261 optimizations based on the type of expressions. In particular, an
7262 object of one type is assumed never to reside at the same address as an
7263 object of a different type, unless the types are almost the same. For
7264 example, an @code{unsigned int} can alias an @code{int}, but not a
7265 @code{void*} or a @code{double}. A character type may alias any other
7268 @anchor{Type-punning}Pay special attention to code like this:
7281 The practice of reading from a different union member than the one most
7282 recently written to (called ``type-punning'') is common. Even with
7283 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7284 is accessed through the union type. So, the code above will work as
7285 expected. @xref{Structures unions enumerations and bit-fields
7286 implementation}. However, this code might not:
7297 Similarly, access by taking the address, casting the resulting pointer
7298 and dereferencing the result has undefined behavior, even if the cast
7299 uses a union type, e.g.:
7303 return ((union a_union *) &d)->i;
7307 The @option{-fstrict-aliasing} option is enabled at levels
7308 @option{-O2}, @option{-O3}, @option{-Os}.
7310 @item -fstrict-overflow
7311 @opindex fstrict-overflow
7312 Allow the compiler to assume strict signed overflow rules, depending
7313 on the language being compiled. For C (and C++) this means that
7314 overflow when doing arithmetic with signed numbers is undefined, which
7315 means that the compiler may assume that it will not happen. This
7316 permits various optimizations. For example, the compiler will assume
7317 that an expression like @code{i + 10 > i} will always be true for
7318 signed @code{i}. This assumption is only valid if signed overflow is
7319 undefined, as the expression is false if @code{i + 10} overflows when
7320 using twos complement arithmetic. When this option is in effect any
7321 attempt to determine whether an operation on signed numbers will
7322 overflow must be written carefully to not actually involve overflow.
7324 This option also allows the compiler to assume strict pointer
7325 semantics: given a pointer to an object, if adding an offset to that
7326 pointer does not produce a pointer to the same object, the addition is
7327 undefined. This permits the compiler to conclude that @code{p + u >
7328 p} is always true for a pointer @code{p} and unsigned integer
7329 @code{u}. This assumption is only valid because pointer wraparound is
7330 undefined, as the expression is false if @code{p + u} overflows using
7331 twos complement arithmetic.
7333 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7334 that integer signed overflow is fully defined: it wraps. When
7335 @option{-fwrapv} is used, there is no difference between
7336 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7337 integers. With @option{-fwrapv} certain types of overflow are
7338 permitted. For example, if the compiler gets an overflow when doing
7339 arithmetic on constants, the overflowed value can still be used with
7340 @option{-fwrapv}, but not otherwise.
7342 The @option{-fstrict-overflow} option is enabled at levels
7343 @option{-O2}, @option{-O3}, @option{-Os}.
7345 @item -falign-functions
7346 @itemx -falign-functions=@var{n}
7347 @opindex falign-functions
7348 Align the start of functions to the next power-of-two greater than
7349 @var{n}, skipping up to @var{n} bytes. For instance,
7350 @option{-falign-functions=32} aligns functions to the next 32-byte
7351 boundary, but @option{-falign-functions=24} would align to the next
7352 32-byte boundary only if this can be done by skipping 23 bytes or less.
7354 @option{-fno-align-functions} and @option{-falign-functions=1} are
7355 equivalent and mean that functions will not be aligned.
7357 Some assemblers only support this flag when @var{n} is a power of two;
7358 in that case, it is rounded up.
7360 If @var{n} is not specified or is zero, use a machine-dependent default.
7362 Enabled at levels @option{-O2}, @option{-O3}.
7364 @item -falign-labels
7365 @itemx -falign-labels=@var{n}
7366 @opindex falign-labels
7367 Align all branch targets to a power-of-two boundary, skipping up to
7368 @var{n} bytes like @option{-falign-functions}. This option can easily
7369 make code slower, because it must insert dummy operations for when the
7370 branch target is reached in the usual flow of the code.
7372 @option{-fno-align-labels} and @option{-falign-labels=1} are
7373 equivalent and mean that labels will not be aligned.
7375 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7376 are greater than this value, then their values are used instead.
7378 If @var{n} is not specified or is zero, use a machine-dependent default
7379 which is very likely to be @samp{1}, meaning no alignment.
7381 Enabled at levels @option{-O2}, @option{-O3}.
7384 @itemx -falign-loops=@var{n}
7385 @opindex falign-loops
7386 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7387 like @option{-falign-functions}. The hope is that the loop will be
7388 executed many times, which will make up for any execution of the dummy
7391 @option{-fno-align-loops} and @option{-falign-loops=1} are
7392 equivalent and mean that loops will not be aligned.
7394 If @var{n} is not specified or is zero, use a machine-dependent default.
7396 Enabled at levels @option{-O2}, @option{-O3}.
7399 @itemx -falign-jumps=@var{n}
7400 @opindex falign-jumps
7401 Align branch targets to a power-of-two boundary, for branch targets
7402 where the targets can only be reached by jumping, skipping up to @var{n}
7403 bytes like @option{-falign-functions}. In this case, no dummy operations
7406 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7407 equivalent and mean that loops will not be aligned.
7409 If @var{n} is not specified or is zero, use a machine-dependent default.
7411 Enabled at levels @option{-O2}, @option{-O3}.
7413 @item -funit-at-a-time
7414 @opindex funit-at-a-time
7415 This option is left for compatibility reasons. @option{-funit-at-a-time}
7416 has no effect, while @option{-fno-unit-at-a-time} implies
7417 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7421 @item -fno-toplevel-reorder
7422 @opindex fno-toplevel-reorder
7423 Do not reorder top-level functions, variables, and @code{asm}
7424 statements. Output them in the same order that they appear in the
7425 input file. When this option is used, unreferenced static variables
7426 will not be removed. This option is intended to support existing code
7427 which relies on a particular ordering. For new code, it is better to
7430 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7431 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7436 Constructs webs as commonly used for register allocation purposes and assign
7437 each web individual pseudo register. This allows the register allocation pass
7438 to operate on pseudos directly, but also strengthens several other optimization
7439 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7440 however, make debugging impossible, since variables will no longer stay in a
7443 Enabled by default with @option{-funroll-loops}.
7445 @item -fwhole-program
7446 @opindex fwhole-program
7447 Assume that the current compilation unit represents the whole program being
7448 compiled. All public functions and variables with the exception of @code{main}
7449 and those merged by attribute @code{externally_visible} become static functions
7450 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.
7451 While this option is equivalent to proper use of the @code{static} keyword for
7452 programs consisting of a single file, in combination with option
7453 @option{-flto} this flag can be used to
7454 compile many smaller scale programs since the functions and variables become
7455 local for the whole combined compilation unit, not for the single source file
7458 This option implies @option{-fwhole-file} for Fortran programs.
7460 @item -flto[=@var{n}]
7462 This option runs the standard link-time optimizer. When invoked
7463 with source code, it generates GIMPLE (one of GCC's internal
7464 representations) and writes it to special ELF sections in the object
7465 file. When the object files are linked together, all the function
7466 bodies are read from these ELF sections and instantiated as if they
7467 had been part of the same translation unit.
7469 To use the link-timer optimizer, @option{-flto} needs to be specified at
7470 compile time and during the final link. For example,
7473 gcc -c -O2 -flto foo.c
7474 gcc -c -O2 -flto bar.c
7475 gcc -o myprog -flto -O2 foo.o bar.o
7478 The first two invocations to GCC will save a bytecode representation
7479 of GIMPLE into special ELF sections inside @file{foo.o} and
7480 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7481 @file{foo.o} and @file{bar.o}, merge the two files into a single
7482 internal image, and compile the result as usual. Since both
7483 @file{foo.o} and @file{bar.o} are merged into a single image, this
7484 causes all the inter-procedural analyses and optimizations in GCC to
7485 work across the two files as if they were a single one. This means,
7486 for example, that the inliner will be able to inline functions in
7487 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7489 Another (simpler) way to enable link-time optimization is,
7492 gcc -o myprog -flto -O2 foo.c bar.c
7495 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7496 merge them together into a single GIMPLE representation and optimize
7497 them as usual to produce @file{myprog}.
7499 The only important thing to keep in mind is that to enable link-time
7500 optimizations the @option{-flto} flag needs to be passed to both the
7501 compile and the link commands.
7503 To make whole program optimization effective, it is necesary to make
7504 certain whole program assumptions. The compiler needs to know
7505 what functions and variables can be accessed by libraries and runtime
7506 outside of the link time optimized unit. When supported by the linker,
7507 the linker plugin (see @option{-fuse-linker-plugin}) passes to the
7508 compiler information about used and externally visible symbols. When
7509 the linker plugin is not available, @option{-fwhole-program} should be
7510 used to allow the compiler to make these assumptions, which will lead
7511 to more aggressive optimization decisions.
7513 Note that when a file is compiled with @option{-flto}, the generated
7514 object file will be larger than a regular object file because it will
7515 contain GIMPLE bytecodes and the usual final code. This means that
7516 object files with LTO information can be linked as a normal object
7517 file. So, in the previous example, if the final link is done with
7520 gcc -o myprog foo.o bar.o
7523 The only difference will be that no inter-procedural optimizations
7524 will be applied to produce @file{myprog}. The two object files
7525 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7528 Additionally, the optimization flags used to compile individual files
7529 are not necessarily related to those used at link-time. For instance,
7532 gcc -c -O0 -flto foo.c
7533 gcc -c -O0 -flto bar.c
7534 gcc -o myprog -flto -O3 foo.o bar.o
7537 This will produce individual object files with unoptimized assembler
7538 code, but the resulting binary @file{myprog} will be optimized at
7539 @option{-O3}. Now, if the final binary is generated without
7540 @option{-flto}, then @file{myprog} will not be optimized.
7542 When producing the final binary with @option{-flto}, GCC will only
7543 apply link-time optimizations to those files that contain bytecode.
7544 Therefore, you can mix and match object files and libraries with
7545 GIMPLE bytecodes and final object code. GCC will automatically select
7546 which files to optimize in LTO mode and which files to link without
7549 There are some code generation flags that GCC will preserve when
7550 generating bytecodes, as they need to be used during the final link
7551 stage. Currently, the following options are saved into the GIMPLE
7552 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7553 @option{-m} target flags.
7555 At link time, these options are read-in and reapplied. Note that the
7556 current implementation makes no attempt at recognizing conflicting
7557 values for these options. If two or more files have a conflicting
7558 value (e.g., one file is compiled with @option{-fPIC} and another
7559 isn't), the compiler will simply use the last value read from the
7560 bytecode files. It is recommended, then, that all the files
7561 participating in the same link be compiled with the same options.
7563 Another feature of LTO is that it is possible to apply interprocedural
7564 optimizations on files written in different languages. This requires
7565 some support in the language front end. Currently, the C, C++ and
7566 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7567 something like this should work
7572 gfortran -c -flto baz.f90
7573 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7576 Notice that the final link is done with @command{g++} to get the C++
7577 runtime libraries and @option{-lgfortran} is added to get the Fortran
7578 runtime libraries. In general, when mixing languages in LTO mode, you
7579 should use the same link command used when mixing languages in a
7580 regular (non-LTO) compilation. This means that if your build process
7581 was mixing languages before, all you need to add is @option{-flto} to
7582 all the compile and link commands.
7584 If LTO encounters objects with C linkage declared with incompatible
7585 types in separate translation units to be linked together (undefined
7586 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7587 issued. The behavior is still undefined at runtime.
7589 If object files containing GIMPLE bytecode are stored in a library
7590 archive, say @file{libfoo.a}, it is possible to extract and use them
7591 in an LTO link if you are using @command{gold} as the linker (which,
7592 in turn requires GCC to be configured with @option{--enable-gold}).
7593 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7597 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7600 With the linker plugin enabled, @command{gold} will extract the needed
7601 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7602 to make them part of the aggregated GIMPLE image to be optimized.
7604 If you are not using @command{gold} and/or do not specify
7605 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7606 will be extracted and linked as usual, but they will not participate
7607 in the LTO optimization process.
7609 Link time optimizations do not require the presence of the whole program to
7610 operate. If the program does not require any symbols to be exported, it is
7611 possible to combine @option{-flto} and with @option{-fwhole-program} to allow
7612 the interprocedural optimizers to use more aggressive assumptions which may
7613 lead to improved optimization opportunities.
7614 Use of @option{-fwhole-program} is not needed when linker plugin is
7615 active (see @option{-fuse-linker-plugin}).
7617 Regarding portability: the current implementation of LTO makes no
7618 attempt at generating bytecode that can be ported between different
7619 types of hosts. The bytecode files are versioned and there is a
7620 strict version check, so bytecode files generated in one version of
7621 GCC will not work with an older/newer version of GCC.
7623 Link time optimization does not play well with generating debugging
7624 information. Combining @option{-flto} with
7625 @option{-g} is currently experimental and expected to produce wrong
7628 If you specify the optional @var{n}, the optimization and code
7629 generation done at link time is executed in parallel using @var{n}
7630 parallel jobs by utilizing an installed @command{make} program. The
7631 environment variable @env{MAKE} may be used to override the program
7632 used. The default value for @var{n} is 1.
7634 You can also specify @option{-flto=jobserver} to use GNU make's
7635 job server mode to determine the number of parallel jobs. This
7636 is useful when the Makefile calling GCC is already executing in parallel.
7637 The parent Makefile will need a @samp{+} prepended to the command recipe
7638 for this to work. This will likely only work if @env{MAKE} is
7641 This option is disabled by default.
7643 @item -flto-partition=@var{alg}
7644 @opindex flto-partition
7645 Specify the partitioning algorithm used by the link time optimizer.
7646 The value is either @code{1to1} to specify a partitioning mirroring
7647 the original source files or @code{balanced} to specify partitioning
7648 into equally sized chunks (whenever possible). Specifying @code{none}
7649 as an algorithm disables partitioning and streaming completely. The
7650 default value is @code{balanced}.
7652 @item -flto-compression-level=@var{n}
7653 This option specifies the level of compression used for intermediate
7654 language written to LTO object files, and is only meaningful in
7655 conjunction with LTO mode (@option{-flto}). Valid
7656 values are 0 (no compression) to 9 (maximum compression). Values
7657 outside this range are clamped to either 0 or 9. If the option is not
7658 given, a default balanced compression setting is used.
7661 Prints a report with internal details on the workings of the link-time
7662 optimizer. The contents of this report vary from version to version,
7663 it is meant to be useful to GCC developers when processing object
7664 files in LTO mode (via @option{-flto}).
7666 Disabled by default.
7668 @item -fuse-linker-plugin
7669 Enables the extraction of objects with GIMPLE bytecode information
7670 from library archives. This option relies on features available only
7671 in @command{gold}, so to use this you must configure GCC with
7672 @option{--enable-gold}. See @option{-flto} for a description on the
7673 effect of this flag and how to use it.
7675 Disabled by default.
7677 @item -fcprop-registers
7678 @opindex fcprop-registers
7679 After register allocation and post-register allocation instruction splitting,
7680 we perform a copy-propagation pass to try to reduce scheduling dependencies
7681 and occasionally eliminate the copy.
7683 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7685 @item -fprofile-correction
7686 @opindex fprofile-correction
7687 Profiles collected using an instrumented binary for multi-threaded programs may
7688 be inconsistent due to missed counter updates. When this option is specified,
7689 GCC will use heuristics to correct or smooth out such inconsistencies. By
7690 default, GCC will emit an error message when an inconsistent profile is detected.
7692 @item -fprofile-dir=@var{path}
7693 @opindex fprofile-dir
7695 Set the directory to search the profile data files in to @var{path}.
7696 This option affects only the profile data generated by
7697 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7698 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7699 and its related options.
7700 By default, GCC will use the current directory as @var{path}
7701 thus the profile data file will appear in the same directory as the object file.
7703 @item -fprofile-generate
7704 @itemx -fprofile-generate=@var{path}
7705 @opindex fprofile-generate
7707 Enable options usually used for instrumenting application to produce
7708 profile useful for later recompilation with profile feedback based
7709 optimization. You must use @option{-fprofile-generate} both when
7710 compiling and when linking your program.
7712 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7714 If @var{path} is specified, GCC will look at the @var{path} to find
7715 the profile feedback data files. See @option{-fprofile-dir}.
7718 @itemx -fprofile-use=@var{path}
7719 @opindex fprofile-use
7720 Enable profile feedback directed optimizations, and optimizations
7721 generally profitable only with profile feedback available.
7723 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7724 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7726 By default, GCC emits an error message if the feedback profiles do not
7727 match the source code. This error can be turned into a warning by using
7728 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7731 If @var{path} is specified, GCC will look at the @var{path} to find
7732 the profile feedback data files. See @option{-fprofile-dir}.
7735 The following options control compiler behavior regarding floating
7736 point arithmetic. These options trade off between speed and
7737 correctness. All must be specifically enabled.
7741 @opindex ffloat-store
7742 Do not store floating point variables in registers, and inhibit other
7743 options that might change whether a floating point value is taken from a
7746 @cindex floating point precision
7747 This option prevents undesirable excess precision on machines such as
7748 the 68000 where the floating registers (of the 68881) keep more
7749 precision than a @code{double} is supposed to have. Similarly for the
7750 x86 architecture. For most programs, the excess precision does only
7751 good, but a few programs rely on the precise definition of IEEE floating
7752 point. Use @option{-ffloat-store} for such programs, after modifying
7753 them to store all pertinent intermediate computations into variables.
7755 @item -fexcess-precision=@var{style}
7756 @opindex fexcess-precision
7757 This option allows further control over excess precision on machines
7758 where floating-point registers have more precision than the IEEE
7759 @code{float} and @code{double} types and the processor does not
7760 support operations rounding to those types. By default,
7761 @option{-fexcess-precision=fast} is in effect; this means that
7762 operations are carried out in the precision of the registers and that
7763 it is unpredictable when rounding to the types specified in the source
7764 code takes place. When compiling C, if
7765 @option{-fexcess-precision=standard} is specified then excess
7766 precision will follow the rules specified in ISO C99; in particular,
7767 both casts and assignments cause values to be rounded to their
7768 semantic types (whereas @option{-ffloat-store} only affects
7769 assignments). This option is enabled by default for C if a strict
7770 conformance option such as @option{-std=c99} is used.
7773 @option{-fexcess-precision=standard} is not implemented for languages
7774 other than C, and has no effect if
7775 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7776 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7777 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7778 semantics apply without excess precision, and in the latter, rounding
7783 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7784 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7785 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7787 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7789 This option is not turned on by any @option{-O} option since
7790 it can result in incorrect output for programs which depend on
7791 an exact implementation of IEEE or ISO rules/specifications for
7792 math functions. It may, however, yield faster code for programs
7793 that do not require the guarantees of these specifications.
7795 @item -fno-math-errno
7796 @opindex fno-math-errno
7797 Do not set ERRNO after calling math functions that are executed
7798 with a single instruction, e.g., sqrt. A program that relies on
7799 IEEE exceptions for math error handling may want to use this flag
7800 for speed while maintaining IEEE arithmetic compatibility.
7802 This option is not turned on by any @option{-O} option since
7803 it can result in incorrect output for programs which depend on
7804 an exact implementation of IEEE or ISO rules/specifications for
7805 math functions. It may, however, yield faster code for programs
7806 that do not require the guarantees of these specifications.
7808 The default is @option{-fmath-errno}.
7810 On Darwin systems, the math library never sets @code{errno}. There is
7811 therefore no reason for the compiler to consider the possibility that
7812 it might, and @option{-fno-math-errno} is the default.
7814 @item -funsafe-math-optimizations
7815 @opindex funsafe-math-optimizations
7817 Allow optimizations for floating-point arithmetic that (a) assume
7818 that arguments and results are valid and (b) may violate IEEE or
7819 ANSI standards. When used at link-time, it may include libraries
7820 or startup files that change the default FPU control word or other
7821 similar optimizations.
7823 This option is not turned on by any @option{-O} option since
7824 it can result in incorrect output for programs which depend on
7825 an exact implementation of IEEE or ISO rules/specifications for
7826 math functions. It may, however, yield faster code for programs
7827 that do not require the guarantees of these specifications.
7828 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7829 @option{-fassociative-math} and @option{-freciprocal-math}.
7831 The default is @option{-fno-unsafe-math-optimizations}.
7833 @item -fassociative-math
7834 @opindex fassociative-math
7836 Allow re-association of operands in series of floating-point operations.
7837 This violates the ISO C and C++ language standard by possibly changing
7838 computation result. NOTE: re-ordering may change the sign of zero as
7839 well as ignore NaNs and inhibit or create underflow or overflow (and
7840 thus cannot be used on a code which relies on rounding behavior like
7841 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7842 and thus may not be used when ordered comparisons are required.
7843 This option requires that both @option{-fno-signed-zeros} and
7844 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7845 much sense with @option{-frounding-math}. For Fortran the option
7846 is automatically enabled when both @option{-fno-signed-zeros} and
7847 @option{-fno-trapping-math} are in effect.
7849 The default is @option{-fno-associative-math}.
7851 @item -freciprocal-math
7852 @opindex freciprocal-math
7854 Allow the reciprocal of a value to be used instead of dividing by
7855 the value if this enables optimizations. For example @code{x / y}
7856 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7857 is subject to common subexpression elimination. Note that this loses
7858 precision and increases the number of flops operating on the value.
7860 The default is @option{-fno-reciprocal-math}.
7862 @item -ffinite-math-only
7863 @opindex ffinite-math-only
7864 Allow optimizations for floating-point arithmetic that assume
7865 that arguments and results are not NaNs or +-Infs.
7867 This option is not turned on by any @option{-O} option since
7868 it can result in incorrect output for programs which depend on
7869 an exact implementation of IEEE or ISO rules/specifications for
7870 math functions. It may, however, yield faster code for programs
7871 that do not require the guarantees of these specifications.
7873 The default is @option{-fno-finite-math-only}.
7875 @item -fno-signed-zeros
7876 @opindex fno-signed-zeros
7877 Allow optimizations for floating point arithmetic that ignore the
7878 signedness of zero. IEEE arithmetic specifies the behavior of
7879 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7880 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7881 This option implies that the sign of a zero result isn't significant.
7883 The default is @option{-fsigned-zeros}.
7885 @item -fno-trapping-math
7886 @opindex fno-trapping-math
7887 Compile code assuming that floating-point operations cannot generate
7888 user-visible traps. These traps include division by zero, overflow,
7889 underflow, inexact result and invalid operation. This option requires
7890 that @option{-fno-signaling-nans} be in effect. Setting this option may
7891 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7893 This option should never be turned on by any @option{-O} option since
7894 it can result in incorrect output for programs which depend on
7895 an exact implementation of IEEE or ISO rules/specifications for
7898 The default is @option{-ftrapping-math}.
7900 @item -frounding-math
7901 @opindex frounding-math
7902 Disable transformations and optimizations that assume default floating
7903 point rounding behavior. This is round-to-zero for all floating point
7904 to integer conversions, and round-to-nearest for all other arithmetic
7905 truncations. This option should be specified for programs that change
7906 the FP rounding mode dynamically, or that may be executed with a
7907 non-default rounding mode. This option disables constant folding of
7908 floating point expressions at compile-time (which may be affected by
7909 rounding mode) and arithmetic transformations that are unsafe in the
7910 presence of sign-dependent rounding modes.
7912 The default is @option{-fno-rounding-math}.
7914 This option is experimental and does not currently guarantee to
7915 disable all GCC optimizations that are affected by rounding mode.
7916 Future versions of GCC may provide finer control of this setting
7917 using C99's @code{FENV_ACCESS} pragma. This command line option
7918 will be used to specify the default state for @code{FENV_ACCESS}.
7920 @item -fsignaling-nans
7921 @opindex fsignaling-nans
7922 Compile code assuming that IEEE signaling NaNs may generate user-visible
7923 traps during floating-point operations. Setting this option disables
7924 optimizations that may change the number of exceptions visible with
7925 signaling NaNs. This option implies @option{-ftrapping-math}.
7927 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7930 The default is @option{-fno-signaling-nans}.
7932 This option is experimental and does not currently guarantee to
7933 disable all GCC optimizations that affect signaling NaN behavior.
7935 @item -fsingle-precision-constant
7936 @opindex fsingle-precision-constant
7937 Treat floating point constant as single precision constant instead of
7938 implicitly converting it to double precision constant.
7940 @item -fcx-limited-range
7941 @opindex fcx-limited-range
7942 When enabled, this option states that a range reduction step is not
7943 needed when performing complex division. Also, there is no checking
7944 whether the result of a complex multiplication or division is @code{NaN
7945 + I*NaN}, with an attempt to rescue the situation in that case. The
7946 default is @option{-fno-cx-limited-range}, but is enabled by
7947 @option{-ffast-math}.
7949 This option controls the default setting of the ISO C99
7950 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7953 @item -fcx-fortran-rules
7954 @opindex fcx-fortran-rules
7955 Complex multiplication and division follow Fortran rules. Range
7956 reduction is done as part of complex division, but there is no checking
7957 whether the result of a complex multiplication or division is @code{NaN
7958 + I*NaN}, with an attempt to rescue the situation in that case.
7960 The default is @option{-fno-cx-fortran-rules}.
7964 The following options control optimizations that may improve
7965 performance, but are not enabled by any @option{-O} options. This
7966 section includes experimental options that may produce broken code.
7969 @item -fbranch-probabilities
7970 @opindex fbranch-probabilities
7971 After running a program compiled with @option{-fprofile-arcs}
7972 (@pxref{Debugging Options,, Options for Debugging Your Program or
7973 @command{gcc}}), you can compile it a second time using
7974 @option{-fbranch-probabilities}, to improve optimizations based on
7975 the number of times each branch was taken. When the program
7976 compiled with @option{-fprofile-arcs} exits it saves arc execution
7977 counts to a file called @file{@var{sourcename}.gcda} for each source
7978 file. The information in this data file is very dependent on the
7979 structure of the generated code, so you must use the same source code
7980 and the same optimization options for both compilations.
7982 With @option{-fbranch-probabilities}, GCC puts a
7983 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7984 These can be used to improve optimization. Currently, they are only
7985 used in one place: in @file{reorg.c}, instead of guessing which path a
7986 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7987 exactly determine which path is taken more often.
7989 @item -fprofile-values
7990 @opindex fprofile-values
7991 If combined with @option{-fprofile-arcs}, it adds code so that some
7992 data about values of expressions in the program is gathered.
7994 With @option{-fbranch-probabilities}, it reads back the data gathered
7995 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7996 notes to instructions for their later usage in optimizations.
7998 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
8002 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
8003 a code to gather information about values of expressions.
8005 With @option{-fbranch-probabilities}, it reads back the data gathered
8006 and actually performs the optimizations based on them.
8007 Currently the optimizations include specialization of division operation
8008 using the knowledge about the value of the denominator.
8010 @item -frename-registers
8011 @opindex frename-registers
8012 Attempt to avoid false dependencies in scheduled code by making use
8013 of registers left over after register allocation. This optimization
8014 will most benefit processors with lots of registers. Depending on the
8015 debug information format adopted by the target, however, it can
8016 make debugging impossible, since variables will no longer stay in
8017 a ``home register''.
8019 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
8023 Perform tail duplication to enlarge superblock size. This transformation
8024 simplifies the control flow of the function allowing other optimizations to do
8027 Enabled with @option{-fprofile-use}.
8029 @item -funroll-loops
8030 @opindex funroll-loops
8031 Unroll loops whose number of iterations can be determined at compile time or
8032 upon entry to the loop. @option{-funroll-loops} implies
8033 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
8034 It also turns on complete loop peeling (i.e.@: complete removal of loops with
8035 small constant number of iterations). This option makes code larger, and may
8036 or may not make it run faster.
8038 Enabled with @option{-fprofile-use}.
8040 @item -funroll-all-loops
8041 @opindex funroll-all-loops
8042 Unroll all loops, even if their number of iterations is uncertain when
8043 the loop is entered. This usually makes programs run more slowly.
8044 @option{-funroll-all-loops} implies the same options as
8045 @option{-funroll-loops}.
8048 @opindex fpeel-loops
8049 Peels the loops for that there is enough information that they do not
8050 roll much (from profile feedback). It also turns on complete loop peeling
8051 (i.e.@: complete removal of loops with small constant number of iterations).
8053 Enabled with @option{-fprofile-use}.
8055 @item -fmove-loop-invariants
8056 @opindex fmove-loop-invariants
8057 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
8058 at level @option{-O1}
8060 @item -funswitch-loops
8061 @opindex funswitch-loops
8062 Move branches with loop invariant conditions out of the loop, with duplicates
8063 of the loop on both branches (modified according to result of the condition).
8065 @item -ffunction-sections
8066 @itemx -fdata-sections
8067 @opindex ffunction-sections
8068 @opindex fdata-sections
8069 Place each function or data item into its own section in the output
8070 file if the target supports arbitrary sections. The name of the
8071 function or the name of the data item determines the section's name
8074 Use these options on systems where the linker can perform optimizations
8075 to improve locality of reference in the instruction space. Most systems
8076 using the ELF object format and SPARC processors running Solaris 2 have
8077 linkers with such optimizations. AIX may have these optimizations in
8080 Only use these options when there are significant benefits from doing
8081 so. When you specify these options, the assembler and linker will
8082 create larger object and executable files and will also be slower.
8083 You will not be able to use @code{gprof} on all systems if you
8084 specify this option and you may have problems with debugging if
8085 you specify both this option and @option{-g}.
8087 @item -fbranch-target-load-optimize
8088 @opindex fbranch-target-load-optimize
8089 Perform branch target register load optimization before prologue / epilogue
8091 The use of target registers can typically be exposed only during reload,
8092 thus hoisting loads out of loops and doing inter-block scheduling needs
8093 a separate optimization pass.
8095 @item -fbranch-target-load-optimize2
8096 @opindex fbranch-target-load-optimize2
8097 Perform branch target register load optimization after prologue / epilogue
8100 @item -fbtr-bb-exclusive
8101 @opindex fbtr-bb-exclusive
8102 When performing branch target register load optimization, don't reuse
8103 branch target registers in within any basic block.
8105 @item -fstack-protector
8106 @opindex fstack-protector
8107 Emit extra code to check for buffer overflows, such as stack smashing
8108 attacks. This is done by adding a guard variable to functions with
8109 vulnerable objects. This includes functions that call alloca, and
8110 functions with buffers larger than 8 bytes. The guards are initialized
8111 when a function is entered and then checked when the function exits.
8112 If a guard check fails, an error message is printed and the program exits.
8114 @item -fstack-protector-all
8115 @opindex fstack-protector-all
8116 Like @option{-fstack-protector} except that all functions are protected.
8118 @item -fsection-anchors
8119 @opindex fsection-anchors
8120 Try to reduce the number of symbolic address calculations by using
8121 shared ``anchor'' symbols to address nearby objects. This transformation
8122 can help to reduce the number of GOT entries and GOT accesses on some
8125 For example, the implementation of the following function @code{foo}:
8129 int foo (void) @{ return a + b + c; @}
8132 would usually calculate the addresses of all three variables, but if you
8133 compile it with @option{-fsection-anchors}, it will access the variables
8134 from a common anchor point instead. The effect is similar to the
8135 following pseudocode (which isn't valid C):
8140 register int *xr = &x;
8141 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
8145 Not all targets support this option.
8147 @item --param @var{name}=@var{value}
8149 In some places, GCC uses various constants to control the amount of
8150 optimization that is done. For example, GCC will not inline functions
8151 that contain more that a certain number of instructions. You can
8152 control some of these constants on the command-line using the
8153 @option{--param} option.
8155 The names of specific parameters, and the meaning of the values, are
8156 tied to the internals of the compiler, and are subject to change
8157 without notice in future releases.
8159 In each case, the @var{value} is an integer. The allowable choices for
8160 @var{name} are given in the following table:
8163 @item struct-reorg-cold-struct-ratio
8164 The threshold ratio (as a percentage) between a structure frequency
8165 and the frequency of the hottest structure in the program. This parameter
8166 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
8167 We say that if the ratio of a structure frequency, calculated by profiling,
8168 to the hottest structure frequency in the program is less than this
8169 parameter, then structure reorganization is not applied to this structure.
8172 @item predictable-branch-outcome
8173 When branch is predicted to be taken with probability lower than this threshold
8174 (in percent), then it is considered well predictable. The default is 10.
8176 @item max-crossjump-edges
8177 The maximum number of incoming edges to consider for crossjumping.
8178 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8179 the number of edges incoming to each block. Increasing values mean
8180 more aggressive optimization, making the compile time increase with
8181 probably small improvement in executable size.
8183 @item min-crossjump-insns
8184 The minimum number of instructions which must be matched at the end
8185 of two blocks before crossjumping will be performed on them. This
8186 value is ignored in the case where all instructions in the block being
8187 crossjumped from are matched. The default value is 5.
8189 @item max-grow-copy-bb-insns
8190 The maximum code size expansion factor when copying basic blocks
8191 instead of jumping. The expansion is relative to a jump instruction.
8192 The default value is 8.
8194 @item max-goto-duplication-insns
8195 The maximum number of instructions to duplicate to a block that jumps
8196 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
8197 passes, GCC factors computed gotos early in the compilation process,
8198 and unfactors them as late as possible. Only computed jumps at the
8199 end of a basic blocks with no more than max-goto-duplication-insns are
8200 unfactored. The default value is 8.
8202 @item max-delay-slot-insn-search
8203 The maximum number of instructions to consider when looking for an
8204 instruction to fill a delay slot. If more than this arbitrary number of
8205 instructions is searched, the time savings from filling the delay slot
8206 will be minimal so stop searching. Increasing values mean more
8207 aggressive optimization, making the compile time increase with probably
8208 small improvement in executable run time.
8210 @item max-delay-slot-live-search
8211 When trying to fill delay slots, the maximum number of instructions to
8212 consider when searching for a block with valid live register
8213 information. Increasing this arbitrarily chosen value means more
8214 aggressive optimization, increasing the compile time. This parameter
8215 should be removed when the delay slot code is rewritten to maintain the
8218 @item max-gcse-memory
8219 The approximate maximum amount of memory that will be allocated in
8220 order to perform the global common subexpression elimination
8221 optimization. If more memory than specified is required, the
8222 optimization will not be done.
8224 @item max-pending-list-length
8225 The maximum number of pending dependencies scheduling will allow
8226 before flushing the current state and starting over. Large functions
8227 with few branches or calls can create excessively large lists which
8228 needlessly consume memory and resources.
8230 @item max-inline-insns-single
8231 Several parameters control the tree inliner used in gcc.
8232 This number sets the maximum number of instructions (counted in GCC's
8233 internal representation) in a single function that the tree inliner
8234 will consider for inlining. This only affects functions declared
8235 inline and methods implemented in a class declaration (C++).
8236 The default value is 300.
8238 @item max-inline-insns-auto
8239 When you use @option{-finline-functions} (included in @option{-O3}),
8240 a lot of functions that would otherwise not be considered for inlining
8241 by the compiler will be investigated. To those functions, a different
8242 (more restrictive) limit compared to functions declared inline can
8244 The default value is 40.
8246 @item large-function-insns
8247 The limit specifying really large functions. For functions larger than this
8248 limit after inlining, inlining is constrained by
8249 @option{--param large-function-growth}. This parameter is useful primarily
8250 to avoid extreme compilation time caused by non-linear algorithms used by the
8252 The default value is 2700.
8254 @item large-function-growth
8255 Specifies maximal growth of large function caused by inlining in percents.
8256 The default value is 100 which limits large function growth to 2.0 times
8259 @item large-unit-insns
8260 The limit specifying large translation unit. Growth caused by inlining of
8261 units larger than this limit is limited by @option{--param inline-unit-growth}.
8262 For small units this might be too tight (consider unit consisting of function A
8263 that is inline and B that just calls A three time. If B is small relative to
8264 A, the growth of unit is 300\% and yet such inlining is very sane. For very
8265 large units consisting of small inlineable functions however the overall unit
8266 growth limit is needed to avoid exponential explosion of code size. Thus for
8267 smaller units, the size is increased to @option{--param large-unit-insns}
8268 before applying @option{--param inline-unit-growth}. The default is 10000
8270 @item inline-unit-growth
8271 Specifies maximal overall growth of the compilation unit caused by inlining.
8272 The default value is 30 which limits unit growth to 1.3 times the original
8275 @item ipcp-unit-growth
8276 Specifies maximal overall growth of the compilation unit caused by
8277 interprocedural constant propagation. The default value is 10 which limits
8278 unit growth to 1.1 times the original size.
8280 @item large-stack-frame
8281 The limit specifying large stack frames. While inlining the algorithm is trying
8282 to not grow past this limit too much. Default value is 256 bytes.
8284 @item large-stack-frame-growth
8285 Specifies maximal growth of large stack frames caused by inlining in percents.
8286 The default value is 1000 which limits large stack frame growth to 11 times
8289 @item max-inline-insns-recursive
8290 @itemx max-inline-insns-recursive-auto
8291 Specifies maximum number of instructions out-of-line copy of self recursive inline
8292 function can grow into by performing recursive inlining.
8294 For functions declared inline @option{--param max-inline-insns-recursive} is
8295 taken into account. For function not declared inline, recursive inlining
8296 happens only when @option{-finline-functions} (included in @option{-O3}) is
8297 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8298 default value is 450.
8300 @item max-inline-recursive-depth
8301 @itemx max-inline-recursive-depth-auto
8302 Specifies maximum recursion depth used by the recursive inlining.
8304 For functions declared inline @option{--param max-inline-recursive-depth} is
8305 taken into account. For function not declared inline, recursive inlining
8306 happens only when @option{-finline-functions} (included in @option{-O3}) is
8307 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8310 @item min-inline-recursive-probability
8311 Recursive inlining is profitable only for function having deep recursion
8312 in average and can hurt for function having little recursion depth by
8313 increasing the prologue size or complexity of function body to other
8316 When profile feedback is available (see @option{-fprofile-generate}) the actual
8317 recursion depth can be guessed from probability that function will recurse via
8318 given call expression. This parameter limits inlining only to call expression
8319 whose probability exceeds given threshold (in percents). The default value is
8322 @item early-inlining-insns
8323 Specify growth that early inliner can make. In effect it increases amount of
8324 inlining for code having large abstraction penalty. The default value is 10.
8326 @item max-early-inliner-iterations
8327 @itemx max-early-inliner-iterations
8328 Limit of iterations of early inliner. This basically bounds number of nested
8329 indirect calls early inliner can resolve. Deeper chains are still handled by
8332 @item comdat-sharing-probability
8333 @itemx comdat-sharing-probability
8334 Probability (in percent) that C++ inline function with comdat visibility
8335 will be shared acroess multiple compilation units. The default value is 20.
8337 @item min-vect-loop-bound
8338 The minimum number of iterations under which a loop will not get vectorized
8339 when @option{-ftree-vectorize} is used. The number of iterations after
8340 vectorization needs to be greater than the value specified by this option
8341 to allow vectorization. The default value is 0.
8343 @item gcse-cost-distance-ratio
8344 Scaling factor in calculation of maximum distance an expression
8345 can be moved by GCSE optimizations. This is currently supported only in
8346 code hoisting pass. The bigger the ratio, the more agressive code hoisting
8347 will be with simple expressions, i.e., the expressions which have cost
8348 less than @option{gcse-unrestricted-cost}. Specifying 0 will disable
8349 hoisting of simple expressions. The default value is 10.
8351 @item gcse-unrestricted-cost
8352 Cost, roughly measured as the cost of a single typical machine
8353 instruction, at which GCSE optimizations will not constrain
8354 the distance an expression can travel. This is currently
8355 supported only in code hoisting pass. The lesser the cost,
8356 the more aggressive code hoisting will be. Specifying 0 will
8357 allow all expressions to travel unrestricted distances.
8358 The default value is 3.
8360 @item max-hoist-depth
8361 The depth of search in the dominator tree for expressions to hoist.
8362 This is used to avoid quadratic behavior in hoisting algorithm.
8363 The value of 0 will avoid limiting the search, but may slow down compilation
8364 of huge functions. The default value is 30.
8366 @item max-unrolled-insns
8367 The maximum number of instructions that a loop should have if that loop
8368 is unrolled, and if the loop is unrolled, it determines how many times
8369 the loop code is unrolled.
8371 @item max-average-unrolled-insns
8372 The maximum number of instructions biased by probabilities of their execution
8373 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8374 it determines how many times the loop code is unrolled.
8376 @item max-unroll-times
8377 The maximum number of unrollings of a single loop.
8379 @item max-peeled-insns
8380 The maximum number of instructions that a loop should have if that loop
8381 is peeled, and if the loop is peeled, it determines how many times
8382 the loop code is peeled.
8384 @item max-peel-times
8385 The maximum number of peelings of a single loop.
8387 @item max-completely-peeled-insns
8388 The maximum number of insns of a completely peeled loop.
8390 @item max-completely-peel-times
8391 The maximum number of iterations of a loop to be suitable for complete peeling.
8393 @item max-completely-peel-loop-nest-depth
8394 The maximum depth of a loop nest suitable for complete peeling.
8396 @item max-unswitch-insns
8397 The maximum number of insns of an unswitched loop.
8399 @item max-unswitch-level
8400 The maximum number of branches unswitched in a single loop.
8403 The minimum cost of an expensive expression in the loop invariant motion.
8405 @item iv-consider-all-candidates-bound
8406 Bound on number of candidates for induction variables below that
8407 all candidates are considered for each use in induction variable
8408 optimizations. Only the most relevant candidates are considered
8409 if there are more candidates, to avoid quadratic time complexity.
8411 @item iv-max-considered-uses
8412 The induction variable optimizations give up on loops that contain more
8413 induction variable uses.
8415 @item iv-always-prune-cand-set-bound
8416 If number of candidates in the set is smaller than this value,
8417 we always try to remove unnecessary ivs from the set during its
8418 optimization when a new iv is added to the set.
8420 @item scev-max-expr-size
8421 Bound on size of expressions used in the scalar evolutions analyzer.
8422 Large expressions slow the analyzer.
8424 @item omega-max-vars
8425 The maximum number of variables in an Omega constraint system.
8426 The default value is 128.
8428 @item omega-max-geqs
8429 The maximum number of inequalities in an Omega constraint system.
8430 The default value is 256.
8433 The maximum number of equalities in an Omega constraint system.
8434 The default value is 128.
8436 @item omega-max-wild-cards
8437 The maximum number of wildcard variables that the Omega solver will
8438 be able to insert. The default value is 18.
8440 @item omega-hash-table-size
8441 The size of the hash table in the Omega solver. The default value is
8444 @item omega-max-keys
8445 The maximal number of keys used by the Omega solver. The default
8448 @item omega-eliminate-redundant-constraints
8449 When set to 1, use expensive methods to eliminate all redundant
8450 constraints. The default value is 0.
8452 @item vect-max-version-for-alignment-checks
8453 The maximum number of runtime checks that can be performed when
8454 doing loop versioning for alignment in the vectorizer. See option
8455 ftree-vect-loop-version for more information.
8457 @item vect-max-version-for-alias-checks
8458 The maximum number of runtime checks that can be performed when
8459 doing loop versioning for alias in the vectorizer. See option
8460 ftree-vect-loop-version for more information.
8462 @item max-iterations-to-track
8464 The maximum number of iterations of a loop the brute force algorithm
8465 for analysis of # of iterations of the loop tries to evaluate.
8467 @item hot-bb-count-fraction
8468 Select fraction of the maximal count of repetitions of basic block in program
8469 given basic block needs to have to be considered hot.
8471 @item hot-bb-frequency-fraction
8472 Select fraction of the maximal frequency of executions of basic block in
8473 function given basic block needs to have to be considered hot
8475 @item max-predicted-iterations
8476 The maximum number of loop iterations we predict statically. This is useful
8477 in cases where function contain single loop with known bound and other loop
8478 with unknown. We predict the known number of iterations correctly, while
8479 the unknown number of iterations average to roughly 10. This means that the
8480 loop without bounds would appear artificially cold relative to the other one.
8482 @item align-threshold
8484 Select fraction of the maximal frequency of executions of basic block in
8485 function given basic block will get aligned.
8487 @item align-loop-iterations
8489 A loop expected to iterate at lest the selected number of iterations will get
8492 @item tracer-dynamic-coverage
8493 @itemx tracer-dynamic-coverage-feedback
8495 This value is used to limit superblock formation once the given percentage of
8496 executed instructions is covered. This limits unnecessary code size
8499 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8500 feedback is available. The real profiles (as opposed to statically estimated
8501 ones) are much less balanced allowing the threshold to be larger value.
8503 @item tracer-max-code-growth
8504 Stop tail duplication once code growth has reached given percentage. This is
8505 rather hokey argument, as most of the duplicates will be eliminated later in
8506 cross jumping, so it may be set to much higher values than is the desired code
8509 @item tracer-min-branch-ratio
8511 Stop reverse growth when the reverse probability of best edge is less than this
8512 threshold (in percent).
8514 @item tracer-min-branch-ratio
8515 @itemx tracer-min-branch-ratio-feedback
8517 Stop forward growth if the best edge do have probability lower than this
8520 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8521 compilation for profile feedback and one for compilation without. The value
8522 for compilation with profile feedback needs to be more conservative (higher) in
8523 order to make tracer effective.
8525 @item max-cse-path-length
8527 Maximum number of basic blocks on path that cse considers. The default is 10.
8530 The maximum instructions CSE process before flushing. The default is 1000.
8532 @item ggc-min-expand
8534 GCC uses a garbage collector to manage its own memory allocation. This
8535 parameter specifies the minimum percentage by which the garbage
8536 collector's heap should be allowed to expand between collections.
8537 Tuning this may improve compilation speed; it has no effect on code
8540 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8541 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8542 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8543 GCC is not able to calculate RAM on a particular platform, the lower
8544 bound of 30% is used. Setting this parameter and
8545 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8546 every opportunity. This is extremely slow, but can be useful for
8549 @item ggc-min-heapsize
8551 Minimum size of the garbage collector's heap before it begins bothering
8552 to collect garbage. The first collection occurs after the heap expands
8553 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8554 tuning this may improve compilation speed, and has no effect on code
8557 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8558 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8559 with a lower bound of 4096 (four megabytes) and an upper bound of
8560 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8561 particular platform, the lower bound is used. Setting this parameter
8562 very large effectively disables garbage collection. Setting this
8563 parameter and @option{ggc-min-expand} to zero causes a full collection
8564 to occur at every opportunity.
8566 @item max-reload-search-insns
8567 The maximum number of instruction reload should look backward for equivalent
8568 register. Increasing values mean more aggressive optimization, making the
8569 compile time increase with probably slightly better performance. The default
8572 @item max-cselib-memory-locations
8573 The maximum number of memory locations cselib should take into account.
8574 Increasing values mean more aggressive optimization, making the compile time
8575 increase with probably slightly better performance. The default value is 500.
8577 @item reorder-blocks-duplicate
8578 @itemx reorder-blocks-duplicate-feedback
8580 Used by basic block reordering pass to decide whether to use unconditional
8581 branch or duplicate the code on its destination. Code is duplicated when its
8582 estimated size is smaller than this value multiplied by the estimated size of
8583 unconditional jump in the hot spots of the program.
8585 The @option{reorder-block-duplicate-feedback} is used only when profile
8586 feedback is available and may be set to higher values than
8587 @option{reorder-block-duplicate} since information about the hot spots is more
8590 @item max-sched-ready-insns
8591 The maximum number of instructions ready to be issued the scheduler should
8592 consider at any given time during the first scheduling pass. Increasing
8593 values mean more thorough searches, making the compilation time increase
8594 with probably little benefit. The default value is 100.
8596 @item max-sched-region-blocks
8597 The maximum number of blocks in a region to be considered for
8598 interblock scheduling. The default value is 10.
8600 @item max-pipeline-region-blocks
8601 The maximum number of blocks in a region to be considered for
8602 pipelining in the selective scheduler. The default value is 15.
8604 @item max-sched-region-insns
8605 The maximum number of insns in a region to be considered for
8606 interblock scheduling. The default value is 100.
8608 @item max-pipeline-region-insns
8609 The maximum number of insns in a region to be considered for
8610 pipelining in the selective scheduler. The default value is 200.
8613 The minimum probability (in percents) of reaching a source block
8614 for interblock speculative scheduling. The default value is 40.
8616 @item max-sched-extend-regions-iters
8617 The maximum number of iterations through CFG to extend regions.
8618 0 - disable region extension,
8619 N - do at most N iterations.
8620 The default value is 0.
8622 @item max-sched-insn-conflict-delay
8623 The maximum conflict delay for an insn to be considered for speculative motion.
8624 The default value is 3.
8626 @item sched-spec-prob-cutoff
8627 The minimal probability of speculation success (in percents), so that
8628 speculative insn will be scheduled.
8629 The default value is 40.
8631 @item sched-mem-true-dep-cost
8632 Minimal distance (in CPU cycles) between store and load targeting same
8633 memory locations. The default value is 1.
8635 @item selsched-max-lookahead
8636 The maximum size of the lookahead window of selective scheduling. It is a
8637 depth of search for available instructions.
8638 The default value is 50.
8640 @item selsched-max-sched-times
8641 The maximum number of times that an instruction will be scheduled during
8642 selective scheduling. This is the limit on the number of iterations
8643 through which the instruction may be pipelined. The default value is 2.
8645 @item selsched-max-insns-to-rename
8646 The maximum number of best instructions in the ready list that are considered
8647 for renaming in the selective scheduler. The default value is 2.
8649 @item max-last-value-rtl
8650 The maximum size measured as number of RTLs that can be recorded in an expression
8651 in combiner for a pseudo register as last known value of that register. The default
8654 @item integer-share-limit
8655 Small integer constants can use a shared data structure, reducing the
8656 compiler's memory usage and increasing its speed. This sets the maximum
8657 value of a shared integer constant. The default value is 256.
8659 @item min-virtual-mappings
8660 Specifies the minimum number of virtual mappings in the incremental
8661 SSA updater that should be registered to trigger the virtual mappings
8662 heuristic defined by virtual-mappings-ratio. The default value is
8665 @item virtual-mappings-ratio
8666 If the number of virtual mappings is virtual-mappings-ratio bigger
8667 than the number of virtual symbols to be updated, then the incremental
8668 SSA updater switches to a full update for those symbols. The default
8671 @item ssp-buffer-size
8672 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8673 protection when @option{-fstack-protection} is used.
8675 @item max-jump-thread-duplication-stmts
8676 Maximum number of statements allowed in a block that needs to be
8677 duplicated when threading jumps.
8679 @item max-fields-for-field-sensitive
8680 Maximum number of fields in a structure we will treat in
8681 a field sensitive manner during pointer analysis. The default is zero
8682 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8684 @item prefetch-latency
8685 Estimate on average number of instructions that are executed before
8686 prefetch finishes. The distance we prefetch ahead is proportional
8687 to this constant. Increasing this number may also lead to less
8688 streams being prefetched (see @option{simultaneous-prefetches}).
8690 @item simultaneous-prefetches
8691 Maximum number of prefetches that can run at the same time.
8693 @item l1-cache-line-size
8694 The size of cache line in L1 cache, in bytes.
8697 The size of L1 cache, in kilobytes.
8700 The size of L2 cache, in kilobytes.
8702 @item min-insn-to-prefetch-ratio
8703 The minimum ratio between the number of instructions and the
8704 number of prefetches to enable prefetching in a loop.
8706 @item prefetch-min-insn-to-mem-ratio
8707 The minimum ratio between the number of instructions and the
8708 number of memory references to enable prefetching in a loop.
8710 @item use-canonical-types
8711 Whether the compiler should use the ``canonical'' type system. By
8712 default, this should always be 1, which uses a more efficient internal
8713 mechanism for comparing types in C++ and Objective-C++. However, if
8714 bugs in the canonical type system are causing compilation failures,
8715 set this value to 0 to disable canonical types.
8717 @item switch-conversion-max-branch-ratio
8718 Switch initialization conversion will refuse to create arrays that are
8719 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8720 branches in the switch.
8722 @item max-partial-antic-length
8723 Maximum length of the partial antic set computed during the tree
8724 partial redundancy elimination optimization (@option{-ftree-pre}) when
8725 optimizing at @option{-O3} and above. For some sorts of source code
8726 the enhanced partial redundancy elimination optimization can run away,
8727 consuming all of the memory available on the host machine. This
8728 parameter sets a limit on the length of the sets that are computed,
8729 which prevents the runaway behavior. Setting a value of 0 for
8730 this parameter will allow an unlimited set length.
8732 @item sccvn-max-scc-size
8733 Maximum size of a strongly connected component (SCC) during SCCVN
8734 processing. If this limit is hit, SCCVN processing for the whole
8735 function will not be done and optimizations depending on it will
8736 be disabled. The default maximum SCC size is 10000.
8738 @item ira-max-loops-num
8739 IRA uses a regional register allocation by default. If a function
8740 contains loops more than number given by the parameter, only at most
8741 given number of the most frequently executed loops will form regions
8742 for the regional register allocation. The default value of the
8745 @item ira-max-conflict-table-size
8746 Although IRA uses a sophisticated algorithm of compression conflict
8747 table, the table can be still big for huge functions. If the conflict
8748 table for a function could be more than size in MB given by the
8749 parameter, the conflict table is not built and faster, simpler, and
8750 lower quality register allocation algorithm will be used. The
8751 algorithm do not use pseudo-register conflicts. The default value of
8752 the parameter is 2000.
8754 @item ira-loop-reserved-regs
8755 IRA can be used to evaluate more accurate register pressure in loops
8756 for decision to move loop invariants (see @option{-O3}). The number
8757 of available registers reserved for some other purposes is described
8758 by this parameter. The default value of the parameter is 2 which is
8759 minimal number of registers needed for execution of typical
8760 instruction. This value is the best found from numerous experiments.
8762 @item loop-invariant-max-bbs-in-loop
8763 Loop invariant motion can be very expensive, both in compile time and
8764 in amount of needed compile time memory, with very large loops. Loops
8765 with more basic blocks than this parameter won't have loop invariant
8766 motion optimization performed on them. The default value of the
8767 parameter is 1000 for -O1 and 10000 for -O2 and above.
8769 @item max-vartrack-size
8770 Sets a maximum number of hash table slots to use during variable
8771 tracking dataflow analysis of any function. If this limit is exceeded
8772 with variable tracking at assignments enabled, analysis for that
8773 function is retried without it, after removing all debug insns from
8774 the function. If the limit is exceeded even without debug insns, var
8775 tracking analysis is completely disabled for the function. Setting
8776 the parameter to zero makes it unlimited.
8778 @item min-nondebug-insn-uid
8779 Use uids starting at this parameter for nondebug insns. The range below
8780 the parameter is reserved exclusively for debug insns created by
8781 @option{-fvar-tracking-assignments}, but debug insns may get
8782 (non-overlapping) uids above it if the reserved range is exhausted.
8784 @item ipa-sra-ptr-growth-factor
8785 IPA-SRA will replace a pointer to an aggregate with one or more new
8786 parameters only when their cumulative size is less or equal to
8787 @option{ipa-sra-ptr-growth-factor} times the size of the original
8790 @item graphite-max-nb-scop-params
8791 To avoid exponential effects in the Graphite loop transforms, the
8792 number of parameters in a Static Control Part (SCoP) is bounded. The
8793 default value is 10 parameters. A variable whose value is unknown at
8794 compile time and defined outside a SCoP is a parameter of the SCoP.
8796 @item graphite-max-bbs-per-function
8797 To avoid exponential effects in the detection of SCoPs, the size of
8798 the functions analyzed by Graphite is bounded. The default value is
8801 @item loop-block-tile-size
8802 Loop blocking or strip mining transforms, enabled with
8803 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
8804 loop in the loop nest by a given number of iterations. The strip
8805 length can be changed using the @option{loop-block-tile-size}
8806 parameter. The default value is 51 iterations.
8808 @item devirt-type-list-size
8809 IPA-CP attempts to track all possible types passed to a function's
8810 parameter in order to perform devirtualization.
8811 @option{devirt-type-list-size} is the maximum number of types it
8812 stores per a single formal parameter of a function.
8814 @item lto-partitions
8815 Specify desired nuber of partitions produced during WHOPR copmilation.
8816 Number of partitions should exceed number of CPUs used for compilatoin.
8817 Default value is 32.
8819 @item lto-minpartition
8820 Size of minimal paritition for WHOPR (in estimated instructions).
8821 This prevents expenses of splitting very small programs into too many
8827 @node Preprocessor Options
8828 @section Options Controlling the Preprocessor
8829 @cindex preprocessor options
8830 @cindex options, preprocessor
8832 These options control the C preprocessor, which is run on each C source
8833 file before actual compilation.
8835 If you use the @option{-E} option, nothing is done except preprocessing.
8836 Some of these options make sense only together with @option{-E} because
8837 they cause the preprocessor output to be unsuitable for actual
8841 @item -Wp,@var{option}
8843 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8844 and pass @var{option} directly through to the preprocessor. If
8845 @var{option} contains commas, it is split into multiple options at the
8846 commas. However, many options are modified, translated or interpreted
8847 by the compiler driver before being passed to the preprocessor, and
8848 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8849 interface is undocumented and subject to change, so whenever possible
8850 you should avoid using @option{-Wp} and let the driver handle the
8853 @item -Xpreprocessor @var{option}
8854 @opindex Xpreprocessor
8855 Pass @var{option} as an option to the preprocessor. You can use this to
8856 supply system-specific preprocessor options which GCC does not know how to
8859 If you want to pass an option that takes an argument, you must use
8860 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8863 @include cppopts.texi
8865 @node Assembler Options
8866 @section Passing Options to the Assembler
8868 @c prevent bad page break with this line
8869 You can pass options to the assembler.
8872 @item -Wa,@var{option}
8874 Pass @var{option} as an option to the assembler. If @var{option}
8875 contains commas, it is split into multiple options at the commas.
8877 @item -Xassembler @var{option}
8879 Pass @var{option} as an option to the assembler. You can use this to
8880 supply system-specific assembler options which GCC does not know how to
8883 If you want to pass an option that takes an argument, you must use
8884 @option{-Xassembler} twice, once for the option and once for the argument.
8889 @section Options for Linking
8890 @cindex link options
8891 @cindex options, linking
8893 These options come into play when the compiler links object files into
8894 an executable output file. They are meaningless if the compiler is
8895 not doing a link step.
8899 @item @var{object-file-name}
8900 A file name that does not end in a special recognized suffix is
8901 considered to name an object file or library. (Object files are
8902 distinguished from libraries by the linker according to the file
8903 contents.) If linking is done, these object files are used as input
8912 If any of these options is used, then the linker is not run, and
8913 object file names should not be used as arguments. @xref{Overall
8917 @item -l@var{library}
8918 @itemx -l @var{library}
8920 Search the library named @var{library} when linking. (The second
8921 alternative with the library as a separate argument is only for
8922 POSIX compliance and is not recommended.)
8924 It makes a difference where in the command you write this option; the
8925 linker searches and processes libraries and object files in the order they
8926 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8927 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8928 to functions in @samp{z}, those functions may not be loaded.
8930 The linker searches a standard list of directories for the library,
8931 which is actually a file named @file{lib@var{library}.a}. The linker
8932 then uses this file as if it had been specified precisely by name.
8934 The directories searched include several standard system directories
8935 plus any that you specify with @option{-L}.
8937 Normally the files found this way are library files---archive files
8938 whose members are object files. The linker handles an archive file by
8939 scanning through it for members which define symbols that have so far
8940 been referenced but not defined. But if the file that is found is an
8941 ordinary object file, it is linked in the usual fashion. The only
8942 difference between using an @option{-l} option and specifying a file name
8943 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8944 and searches several directories.
8948 You need this special case of the @option{-l} option in order to
8949 link an Objective-C or Objective-C++ program.
8952 @opindex nostartfiles
8953 Do not use the standard system startup files when linking.
8954 The standard system libraries are used normally, unless @option{-nostdlib}
8955 or @option{-nodefaultlibs} is used.
8957 @item -nodefaultlibs
8958 @opindex nodefaultlibs
8959 Do not use the standard system libraries when linking.
8960 Only the libraries you specify will be passed to the linker, options
8961 specifying linkage of the system libraries, such as @code{-static-libgcc}
8962 or @code{-shared-libgcc}, will be ignored.
8963 The standard startup files are used normally, unless @option{-nostartfiles}
8964 is used. The compiler may generate calls to @code{memcmp},
8965 @code{memset}, @code{memcpy} and @code{memmove}.
8966 These entries are usually resolved by entries in
8967 libc. These entry points should be supplied through some other
8968 mechanism when this option is specified.
8972 Do not use the standard system startup files or libraries when linking.
8973 No startup files and only the libraries you specify will be passed to
8974 the linker, options specifying linkage of the system libraries, such as
8975 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8976 The compiler may generate calls to @code{memcmp}, @code{memset},
8977 @code{memcpy} and @code{memmove}.
8978 These entries are usually resolved by entries in
8979 libc. These entry points should be supplied through some other
8980 mechanism when this option is specified.
8982 @cindex @option{-lgcc}, use with @option{-nostdlib}
8983 @cindex @option{-nostdlib} and unresolved references
8984 @cindex unresolved references and @option{-nostdlib}
8985 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8986 @cindex @option{-nodefaultlibs} and unresolved references
8987 @cindex unresolved references and @option{-nodefaultlibs}
8988 One of the standard libraries bypassed by @option{-nostdlib} and
8989 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8990 that GCC uses to overcome shortcomings of particular machines, or special
8991 needs for some languages.
8992 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8993 Collection (GCC) Internals},
8994 for more discussion of @file{libgcc.a}.)
8995 In most cases, you need @file{libgcc.a} even when you want to avoid
8996 other standard libraries. In other words, when you specify @option{-nostdlib}
8997 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8998 This ensures that you have no unresolved references to internal GCC
8999 library subroutines. (For example, @samp{__main}, used to ensure C++
9000 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
9001 GNU Compiler Collection (GCC) Internals}.)
9005 Produce a position independent executable on targets which support it.
9006 For predictable results, you must also specify the same set of options
9007 that were used to generate code (@option{-fpie}, @option{-fPIE},
9008 or model suboptions) when you specify this option.
9012 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
9013 that support it. This instructs the linker to add all symbols, not
9014 only used ones, to the dynamic symbol table. This option is needed
9015 for some uses of @code{dlopen} or to allow obtaining backtraces
9016 from within a program.
9020 Remove all symbol table and relocation information from the executable.
9024 On systems that support dynamic linking, this prevents linking with the shared
9025 libraries. On other systems, this option has no effect.
9029 Produce a shared object which can then be linked with other objects to
9030 form an executable. Not all systems support this option. For predictable
9031 results, you must also specify the same set of options that were used to
9032 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
9033 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
9034 needs to build supplementary stub code for constructors to work. On
9035 multi-libbed systems, @samp{gcc -shared} must select the correct support
9036 libraries to link against. Failing to supply the correct flags may lead
9037 to subtle defects. Supplying them in cases where they are not necessary
9040 @item -shared-libgcc
9041 @itemx -static-libgcc
9042 @opindex shared-libgcc
9043 @opindex static-libgcc
9044 On systems that provide @file{libgcc} as a shared library, these options
9045 force the use of either the shared or static version respectively.
9046 If no shared version of @file{libgcc} was built when the compiler was
9047 configured, these options have no effect.
9049 There are several situations in which an application should use the
9050 shared @file{libgcc} instead of the static version. The most common
9051 of these is when the application wishes to throw and catch exceptions
9052 across different shared libraries. In that case, each of the libraries
9053 as well as the application itself should use the shared @file{libgcc}.
9055 Therefore, the G++ and GCJ drivers automatically add
9056 @option{-shared-libgcc} whenever you build a shared library or a main
9057 executable, because C++ and Java programs typically use exceptions, so
9058 this is the right thing to do.
9060 If, instead, you use the GCC driver to create shared libraries, you may
9061 find that they will not always be linked with the shared @file{libgcc}.
9062 If GCC finds, at its configuration time, that you have a non-GNU linker
9063 or a GNU linker that does not support option @option{--eh-frame-hdr},
9064 it will link the shared version of @file{libgcc} into shared libraries
9065 by default. Otherwise, it will take advantage of the linker and optimize
9066 away the linking with the shared version of @file{libgcc}, linking with
9067 the static version of libgcc by default. This allows exceptions to
9068 propagate through such shared libraries, without incurring relocation
9069 costs at library load time.
9071 However, if a library or main executable is supposed to throw or catch
9072 exceptions, you must link it using the G++ or GCJ driver, as appropriate
9073 for the languages used in the program, or using the option
9074 @option{-shared-libgcc}, such that it is linked with the shared
9077 @item -static-libstdc++
9078 When the @command{g++} program is used to link a C++ program, it will
9079 normally automatically link against @option{libstdc++}. If
9080 @file{libstdc++} is available as a shared library, and the
9081 @option{-static} option is not used, then this will link against the
9082 shared version of @file{libstdc++}. That is normally fine. However, it
9083 is sometimes useful to freeze the version of @file{libstdc++} used by
9084 the program without going all the way to a fully static link. The
9085 @option{-static-libstdc++} option directs the @command{g++} driver to
9086 link @file{libstdc++} statically, without necessarily linking other
9087 libraries statically.
9091 Bind references to global symbols when building a shared object. Warn
9092 about any unresolved references (unless overridden by the link editor
9093 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
9096 @item -T @var{script}
9098 @cindex linker script
9099 Use @var{script} as the linker script. This option is supported by most
9100 systems using the GNU linker. On some targets, such as bare-board
9101 targets without an operating system, the @option{-T} option may be required
9102 when linking to avoid references to undefined symbols.
9104 @item -Xlinker @var{option}
9106 Pass @var{option} as an option to the linker. You can use this to
9107 supply system-specific linker options which GCC does not know how to
9110 If you want to pass an option that takes a separate argument, you must use
9111 @option{-Xlinker} twice, once for the option and once for the argument.
9112 For example, to pass @option{-assert definitions}, you must write
9113 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
9114 @option{-Xlinker "-assert definitions"}, because this passes the entire
9115 string as a single argument, which is not what the linker expects.
9117 When using the GNU linker, it is usually more convenient to pass
9118 arguments to linker options using the @option{@var{option}=@var{value}}
9119 syntax than as separate arguments. For example, you can specify
9120 @samp{-Xlinker -Map=output.map} rather than
9121 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
9122 this syntax for command-line options.
9124 @item -Wl,@var{option}
9126 Pass @var{option} as an option to the linker. If @var{option} contains
9127 commas, it is split into multiple options at the commas. You can use this
9128 syntax to pass an argument to the option.
9129 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
9130 linker. When using the GNU linker, you can also get the same effect with
9131 @samp{-Wl,-Map=output.map}.
9133 @item -u @var{symbol}
9135 Pretend the symbol @var{symbol} is undefined, to force linking of
9136 library modules to define it. You can use @option{-u} multiple times with
9137 different symbols to force loading of additional library modules.
9140 @node Directory Options
9141 @section Options for Directory Search
9142 @cindex directory options
9143 @cindex options, directory search
9146 These options specify directories to search for header files, for
9147 libraries and for parts of the compiler:
9152 Add the directory @var{dir} to the head of the list of directories to be
9153 searched for header files. This can be used to override a system header
9154 file, substituting your own version, since these directories are
9155 searched before the system header file directories. However, you should
9156 not use this option to add directories that contain vendor-supplied
9157 system header files (use @option{-isystem} for that). If you use more than
9158 one @option{-I} option, the directories are scanned in left-to-right
9159 order; the standard system directories come after.
9161 If a standard system include directory, or a directory specified with
9162 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
9163 option will be ignored. The directory will still be searched but as a
9164 system directory at its normal position in the system include chain.
9165 This is to ensure that GCC's procedure to fix buggy system headers and
9166 the ordering for the include_next directive are not inadvertently changed.
9167 If you really need to change the search order for system directories,
9168 use the @option{-nostdinc} and/or @option{-isystem} options.
9170 @item -iplugindir=@var{dir}
9171 Set the directory to search for plugins which are passed
9172 by @option{-fplugin=@var{name}} instead of
9173 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
9174 to be used by the user, but only passed by the driver.
9176 @item -iquote@var{dir}
9178 Add the directory @var{dir} to the head of the list of directories to
9179 be searched for header files only for the case of @samp{#include
9180 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
9181 otherwise just like @option{-I}.
9185 Add directory @var{dir} to the list of directories to be searched
9188 @item -B@var{prefix}
9190 This option specifies where to find the executables, libraries,
9191 include files, and data files of the compiler itself.
9193 The compiler driver program runs one or more of the subprograms
9194 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
9195 @var{prefix} as a prefix for each program it tries to run, both with and
9196 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
9198 For each subprogram to be run, the compiler driver first tries the
9199 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
9200 was not specified, the driver tries two standard prefixes, which are
9201 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
9202 those results in a file name that is found, the unmodified program
9203 name is searched for using the directories specified in your
9204 @env{PATH} environment variable.
9206 The compiler will check to see if the path provided by the @option{-B}
9207 refers to a directory, and if necessary it will add a directory
9208 separator character at the end of the path.
9210 @option{-B} prefixes that effectively specify directory names also apply
9211 to libraries in the linker, because the compiler translates these
9212 options into @option{-L} options for the linker. They also apply to
9213 includes files in the preprocessor, because the compiler translates these
9214 options into @option{-isystem} options for the preprocessor. In this case,
9215 the compiler appends @samp{include} to the prefix.
9217 The run-time support file @file{libgcc.a} can also be searched for using
9218 the @option{-B} prefix, if needed. If it is not found there, the two
9219 standard prefixes above are tried, and that is all. The file is left
9220 out of the link if it is not found by those means.
9222 Another way to specify a prefix much like the @option{-B} prefix is to use
9223 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
9226 As a special kludge, if the path provided by @option{-B} is
9227 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
9228 9, then it will be replaced by @file{[dir/]include}. This is to help
9229 with boot-strapping the compiler.
9231 @item -specs=@var{file}
9233 Process @var{file} after the compiler reads in the standard @file{specs}
9234 file, in order to override the defaults that the @file{gcc} driver
9235 program uses when determining what switches to pass to @file{cc1},
9236 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
9237 @option{-specs=@var{file}} can be specified on the command line, and they
9238 are processed in order, from left to right.
9240 @item --sysroot=@var{dir}
9242 Use @var{dir} as the logical root directory for headers and libraries.
9243 For example, if the compiler would normally search for headers in
9244 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
9245 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
9247 If you use both this option and the @option{-isysroot} option, then
9248 the @option{--sysroot} option will apply to libraries, but the
9249 @option{-isysroot} option will apply to header files.
9251 The GNU linker (beginning with version 2.16) has the necessary support
9252 for this option. If your linker does not support this option, the
9253 header file aspect of @option{--sysroot} will still work, but the
9254 library aspect will not.
9258 This option has been deprecated. Please use @option{-iquote} instead for
9259 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
9260 Any directories you specify with @option{-I} options before the @option{-I-}
9261 option are searched only for the case of @samp{#include "@var{file}"};
9262 they are not searched for @samp{#include <@var{file}>}.
9264 If additional directories are specified with @option{-I} options after
9265 the @option{-I-}, these directories are searched for all @samp{#include}
9266 directives. (Ordinarily @emph{all} @option{-I} directories are used
9269 In addition, the @option{-I-} option inhibits the use of the current
9270 directory (where the current input file came from) as the first search
9271 directory for @samp{#include "@var{file}"}. There is no way to
9272 override this effect of @option{-I-}. With @option{-I.} you can specify
9273 searching the directory which was current when the compiler was
9274 invoked. That is not exactly the same as what the preprocessor does
9275 by default, but it is often satisfactory.
9277 @option{-I-} does not inhibit the use of the standard system directories
9278 for header files. Thus, @option{-I-} and @option{-nostdinc} are
9285 @section Specifying subprocesses and the switches to pass to them
9288 @command{gcc} is a driver program. It performs its job by invoking a
9289 sequence of other programs to do the work of compiling, assembling and
9290 linking. GCC interprets its command-line parameters and uses these to
9291 deduce which programs it should invoke, and which command-line options
9292 it ought to place on their command lines. This behavior is controlled
9293 by @dfn{spec strings}. In most cases there is one spec string for each
9294 program that GCC can invoke, but a few programs have multiple spec
9295 strings to control their behavior. The spec strings built into GCC can
9296 be overridden by using the @option{-specs=} command-line switch to specify
9299 @dfn{Spec files} are plaintext files that are used to construct spec
9300 strings. They consist of a sequence of directives separated by blank
9301 lines. The type of directive is determined by the first non-whitespace
9302 character on the line and it can be one of the following:
9305 @item %@var{command}
9306 Issues a @var{command} to the spec file processor. The commands that can
9310 @item %include <@var{file}>
9311 @cindex @code{%include}
9312 Search for @var{file} and insert its text at the current point in the
9315 @item %include_noerr <@var{file}>
9316 @cindex @code{%include_noerr}
9317 Just like @samp{%include}, but do not generate an error message if the include
9318 file cannot be found.
9320 @item %rename @var{old_name} @var{new_name}
9321 @cindex @code{%rename}
9322 Rename the spec string @var{old_name} to @var{new_name}.
9326 @item *[@var{spec_name}]:
9327 This tells the compiler to create, override or delete the named spec
9328 string. All lines after this directive up to the next directive or
9329 blank line are considered to be the text for the spec string. If this
9330 results in an empty string then the spec will be deleted. (Or, if the
9331 spec did not exist, then nothing will happened.) Otherwise, if the spec
9332 does not currently exist a new spec will be created. If the spec does
9333 exist then its contents will be overridden by the text of this
9334 directive, unless the first character of that text is the @samp{+}
9335 character, in which case the text will be appended to the spec.
9337 @item [@var{suffix}]:
9338 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
9339 and up to the next directive or blank line are considered to make up the
9340 spec string for the indicated suffix. When the compiler encounters an
9341 input file with the named suffix, it will processes the spec string in
9342 order to work out how to compile that file. For example:
9349 This says that any input file whose name ends in @samp{.ZZ} should be
9350 passed to the program @samp{z-compile}, which should be invoked with the
9351 command-line switch @option{-input} and with the result of performing the
9352 @samp{%i} substitution. (See below.)
9354 As an alternative to providing a spec string, the text that follows a
9355 suffix directive can be one of the following:
9358 @item @@@var{language}
9359 This says that the suffix is an alias for a known @var{language}. This is
9360 similar to using the @option{-x} command-line switch to GCC to specify a
9361 language explicitly. For example:
9368 Says that .ZZ files are, in fact, C++ source files.
9371 This causes an error messages saying:
9374 @var{name} compiler not installed on this system.
9378 GCC already has an extensive list of suffixes built into it.
9379 This directive will add an entry to the end of the list of suffixes, but
9380 since the list is searched from the end backwards, it is effectively
9381 possible to override earlier entries using this technique.
9385 GCC has the following spec strings built into it. Spec files can
9386 override these strings or create their own. Note that individual
9387 targets can also add their own spec strings to this list.
9390 asm Options to pass to the assembler
9391 asm_final Options to pass to the assembler post-processor
9392 cpp Options to pass to the C preprocessor
9393 cc1 Options to pass to the C compiler
9394 cc1plus Options to pass to the C++ compiler
9395 endfile Object files to include at the end of the link
9396 link Options to pass to the linker
9397 lib Libraries to include on the command line to the linker
9398 libgcc Decides which GCC support library to pass to the linker
9399 linker Sets the name of the linker
9400 predefines Defines to be passed to the C preprocessor
9401 signed_char Defines to pass to CPP to say whether @code{char} is signed
9403 startfile Object files to include at the start of the link
9406 Here is a small example of a spec file:
9412 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9415 This example renames the spec called @samp{lib} to @samp{old_lib} and
9416 then overrides the previous definition of @samp{lib} with a new one.
9417 The new definition adds in some extra command-line options before
9418 including the text of the old definition.
9420 @dfn{Spec strings} are a list of command-line options to be passed to their
9421 corresponding program. In addition, the spec strings can contain
9422 @samp{%}-prefixed sequences to substitute variable text or to
9423 conditionally insert text into the command line. Using these constructs
9424 it is possible to generate quite complex command lines.
9426 Here is a table of all defined @samp{%}-sequences for spec
9427 strings. Note that spaces are not generated automatically around the
9428 results of expanding these sequences. Therefore you can concatenate them
9429 together or combine them with constant text in a single argument.
9433 Substitute one @samp{%} into the program name or argument.
9436 Substitute the name of the input file being processed.
9439 Substitute the basename of the input file being processed.
9440 This is the substring up to (and not including) the last period
9441 and not including the directory.
9444 This is the same as @samp{%b}, but include the file suffix (text after
9448 Marks the argument containing or following the @samp{%d} as a
9449 temporary file name, so that that file will be deleted if GCC exits
9450 successfully. Unlike @samp{%g}, this contributes no text to the
9453 @item %g@var{suffix}
9454 Substitute a file name that has suffix @var{suffix} and is chosen
9455 once per compilation, and mark the argument in the same way as
9456 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9457 name is now chosen in a way that is hard to predict even when previously
9458 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9459 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9460 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9461 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9462 was simply substituted with a file name chosen once per compilation,
9463 without regard to any appended suffix (which was therefore treated
9464 just like ordinary text), making such attacks more likely to succeed.
9466 @item %u@var{suffix}
9467 Like @samp{%g}, but generates a new temporary file name even if
9468 @samp{%u@var{suffix}} was already seen.
9470 @item %U@var{suffix}
9471 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9472 new one if there is no such last file name. In the absence of any
9473 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9474 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9475 would involve the generation of two distinct file names, one
9476 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9477 simply substituted with a file name chosen for the previous @samp{%u},
9478 without regard to any appended suffix.
9480 @item %j@var{suffix}
9481 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9482 writable, and if save-temps is off; otherwise, substitute the name
9483 of a temporary file, just like @samp{%u}. This temporary file is not
9484 meant for communication between processes, but rather as a junk
9487 @item %|@var{suffix}
9488 @itemx %m@var{suffix}
9489 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9490 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9491 all. These are the two most common ways to instruct a program that it
9492 should read from standard input or write to standard output. If you
9493 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9494 construct: see for example @file{f/lang-specs.h}.
9496 @item %.@var{SUFFIX}
9497 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9498 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9499 terminated by the next space or %.
9502 Marks the argument containing or following the @samp{%w} as the
9503 designated output file of this compilation. This puts the argument
9504 into the sequence of arguments that @samp{%o} will substitute later.
9507 Substitutes the names of all the output files, with spaces
9508 automatically placed around them. You should write spaces
9509 around the @samp{%o} as well or the results are undefined.
9510 @samp{%o} is for use in the specs for running the linker.
9511 Input files whose names have no recognized suffix are not compiled
9512 at all, but they are included among the output files, so they will
9516 Substitutes the suffix for object files. Note that this is
9517 handled specially when it immediately follows @samp{%g, %u, or %U},
9518 because of the need for those to form complete file names. The
9519 handling is such that @samp{%O} is treated exactly as if it had already
9520 been substituted, except that @samp{%g, %u, and %U} do not currently
9521 support additional @var{suffix} characters following @samp{%O} as they would
9522 following, for example, @samp{.o}.
9525 Substitutes the standard macro predefinitions for the
9526 current target machine. Use this when running @code{cpp}.
9529 Like @samp{%p}, but puts @samp{__} before and after the name of each
9530 predefined macro, except for macros that start with @samp{__} or with
9531 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9535 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9536 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9537 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9538 and @option{-imultilib} as necessary.
9541 Current argument is the name of a library or startup file of some sort.
9542 Search for that file in a standard list of directories and substitute
9543 the full name found. The current working directory is included in the
9544 list of directories scanned.
9547 Current argument is the name of a linker script. Search for that file
9548 in the current list of directories to scan for libraries. If the file
9549 is located insert a @option{--script} option into the command line
9550 followed by the full path name found. If the file is not found then
9551 generate an error message. Note: the current working directory is not
9555 Print @var{str} as an error message. @var{str} is terminated by a newline.
9556 Use this when inconsistent options are detected.
9559 Substitute the contents of spec string @var{name} at this point.
9562 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9564 @item %x@{@var{option}@}
9565 Accumulate an option for @samp{%X}.
9568 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9572 Output the accumulated assembler options specified by @option{-Wa}.
9575 Output the accumulated preprocessor options specified by @option{-Wp}.
9578 Process the @code{asm} spec. This is used to compute the
9579 switches to be passed to the assembler.
9582 Process the @code{asm_final} spec. This is a spec string for
9583 passing switches to an assembler post-processor, if such a program is
9587 Process the @code{link} spec. This is the spec for computing the
9588 command line passed to the linker. Typically it will make use of the
9589 @samp{%L %G %S %D and %E} sequences.
9592 Dump out a @option{-L} option for each directory that GCC believes might
9593 contain startup files. If the target supports multilibs then the
9594 current multilib directory will be prepended to each of these paths.
9597 Process the @code{lib} spec. This is a spec string for deciding which
9598 libraries should be included on the command line to the linker.
9601 Process the @code{libgcc} spec. This is a spec string for deciding
9602 which GCC support library should be included on the command line to the linker.
9605 Process the @code{startfile} spec. This is a spec for deciding which
9606 object files should be the first ones passed to the linker. Typically
9607 this might be a file named @file{crt0.o}.
9610 Process the @code{endfile} spec. This is a spec string that specifies
9611 the last object files that will be passed to the linker.
9614 Process the @code{cpp} spec. This is used to construct the arguments
9615 to be passed to the C preprocessor.
9618 Process the @code{cc1} spec. This is used to construct the options to be
9619 passed to the actual C compiler (@samp{cc1}).
9622 Process the @code{cc1plus} spec. This is used to construct the options to be
9623 passed to the actual C++ compiler (@samp{cc1plus}).
9626 Substitute the variable part of a matched option. See below.
9627 Note that each comma in the substituted string is replaced by
9631 Remove all occurrences of @code{-S} from the command line. Note---this
9632 command is position dependent. @samp{%} commands in the spec string
9633 before this one will see @code{-S}, @samp{%} commands in the spec string
9634 after this one will not.
9636 @item %:@var{function}(@var{args})
9637 Call the named function @var{function}, passing it @var{args}.
9638 @var{args} is first processed as a nested spec string, then split
9639 into an argument vector in the usual fashion. The function returns
9640 a string which is processed as if it had appeared literally as part
9641 of the current spec.
9643 The following built-in spec functions are provided:
9647 The @code{getenv} spec function takes two arguments: an environment
9648 variable name and a string. If the environment variable is not
9649 defined, a fatal error is issued. Otherwise, the return value is the
9650 value of the environment variable concatenated with the string. For
9651 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9654 %:getenv(TOPDIR /include)
9657 expands to @file{/path/to/top/include}.
9659 @item @code{if-exists}
9660 The @code{if-exists} spec function takes one argument, an absolute
9661 pathname to a file. If the file exists, @code{if-exists} returns the
9662 pathname. Here is a small example of its usage:
9666 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9669 @item @code{if-exists-else}
9670 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9671 spec function, except that it takes two arguments. The first argument is
9672 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9673 returns the pathname. If it does not exist, it returns the second argument.
9674 This way, @code{if-exists-else} can be used to select one file or another,
9675 based on the existence of the first. Here is a small example of its usage:
9679 crt0%O%s %:if-exists(crti%O%s) \
9680 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9683 @item @code{replace-outfile}
9684 The @code{replace-outfile} spec function takes two arguments. It looks for the
9685 first argument in the outfiles array and replaces it with the second argument. Here
9686 is a small example of its usage:
9689 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9692 @item @code{remove-outfile}
9693 The @code{remove-outfile} spec function takes one argument. It looks for the
9694 first argument in the outfiles array and removes it. Here is a small example
9698 %:remove-outfile(-lm)
9701 @item @code{pass-through-libs}
9702 The @code{pass-through-libs} spec function takes any number of arguments. It
9703 finds any @option{-l} options and any non-options ending in ".a" (which it
9704 assumes are the names of linker input library archive files) and returns a
9705 result containing all the found arguments each prepended by
9706 @option{-plugin-opt=-pass-through=} and joined by spaces. This list is
9707 intended to be passed to the LTO linker plugin.
9710 %:pass-through-libs(%G %L %G)
9713 @item @code{print-asm-header}
9714 The @code{print-asm-header} function takes no arguments and simply
9715 prints a banner like:
9721 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9724 It is used to separate compiler options from assembler options
9725 in the @option{--target-help} output.
9729 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9730 If that switch was not specified, this substitutes nothing. Note that
9731 the leading dash is omitted when specifying this option, and it is
9732 automatically inserted if the substitution is performed. Thus the spec
9733 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9734 and would output the command line option @option{-foo}.
9736 @item %W@{@code{S}@}
9737 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9740 @item %@{@code{S}*@}
9741 Substitutes all the switches specified to GCC whose names start
9742 with @code{-S}, but which also take an argument. This is used for
9743 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9744 GCC considers @option{-o foo} as being
9745 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9746 text, including the space. Thus two arguments would be generated.
9748 @item %@{@code{S}*&@code{T}*@}
9749 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9750 (the order of @code{S} and @code{T} in the spec is not significant).
9751 There can be any number of ampersand-separated variables; for each the
9752 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9754 @item %@{@code{S}:@code{X}@}
9755 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9757 @item %@{!@code{S}:@code{X}@}
9758 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9760 @item %@{@code{S}*:@code{X}@}
9761 Substitutes @code{X} if one or more switches whose names start with
9762 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9763 once, no matter how many such switches appeared. However, if @code{%*}
9764 appears somewhere in @code{X}, then @code{X} will be substituted once
9765 for each matching switch, with the @code{%*} replaced by the part of
9766 that switch that matched the @code{*}.
9768 @item %@{.@code{S}:@code{X}@}
9769 Substitutes @code{X}, if processing a file with suffix @code{S}.
9771 @item %@{!.@code{S}:@code{X}@}
9772 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9774 @item %@{,@code{S}:@code{X}@}
9775 Substitutes @code{X}, if processing a file for language @code{S}.
9777 @item %@{!,@code{S}:@code{X}@}
9778 Substitutes @code{X}, if not processing a file for language @code{S}.
9780 @item %@{@code{S}|@code{P}:@code{X}@}
9781 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9782 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9783 @code{*} sequences as well, although they have a stronger binding than
9784 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9785 alternatives must be starred, and only the first matching alternative
9788 For example, a spec string like this:
9791 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9794 will output the following command-line options from the following input
9795 command-line options:
9800 -d fred.c -foo -baz -boggle
9801 -d jim.d -bar -baz -boggle
9804 @item %@{S:X; T:Y; :D@}
9806 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9807 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9808 be as many clauses as you need. This may be combined with @code{.},
9809 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9814 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9815 construct may contain other nested @samp{%} constructs or spaces, or
9816 even newlines. They are processed as usual, as described above.
9817 Trailing white space in @code{X} is ignored. White space may also
9818 appear anywhere on the left side of the colon in these constructs,
9819 except between @code{.} or @code{*} and the corresponding word.
9821 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9822 handled specifically in these constructs. If another value of
9823 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9824 @option{-W} switch is found later in the command line, the earlier
9825 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9826 just one letter, which passes all matching options.
9828 The character @samp{|} at the beginning of the predicate text is used to
9829 indicate that a command should be piped to the following command, but
9830 only if @option{-pipe} is specified.
9832 It is built into GCC which switches take arguments and which do not.
9833 (You might think it would be useful to generalize this to allow each
9834 compiler's spec to say which switches take arguments. But this cannot
9835 be done in a consistent fashion. GCC cannot even decide which input
9836 files have been specified without knowing which switches take arguments,
9837 and it must know which input files to compile in order to tell which
9840 GCC also knows implicitly that arguments starting in @option{-l} are to be
9841 treated as compiler output files, and passed to the linker in their
9842 proper position among the other output files.
9844 @c man begin OPTIONS
9846 @node Target Options
9847 @section Specifying Target Machine and Compiler Version
9848 @cindex target options
9849 @cindex cross compiling
9850 @cindex specifying machine version
9851 @cindex specifying compiler version and target machine
9852 @cindex compiler version, specifying
9853 @cindex target machine, specifying
9855 The usual way to run GCC is to run the executable called @file{gcc}, or
9856 @file{<machine>-gcc} when cross-compiling, or
9857 @file{<machine>-gcc-<version>} to run a version other than the one that
9860 @node Submodel Options
9861 @section Hardware Models and Configurations
9862 @cindex submodel options
9863 @cindex specifying hardware config
9864 @cindex hardware models and configurations, specifying
9865 @cindex machine dependent options
9867 Each target machine types can have its own
9868 special options, starting with @samp{-m}, to choose among various
9869 hardware models or configurations---for example, 68010 vs 68020,
9870 floating coprocessor or none. A single installed version of the
9871 compiler can compile for any model or configuration, according to the
9874 Some configurations of the compiler also support additional special
9875 options, usually for compatibility with other compilers on the same
9878 @c This list is ordered alphanumerically by subsection name.
9879 @c It should be the same order and spelling as these options are listed
9880 @c in Machine Dependent Options
9886 * Blackfin Options::
9890 * DEC Alpha Options::
9891 * DEC Alpha/VMS Options::
9894 * GNU/Linux Options::
9897 * i386 and x86-64 Options::
9898 * i386 and x86-64 Windows Options::
9900 * IA-64/VMS Options::
9908 * MicroBlaze Options::
9913 * picoChip Options::
9915 * RS/6000 and PowerPC Options::
9917 * S/390 and zSeries Options::
9920 * Solaris 2 Options::
9923 * System V Options::
9928 * Xstormy16 Options::
9934 @subsection ARC Options
9937 These options are defined for ARC implementations:
9942 Compile code for little endian mode. This is the default.
9946 Compile code for big endian mode.
9949 @opindex mmangle-cpu
9950 Prepend the name of the cpu to all public symbol names.
9951 In multiple-processor systems, there are many ARC variants with different
9952 instruction and register set characteristics. This flag prevents code
9953 compiled for one cpu to be linked with code compiled for another.
9954 No facility exists for handling variants that are ``almost identical''.
9955 This is an all or nothing option.
9957 @item -mcpu=@var{cpu}
9959 Compile code for ARC variant @var{cpu}.
9960 Which variants are supported depend on the configuration.
9961 All variants support @option{-mcpu=base}, this is the default.
9963 @item -mtext=@var{text-section}
9964 @itemx -mdata=@var{data-section}
9965 @itemx -mrodata=@var{readonly-data-section}
9969 Put functions, data, and readonly data in @var{text-section},
9970 @var{data-section}, and @var{readonly-data-section} respectively
9971 by default. This can be overridden with the @code{section} attribute.
9972 @xref{Variable Attributes}.
9977 @subsection ARM Options
9980 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9984 @item -mabi=@var{name}
9986 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9987 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9990 @opindex mapcs-frame
9991 Generate a stack frame that is compliant with the ARM Procedure Call
9992 Standard for all functions, even if this is not strictly necessary for
9993 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9994 with this option will cause the stack frames not to be generated for
9995 leaf functions. The default is @option{-mno-apcs-frame}.
9999 This is a synonym for @option{-mapcs-frame}.
10002 @c not currently implemented
10003 @item -mapcs-stack-check
10004 @opindex mapcs-stack-check
10005 Generate code to check the amount of stack space available upon entry to
10006 every function (that actually uses some stack space). If there is
10007 insufficient space available then either the function
10008 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
10009 called, depending upon the amount of stack space required. The run time
10010 system is required to provide these functions. The default is
10011 @option{-mno-apcs-stack-check}, since this produces smaller code.
10013 @c not currently implemented
10015 @opindex mapcs-float
10016 Pass floating point arguments using the float point registers. This is
10017 one of the variants of the APCS@. This option is recommended if the
10018 target hardware has a floating point unit or if a lot of floating point
10019 arithmetic is going to be performed by the code. The default is
10020 @option{-mno-apcs-float}, since integer only code is slightly increased in
10021 size if @option{-mapcs-float} is used.
10023 @c not currently implemented
10024 @item -mapcs-reentrant
10025 @opindex mapcs-reentrant
10026 Generate reentrant, position independent code. The default is
10027 @option{-mno-apcs-reentrant}.
10030 @item -mthumb-interwork
10031 @opindex mthumb-interwork
10032 Generate code which supports calling between the ARM and Thumb
10033 instruction sets. Without this option the two instruction sets cannot
10034 be reliably used inside one program. The default is
10035 @option{-mno-thumb-interwork}, since slightly larger code is generated
10036 when @option{-mthumb-interwork} is specified.
10038 @item -mno-sched-prolog
10039 @opindex mno-sched-prolog
10040 Prevent the reordering of instructions in the function prolog, or the
10041 merging of those instruction with the instructions in the function's
10042 body. This means that all functions will start with a recognizable set
10043 of instructions (or in fact one of a choice from a small set of
10044 different function prologues), and this information can be used to
10045 locate the start if functions inside an executable piece of code. The
10046 default is @option{-msched-prolog}.
10048 @item -mfloat-abi=@var{name}
10049 @opindex mfloat-abi
10050 Specifies which floating-point ABI to use. Permissible values
10051 are: @samp{soft}, @samp{softfp} and @samp{hard}.
10053 Specifying @samp{soft} causes GCC to generate output containing
10054 library calls for floating-point operations.
10055 @samp{softfp} allows the generation of code using hardware floating-point
10056 instructions, but still uses the soft-float calling conventions.
10057 @samp{hard} allows generation of floating-point instructions
10058 and uses FPU-specific calling conventions.
10060 The default depends on the specific target configuration. Note that
10061 the hard-float and soft-float ABIs are not link-compatible; you must
10062 compile your entire program with the same ABI, and link with a
10063 compatible set of libraries.
10066 @opindex mhard-float
10067 Equivalent to @option{-mfloat-abi=hard}.
10070 @opindex msoft-float
10071 Equivalent to @option{-mfloat-abi=soft}.
10073 @item -mlittle-endian
10074 @opindex mlittle-endian
10075 Generate code for a processor running in little-endian mode. This is
10076 the default for all standard configurations.
10079 @opindex mbig-endian
10080 Generate code for a processor running in big-endian mode; the default is
10081 to compile code for a little-endian processor.
10083 @item -mwords-little-endian
10084 @opindex mwords-little-endian
10085 This option only applies when generating code for big-endian processors.
10086 Generate code for a little-endian word order but a big-endian byte
10087 order. That is, a byte order of the form @samp{32107654}. Note: this
10088 option should only be used if you require compatibility with code for
10089 big-endian ARM processors generated by versions of the compiler prior to
10092 @item -mcpu=@var{name}
10094 This specifies the name of the target ARM processor. GCC uses this name
10095 to determine what kind of instructions it can emit when generating
10096 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
10097 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
10098 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
10099 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
10100 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
10102 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
10103 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
10104 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
10105 @samp{strongarm1110},
10106 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
10107 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
10108 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
10109 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
10110 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
10111 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
10112 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
10113 @samp{cortex-a5}, @samp{cortex-a8}, @samp{cortex-a9}, @samp{cortex-a15},
10114 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m4}, @samp{cortex-m3},
10117 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
10119 @item -mtune=@var{name}
10121 This option is very similar to the @option{-mcpu=} option, except that
10122 instead of specifying the actual target processor type, and hence
10123 restricting which instructions can be used, it specifies that GCC should
10124 tune the performance of the code as if the target were of the type
10125 specified in this option, but still choosing the instructions that it
10126 will generate based on the cpu specified by a @option{-mcpu=} option.
10127 For some ARM implementations better performance can be obtained by using
10130 @item -march=@var{name}
10132 This specifies the name of the target ARM architecture. GCC uses this
10133 name to determine what kind of instructions it can emit when generating
10134 assembly code. This option can be used in conjunction with or instead
10135 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
10136 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
10137 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
10138 @samp{armv6}, @samp{armv6j},
10139 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
10140 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
10141 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
10143 @item -mfpu=@var{name}
10144 @itemx -mfpe=@var{number}
10145 @itemx -mfp=@var{number}
10149 This specifies what floating point hardware (or hardware emulation) is
10150 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
10151 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
10152 @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
10153 @samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
10154 @samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
10155 @option{-mfp} and @option{-mfpe} are synonyms for
10156 @option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
10159 If @option{-msoft-float} is specified this specifies the format of
10160 floating point values.
10162 If the selected floating-point hardware includes the NEON extension
10163 (e.g. @option{-mfpu}=@samp{neon}), note that floating-point
10164 operations will not be used by GCC's auto-vectorization pass unless
10165 @option{-funsafe-math-optimizations} is also specified. This is
10166 because NEON hardware does not fully implement the IEEE 754 standard for
10167 floating-point arithmetic (in particular denormal values are treated as
10168 zero), so the use of NEON instructions may lead to a loss of precision.
10170 @item -mfp16-format=@var{name}
10171 @opindex mfp16-format
10172 Specify the format of the @code{__fp16} half-precision floating-point type.
10173 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
10174 the default is @samp{none}, in which case the @code{__fp16} type is not
10175 defined. @xref{Half-Precision}, for more information.
10177 @item -mstructure-size-boundary=@var{n}
10178 @opindex mstructure-size-boundary
10179 The size of all structures and unions will be rounded up to a multiple
10180 of the number of bits set by this option. Permissible values are 8, 32
10181 and 64. The default value varies for different toolchains. For the COFF
10182 targeted toolchain the default value is 8. A value of 64 is only allowed
10183 if the underlying ABI supports it.
10185 Specifying the larger number can produce faster, more efficient code, but
10186 can also increase the size of the program. Different values are potentially
10187 incompatible. Code compiled with one value cannot necessarily expect to
10188 work with code or libraries compiled with another value, if they exchange
10189 information using structures or unions.
10191 @item -mabort-on-noreturn
10192 @opindex mabort-on-noreturn
10193 Generate a call to the function @code{abort} at the end of a
10194 @code{noreturn} function. It will be executed if the function tries to
10198 @itemx -mno-long-calls
10199 @opindex mlong-calls
10200 @opindex mno-long-calls
10201 Tells the compiler to perform function calls by first loading the
10202 address of the function into a register and then performing a subroutine
10203 call on this register. This switch is needed if the target function
10204 will lie outside of the 64 megabyte addressing range of the offset based
10205 version of subroutine call instruction.
10207 Even if this switch is enabled, not all function calls will be turned
10208 into long calls. The heuristic is that static functions, functions
10209 which have the @samp{short-call} attribute, functions that are inside
10210 the scope of a @samp{#pragma no_long_calls} directive and functions whose
10211 definitions have already been compiled within the current compilation
10212 unit, will not be turned into long calls. The exception to this rule is
10213 that weak function definitions, functions with the @samp{long-call}
10214 attribute or the @samp{section} attribute, and functions that are within
10215 the scope of a @samp{#pragma long_calls} directive, will always be
10216 turned into long calls.
10218 This feature is not enabled by default. Specifying
10219 @option{-mno-long-calls} will restore the default behavior, as will
10220 placing the function calls within the scope of a @samp{#pragma
10221 long_calls_off} directive. Note these switches have no effect on how
10222 the compiler generates code to handle function calls via function
10225 @item -msingle-pic-base
10226 @opindex msingle-pic-base
10227 Treat the register used for PIC addressing as read-only, rather than
10228 loading it in the prologue for each function. The run-time system is
10229 responsible for initializing this register with an appropriate value
10230 before execution begins.
10232 @item -mpic-register=@var{reg}
10233 @opindex mpic-register
10234 Specify the register to be used for PIC addressing. The default is R10
10235 unless stack-checking is enabled, when R9 is used.
10237 @item -mcirrus-fix-invalid-insns
10238 @opindex mcirrus-fix-invalid-insns
10239 @opindex mno-cirrus-fix-invalid-insns
10240 Insert NOPs into the instruction stream to in order to work around
10241 problems with invalid Maverick instruction combinations. This option
10242 is only valid if the @option{-mcpu=ep9312} option has been used to
10243 enable generation of instructions for the Cirrus Maverick floating
10244 point co-processor. This option is not enabled by default, since the
10245 problem is only present in older Maverick implementations. The default
10246 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
10249 @item -mpoke-function-name
10250 @opindex mpoke-function-name
10251 Write the name of each function into the text section, directly
10252 preceding the function prologue. The generated code is similar to this:
10256 .ascii "arm_poke_function_name", 0
10259 .word 0xff000000 + (t1 - t0)
10260 arm_poke_function_name
10262 stmfd sp!, @{fp, ip, lr, pc@}
10266 When performing a stack backtrace, code can inspect the value of
10267 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
10268 location @code{pc - 12} and the top 8 bits are set, then we know that
10269 there is a function name embedded immediately preceding this location
10270 and has length @code{((pc[-3]) & 0xff000000)}.
10274 Generate code for the Thumb instruction set. The default is to
10275 use the 32-bit ARM instruction set.
10276 This option automatically enables either 16-bit Thumb-1 or
10277 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
10278 and @option{-march=@var{name}} options. This option is not passed to the
10279 assembler. If you want to force assembler files to be interpreted as Thumb code,
10280 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
10281 option directly to the assembler by prefixing it with @option{-Wa}.
10284 @opindex mtpcs-frame
10285 Generate a stack frame that is compliant with the Thumb Procedure Call
10286 Standard for all non-leaf functions. (A leaf function is one that does
10287 not call any other functions.) The default is @option{-mno-tpcs-frame}.
10289 @item -mtpcs-leaf-frame
10290 @opindex mtpcs-leaf-frame
10291 Generate a stack frame that is compliant with the Thumb Procedure Call
10292 Standard for all leaf functions. (A leaf function is one that does
10293 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
10295 @item -mcallee-super-interworking
10296 @opindex mcallee-super-interworking
10297 Gives all externally visible functions in the file being compiled an ARM
10298 instruction set header which switches to Thumb mode before executing the
10299 rest of the function. This allows these functions to be called from
10300 non-interworking code. This option is not valid in AAPCS configurations
10301 because interworking is enabled by default.
10303 @item -mcaller-super-interworking
10304 @opindex mcaller-super-interworking
10305 Allows calls via function pointers (including virtual functions) to
10306 execute correctly regardless of whether the target code has been
10307 compiled for interworking or not. There is a small overhead in the cost
10308 of executing a function pointer if this option is enabled. This option
10309 is not valid in AAPCS configurations because interworking is enabled
10312 @item -mtp=@var{name}
10314 Specify the access model for the thread local storage pointer. The valid
10315 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10316 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
10317 (supported in the arm6k architecture), and @option{auto}, which uses the
10318 best available method for the selected processor. The default setting is
10321 @item -mword-relocations
10322 @opindex mword-relocations
10323 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
10324 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10325 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10328 @item -mfix-cortex-m3-ldrd
10329 @opindex mfix-cortex-m3-ldrd
10330 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
10331 with overlapping destination and base registers are used. This option avoids
10332 generating these instructions. This option is enabled by default when
10333 @option{-mcpu=cortex-m3} is specified.
10338 @subsection AVR Options
10339 @cindex AVR Options
10341 These options are defined for AVR implementations:
10344 @item -mmcu=@var{mcu}
10346 Specify ATMEL AVR instruction set or MCU type.
10348 Instruction set avr1 is for the minimal AVR core, not supported by the C
10349 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10350 attiny11, attiny12, attiny15, attiny28).
10352 Instruction set avr2 (default) is for the classic AVR core with up to
10353 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10354 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10355 at90c8534, at90s8535).
10357 Instruction set avr3 is for the classic AVR core with up to 128K program
10358 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10360 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10361 memory space (MCU types: atmega8, atmega83, atmega85).
10363 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10364 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10365 atmega64, atmega128, at43usb355, at94k).
10367 @item -mno-interrupts
10368 @opindex mno-interrupts
10369 Generated code is not compatible with hardware interrupts.
10370 Code size will be smaller.
10372 @item -mcall-prologues
10373 @opindex mcall-prologues
10374 Functions prologues/epilogues expanded as call to appropriate
10375 subroutines. Code size will be smaller.
10378 @opindex mtiny-stack
10379 Change only the low 8 bits of the stack pointer.
10383 Assume int to be 8 bit integer. This affects the sizes of all types: A
10384 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10385 and long long will be 4 bytes. Please note that this option does not
10386 comply to the C standards, but it will provide you with smaller code
10390 @node Blackfin Options
10391 @subsection Blackfin Options
10392 @cindex Blackfin Options
10395 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10397 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10398 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10399 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10400 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10401 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10402 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10403 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10405 The optional @var{sirevision} specifies the silicon revision of the target
10406 Blackfin processor. Any workarounds available for the targeted silicon revision
10407 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10408 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10409 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10410 hexadecimal digits representing the major and minor numbers in the silicon
10411 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10412 is not defined. If @var{sirevision} is @samp{any}, the
10413 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10414 If this optional @var{sirevision} is not used, GCC assumes the latest known
10415 silicon revision of the targeted Blackfin processor.
10417 Support for @samp{bf561} is incomplete. For @samp{bf561},
10418 Only the processor macro is defined.
10419 Without this option, @samp{bf532} is used as the processor by default.
10420 The corresponding predefined processor macros for @var{cpu} is to
10421 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10422 provided by libgloss to be linked in if @option{-msim} is not given.
10426 Specifies that the program will be run on the simulator. This causes
10427 the simulator BSP provided by libgloss to be linked in. This option
10428 has effect only for @samp{bfin-elf} toolchain.
10429 Certain other options, such as @option{-mid-shared-library} and
10430 @option{-mfdpic}, imply @option{-msim}.
10432 @item -momit-leaf-frame-pointer
10433 @opindex momit-leaf-frame-pointer
10434 Don't keep the frame pointer in a register for leaf functions. This
10435 avoids the instructions to save, set up and restore frame pointers and
10436 makes an extra register available in leaf functions. The option
10437 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10438 which might make debugging harder.
10440 @item -mspecld-anomaly
10441 @opindex mspecld-anomaly
10442 When enabled, the compiler will ensure that the generated code does not
10443 contain speculative loads after jump instructions. If this option is used,
10444 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10446 @item -mno-specld-anomaly
10447 @opindex mno-specld-anomaly
10448 Don't generate extra code to prevent speculative loads from occurring.
10450 @item -mcsync-anomaly
10451 @opindex mcsync-anomaly
10452 When enabled, the compiler will ensure that the generated code does not
10453 contain CSYNC or SSYNC instructions too soon after conditional branches.
10454 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10456 @item -mno-csync-anomaly
10457 @opindex mno-csync-anomaly
10458 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10459 occurring too soon after a conditional branch.
10463 When enabled, the compiler is free to take advantage of the knowledge that
10464 the entire program fits into the low 64k of memory.
10467 @opindex mno-low-64k
10468 Assume that the program is arbitrarily large. This is the default.
10470 @item -mstack-check-l1
10471 @opindex mstack-check-l1
10472 Do stack checking using information placed into L1 scratchpad memory by the
10475 @item -mid-shared-library
10476 @opindex mid-shared-library
10477 Generate code that supports shared libraries via the library ID method.
10478 This allows for execute in place and shared libraries in an environment
10479 without virtual memory management. This option implies @option{-fPIC}.
10480 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10482 @item -mno-id-shared-library
10483 @opindex mno-id-shared-library
10484 Generate code that doesn't assume ID based shared libraries are being used.
10485 This is the default.
10487 @item -mleaf-id-shared-library
10488 @opindex mleaf-id-shared-library
10489 Generate code that supports shared libraries via the library ID method,
10490 but assumes that this library or executable won't link against any other
10491 ID shared libraries. That allows the compiler to use faster code for jumps
10494 @item -mno-leaf-id-shared-library
10495 @opindex mno-leaf-id-shared-library
10496 Do not assume that the code being compiled won't link against any ID shared
10497 libraries. Slower code will be generated for jump and call insns.
10499 @item -mshared-library-id=n
10500 @opindex mshared-library-id
10501 Specified the identification number of the ID based shared library being
10502 compiled. Specifying a value of 0 will generate more compact code, specifying
10503 other values will force the allocation of that number to the current
10504 library but is no more space or time efficient than omitting this option.
10508 Generate code that allows the data segment to be located in a different
10509 area of memory from the text segment. This allows for execute in place in
10510 an environment without virtual memory management by eliminating relocations
10511 against the text section.
10513 @item -mno-sep-data
10514 @opindex mno-sep-data
10515 Generate code that assumes that the data segment follows the text segment.
10516 This is the default.
10519 @itemx -mno-long-calls
10520 @opindex mlong-calls
10521 @opindex mno-long-calls
10522 Tells the compiler to perform function calls by first loading the
10523 address of the function into a register and then performing a subroutine
10524 call on this register. This switch is needed if the target function
10525 will lie outside of the 24 bit addressing range of the offset based
10526 version of subroutine call instruction.
10528 This feature is not enabled by default. Specifying
10529 @option{-mno-long-calls} will restore the default behavior. Note these
10530 switches have no effect on how the compiler generates code to handle
10531 function calls via function pointers.
10535 Link with the fast floating-point library. This library relaxes some of
10536 the IEEE floating-point standard's rules for checking inputs against
10537 Not-a-Number (NAN), in the interest of performance.
10540 @opindex minline-plt
10541 Enable inlining of PLT entries in function calls to functions that are
10542 not known to bind locally. It has no effect without @option{-mfdpic}.
10545 @opindex mmulticore
10546 Build standalone application for multicore Blackfin processor. Proper
10547 start files and link scripts will be used to support multicore.
10548 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10549 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10550 @option{-mcorea} or @option{-mcoreb}. If it's used without
10551 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10552 programming model is used. In this model, the main function of Core B
10553 should be named as coreb_main. If it's used with @option{-mcorea} or
10554 @option{-mcoreb}, one application per core programming model is used.
10555 If this option is not used, single core application programming
10560 Build standalone application for Core A of BF561 when using
10561 one application per core programming model. Proper start files
10562 and link scripts will be used to support Core A. This option
10563 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10567 Build standalone application for Core B of BF561 when using
10568 one application per core programming model. Proper start files
10569 and link scripts will be used to support Core B. This option
10570 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10571 should be used instead of main. It must be used with
10572 @option{-mmulticore}.
10576 Build standalone application for SDRAM. Proper start files and
10577 link scripts will be used to put the application into SDRAM.
10578 Loader should initialize SDRAM before loading the application
10579 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10583 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10584 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10585 are enabled; for standalone applications the default is off.
10589 @subsection CRIS Options
10590 @cindex CRIS Options
10592 These options are defined specifically for the CRIS ports.
10595 @item -march=@var{architecture-type}
10596 @itemx -mcpu=@var{architecture-type}
10599 Generate code for the specified architecture. The choices for
10600 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10601 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10602 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10605 @item -mtune=@var{architecture-type}
10607 Tune to @var{architecture-type} everything applicable about the generated
10608 code, except for the ABI and the set of available instructions. The
10609 choices for @var{architecture-type} are the same as for
10610 @option{-march=@var{architecture-type}}.
10612 @item -mmax-stack-frame=@var{n}
10613 @opindex mmax-stack-frame
10614 Warn when the stack frame of a function exceeds @var{n} bytes.
10620 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10621 @option{-march=v3} and @option{-march=v8} respectively.
10623 @item -mmul-bug-workaround
10624 @itemx -mno-mul-bug-workaround
10625 @opindex mmul-bug-workaround
10626 @opindex mno-mul-bug-workaround
10627 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10628 models where it applies. This option is active by default.
10632 Enable CRIS-specific verbose debug-related information in the assembly
10633 code. This option also has the effect to turn off the @samp{#NO_APP}
10634 formatted-code indicator to the assembler at the beginning of the
10639 Do not use condition-code results from previous instruction; always emit
10640 compare and test instructions before use of condition codes.
10642 @item -mno-side-effects
10643 @opindex mno-side-effects
10644 Do not emit instructions with side-effects in addressing modes other than
10647 @item -mstack-align
10648 @itemx -mno-stack-align
10649 @itemx -mdata-align
10650 @itemx -mno-data-align
10651 @itemx -mconst-align
10652 @itemx -mno-const-align
10653 @opindex mstack-align
10654 @opindex mno-stack-align
10655 @opindex mdata-align
10656 @opindex mno-data-align
10657 @opindex mconst-align
10658 @opindex mno-const-align
10659 These options (no-options) arranges (eliminate arrangements) for the
10660 stack-frame, individual data and constants to be aligned for the maximum
10661 single data access size for the chosen CPU model. The default is to
10662 arrange for 32-bit alignment. ABI details such as structure layout are
10663 not affected by these options.
10671 Similar to the stack- data- and const-align options above, these options
10672 arrange for stack-frame, writable data and constants to all be 32-bit,
10673 16-bit or 8-bit aligned. The default is 32-bit alignment.
10675 @item -mno-prologue-epilogue
10676 @itemx -mprologue-epilogue
10677 @opindex mno-prologue-epilogue
10678 @opindex mprologue-epilogue
10679 With @option{-mno-prologue-epilogue}, the normal function prologue and
10680 epilogue that sets up the stack-frame are omitted and no return
10681 instructions or return sequences are generated in the code. Use this
10682 option only together with visual inspection of the compiled code: no
10683 warnings or errors are generated when call-saved registers must be saved,
10684 or storage for local variable needs to be allocated.
10688 @opindex mno-gotplt
10690 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10691 instruction sequences that load addresses for functions from the PLT part
10692 of the GOT rather than (traditional on other architectures) calls to the
10693 PLT@. The default is @option{-mgotplt}.
10697 Legacy no-op option only recognized with the cris-axis-elf and
10698 cris-axis-linux-gnu targets.
10702 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10706 This option, recognized for the cris-axis-elf arranges
10707 to link with input-output functions from a simulator library. Code,
10708 initialized data and zero-initialized data are allocated consecutively.
10712 Like @option{-sim}, but pass linker options to locate initialized data at
10713 0x40000000 and zero-initialized data at 0x80000000.
10717 @subsection CRX Options
10718 @cindex CRX Options
10720 These options are defined specifically for the CRX ports.
10726 Enable the use of multiply-accumulate instructions. Disabled by default.
10729 @opindex mpush-args
10730 Push instructions will be used to pass outgoing arguments when functions
10731 are called. Enabled by default.
10734 @node Darwin Options
10735 @subsection Darwin Options
10736 @cindex Darwin options
10738 These options are defined for all architectures running the Darwin operating
10741 FSF GCC on Darwin does not create ``fat'' object files; it will create
10742 an object file for the single architecture that it was built to
10743 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10744 @option{-arch} options are used; it does so by running the compiler or
10745 linker multiple times and joining the results together with
10748 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10749 @samp{i686}) is determined by the flags that specify the ISA
10750 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10751 @option{-force_cpusubtype_ALL} option can be used to override this.
10753 The Darwin tools vary in their behavior when presented with an ISA
10754 mismatch. The assembler, @file{as}, will only permit instructions to
10755 be used that are valid for the subtype of the file it is generating,
10756 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10757 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10758 and print an error if asked to create a shared library with a less
10759 restrictive subtype than its input files (for instance, trying to put
10760 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10761 for executables, @file{ld}, will quietly give the executable the most
10762 restrictive subtype of any of its input files.
10767 Add the framework directory @var{dir} to the head of the list of
10768 directories to be searched for header files. These directories are
10769 interleaved with those specified by @option{-I} options and are
10770 scanned in a left-to-right order.
10772 A framework directory is a directory with frameworks in it. A
10773 framework is a directory with a @samp{"Headers"} and/or
10774 @samp{"PrivateHeaders"} directory contained directly in it that ends
10775 in @samp{".framework"}. The name of a framework is the name of this
10776 directory excluding the @samp{".framework"}. Headers associated with
10777 the framework are found in one of those two directories, with
10778 @samp{"Headers"} being searched first. A subframework is a framework
10779 directory that is in a framework's @samp{"Frameworks"} directory.
10780 Includes of subframework headers can only appear in a header of a
10781 framework that contains the subframework, or in a sibling subframework
10782 header. Two subframeworks are siblings if they occur in the same
10783 framework. A subframework should not have the same name as a
10784 framework, a warning will be issued if this is violated. Currently a
10785 subframework cannot have subframeworks, in the future, the mechanism
10786 may be extended to support this. The standard frameworks can be found
10787 in @samp{"/System/Library/Frameworks"} and
10788 @samp{"/Library/Frameworks"}. An example include looks like
10789 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10790 the name of the framework and header.h is found in the
10791 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10793 @item -iframework@var{dir}
10794 @opindex iframework
10795 Like @option{-F} except the directory is a treated as a system
10796 directory. The main difference between this @option{-iframework} and
10797 @option{-F} is that with @option{-iframework} the compiler does not
10798 warn about constructs contained within header files found via
10799 @var{dir}. This option is valid only for the C family of languages.
10803 Emit debugging information for symbols that are used. For STABS
10804 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10805 This is by default ON@.
10809 Emit debugging information for all symbols and types.
10811 @item -mmacosx-version-min=@var{version}
10812 The earliest version of MacOS X that this executable will run on
10813 is @var{version}. Typical values of @var{version} include @code{10.1},
10814 @code{10.2}, and @code{10.3.9}.
10816 If the compiler was built to use the system's headers by default,
10817 then the default for this option is the system version on which the
10818 compiler is running, otherwise the default is to make choices which
10819 are compatible with as many systems and code bases as possible.
10823 Enable kernel development mode. The @option{-mkernel} option sets
10824 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10825 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10826 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10827 applicable. This mode also sets @option{-mno-altivec},
10828 @option{-msoft-float}, @option{-fno-builtin} and
10829 @option{-mlong-branch} for PowerPC targets.
10831 @item -mone-byte-bool
10832 @opindex mone-byte-bool
10833 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10834 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10835 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10836 option has no effect on x86.
10838 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10839 to generate code that is not binary compatible with code generated
10840 without that switch. Using this switch may require recompiling all
10841 other modules in a program, including system libraries. Use this
10842 switch to conform to a non-default data model.
10844 @item -mfix-and-continue
10845 @itemx -ffix-and-continue
10846 @itemx -findirect-data
10847 @opindex mfix-and-continue
10848 @opindex ffix-and-continue
10849 @opindex findirect-data
10850 Generate code suitable for fast turn around development. Needed to
10851 enable gdb to dynamically load @code{.o} files into already running
10852 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10853 are provided for backwards compatibility.
10857 Loads all members of static archive libraries.
10858 See man ld(1) for more information.
10860 @item -arch_errors_fatal
10861 @opindex arch_errors_fatal
10862 Cause the errors having to do with files that have the wrong architecture
10865 @item -bind_at_load
10866 @opindex bind_at_load
10867 Causes the output file to be marked such that the dynamic linker will
10868 bind all undefined references when the file is loaded or launched.
10872 Produce a Mach-o bundle format file.
10873 See man ld(1) for more information.
10875 @item -bundle_loader @var{executable}
10876 @opindex bundle_loader
10877 This option specifies the @var{executable} that will be loading the build
10878 output file being linked. See man ld(1) for more information.
10881 @opindex dynamiclib
10882 When passed this option, GCC will produce a dynamic library instead of
10883 an executable when linking, using the Darwin @file{libtool} command.
10885 @item -force_cpusubtype_ALL
10886 @opindex force_cpusubtype_ALL
10887 This causes GCC's output file to have the @var{ALL} subtype, instead of
10888 one controlled by the @option{-mcpu} or @option{-march} option.
10890 @item -allowable_client @var{client_name}
10891 @itemx -client_name
10892 @itemx -compatibility_version
10893 @itemx -current_version
10895 @itemx -dependency-file
10897 @itemx -dylinker_install_name
10899 @itemx -exported_symbols_list
10902 @itemx -flat_namespace
10903 @itemx -force_flat_namespace
10904 @itemx -headerpad_max_install_names
10907 @itemx -install_name
10908 @itemx -keep_private_externs
10909 @itemx -multi_module
10910 @itemx -multiply_defined
10911 @itemx -multiply_defined_unused
10914 @itemx -no_dead_strip_inits_and_terms
10915 @itemx -nofixprebinding
10916 @itemx -nomultidefs
10918 @itemx -noseglinkedit
10919 @itemx -pagezero_size
10921 @itemx -prebind_all_twolevel_modules
10922 @itemx -private_bundle
10924 @itemx -read_only_relocs
10926 @itemx -sectobjectsymbols
10930 @itemx -sectobjectsymbols
10933 @itemx -segs_read_only_addr
10935 @itemx -segs_read_write_addr
10936 @itemx -seg_addr_table
10937 @itemx -seg_addr_table_filename
10938 @itemx -seglinkedit
10940 @itemx -segs_read_only_addr
10941 @itemx -segs_read_write_addr
10942 @itemx -single_module
10944 @itemx -sub_library
10946 @itemx -sub_umbrella
10947 @itemx -twolevel_namespace
10950 @itemx -unexported_symbols_list
10951 @itemx -weak_reference_mismatches
10952 @itemx -whatsloaded
10953 @opindex allowable_client
10954 @opindex client_name
10955 @opindex compatibility_version
10956 @opindex current_version
10957 @opindex dead_strip
10958 @opindex dependency-file
10959 @opindex dylib_file
10960 @opindex dylinker_install_name
10962 @opindex exported_symbols_list
10964 @opindex flat_namespace
10965 @opindex force_flat_namespace
10966 @opindex headerpad_max_install_names
10967 @opindex image_base
10969 @opindex install_name
10970 @opindex keep_private_externs
10971 @opindex multi_module
10972 @opindex multiply_defined
10973 @opindex multiply_defined_unused
10974 @opindex noall_load
10975 @opindex no_dead_strip_inits_and_terms
10976 @opindex nofixprebinding
10977 @opindex nomultidefs
10979 @opindex noseglinkedit
10980 @opindex pagezero_size
10982 @opindex prebind_all_twolevel_modules
10983 @opindex private_bundle
10984 @opindex read_only_relocs
10986 @opindex sectobjectsymbols
10989 @opindex sectcreate
10990 @opindex sectobjectsymbols
10993 @opindex segs_read_only_addr
10994 @opindex segs_read_write_addr
10995 @opindex seg_addr_table
10996 @opindex seg_addr_table_filename
10997 @opindex seglinkedit
10999 @opindex segs_read_only_addr
11000 @opindex segs_read_write_addr
11001 @opindex single_module
11003 @opindex sub_library
11004 @opindex sub_umbrella
11005 @opindex twolevel_namespace
11008 @opindex unexported_symbols_list
11009 @opindex weak_reference_mismatches
11010 @opindex whatsloaded
11011 These options are passed to the Darwin linker. The Darwin linker man page
11012 describes them in detail.
11015 @node DEC Alpha Options
11016 @subsection DEC Alpha Options
11018 These @samp{-m} options are defined for the DEC Alpha implementations:
11021 @item -mno-soft-float
11022 @itemx -msoft-float
11023 @opindex mno-soft-float
11024 @opindex msoft-float
11025 Use (do not use) the hardware floating-point instructions for
11026 floating-point operations. When @option{-msoft-float} is specified,
11027 functions in @file{libgcc.a} will be used to perform floating-point
11028 operations. Unless they are replaced by routines that emulate the
11029 floating-point operations, or compiled in such a way as to call such
11030 emulations routines, these routines will issue floating-point
11031 operations. If you are compiling for an Alpha without floating-point
11032 operations, you must ensure that the library is built so as not to call
11035 Note that Alpha implementations without floating-point operations are
11036 required to have floating-point registers.
11039 @itemx -mno-fp-regs
11041 @opindex mno-fp-regs
11042 Generate code that uses (does not use) the floating-point register set.
11043 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
11044 register set is not used, floating point operands are passed in integer
11045 registers as if they were integers and floating-point results are passed
11046 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
11047 so any function with a floating-point argument or return value called by code
11048 compiled with @option{-mno-fp-regs} must also be compiled with that
11051 A typical use of this option is building a kernel that does not use,
11052 and hence need not save and restore, any floating-point registers.
11056 The Alpha architecture implements floating-point hardware optimized for
11057 maximum performance. It is mostly compliant with the IEEE floating
11058 point standard. However, for full compliance, software assistance is
11059 required. This option generates code fully IEEE compliant code
11060 @emph{except} that the @var{inexact-flag} is not maintained (see below).
11061 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
11062 defined during compilation. The resulting code is less efficient but is
11063 able to correctly support denormalized numbers and exceptional IEEE
11064 values such as not-a-number and plus/minus infinity. Other Alpha
11065 compilers call this option @option{-ieee_with_no_inexact}.
11067 @item -mieee-with-inexact
11068 @opindex mieee-with-inexact
11069 This is like @option{-mieee} except the generated code also maintains
11070 the IEEE @var{inexact-flag}. Turning on this option causes the
11071 generated code to implement fully-compliant IEEE math. In addition to
11072 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
11073 macro. On some Alpha implementations the resulting code may execute
11074 significantly slower than the code generated by default. Since there is
11075 very little code that depends on the @var{inexact-flag}, you should
11076 normally not specify this option. Other Alpha compilers call this
11077 option @option{-ieee_with_inexact}.
11079 @item -mfp-trap-mode=@var{trap-mode}
11080 @opindex mfp-trap-mode
11081 This option controls what floating-point related traps are enabled.
11082 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
11083 The trap mode can be set to one of four values:
11087 This is the default (normal) setting. The only traps that are enabled
11088 are the ones that cannot be disabled in software (e.g., division by zero
11092 In addition to the traps enabled by @samp{n}, underflow traps are enabled
11096 Like @samp{u}, but the instructions are marked to be safe for software
11097 completion (see Alpha architecture manual for details).
11100 Like @samp{su}, but inexact traps are enabled as well.
11103 @item -mfp-rounding-mode=@var{rounding-mode}
11104 @opindex mfp-rounding-mode
11105 Selects the IEEE rounding mode. Other Alpha compilers call this option
11106 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
11111 Normal IEEE rounding mode. Floating point numbers are rounded towards
11112 the nearest machine number or towards the even machine number in case
11116 Round towards minus infinity.
11119 Chopped rounding mode. Floating point numbers are rounded towards zero.
11122 Dynamic rounding mode. A field in the floating point control register
11123 (@var{fpcr}, see Alpha architecture reference manual) controls the
11124 rounding mode in effect. The C library initializes this register for
11125 rounding towards plus infinity. Thus, unless your program modifies the
11126 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
11129 @item -mtrap-precision=@var{trap-precision}
11130 @opindex mtrap-precision
11131 In the Alpha architecture, floating point traps are imprecise. This
11132 means without software assistance it is impossible to recover from a
11133 floating trap and program execution normally needs to be terminated.
11134 GCC can generate code that can assist operating system trap handlers
11135 in determining the exact location that caused a floating point trap.
11136 Depending on the requirements of an application, different levels of
11137 precisions can be selected:
11141 Program precision. This option is the default and means a trap handler
11142 can only identify which program caused a floating point exception.
11145 Function precision. The trap handler can determine the function that
11146 caused a floating point exception.
11149 Instruction precision. The trap handler can determine the exact
11150 instruction that caused a floating point exception.
11153 Other Alpha compilers provide the equivalent options called
11154 @option{-scope_safe} and @option{-resumption_safe}.
11156 @item -mieee-conformant
11157 @opindex mieee-conformant
11158 This option marks the generated code as IEEE conformant. You must not
11159 use this option unless you also specify @option{-mtrap-precision=i} and either
11160 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
11161 is to emit the line @samp{.eflag 48} in the function prologue of the
11162 generated assembly file. Under DEC Unix, this has the effect that
11163 IEEE-conformant math library routines will be linked in.
11165 @item -mbuild-constants
11166 @opindex mbuild-constants
11167 Normally GCC examines a 32- or 64-bit integer constant to
11168 see if it can construct it from smaller constants in two or three
11169 instructions. If it cannot, it will output the constant as a literal and
11170 generate code to load it from the data segment at runtime.
11172 Use this option to require GCC to construct @emph{all} integer constants
11173 using code, even if it takes more instructions (the maximum is six).
11175 You would typically use this option to build a shared library dynamic
11176 loader. Itself a shared library, it must relocate itself in memory
11177 before it can find the variables and constants in its own data segment.
11183 Select whether to generate code to be assembled by the vendor-supplied
11184 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
11202 Indicate whether GCC should generate code to use the optional BWX,
11203 CIX, FIX and MAX instruction sets. The default is to use the instruction
11204 sets supported by the CPU type specified via @option{-mcpu=} option or that
11205 of the CPU on which GCC was built if none was specified.
11208 @itemx -mfloat-ieee
11209 @opindex mfloat-vax
11210 @opindex mfloat-ieee
11211 Generate code that uses (does not use) VAX F and G floating point
11212 arithmetic instead of IEEE single and double precision.
11214 @item -mexplicit-relocs
11215 @itemx -mno-explicit-relocs
11216 @opindex mexplicit-relocs
11217 @opindex mno-explicit-relocs
11218 Older Alpha assemblers provided no way to generate symbol relocations
11219 except via assembler macros. Use of these macros does not allow
11220 optimal instruction scheduling. GNU binutils as of version 2.12
11221 supports a new syntax that allows the compiler to explicitly mark
11222 which relocations should apply to which instructions. This option
11223 is mostly useful for debugging, as GCC detects the capabilities of
11224 the assembler when it is built and sets the default accordingly.
11227 @itemx -mlarge-data
11228 @opindex msmall-data
11229 @opindex mlarge-data
11230 When @option{-mexplicit-relocs} is in effect, static data is
11231 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
11232 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
11233 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
11234 16-bit relocations off of the @code{$gp} register. This limits the
11235 size of the small data area to 64KB, but allows the variables to be
11236 directly accessed via a single instruction.
11238 The default is @option{-mlarge-data}. With this option the data area
11239 is limited to just below 2GB@. Programs that require more than 2GB of
11240 data must use @code{malloc} or @code{mmap} to allocate the data in the
11241 heap instead of in the program's data segment.
11243 When generating code for shared libraries, @option{-fpic} implies
11244 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
11247 @itemx -mlarge-text
11248 @opindex msmall-text
11249 @opindex mlarge-text
11250 When @option{-msmall-text} is used, the compiler assumes that the
11251 code of the entire program (or shared library) fits in 4MB, and is
11252 thus reachable with a branch instruction. When @option{-msmall-data}
11253 is used, the compiler can assume that all local symbols share the
11254 same @code{$gp} value, and thus reduce the number of instructions
11255 required for a function call from 4 to 1.
11257 The default is @option{-mlarge-text}.
11259 @item -mcpu=@var{cpu_type}
11261 Set the instruction set and instruction scheduling parameters for
11262 machine type @var{cpu_type}. You can specify either the @samp{EV}
11263 style name or the corresponding chip number. GCC supports scheduling
11264 parameters for the EV4, EV5 and EV6 family of processors and will
11265 choose the default values for the instruction set from the processor
11266 you specify. If you do not specify a processor type, GCC will default
11267 to the processor on which the compiler was built.
11269 Supported values for @var{cpu_type} are
11275 Schedules as an EV4 and has no instruction set extensions.
11279 Schedules as an EV5 and has no instruction set extensions.
11283 Schedules as an EV5 and supports the BWX extension.
11288 Schedules as an EV5 and supports the BWX and MAX extensions.
11292 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
11296 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
11299 Native Linux/GNU toolchains also support the value @samp{native},
11300 which selects the best architecture option for the host processor.
11301 @option{-mcpu=native} has no effect if GCC does not recognize
11304 @item -mtune=@var{cpu_type}
11306 Set only the instruction scheduling parameters for machine type
11307 @var{cpu_type}. The instruction set is not changed.
11309 Native Linux/GNU toolchains also support the value @samp{native},
11310 which selects the best architecture option for the host processor.
11311 @option{-mtune=native} has no effect if GCC does not recognize
11314 @item -mmemory-latency=@var{time}
11315 @opindex mmemory-latency
11316 Sets the latency the scheduler should assume for typical memory
11317 references as seen by the application. This number is highly
11318 dependent on the memory access patterns used by the application
11319 and the size of the external cache on the machine.
11321 Valid options for @var{time} are
11325 A decimal number representing clock cycles.
11331 The compiler contains estimates of the number of clock cycles for
11332 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
11333 (also called Dcache, Scache, and Bcache), as well as to main memory.
11334 Note that L3 is only valid for EV5.
11339 @node DEC Alpha/VMS Options
11340 @subsection DEC Alpha/VMS Options
11342 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
11345 @item -mvms-return-codes
11346 @opindex mvms-return-codes
11347 Return VMS condition codes from main. The default is to return POSIX
11348 style condition (e.g.@: error) codes.
11350 @item -mdebug-main=@var{prefix}
11351 @opindex mdebug-main=@var{prefix}
11352 Flag the first routine whose name starts with @var{prefix} as the main
11353 routine for the debugger.
11357 Default to 64bit memory allocation routines.
11361 @subsection FR30 Options
11362 @cindex FR30 Options
11364 These options are defined specifically for the FR30 port.
11368 @item -msmall-model
11369 @opindex msmall-model
11370 Use the small address space model. This can produce smaller code, but
11371 it does assume that all symbolic values and addresses will fit into a
11376 Assume that run-time support has been provided and so there is no need
11377 to include the simulator library (@file{libsim.a}) on the linker
11383 @subsection FRV Options
11384 @cindex FRV Options
11390 Only use the first 32 general purpose registers.
11395 Use all 64 general purpose registers.
11400 Use only the first 32 floating point registers.
11405 Use all 64 floating point registers
11408 @opindex mhard-float
11410 Use hardware instructions for floating point operations.
11413 @opindex msoft-float
11415 Use library routines for floating point operations.
11420 Dynamically allocate condition code registers.
11425 Do not try to dynamically allocate condition code registers, only
11426 use @code{icc0} and @code{fcc0}.
11431 Change ABI to use double word insns.
11436 Do not use double word instructions.
11441 Use floating point double instructions.
11444 @opindex mno-double
11446 Do not use floating point double instructions.
11451 Use media instructions.
11456 Do not use media instructions.
11461 Use multiply and add/subtract instructions.
11464 @opindex mno-muladd
11466 Do not use multiply and add/subtract instructions.
11471 Select the FDPIC ABI, that uses function descriptors to represent
11472 pointers to functions. Without any PIC/PIE-related options, it
11473 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11474 assumes GOT entries and small data are within a 12-bit range from the
11475 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11476 are computed with 32 bits.
11477 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11480 @opindex minline-plt
11482 Enable inlining of PLT entries in function calls to functions that are
11483 not known to bind locally. It has no effect without @option{-mfdpic}.
11484 It's enabled by default if optimizing for speed and compiling for
11485 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11486 optimization option such as @option{-O3} or above is present in the
11492 Assume a large TLS segment when generating thread-local code.
11497 Do not assume a large TLS segment when generating thread-local code.
11502 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11503 that is known to be in read-only sections. It's enabled by default,
11504 except for @option{-fpic} or @option{-fpie}: even though it may help
11505 make the global offset table smaller, it trades 1 instruction for 4.
11506 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11507 one of which may be shared by multiple symbols, and it avoids the need
11508 for a GOT entry for the referenced symbol, so it's more likely to be a
11509 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11511 @item -multilib-library-pic
11512 @opindex multilib-library-pic
11514 Link with the (library, not FD) pic libraries. It's implied by
11515 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11516 @option{-fpic} without @option{-mfdpic}. You should never have to use
11520 @opindex mlinked-fp
11522 Follow the EABI requirement of always creating a frame pointer whenever
11523 a stack frame is allocated. This option is enabled by default and can
11524 be disabled with @option{-mno-linked-fp}.
11527 @opindex mlong-calls
11529 Use indirect addressing to call functions outside the current
11530 compilation unit. This allows the functions to be placed anywhere
11531 within the 32-bit address space.
11533 @item -malign-labels
11534 @opindex malign-labels
11536 Try to align labels to an 8-byte boundary by inserting nops into the
11537 previous packet. This option only has an effect when VLIW packing
11538 is enabled. It doesn't create new packets; it merely adds nops to
11541 @item -mlibrary-pic
11542 @opindex mlibrary-pic
11544 Generate position-independent EABI code.
11549 Use only the first four media accumulator registers.
11554 Use all eight media accumulator registers.
11559 Pack VLIW instructions.
11564 Do not pack VLIW instructions.
11567 @opindex mno-eflags
11569 Do not mark ABI switches in e_flags.
11572 @opindex mcond-move
11574 Enable the use of conditional-move instructions (default).
11576 This switch is mainly for debugging the compiler and will likely be removed
11577 in a future version.
11579 @item -mno-cond-move
11580 @opindex mno-cond-move
11582 Disable the use of conditional-move instructions.
11584 This switch is mainly for debugging the compiler and will likely be removed
11585 in a future version.
11590 Enable the use of conditional set instructions (default).
11592 This switch is mainly for debugging the compiler and will likely be removed
11593 in a future version.
11598 Disable the use of conditional set instructions.
11600 This switch is mainly for debugging the compiler and will likely be removed
11601 in a future version.
11604 @opindex mcond-exec
11606 Enable the use of conditional execution (default).
11608 This switch is mainly for debugging the compiler and will likely be removed
11609 in a future version.
11611 @item -mno-cond-exec
11612 @opindex mno-cond-exec
11614 Disable the use of conditional execution.
11616 This switch is mainly for debugging the compiler and will likely be removed
11617 in a future version.
11619 @item -mvliw-branch
11620 @opindex mvliw-branch
11622 Run a pass to pack branches into VLIW instructions (default).
11624 This switch is mainly for debugging the compiler and will likely be removed
11625 in a future version.
11627 @item -mno-vliw-branch
11628 @opindex mno-vliw-branch
11630 Do not run a pass to pack branches into VLIW instructions.
11632 This switch is mainly for debugging the compiler and will likely be removed
11633 in a future version.
11635 @item -mmulti-cond-exec
11636 @opindex mmulti-cond-exec
11638 Enable optimization of @code{&&} and @code{||} in conditional execution
11641 This switch is mainly for debugging the compiler and will likely be removed
11642 in a future version.
11644 @item -mno-multi-cond-exec
11645 @opindex mno-multi-cond-exec
11647 Disable optimization of @code{&&} and @code{||} in conditional execution.
11649 This switch is mainly for debugging the compiler and will likely be removed
11650 in a future version.
11652 @item -mnested-cond-exec
11653 @opindex mnested-cond-exec
11655 Enable nested conditional execution optimizations (default).
11657 This switch is mainly for debugging the compiler and will likely be removed
11658 in a future version.
11660 @item -mno-nested-cond-exec
11661 @opindex mno-nested-cond-exec
11663 Disable nested conditional execution optimizations.
11665 This switch is mainly for debugging the compiler and will likely be removed
11666 in a future version.
11668 @item -moptimize-membar
11669 @opindex moptimize-membar
11671 This switch removes redundant @code{membar} instructions from the
11672 compiler generated code. It is enabled by default.
11674 @item -mno-optimize-membar
11675 @opindex mno-optimize-membar
11677 This switch disables the automatic removal of redundant @code{membar}
11678 instructions from the generated code.
11680 @item -mtomcat-stats
11681 @opindex mtomcat-stats
11683 Cause gas to print out tomcat statistics.
11685 @item -mcpu=@var{cpu}
11688 Select the processor type for which to generate code. Possible values are
11689 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11690 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11694 @node GNU/Linux Options
11695 @subsection GNU/Linux Options
11697 These @samp{-m} options are defined for GNU/Linux targets:
11702 Use the GNU C library. This is the default except
11703 on @samp{*-*-linux-*uclibc*} and @samp{*-*-linux-*android*} targets.
11707 Use uClibc C library. This is the default on
11708 @samp{*-*-linux-*uclibc*} targets.
11712 Use Bionic C library. This is the default on
11713 @samp{*-*-linux-*android*} targets.
11717 Compile code compatible with Android platform. This is the default on
11718 @samp{*-*-linux-*android*} targets.
11720 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
11721 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
11722 this option makes the GCC driver pass Android-specific options to the linker.
11723 Finally, this option causes the preprocessor macro @code{__ANDROID__}
11726 @item -tno-android-cc
11727 @opindex tno-android-cc
11728 Disable compilation effects of @option{-mandroid}, i.e., do not enable
11729 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
11730 @option{-fno-rtti} by default.
11732 @item -tno-android-ld
11733 @opindex tno-android-ld
11734 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
11735 linking options to the linker.
11739 @node H8/300 Options
11740 @subsection H8/300 Options
11742 These @samp{-m} options are defined for the H8/300 implementations:
11747 Shorten some address references at link time, when possible; uses the
11748 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11749 ld, Using ld}, for a fuller description.
11753 Generate code for the H8/300H@.
11757 Generate code for the H8S@.
11761 Generate code for the H8S and H8/300H in the normal mode. This switch
11762 must be used either with @option{-mh} or @option{-ms}.
11766 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11770 Make @code{int} data 32 bits by default.
11773 @opindex malign-300
11774 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11775 The default for the H8/300H and H8S is to align longs and floats on 4
11777 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11778 This option has no effect on the H8/300.
11782 @subsection HPPA Options
11783 @cindex HPPA Options
11785 These @samp{-m} options are defined for the HPPA family of computers:
11788 @item -march=@var{architecture-type}
11790 Generate code for the specified architecture. The choices for
11791 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11792 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11793 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11794 architecture option for your machine. Code compiled for lower numbered
11795 architectures will run on higher numbered architectures, but not the
11798 @item -mpa-risc-1-0
11799 @itemx -mpa-risc-1-1
11800 @itemx -mpa-risc-2-0
11801 @opindex mpa-risc-1-0
11802 @opindex mpa-risc-1-1
11803 @opindex mpa-risc-2-0
11804 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11807 @opindex mbig-switch
11808 Generate code suitable for big switch tables. Use this option only if
11809 the assembler/linker complain about out of range branches within a switch
11812 @item -mjump-in-delay
11813 @opindex mjump-in-delay
11814 Fill delay slots of function calls with unconditional jump instructions
11815 by modifying the return pointer for the function call to be the target
11816 of the conditional jump.
11818 @item -mdisable-fpregs
11819 @opindex mdisable-fpregs
11820 Prevent floating point registers from being used in any manner. This is
11821 necessary for compiling kernels which perform lazy context switching of
11822 floating point registers. If you use this option and attempt to perform
11823 floating point operations, the compiler will abort.
11825 @item -mdisable-indexing
11826 @opindex mdisable-indexing
11827 Prevent the compiler from using indexing address modes. This avoids some
11828 rather obscure problems when compiling MIG generated code under MACH@.
11830 @item -mno-space-regs
11831 @opindex mno-space-regs
11832 Generate code that assumes the target has no space registers. This allows
11833 GCC to generate faster indirect calls and use unscaled index address modes.
11835 Such code is suitable for level 0 PA systems and kernels.
11837 @item -mfast-indirect-calls
11838 @opindex mfast-indirect-calls
11839 Generate code that assumes calls never cross space boundaries. This
11840 allows GCC to emit code which performs faster indirect calls.
11842 This option will not work in the presence of shared libraries or nested
11845 @item -mfixed-range=@var{register-range}
11846 @opindex mfixed-range
11847 Generate code treating the given register range as fixed registers.
11848 A fixed register is one that the register allocator can not use. This is
11849 useful when compiling kernel code. A register range is specified as
11850 two registers separated by a dash. Multiple register ranges can be
11851 specified separated by a comma.
11853 @item -mlong-load-store
11854 @opindex mlong-load-store
11855 Generate 3-instruction load and store sequences as sometimes required by
11856 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11859 @item -mportable-runtime
11860 @opindex mportable-runtime
11861 Use the portable calling conventions proposed by HP for ELF systems.
11865 Enable the use of assembler directives only GAS understands.
11867 @item -mschedule=@var{cpu-type}
11869 Schedule code according to the constraints for the machine type
11870 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11871 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11872 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11873 proper scheduling option for your machine. The default scheduling is
11877 @opindex mlinker-opt
11878 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11879 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11880 linkers in which they give bogus error messages when linking some programs.
11883 @opindex msoft-float
11884 Generate output containing library calls for floating point.
11885 @strong{Warning:} the requisite libraries are not available for all HPPA
11886 targets. Normally the facilities of the machine's usual C compiler are
11887 used, but this cannot be done directly in cross-compilation. You must make
11888 your own arrangements to provide suitable library functions for
11891 @option{-msoft-float} changes the calling convention in the output file;
11892 therefore, it is only useful if you compile @emph{all} of a program with
11893 this option. In particular, you need to compile @file{libgcc.a}, the
11894 library that comes with GCC, with @option{-msoft-float} in order for
11899 Generate the predefine, @code{_SIO}, for server IO@. The default is
11900 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11901 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11902 options are available under HP-UX and HI-UX@.
11906 Use GNU ld specific options. This passes @option{-shared} to ld when
11907 building a shared library. It is the default when GCC is configured,
11908 explicitly or implicitly, with the GNU linker. This option does not
11909 have any affect on which ld is called, it only changes what parameters
11910 are passed to that ld. The ld that is called is determined by the
11911 @option{--with-ld} configure option, GCC's program search path, and
11912 finally by the user's @env{PATH}. The linker used by GCC can be printed
11913 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11914 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11918 Use HP ld specific options. This passes @option{-b} to ld when building
11919 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11920 links. It is the default when GCC is configured, explicitly or
11921 implicitly, with the HP linker. This option does not have any affect on
11922 which ld is called, it only changes what parameters are passed to that
11923 ld. The ld that is called is determined by the @option{--with-ld}
11924 configure option, GCC's program search path, and finally by the user's
11925 @env{PATH}. The linker used by GCC can be printed using @samp{which
11926 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11927 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11930 @opindex mno-long-calls
11931 Generate code that uses long call sequences. This ensures that a call
11932 is always able to reach linker generated stubs. The default is to generate
11933 long calls only when the distance from the call site to the beginning
11934 of the function or translation unit, as the case may be, exceeds a
11935 predefined limit set by the branch type being used. The limits for
11936 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11937 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11940 Distances are measured from the beginning of functions when using the
11941 @option{-ffunction-sections} option, or when using the @option{-mgas}
11942 and @option{-mno-portable-runtime} options together under HP-UX with
11945 It is normally not desirable to use this option as it will degrade
11946 performance. However, it may be useful in large applications,
11947 particularly when partial linking is used to build the application.
11949 The types of long calls used depends on the capabilities of the
11950 assembler and linker, and the type of code being generated. The
11951 impact on systems that support long absolute calls, and long pic
11952 symbol-difference or pc-relative calls should be relatively small.
11953 However, an indirect call is used on 32-bit ELF systems in pic code
11954 and it is quite long.
11956 @item -munix=@var{unix-std}
11958 Generate compiler predefines and select a startfile for the specified
11959 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11960 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11961 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11962 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11963 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11966 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11967 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11968 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11969 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11970 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11971 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11973 It is @emph{important} to note that this option changes the interfaces
11974 for various library routines. It also affects the operational behavior
11975 of the C library. Thus, @emph{extreme} care is needed in using this
11978 Library code that is intended to operate with more than one UNIX
11979 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11980 as appropriate. Most GNU software doesn't provide this capability.
11984 Suppress the generation of link options to search libdld.sl when the
11985 @option{-static} option is specified on HP-UX 10 and later.
11989 The HP-UX implementation of setlocale in libc has a dependency on
11990 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11991 when the @option{-static} option is specified, special link options
11992 are needed to resolve this dependency.
11994 On HP-UX 10 and later, the GCC driver adds the necessary options to
11995 link with libdld.sl when the @option{-static} option is specified.
11996 This causes the resulting binary to be dynamic. On the 64-bit port,
11997 the linkers generate dynamic binaries by default in any case. The
11998 @option{-nolibdld} option can be used to prevent the GCC driver from
11999 adding these link options.
12003 Add support for multithreading with the @dfn{dce thread} library
12004 under HP-UX@. This option sets flags for both the preprocessor and
12008 @node i386 and x86-64 Options
12009 @subsection Intel 386 and AMD x86-64 Options
12010 @cindex i386 Options
12011 @cindex x86-64 Options
12012 @cindex Intel 386 Options
12013 @cindex AMD x86-64 Options
12015 These @samp{-m} options are defined for the i386 and x86-64 family of
12019 @item -mtune=@var{cpu-type}
12021 Tune to @var{cpu-type} everything applicable about the generated code, except
12022 for the ABI and the set of available instructions. The choices for
12023 @var{cpu-type} are:
12026 Produce code optimized for the most common IA32/@/AMD64/@/EM64T processors.
12027 If you know the CPU on which your code will run, then you should use
12028 the corresponding @option{-mtune} option instead of
12029 @option{-mtune=generic}. But, if you do not know exactly what CPU users
12030 of your application will have, then you should use this option.
12032 As new processors are deployed in the marketplace, the behavior of this
12033 option will change. Therefore, if you upgrade to a newer version of
12034 GCC, the code generated option will change to reflect the processors
12035 that were most common when that version of GCC was released.
12037 There is no @option{-march=generic} option because @option{-march}
12038 indicates the instruction set the compiler can use, and there is no
12039 generic instruction set applicable to all processors. In contrast,
12040 @option{-mtune} indicates the processor (or, in this case, collection of
12041 processors) for which the code is optimized.
12043 This selects the CPU to tune for at compilation time by determining
12044 the processor type of the compiling machine. Using @option{-mtune=native}
12045 will produce code optimized for the local machine under the constraints
12046 of the selected instruction set. Using @option{-march=native} will
12047 enable all instruction subsets supported by the local machine (hence
12048 the result might not run on different machines).
12050 Original Intel's i386 CPU@.
12052 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
12053 @item i586, pentium
12054 Intel Pentium CPU with no MMX support.
12056 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
12058 Intel PentiumPro CPU@.
12060 Same as @code{generic}, but when used as @code{march} option, PentiumPro
12061 instruction set will be used, so the code will run on all i686 family chips.
12063 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
12064 @item pentium3, pentium3m
12065 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
12068 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
12069 support. Used by Centrino notebooks.
12070 @item pentium4, pentium4m
12071 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
12073 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
12076 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
12077 SSE2 and SSE3 instruction set support.
12079 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
12080 instruction set support.
12082 Intel Core i7 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1
12083 and SSE4.2 instruction set support.
12085 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
12086 instruction set support.
12088 AMD K6 CPU with MMX instruction set support.
12090 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
12091 @item athlon, athlon-tbird
12092 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
12094 @item athlon-4, athlon-xp, athlon-mp
12095 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
12096 instruction set support.
12097 @item k8, opteron, athlon64, athlon-fx
12098 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
12099 MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
12100 @item k8-sse3, opteron-sse3, athlon64-sse3
12101 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
12102 @item amdfam10, barcelona
12103 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
12104 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
12105 instruction set extensions.)
12107 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
12110 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
12111 instruction set support.
12113 Via C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
12114 implemented for this chip.)
12116 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
12117 implemented for this chip.)
12119 Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
12122 While picking a specific @var{cpu-type} will schedule things appropriately
12123 for that particular chip, the compiler will not generate any code that
12124 does not run on the i386 without the @option{-march=@var{cpu-type}} option
12127 @item -march=@var{cpu-type}
12129 Generate instructions for the machine type @var{cpu-type}. The choices
12130 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
12131 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
12133 @item -mcpu=@var{cpu-type}
12135 A deprecated synonym for @option{-mtune}.
12137 @item -mfpmath=@var{unit}
12139 Generate floating point arithmetics for selected unit @var{unit}. The choices
12140 for @var{unit} are:
12144 Use the standard 387 floating point coprocessor present majority of chips and
12145 emulated otherwise. Code compiled with this option will run almost everywhere.
12146 The temporary results are computed in 80bit precision instead of precision
12147 specified by the type resulting in slightly different results compared to most
12148 of other chips. See @option{-ffloat-store} for more detailed description.
12150 This is the default choice for i386 compiler.
12153 Use scalar floating point instructions present in the SSE instruction set.
12154 This instruction set is supported by Pentium3 and newer chips, in the AMD line
12155 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
12156 instruction set supports only single precision arithmetics, thus the double and
12157 extended precision arithmetics is still done using 387. Later version, present
12158 only in Pentium4 and the future AMD x86-64 chips supports double precision
12161 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
12162 or @option{-msse2} switches to enable SSE extensions and make this option
12163 effective. For the x86-64 compiler, these extensions are enabled by default.
12165 The resulting code should be considerably faster in the majority of cases and avoid
12166 the numerical instability problems of 387 code, but may break some existing
12167 code that expects temporaries to be 80bit.
12169 This is the default choice for the x86-64 compiler.
12174 Attempt to utilize both instruction sets at once. This effectively double the
12175 amount of available registers and on chips with separate execution units for
12176 387 and SSE the execution resources too. Use this option with care, as it is
12177 still experimental, because the GCC register allocator does not model separate
12178 functional units well resulting in instable performance.
12181 @item -masm=@var{dialect}
12182 @opindex masm=@var{dialect}
12183 Output asm instructions using selected @var{dialect}. Supported
12184 choices are @samp{intel} or @samp{att} (the default one). Darwin does
12185 not support @samp{intel}.
12188 @itemx -mno-ieee-fp
12190 @opindex mno-ieee-fp
12191 Control whether or not the compiler uses IEEE floating point
12192 comparisons. These handle correctly the case where the result of a
12193 comparison is unordered.
12196 @opindex msoft-float
12197 Generate output containing library calls for floating point.
12198 @strong{Warning:} the requisite libraries are not part of GCC@.
12199 Normally the facilities of the machine's usual C compiler are used, but
12200 this can't be done directly in cross-compilation. You must make your
12201 own arrangements to provide suitable library functions for
12204 On machines where a function returns floating point results in the 80387
12205 register stack, some floating point opcodes may be emitted even if
12206 @option{-msoft-float} is used.
12208 @item -mno-fp-ret-in-387
12209 @opindex mno-fp-ret-in-387
12210 Do not use the FPU registers for return values of functions.
12212 The usual calling convention has functions return values of types
12213 @code{float} and @code{double} in an FPU register, even if there
12214 is no FPU@. The idea is that the operating system should emulate
12217 The option @option{-mno-fp-ret-in-387} causes such values to be returned
12218 in ordinary CPU registers instead.
12220 @item -mno-fancy-math-387
12221 @opindex mno-fancy-math-387
12222 Some 387 emulators do not support the @code{sin}, @code{cos} and
12223 @code{sqrt} instructions for the 387. Specify this option to avoid
12224 generating those instructions. This option is the default on FreeBSD,
12225 OpenBSD and NetBSD@. This option is overridden when @option{-march}
12226 indicates that the target cpu will always have an FPU and so the
12227 instruction will not need emulation. As of revision 2.6.1, these
12228 instructions are not generated unless you also use the
12229 @option{-funsafe-math-optimizations} switch.
12231 @item -malign-double
12232 @itemx -mno-align-double
12233 @opindex malign-double
12234 @opindex mno-align-double
12235 Control whether GCC aligns @code{double}, @code{long double}, and
12236 @code{long long} variables on a two word boundary or a one word
12237 boundary. Aligning @code{double} variables on a two word boundary will
12238 produce code that runs somewhat faster on a @samp{Pentium} at the
12239 expense of more memory.
12241 On x86-64, @option{-malign-double} is enabled by default.
12243 @strong{Warning:} if you use the @option{-malign-double} switch,
12244 structures containing the above types will be aligned differently than
12245 the published application binary interface specifications for the 386
12246 and will not be binary compatible with structures in code compiled
12247 without that switch.
12249 @item -m96bit-long-double
12250 @itemx -m128bit-long-double
12251 @opindex m96bit-long-double
12252 @opindex m128bit-long-double
12253 These switches control the size of @code{long double} type. The i386
12254 application binary interface specifies the size to be 96 bits,
12255 so @option{-m96bit-long-double} is the default in 32 bit mode.
12257 Modern architectures (Pentium and newer) would prefer @code{long double}
12258 to be aligned to an 8 or 16 byte boundary. In arrays or structures
12259 conforming to the ABI, this would not be possible. So specifying a
12260 @option{-m128bit-long-double} will align @code{long double}
12261 to a 16 byte boundary by padding the @code{long double} with an additional
12264 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
12265 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
12267 Notice that neither of these options enable any extra precision over the x87
12268 standard of 80 bits for a @code{long double}.
12270 @strong{Warning:} if you override the default value for your target ABI, the
12271 structures and arrays containing @code{long double} variables will change
12272 their size as well as function calling convention for function taking
12273 @code{long double} will be modified. Hence they will not be binary
12274 compatible with arrays or structures in code compiled without that switch.
12276 @item -mlarge-data-threshold=@var{number}
12277 @opindex mlarge-data-threshold=@var{number}
12278 When @option{-mcmodel=medium} is specified, the data greater than
12279 @var{threshold} are placed in large data section. This value must be the
12280 same across all object linked into the binary and defaults to 65535.
12284 Use a different function-calling convention, in which functions that
12285 take a fixed number of arguments return with the @code{ret} @var{num}
12286 instruction, which pops their arguments while returning. This saves one
12287 instruction in the caller since there is no need to pop the arguments
12290 You can specify that an individual function is called with this calling
12291 sequence with the function attribute @samp{stdcall}. You can also
12292 override the @option{-mrtd} option by using the function attribute
12293 @samp{cdecl}. @xref{Function Attributes}.
12295 @strong{Warning:} this calling convention is incompatible with the one
12296 normally used on Unix, so you cannot use it if you need to call
12297 libraries compiled with the Unix compiler.
12299 Also, you must provide function prototypes for all functions that
12300 take variable numbers of arguments (including @code{printf});
12301 otherwise incorrect code will be generated for calls to those
12304 In addition, seriously incorrect code will result if you call a
12305 function with too many arguments. (Normally, extra arguments are
12306 harmlessly ignored.)
12308 @item -mregparm=@var{num}
12310 Control how many registers are used to pass integer arguments. By
12311 default, no registers are used to pass arguments, and at most 3
12312 registers can be used. You can control this behavior for a specific
12313 function by using the function attribute @samp{regparm}.
12314 @xref{Function Attributes}.
12316 @strong{Warning:} if you use this switch, and
12317 @var{num} is nonzero, then you must build all modules with the same
12318 value, including any libraries. This includes the system libraries and
12322 @opindex msseregparm
12323 Use SSE register passing conventions for float and double arguments
12324 and return values. You can control this behavior for a specific
12325 function by using the function attribute @samp{sseregparm}.
12326 @xref{Function Attributes}.
12328 @strong{Warning:} if you use this switch then you must build all
12329 modules with the same value, including any libraries. This includes
12330 the system libraries and startup modules.
12332 @item -mvect8-ret-in-mem
12333 @opindex mvect8-ret-in-mem
12334 Return 8-byte vectors in memory instead of MMX registers. This is the
12335 default on Solaris~8 and 9 and VxWorks to match the ABI of the Sun
12336 Studio compilers until version 12. Later compiler versions (starting
12337 with Studio 12 Update~1) follow the ABI used by other x86 targets, which
12338 is the default on Solaris~10 and later. @emph{Only} use this option if
12339 you need to remain compatible with existing code produced by those
12340 previous compiler versions or older versions of GCC.
12349 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
12350 is specified, the significands of results of floating-point operations are
12351 rounded to 24 bits (single precision); @option{-mpc64} rounds the
12352 significands of results of floating-point operations to 53 bits (double
12353 precision) and @option{-mpc80} rounds the significands of results of
12354 floating-point operations to 64 bits (extended double precision), which is
12355 the default. When this option is used, floating-point operations in higher
12356 precisions are not available to the programmer without setting the FPU
12357 control word explicitly.
12359 Setting the rounding of floating-point operations to less than the default
12360 80 bits can speed some programs by 2% or more. Note that some mathematical
12361 libraries assume that extended precision (80 bit) floating-point operations
12362 are enabled by default; routines in such libraries could suffer significant
12363 loss of accuracy, typically through so-called "catastrophic cancellation",
12364 when this option is used to set the precision to less than extended precision.
12366 @item -mstackrealign
12367 @opindex mstackrealign
12368 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
12369 option will generate an alternate prologue and epilogue that realigns the
12370 runtime stack if necessary. This supports mixing legacy codes that keep
12371 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
12372 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
12373 applicable to individual functions.
12375 @item -mpreferred-stack-boundary=@var{num}
12376 @opindex mpreferred-stack-boundary
12377 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
12378 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
12379 the default is 4 (16 bytes or 128 bits).
12381 @item -mincoming-stack-boundary=@var{num}
12382 @opindex mincoming-stack-boundary
12383 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
12384 boundary. If @option{-mincoming-stack-boundary} is not specified,
12385 the one specified by @option{-mpreferred-stack-boundary} will be used.
12387 On Pentium and PentiumPro, @code{double} and @code{long double} values
12388 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
12389 suffer significant run time performance penalties. On Pentium III, the
12390 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
12391 properly if it is not 16 byte aligned.
12393 To ensure proper alignment of this values on the stack, the stack boundary
12394 must be as aligned as that required by any value stored on the stack.
12395 Further, every function must be generated such that it keeps the stack
12396 aligned. Thus calling a function compiled with a higher preferred
12397 stack boundary from a function compiled with a lower preferred stack
12398 boundary will most likely misalign the stack. It is recommended that
12399 libraries that use callbacks always use the default setting.
12401 This extra alignment does consume extra stack space, and generally
12402 increases code size. Code that is sensitive to stack space usage, such
12403 as embedded systems and operating system kernels, may want to reduce the
12404 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12431 @itemx -mno-fsgsbase
12461 These switches enable or disable the use of instructions in the MMX,
12462 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, FSGSBASE, RDRND,
12463 F16C, SSE4A, FMA4, XOP, LWP, ABM, BMI, or 3DNow!@: extended instruction sets.
12464 These extensions are also available as built-in functions: see
12465 @ref{X86 Built-in Functions}, for details of the functions enabled and
12466 disabled by these switches.
12468 To have SSE/SSE2 instructions generated automatically from floating-point
12469 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12471 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12472 generates new AVX instructions or AVX equivalence for all SSEx instructions
12475 These options will enable GCC to use these extended instructions in
12476 generated code, even without @option{-mfpmath=sse}. Applications which
12477 perform runtime CPU detection must compile separate files for each
12478 supported architecture, using the appropriate flags. In particular,
12479 the file containing the CPU detection code should be compiled without
12483 @itemx -mno-fused-madd
12484 @opindex mfused-madd
12485 @opindex mno-fused-madd
12486 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12487 instructions. The default is to use these instructions.
12491 This option instructs GCC to emit a @code{cld} instruction in the prologue
12492 of functions that use string instructions. String instructions depend on
12493 the DF flag to select between autoincrement or autodecrement mode. While the
12494 ABI specifies the DF flag to be cleared on function entry, some operating
12495 systems violate this specification by not clearing the DF flag in their
12496 exception dispatchers. The exception handler can be invoked with the DF flag
12497 set which leads to wrong direction mode, when string instructions are used.
12498 This option can be enabled by default on 32-bit x86 targets by configuring
12499 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12500 instructions can be suppressed with the @option{-mno-cld} compiler option
12504 @opindex mvzeroupper
12505 This option instructs GCC to emit a @code{vzeroupper} instruction
12506 before a transfer of control flow out of the function to minimize
12507 AVX to SSE transition penalty as well as remove unnecessary zeroupper
12512 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12513 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12514 data types. This is useful for high resolution counters that could be updated
12515 by multiple processors (or cores). This instruction is generated as part of
12516 atomic built-in functions: see @ref{Atomic Builtins} for details.
12520 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12521 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12522 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12523 SAHF are load and store instructions, respectively, for certain status flags.
12524 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12525 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12529 This option will enable GCC to use movbe instruction to implement
12530 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12534 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12535 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12536 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12540 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12541 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12542 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12543 variants) for single precision floating point arguments. These instructions
12544 are generated only when @option{-funsafe-math-optimizations} is enabled
12545 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12546 Note that while the throughput of the sequence is higher than the throughput
12547 of the non-reciprocal instruction, the precision of the sequence can be
12548 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12550 Note that GCC implements 1.0f/sqrtf(x) in terms of RSQRTSS (or RSQRTPS)
12551 already with @option{-ffast-math} (or the above option combination), and
12552 doesn't need @option{-mrecip}.
12554 @item -mveclibabi=@var{type}
12555 @opindex mveclibabi
12556 Specifies the ABI type to use for vectorizing intrinsics using an
12557 external library. Supported types are @code{svml} for the Intel short
12558 vector math library and @code{acml} for the AMD math core library style
12559 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12560 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12561 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12562 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12563 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12564 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12565 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12566 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12567 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12568 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12569 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12570 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12571 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12572 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12573 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12574 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12575 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12576 compatible library will have to be specified at link time.
12578 @item -mabi=@var{name}
12580 Generate code for the specified calling convention. Permissible values
12581 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12582 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12583 ABI when targeting Windows. On all other systems, the default is the
12584 SYSV ABI. You can control this behavior for a specific function by
12585 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12586 @xref{Function Attributes}.
12589 @itemx -mno-push-args
12590 @opindex mpush-args
12591 @opindex mno-push-args
12592 Use PUSH operations to store outgoing parameters. This method is shorter
12593 and usually equally fast as method using SUB/MOV operations and is enabled
12594 by default. In some cases disabling it may improve performance because of
12595 improved scheduling and reduced dependencies.
12597 @item -maccumulate-outgoing-args
12598 @opindex maccumulate-outgoing-args
12599 If enabled, the maximum amount of space required for outgoing arguments will be
12600 computed in the function prologue. This is faster on most modern CPUs
12601 because of reduced dependencies, improved scheduling and reduced stack usage
12602 when preferred stack boundary is not equal to 2. The drawback is a notable
12603 increase in code size. This switch implies @option{-mno-push-args}.
12607 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12608 on thread-safe exception handling must compile and link all code with the
12609 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12610 @option{-D_MT}; when linking, it links in a special thread helper library
12611 @option{-lmingwthrd} which cleans up per thread exception handling data.
12613 @item -mno-align-stringops
12614 @opindex mno-align-stringops
12615 Do not align destination of inlined string operations. This switch reduces
12616 code size and improves performance in case the destination is already aligned,
12617 but GCC doesn't know about it.
12619 @item -minline-all-stringops
12620 @opindex minline-all-stringops
12621 By default GCC inlines string operations only when destination is known to be
12622 aligned at least to 4 byte boundary. This enables more inlining, increase code
12623 size, but may improve performance of code that depends on fast memcpy, strlen
12624 and memset for short lengths.
12626 @item -minline-stringops-dynamically
12627 @opindex minline-stringops-dynamically
12628 For string operation of unknown size, inline runtime checks so for small
12629 blocks inline code is used, while for large blocks library call is used.
12631 @item -mstringop-strategy=@var{alg}
12632 @opindex mstringop-strategy=@var{alg}
12633 Overwrite internal decision heuristic about particular algorithm to inline
12634 string operation with. The allowed values are @code{rep_byte},
12635 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12636 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12637 expanding inline loop, @code{libcall} for always expanding library call.
12639 @item -momit-leaf-frame-pointer
12640 @opindex momit-leaf-frame-pointer
12641 Don't keep the frame pointer in a register for leaf functions. This
12642 avoids the instructions to save, set up and restore frame pointers and
12643 makes an extra register available in leaf functions. The option
12644 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12645 which might make debugging harder.
12647 @item -mtls-direct-seg-refs
12648 @itemx -mno-tls-direct-seg-refs
12649 @opindex mtls-direct-seg-refs
12650 Controls whether TLS variables may be accessed with offsets from the
12651 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12652 or whether the thread base pointer must be added. Whether or not this
12653 is legal depends on the operating system, and whether it maps the
12654 segment to cover the entire TLS area.
12656 For systems that use GNU libc, the default is on.
12659 @itemx -mno-sse2avx
12661 Specify that the assembler should encode SSE instructions with VEX
12662 prefix. The option @option{-mavx} turns this on by default.
12667 If profiling is active @option{-pg} put the profiling
12668 counter call before prologue.
12669 Note: On x86 architectures the attribute @code{ms_hook_prologue}
12670 isn't possible at the moment for @option{-mfentry} and @option{-pg}.
12673 @itemx -mno-8bit-idiv
12675 On some processors, like Intel Atom, 8bit unsigned integer divide is
12676 much faster than 32bit/64bit integer divide. This option will generate a
12677 runt-time check. If both dividend and divisor are within range of 0
12678 to 255, 8bit unsigned integer divide will be used instead of
12679 32bit/64bit integer divide.
12683 These @samp{-m} switches are supported in addition to the above
12684 on AMD x86-64 processors in 64-bit environments.
12691 Generate code for a 32-bit or 64-bit environment.
12692 The 32-bit environment sets int, long and pointer to 32 bits and
12693 generates code that runs on any i386 system.
12694 The 64-bit environment sets int to 32 bits and long and pointer
12695 to 64 bits and generates code for AMD's x86-64 architecture. For
12696 darwin only the -m64 option turns off the @option{-fno-pic} and
12697 @option{-mdynamic-no-pic} options.
12699 @item -mno-red-zone
12700 @opindex mno-red-zone
12701 Do not use a so called red zone for x86-64 code. The red zone is mandated
12702 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12703 stack pointer that will not be modified by signal or interrupt handlers
12704 and therefore can be used for temporary data without adjusting the stack
12705 pointer. The flag @option{-mno-red-zone} disables this red zone.
12707 @item -mcmodel=small
12708 @opindex mcmodel=small
12709 Generate code for the small code model: the program and its symbols must
12710 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12711 Programs can be statically or dynamically linked. This is the default
12714 @item -mcmodel=kernel
12715 @opindex mcmodel=kernel
12716 Generate code for the kernel code model. The kernel runs in the
12717 negative 2 GB of the address space.
12718 This model has to be used for Linux kernel code.
12720 @item -mcmodel=medium
12721 @opindex mcmodel=medium
12722 Generate code for the medium model: The program is linked in the lower 2
12723 GB of the address space. Small symbols are also placed there. Symbols
12724 with sizes larger than @option{-mlarge-data-threshold} are put into
12725 large data or bss sections and can be located above 2GB. Programs can
12726 be statically or dynamically linked.
12728 @item -mcmodel=large
12729 @opindex mcmodel=large
12730 Generate code for the large model: This model makes no assumptions
12731 about addresses and sizes of sections.
12734 @node IA-64 Options
12735 @subsection IA-64 Options
12736 @cindex IA-64 Options
12738 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12742 @opindex mbig-endian
12743 Generate code for a big endian target. This is the default for HP-UX@.
12745 @item -mlittle-endian
12746 @opindex mlittle-endian
12747 Generate code for a little endian target. This is the default for AIX5
12753 @opindex mno-gnu-as
12754 Generate (or don't) code for the GNU assembler. This is the default.
12755 @c Also, this is the default if the configure option @option{--with-gnu-as}
12761 @opindex mno-gnu-ld
12762 Generate (or don't) code for the GNU linker. This is the default.
12763 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12768 Generate code that does not use a global pointer register. The result
12769 is not position independent code, and violates the IA-64 ABI@.
12771 @item -mvolatile-asm-stop
12772 @itemx -mno-volatile-asm-stop
12773 @opindex mvolatile-asm-stop
12774 @opindex mno-volatile-asm-stop
12775 Generate (or don't) a stop bit immediately before and after volatile asm
12778 @item -mregister-names
12779 @itemx -mno-register-names
12780 @opindex mregister-names
12781 @opindex mno-register-names
12782 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12783 the stacked registers. This may make assembler output more readable.
12789 Disable (or enable) optimizations that use the small data section. This may
12790 be useful for working around optimizer bugs.
12792 @item -mconstant-gp
12793 @opindex mconstant-gp
12794 Generate code that uses a single constant global pointer value. This is
12795 useful when compiling kernel code.
12799 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12800 This is useful when compiling firmware code.
12802 @item -minline-float-divide-min-latency
12803 @opindex minline-float-divide-min-latency
12804 Generate code for inline divides of floating point values
12805 using the minimum latency algorithm.
12807 @item -minline-float-divide-max-throughput
12808 @opindex minline-float-divide-max-throughput
12809 Generate code for inline divides of floating point values
12810 using the maximum throughput algorithm.
12812 @item -mno-inline-float-divide
12813 @opindex mno-inline-float-divide
12814 Do not generate inline code for divides of floating point values.
12816 @item -minline-int-divide-min-latency
12817 @opindex minline-int-divide-min-latency
12818 Generate code for inline divides of integer values
12819 using the minimum latency algorithm.
12821 @item -minline-int-divide-max-throughput
12822 @opindex minline-int-divide-max-throughput
12823 Generate code for inline divides of integer values
12824 using the maximum throughput algorithm.
12826 @item -mno-inline-int-divide
12827 @opindex mno-inline-int-divide
12828 Do not generate inline code for divides of integer values.
12830 @item -minline-sqrt-min-latency
12831 @opindex minline-sqrt-min-latency
12832 Generate code for inline square roots
12833 using the minimum latency algorithm.
12835 @item -minline-sqrt-max-throughput
12836 @opindex minline-sqrt-max-throughput
12837 Generate code for inline square roots
12838 using the maximum throughput algorithm.
12840 @item -mno-inline-sqrt
12841 @opindex mno-inline-sqrt
12842 Do not generate inline code for sqrt.
12845 @itemx -mno-fused-madd
12846 @opindex mfused-madd
12847 @opindex mno-fused-madd
12848 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12849 instructions. The default is to use these instructions.
12851 @item -mno-dwarf2-asm
12852 @itemx -mdwarf2-asm
12853 @opindex mno-dwarf2-asm
12854 @opindex mdwarf2-asm
12855 Don't (or do) generate assembler code for the DWARF2 line number debugging
12856 info. This may be useful when not using the GNU assembler.
12858 @item -mearly-stop-bits
12859 @itemx -mno-early-stop-bits
12860 @opindex mearly-stop-bits
12861 @opindex mno-early-stop-bits
12862 Allow stop bits to be placed earlier than immediately preceding the
12863 instruction that triggered the stop bit. This can improve instruction
12864 scheduling, but does not always do so.
12866 @item -mfixed-range=@var{register-range}
12867 @opindex mfixed-range
12868 Generate code treating the given register range as fixed registers.
12869 A fixed register is one that the register allocator can not use. This is
12870 useful when compiling kernel code. A register range is specified as
12871 two registers separated by a dash. Multiple register ranges can be
12872 specified separated by a comma.
12874 @item -mtls-size=@var{tls-size}
12876 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12879 @item -mtune=@var{cpu-type}
12881 Tune the instruction scheduling for a particular CPU, Valid values are
12882 itanium, itanium1, merced, itanium2, and mckinley.
12888 Generate code for a 32-bit or 64-bit environment.
12889 The 32-bit environment sets int, long and pointer to 32 bits.
12890 The 64-bit environment sets int to 32 bits and long and pointer
12891 to 64 bits. These are HP-UX specific flags.
12893 @item -mno-sched-br-data-spec
12894 @itemx -msched-br-data-spec
12895 @opindex mno-sched-br-data-spec
12896 @opindex msched-br-data-spec
12897 (Dis/En)able data speculative scheduling before reload.
12898 This will result in generation of the ld.a instructions and
12899 the corresponding check instructions (ld.c / chk.a).
12900 The default is 'disable'.
12902 @item -msched-ar-data-spec
12903 @itemx -mno-sched-ar-data-spec
12904 @opindex msched-ar-data-spec
12905 @opindex mno-sched-ar-data-spec
12906 (En/Dis)able data speculative scheduling after reload.
12907 This will result in generation of the ld.a instructions and
12908 the corresponding check instructions (ld.c / chk.a).
12909 The default is 'enable'.
12911 @item -mno-sched-control-spec
12912 @itemx -msched-control-spec
12913 @opindex mno-sched-control-spec
12914 @opindex msched-control-spec
12915 (Dis/En)able control speculative scheduling. This feature is
12916 available only during region scheduling (i.e.@: before reload).
12917 This will result in generation of the ld.s instructions and
12918 the corresponding check instructions chk.s .
12919 The default is 'disable'.
12921 @item -msched-br-in-data-spec
12922 @itemx -mno-sched-br-in-data-spec
12923 @opindex msched-br-in-data-spec
12924 @opindex mno-sched-br-in-data-spec
12925 (En/Dis)able speculative scheduling of the instructions that
12926 are dependent on the data speculative loads before reload.
12927 This is effective only with @option{-msched-br-data-spec} enabled.
12928 The default is 'enable'.
12930 @item -msched-ar-in-data-spec
12931 @itemx -mno-sched-ar-in-data-spec
12932 @opindex msched-ar-in-data-spec
12933 @opindex mno-sched-ar-in-data-spec
12934 (En/Dis)able speculative scheduling of the instructions that
12935 are dependent on the data speculative loads after reload.
12936 This is effective only with @option{-msched-ar-data-spec} enabled.
12937 The default is 'enable'.
12939 @item -msched-in-control-spec
12940 @itemx -mno-sched-in-control-spec
12941 @opindex msched-in-control-spec
12942 @opindex mno-sched-in-control-spec
12943 (En/Dis)able speculative scheduling of the instructions that
12944 are dependent on the control speculative loads.
12945 This is effective only with @option{-msched-control-spec} enabled.
12946 The default is 'enable'.
12948 @item -mno-sched-prefer-non-data-spec-insns
12949 @itemx -msched-prefer-non-data-spec-insns
12950 @opindex mno-sched-prefer-non-data-spec-insns
12951 @opindex msched-prefer-non-data-spec-insns
12952 If enabled, data speculative instructions will be chosen for schedule
12953 only if there are no other choices at the moment. This will make
12954 the use of the data speculation much more conservative.
12955 The default is 'disable'.
12957 @item -mno-sched-prefer-non-control-spec-insns
12958 @itemx -msched-prefer-non-control-spec-insns
12959 @opindex mno-sched-prefer-non-control-spec-insns
12960 @opindex msched-prefer-non-control-spec-insns
12961 If enabled, control speculative instructions will be chosen for schedule
12962 only if there are no other choices at the moment. This will make
12963 the use of the control speculation much more conservative.
12964 The default is 'disable'.
12966 @item -mno-sched-count-spec-in-critical-path
12967 @itemx -msched-count-spec-in-critical-path
12968 @opindex mno-sched-count-spec-in-critical-path
12969 @opindex msched-count-spec-in-critical-path
12970 If enabled, speculative dependencies will be considered during
12971 computation of the instructions priorities. This will make the use of the
12972 speculation a bit more conservative.
12973 The default is 'disable'.
12975 @item -msched-spec-ldc
12976 @opindex msched-spec-ldc
12977 Use a simple data speculation check. This option is on by default.
12979 @item -msched-control-spec-ldc
12980 @opindex msched-spec-ldc
12981 Use a simple check for control speculation. This option is on by default.
12983 @item -msched-stop-bits-after-every-cycle
12984 @opindex msched-stop-bits-after-every-cycle
12985 Place a stop bit after every cycle when scheduling. This option is on
12988 @item -msched-fp-mem-deps-zero-cost
12989 @opindex msched-fp-mem-deps-zero-cost
12990 Assume that floating-point stores and loads are not likely to cause a conflict
12991 when placed into the same instruction group. This option is disabled by
12994 @item -msel-sched-dont-check-control-spec
12995 @opindex msel-sched-dont-check-control-spec
12996 Generate checks for control speculation in selective scheduling.
12997 This flag is disabled by default.
12999 @item -msched-max-memory-insns=@var{max-insns}
13000 @opindex msched-max-memory-insns
13001 Limit on the number of memory insns per instruction group, giving lower
13002 priority to subsequent memory insns attempting to schedule in the same
13003 instruction group. Frequently useful to prevent cache bank conflicts.
13004 The default value is 1.
13006 @item -msched-max-memory-insns-hard-limit
13007 @opindex msched-max-memory-insns-hard-limit
13008 Disallow more than `msched-max-memory-insns' in instruction group.
13009 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
13010 when limit is reached but may still schedule memory operations.
13014 @node IA-64/VMS Options
13015 @subsection IA-64/VMS Options
13017 These @samp{-m} options are defined for the IA-64/VMS implementations:
13020 @item -mvms-return-codes
13021 @opindex mvms-return-codes
13022 Return VMS condition codes from main. The default is to return POSIX
13023 style condition (e.g.@ error) codes.
13025 @item -mdebug-main=@var{prefix}
13026 @opindex mdebug-main=@var{prefix}
13027 Flag the first routine whose name starts with @var{prefix} as the main
13028 routine for the debugger.
13032 Default to 64bit memory allocation routines.
13036 @subsection LM32 Options
13037 @cindex LM32 options
13039 These @option{-m} options are defined for the Lattice Mico32 architecture:
13042 @item -mbarrel-shift-enabled
13043 @opindex mbarrel-shift-enabled
13044 Enable barrel-shift instructions.
13046 @item -mdivide-enabled
13047 @opindex mdivide-enabled
13048 Enable divide and modulus instructions.
13050 @item -mmultiply-enabled
13051 @opindex multiply-enabled
13052 Enable multiply instructions.
13054 @item -msign-extend-enabled
13055 @opindex msign-extend-enabled
13056 Enable sign extend instructions.
13058 @item -muser-enabled
13059 @opindex muser-enabled
13060 Enable user-defined instructions.
13065 @subsection M32C Options
13066 @cindex M32C options
13069 @item -mcpu=@var{name}
13071 Select the CPU for which code is generated. @var{name} may be one of
13072 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
13073 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
13074 the M32C/80 series.
13078 Specifies that the program will be run on the simulator. This causes
13079 an alternate runtime library to be linked in which supports, for
13080 example, file I/O@. You must not use this option when generating
13081 programs that will run on real hardware; you must provide your own
13082 runtime library for whatever I/O functions are needed.
13084 @item -memregs=@var{number}
13086 Specifies the number of memory-based pseudo-registers GCC will use
13087 during code generation. These pseudo-registers will be used like real
13088 registers, so there is a tradeoff between GCC's ability to fit the
13089 code into available registers, and the performance penalty of using
13090 memory instead of registers. Note that all modules in a program must
13091 be compiled with the same value for this option. Because of that, you
13092 must not use this option with the default runtime libraries gcc
13097 @node M32R/D Options
13098 @subsection M32R/D Options
13099 @cindex M32R/D options
13101 These @option{-m} options are defined for Renesas M32R/D architectures:
13106 Generate code for the M32R/2@.
13110 Generate code for the M32R/X@.
13114 Generate code for the M32R@. This is the default.
13116 @item -mmodel=small
13117 @opindex mmodel=small
13118 Assume all objects live in the lower 16MB of memory (so that their addresses
13119 can be loaded with the @code{ld24} instruction), and assume all subroutines
13120 are reachable with the @code{bl} instruction.
13121 This is the default.
13123 The addressability of a particular object can be set with the
13124 @code{model} attribute.
13126 @item -mmodel=medium
13127 @opindex mmodel=medium
13128 Assume objects may be anywhere in the 32-bit address space (the compiler
13129 will generate @code{seth/add3} instructions to load their addresses), and
13130 assume all subroutines are reachable with the @code{bl} instruction.
13132 @item -mmodel=large
13133 @opindex mmodel=large
13134 Assume objects may be anywhere in the 32-bit address space (the compiler
13135 will generate @code{seth/add3} instructions to load their addresses), and
13136 assume subroutines may not be reachable with the @code{bl} instruction
13137 (the compiler will generate the much slower @code{seth/add3/jl}
13138 instruction sequence).
13141 @opindex msdata=none
13142 Disable use of the small data area. Variables will be put into
13143 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
13144 @code{section} attribute has been specified).
13145 This is the default.
13147 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
13148 Objects may be explicitly put in the small data area with the
13149 @code{section} attribute using one of these sections.
13151 @item -msdata=sdata
13152 @opindex msdata=sdata
13153 Put small global and static data in the small data area, but do not
13154 generate special code to reference them.
13157 @opindex msdata=use
13158 Put small global and static data in the small data area, and generate
13159 special instructions to reference them.
13163 @cindex smaller data references
13164 Put global and static objects less than or equal to @var{num} bytes
13165 into the small data or bss sections instead of the normal data or bss
13166 sections. The default value of @var{num} is 8.
13167 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
13168 for this option to have any effect.
13170 All modules should be compiled with the same @option{-G @var{num}} value.
13171 Compiling with different values of @var{num} may or may not work; if it
13172 doesn't the linker will give an error message---incorrect code will not be
13177 Makes the M32R specific code in the compiler display some statistics
13178 that might help in debugging programs.
13180 @item -malign-loops
13181 @opindex malign-loops
13182 Align all loops to a 32-byte boundary.
13184 @item -mno-align-loops
13185 @opindex mno-align-loops
13186 Do not enforce a 32-byte alignment for loops. This is the default.
13188 @item -missue-rate=@var{number}
13189 @opindex missue-rate=@var{number}
13190 Issue @var{number} instructions per cycle. @var{number} can only be 1
13193 @item -mbranch-cost=@var{number}
13194 @opindex mbranch-cost=@var{number}
13195 @var{number} can only be 1 or 2. If it is 1 then branches will be
13196 preferred over conditional code, if it is 2, then the opposite will
13199 @item -mflush-trap=@var{number}
13200 @opindex mflush-trap=@var{number}
13201 Specifies the trap number to use to flush the cache. The default is
13202 12. Valid numbers are between 0 and 15 inclusive.
13204 @item -mno-flush-trap
13205 @opindex mno-flush-trap
13206 Specifies that the cache cannot be flushed by using a trap.
13208 @item -mflush-func=@var{name}
13209 @opindex mflush-func=@var{name}
13210 Specifies the name of the operating system function to call to flush
13211 the cache. The default is @emph{_flush_cache}, but a function call
13212 will only be used if a trap is not available.
13214 @item -mno-flush-func
13215 @opindex mno-flush-func
13216 Indicates that there is no OS function for flushing the cache.
13220 @node M680x0 Options
13221 @subsection M680x0 Options
13222 @cindex M680x0 options
13224 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
13225 The default settings depend on which architecture was selected when
13226 the compiler was configured; the defaults for the most common choices
13230 @item -march=@var{arch}
13232 Generate code for a specific M680x0 or ColdFire instruction set
13233 architecture. Permissible values of @var{arch} for M680x0
13234 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
13235 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
13236 architectures are selected according to Freescale's ISA classification
13237 and the permissible values are: @samp{isaa}, @samp{isaaplus},
13238 @samp{isab} and @samp{isac}.
13240 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
13241 code for a ColdFire target. The @var{arch} in this macro is one of the
13242 @option{-march} arguments given above.
13244 When used together, @option{-march} and @option{-mtune} select code
13245 that runs on a family of similar processors but that is optimized
13246 for a particular microarchitecture.
13248 @item -mcpu=@var{cpu}
13250 Generate code for a specific M680x0 or ColdFire processor.
13251 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
13252 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
13253 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
13254 below, which also classifies the CPUs into families:
13256 @multitable @columnfractions 0.20 0.80
13257 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
13258 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
13259 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
13260 @item @samp{5206e} @tab @samp{5206e}
13261 @item @samp{5208} @tab @samp{5207} @samp{5208}
13262 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
13263 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
13264 @item @samp{5216} @tab @samp{5214} @samp{5216}
13265 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
13266 @item @samp{5225} @tab @samp{5224} @samp{5225}
13267 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
13268 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
13269 @item @samp{5249} @tab @samp{5249}
13270 @item @samp{5250} @tab @samp{5250}
13271 @item @samp{5271} @tab @samp{5270} @samp{5271}
13272 @item @samp{5272} @tab @samp{5272}
13273 @item @samp{5275} @tab @samp{5274} @samp{5275}
13274 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
13275 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
13276 @item @samp{5307} @tab @samp{5307}
13277 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
13278 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
13279 @item @samp{5407} @tab @samp{5407}
13280 @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}
13283 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
13284 @var{arch} is compatible with @var{cpu}. Other combinations of
13285 @option{-mcpu} and @option{-march} are rejected.
13287 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
13288 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
13289 where the value of @var{family} is given by the table above.
13291 @item -mtune=@var{tune}
13293 Tune the code for a particular microarchitecture, within the
13294 constraints set by @option{-march} and @option{-mcpu}.
13295 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
13296 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
13297 and @samp{cpu32}. The ColdFire microarchitectures
13298 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
13300 You can also use @option{-mtune=68020-40} for code that needs
13301 to run relatively well on 68020, 68030 and 68040 targets.
13302 @option{-mtune=68020-60} is similar but includes 68060 targets
13303 as well. These two options select the same tuning decisions as
13304 @option{-m68020-40} and @option{-m68020-60} respectively.
13306 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
13307 when tuning for 680x0 architecture @var{arch}. It also defines
13308 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
13309 option is used. If gcc is tuning for a range of architectures,
13310 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
13311 it defines the macros for every architecture in the range.
13313 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
13314 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
13315 of the arguments given above.
13321 Generate output for a 68000. This is the default
13322 when the compiler is configured for 68000-based systems.
13323 It is equivalent to @option{-march=68000}.
13325 Use this option for microcontrollers with a 68000 or EC000 core,
13326 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
13330 Generate output for a 68010. This is the default
13331 when the compiler is configured for 68010-based systems.
13332 It is equivalent to @option{-march=68010}.
13338 Generate output for a 68020. This is the default
13339 when the compiler is configured for 68020-based systems.
13340 It is equivalent to @option{-march=68020}.
13344 Generate output for a 68030. This is the default when the compiler is
13345 configured for 68030-based systems. It is equivalent to
13346 @option{-march=68030}.
13350 Generate output for a 68040. This is the default when the compiler is
13351 configured for 68040-based systems. It is equivalent to
13352 @option{-march=68040}.
13354 This option inhibits the use of 68881/68882 instructions that have to be
13355 emulated by software on the 68040. Use this option if your 68040 does not
13356 have code to emulate those instructions.
13360 Generate output for a 68060. This is the default when the compiler is
13361 configured for 68060-based systems. It is equivalent to
13362 @option{-march=68060}.
13364 This option inhibits the use of 68020 and 68881/68882 instructions that
13365 have to be emulated by software on the 68060. Use this option if your 68060
13366 does not have code to emulate those instructions.
13370 Generate output for a CPU32. This is the default
13371 when the compiler is configured for CPU32-based systems.
13372 It is equivalent to @option{-march=cpu32}.
13374 Use this option for microcontrollers with a
13375 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
13376 68336, 68340, 68341, 68349 and 68360.
13380 Generate output for a 520X ColdFire CPU@. This is the default
13381 when the compiler is configured for 520X-based systems.
13382 It is equivalent to @option{-mcpu=5206}, and is now deprecated
13383 in favor of that option.
13385 Use this option for microcontroller with a 5200 core, including
13386 the MCF5202, MCF5203, MCF5204 and MCF5206.
13390 Generate output for a 5206e ColdFire CPU@. The option is now
13391 deprecated in favor of the equivalent @option{-mcpu=5206e}.
13395 Generate output for a member of the ColdFire 528X family.
13396 The option is now deprecated in favor of the equivalent
13397 @option{-mcpu=528x}.
13401 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
13402 in favor of the equivalent @option{-mcpu=5307}.
13406 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
13407 in favor of the equivalent @option{-mcpu=5407}.
13411 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
13412 This includes use of hardware floating point instructions.
13413 The option is equivalent to @option{-mcpu=547x}, and is now
13414 deprecated in favor of that option.
13418 Generate output for a 68040, without using any of the new instructions.
13419 This results in code which can run relatively efficiently on either a
13420 68020/68881 or a 68030 or a 68040. The generated code does use the
13421 68881 instructions that are emulated on the 68040.
13423 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
13427 Generate output for a 68060, without using any of the new instructions.
13428 This results in code which can run relatively efficiently on either a
13429 68020/68881 or a 68030 or a 68040. The generated code does use the
13430 68881 instructions that are emulated on the 68060.
13432 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
13436 @opindex mhard-float
13438 Generate floating-point instructions. This is the default for 68020
13439 and above, and for ColdFire devices that have an FPU@. It defines the
13440 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
13441 on ColdFire targets.
13444 @opindex msoft-float
13445 Do not generate floating-point instructions; use library calls instead.
13446 This is the default for 68000, 68010, and 68832 targets. It is also
13447 the default for ColdFire devices that have no FPU.
13453 Generate (do not generate) ColdFire hardware divide and remainder
13454 instructions. If @option{-march} is used without @option{-mcpu},
13455 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
13456 architectures. Otherwise, the default is taken from the target CPU
13457 (either the default CPU, or the one specified by @option{-mcpu}). For
13458 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
13459 @option{-mcpu=5206e}.
13461 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
13465 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13466 Additionally, parameters passed on the stack are also aligned to a
13467 16-bit boundary even on targets whose API mandates promotion to 32-bit.
13471 Do not consider type @code{int} to be 16 bits wide. This is the default.
13474 @itemx -mno-bitfield
13475 @opindex mnobitfield
13476 @opindex mno-bitfield
13477 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
13478 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
13482 Do use the bit-field instructions. The @option{-m68020} option implies
13483 @option{-mbitfield}. This is the default if you use a configuration
13484 designed for a 68020.
13488 Use a different function-calling convention, in which functions
13489 that take a fixed number of arguments return with the @code{rtd}
13490 instruction, which pops their arguments while returning. This
13491 saves one instruction in the caller since there is no need to pop
13492 the arguments there.
13494 This calling convention is incompatible with the one normally
13495 used on Unix, so you cannot use it if you need to call libraries
13496 compiled with the Unix compiler.
13498 Also, you must provide function prototypes for all functions that
13499 take variable numbers of arguments (including @code{printf});
13500 otherwise incorrect code will be generated for calls to those
13503 In addition, seriously incorrect code will result if you call a
13504 function with too many arguments. (Normally, extra arguments are
13505 harmlessly ignored.)
13507 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13508 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13512 Do not use the calling conventions selected by @option{-mrtd}.
13513 This is the default.
13516 @itemx -mno-align-int
13517 @opindex malign-int
13518 @opindex mno-align-int
13519 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13520 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13521 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13522 Aligning variables on 32-bit boundaries produces code that runs somewhat
13523 faster on processors with 32-bit busses at the expense of more memory.
13525 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13526 align structures containing the above types differently than
13527 most published application binary interface specifications for the m68k.
13531 Use the pc-relative addressing mode of the 68000 directly, instead of
13532 using a global offset table. At present, this option implies @option{-fpic},
13533 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13534 not presently supported with @option{-mpcrel}, though this could be supported for
13535 68020 and higher processors.
13537 @item -mno-strict-align
13538 @itemx -mstrict-align
13539 @opindex mno-strict-align
13540 @opindex mstrict-align
13541 Do not (do) assume that unaligned memory references will be handled by
13545 Generate code that allows the data segment to be located in a different
13546 area of memory from the text segment. This allows for execute in place in
13547 an environment without virtual memory management. This option implies
13550 @item -mno-sep-data
13551 Generate code that assumes that the data segment follows the text segment.
13552 This is the default.
13554 @item -mid-shared-library
13555 Generate code that supports shared libraries via the library ID method.
13556 This allows for execute in place and shared libraries in an environment
13557 without virtual memory management. This option implies @option{-fPIC}.
13559 @item -mno-id-shared-library
13560 Generate code that doesn't assume ID based shared libraries are being used.
13561 This is the default.
13563 @item -mshared-library-id=n
13564 Specified the identification number of the ID based shared library being
13565 compiled. Specifying a value of 0 will generate more compact code, specifying
13566 other values will force the allocation of that number to the current
13567 library but is no more space or time efficient than omitting this option.
13573 When generating position-independent code for ColdFire, generate code
13574 that works if the GOT has more than 8192 entries. This code is
13575 larger and slower than code generated without this option. On M680x0
13576 processors, this option is not needed; @option{-fPIC} suffices.
13578 GCC normally uses a single instruction to load values from the GOT@.
13579 While this is relatively efficient, it only works if the GOT
13580 is smaller than about 64k. Anything larger causes the linker
13581 to report an error such as:
13583 @cindex relocation truncated to fit (ColdFire)
13585 relocation truncated to fit: R_68K_GOT16O foobar
13588 If this happens, you should recompile your code with @option{-mxgot}.
13589 It should then work with very large GOTs. However, code generated with
13590 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13591 the value of a global symbol.
13593 Note that some linkers, including newer versions of the GNU linker,
13594 can create multiple GOTs and sort GOT entries. If you have such a linker,
13595 you should only need to use @option{-mxgot} when compiling a single
13596 object file that accesses more than 8192 GOT entries. Very few do.
13598 These options have no effect unless GCC is generating
13599 position-independent code.
13603 @node M68hc1x Options
13604 @subsection M68hc1x Options
13605 @cindex M68hc1x options
13607 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13608 microcontrollers. The default values for these options depends on
13609 which style of microcontroller was selected when the compiler was configured;
13610 the defaults for the most common choices are given below.
13617 Generate output for a 68HC11. This is the default
13618 when the compiler is configured for 68HC11-based systems.
13624 Generate output for a 68HC12. This is the default
13625 when the compiler is configured for 68HC12-based systems.
13631 Generate output for a 68HCS12.
13633 @item -mauto-incdec
13634 @opindex mauto-incdec
13635 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13642 Enable the use of 68HC12 min and max instructions.
13645 @itemx -mno-long-calls
13646 @opindex mlong-calls
13647 @opindex mno-long-calls
13648 Treat all calls as being far away (near). If calls are assumed to be
13649 far away, the compiler will use the @code{call} instruction to
13650 call a function and the @code{rtc} instruction for returning.
13654 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13656 @item -msoft-reg-count=@var{count}
13657 @opindex msoft-reg-count
13658 Specify the number of pseudo-soft registers which are used for the
13659 code generation. The maximum number is 32. Using more pseudo-soft
13660 register may or may not result in better code depending on the program.
13661 The default is 4 for 68HC11 and 2 for 68HC12.
13665 @node MCore Options
13666 @subsection MCore Options
13667 @cindex MCore options
13669 These are the @samp{-m} options defined for the Motorola M*Core
13675 @itemx -mno-hardlit
13677 @opindex mno-hardlit
13678 Inline constants into the code stream if it can be done in two
13679 instructions or less.
13685 Use the divide instruction. (Enabled by default).
13687 @item -mrelax-immediate
13688 @itemx -mno-relax-immediate
13689 @opindex mrelax-immediate
13690 @opindex mno-relax-immediate
13691 Allow arbitrary sized immediates in bit operations.
13693 @item -mwide-bitfields
13694 @itemx -mno-wide-bitfields
13695 @opindex mwide-bitfields
13696 @opindex mno-wide-bitfields
13697 Always treat bit-fields as int-sized.
13699 @item -m4byte-functions
13700 @itemx -mno-4byte-functions
13701 @opindex m4byte-functions
13702 @opindex mno-4byte-functions
13703 Force all functions to be aligned to a four byte boundary.
13705 @item -mcallgraph-data
13706 @itemx -mno-callgraph-data
13707 @opindex mcallgraph-data
13708 @opindex mno-callgraph-data
13709 Emit callgraph information.
13712 @itemx -mno-slow-bytes
13713 @opindex mslow-bytes
13714 @opindex mno-slow-bytes
13715 Prefer word access when reading byte quantities.
13717 @item -mlittle-endian
13718 @itemx -mbig-endian
13719 @opindex mlittle-endian
13720 @opindex mbig-endian
13721 Generate code for a little endian target.
13727 Generate code for the 210 processor.
13731 Assume that run-time support has been provided and so omit the
13732 simulator library (@file{libsim.a)} from the linker command line.
13734 @item -mstack-increment=@var{size}
13735 @opindex mstack-increment
13736 Set the maximum amount for a single stack increment operation. Large
13737 values can increase the speed of programs which contain functions
13738 that need a large amount of stack space, but they can also trigger a
13739 segmentation fault if the stack is extended too much. The default
13745 @subsection MeP Options
13746 @cindex MeP options
13752 Enables the @code{abs} instruction, which is the absolute difference
13753 between two registers.
13757 Enables all the optional instructions - average, multiply, divide, bit
13758 operations, leading zero, absolute difference, min/max, clip, and
13764 Enables the @code{ave} instruction, which computes the average of two
13767 @item -mbased=@var{n}
13769 Variables of size @var{n} bytes or smaller will be placed in the
13770 @code{.based} section by default. Based variables use the @code{$tp}
13771 register as a base register, and there is a 128 byte limit to the
13772 @code{.based} section.
13776 Enables the bit operation instructions - bit test (@code{btstm}), set
13777 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13778 test-and-set (@code{tas}).
13780 @item -mc=@var{name}
13782 Selects which section constant data will be placed in. @var{name} may
13783 be @code{tiny}, @code{near}, or @code{far}.
13787 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13788 useful unless you also provide @code{-mminmax}.
13790 @item -mconfig=@var{name}
13792 Selects one of the build-in core configurations. Each MeP chip has
13793 one or more modules in it; each module has a core CPU and a variety of
13794 coprocessors, optional instructions, and peripherals. The
13795 @code{MeP-Integrator} tool, not part of GCC, provides these
13796 configurations through this option; using this option is the same as
13797 using all the corresponding command line options. The default
13798 configuration is @code{default}.
13802 Enables the coprocessor instructions. By default, this is a 32-bit
13803 coprocessor. Note that the coprocessor is normally enabled via the
13804 @code{-mconfig=} option.
13808 Enables the 32-bit coprocessor's instructions.
13812 Enables the 64-bit coprocessor's instructions.
13816 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13820 Causes constant variables to be placed in the @code{.near} section.
13824 Enables the @code{div} and @code{divu} instructions.
13828 Generate big-endian code.
13832 Generate little-endian code.
13834 @item -mio-volatile
13835 @opindex mio-volatile
13836 Tells the compiler that any variable marked with the @code{io}
13837 attribute is to be considered volatile.
13841 Causes variables to be assigned to the @code{.far} section by default.
13845 Enables the @code{leadz} (leading zero) instruction.
13849 Causes variables to be assigned to the @code{.near} section by default.
13853 Enables the @code{min} and @code{max} instructions.
13857 Enables the multiplication and multiply-accumulate instructions.
13861 Disables all the optional instructions enabled by @code{-mall-opts}.
13865 Enables the @code{repeat} and @code{erepeat} instructions, used for
13866 low-overhead looping.
13870 Causes all variables to default to the @code{.tiny} section. Note
13871 that there is a 65536 byte limit to this section. Accesses to these
13872 variables use the @code{%gp} base register.
13876 Enables the saturation instructions. Note that the compiler does not
13877 currently generate these itself, but this option is included for
13878 compatibility with other tools, like @code{as}.
13882 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13886 Link the simulator runtime libraries.
13890 Link the simulator runtime libraries, excluding built-in support
13891 for reset and exception vectors and tables.
13895 Causes all functions to default to the @code{.far} section. Without
13896 this option, functions default to the @code{.near} section.
13898 @item -mtiny=@var{n}
13900 Variables that are @var{n} bytes or smaller will be allocated to the
13901 @code{.tiny} section. These variables use the @code{$gp} base
13902 register. The default for this option is 4, but note that there's a
13903 65536 byte limit to the @code{.tiny} section.
13907 @node MicroBlaze Options
13908 @subsection MicroBlaze Options
13909 @cindex MicroBlaze Options
13914 @opindex msoft-float
13915 Use software emulation for floating point (default).
13918 @opindex mhard-float
13919 Use hardware floating point instructions.
13923 Do not optimize block moves, use @code{memcpy}.
13925 @item -mno-clearbss
13926 @opindex mno-clearbss
13927 This option is deprecated. Use @option{-fno-zero-initialized-in-bss} instead.
13929 @item -mcpu=@var{cpu-type}
13931 Use features of and schedule code for given CPU.
13932 Supported values are in the format @samp{v@var{X}.@var{YY}.@var{Z}},
13933 where @var{X} is a major version, @var{YY} is the minor version, and
13934 @var{Z} is compatiblity code. Example values are @samp{v3.00.a},
13935 @samp{v4.00.b}, @samp{v5.00.a}, @samp{v5.00.b}, @samp{v5.00.b}, @samp{v6.00.a}.
13937 @item -mxl-soft-mul
13938 @opindex mxl-soft-mul
13939 Use software multiply emulation (default).
13941 @item -mxl-soft-div
13942 @opindex mxl-soft-div
13943 Use software emulation for divides (default).
13945 @item -mxl-barrel-shift
13946 @opindex mxl-barrel-shift
13947 Use the hardware barrel shifter.
13949 @item -mxl-pattern-compare
13950 @opindex mxl-pattern-compare
13951 Use pattern compare instructions.
13953 @item -msmall-divides
13954 @opindex msmall-divides
13955 Use table lookup optimization for small signed integer divisions.
13957 @item -mxl-stack-check
13958 @opindex mxl-stack-check
13959 This option is deprecated. Use -fstack-check instead.
13962 @opindex mxl-gp-opt
13963 Use GP relative sdata/sbss sections.
13965 @item -mxl-multiply-high
13966 @opindex mxl-multiply-high
13967 Use multiply high instructions for high part of 32x32 multiply.
13969 @item -mxl-float-convert
13970 @opindex mxl-float-convert
13971 Use hardware floating point converstion instructions.
13973 @item -mxl-float-sqrt
13974 @opindex mxl-float-sqrt
13975 Use hardware floating point square root instruction.
13977 @item -mxl-mode-@var{app-model}
13978 Select application model @var{app-model}. Valid models are
13981 normal executable (default), uses startup code @file{crt0.o}.
13984 for use with Xilinx Microprocessor Debugger (XMD) based
13985 software intrusive debug agent called xmdstub. This uses startup file
13986 @file{crt1.o} and sets the start address of the program to be 0x800.
13989 for applications that are loaded using a bootloader.
13990 This model uses startup file @file{crt2.o} which does not contain a processor
13991 reset vector handler. This is suitable for transferring control on a
13992 processor reset to the bootloader rather than the application.
13995 for applications that do not require any of the
13996 MicroBlaze vectors. This option may be useful for applications running
13997 within a monitoring application. This model uses @file{crt3.o} as a startup file.
14000 Option @option{-xl-mode-@var{app-model}} is a deprecated alias for
14001 @option{-mxl-mode-@var{app-model}}.
14006 @subsection MIPS Options
14007 @cindex MIPS options
14013 Generate big-endian code.
14017 Generate little-endian code. This is the default for @samp{mips*el-*-*}
14020 @item -march=@var{arch}
14022 Generate code that will run on @var{arch}, which can be the name of a
14023 generic MIPS ISA, or the name of a particular processor.
14025 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
14026 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
14027 The processor names are:
14028 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
14029 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
14030 @samp{5kc}, @samp{5kf},
14032 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
14033 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
14034 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
14035 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
14036 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
14037 @samp{loongson2e}, @samp{loongson2f}, @samp{loongson3a},
14041 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
14042 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
14043 @samp{rm7000}, @samp{rm9000},
14044 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
14047 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
14048 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
14050 The special value @samp{from-abi} selects the
14051 most compatible architecture for the selected ABI (that is,
14052 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
14054 Native Linux/GNU toolchains also support the value @samp{native},
14055 which selects the best architecture option for the host processor.
14056 @option{-march=native} has no effect if GCC does not recognize
14059 In processor names, a final @samp{000} can be abbreviated as @samp{k}
14060 (for example, @samp{-march=r2k}). Prefixes are optional, and
14061 @samp{vr} may be written @samp{r}.
14063 Names of the form @samp{@var{n}f2_1} refer to processors with
14064 FPUs clocked at half the rate of the core, names of the form
14065 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
14066 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
14067 processors with FPUs clocked a ratio of 3:2 with respect to the core.
14068 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
14069 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
14070 accepted as synonyms for @samp{@var{n}f1_1}.
14072 GCC defines two macros based on the value of this option. The first
14073 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
14074 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
14075 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
14076 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
14077 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
14079 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
14080 above. In other words, it will have the full prefix and will not
14081 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
14082 the macro names the resolved architecture (either @samp{"mips1"} or
14083 @samp{"mips3"}). It names the default architecture when no
14084 @option{-march} option is given.
14086 @item -mtune=@var{arch}
14088 Optimize for @var{arch}. Among other things, this option controls
14089 the way instructions are scheduled, and the perceived cost of arithmetic
14090 operations. The list of @var{arch} values is the same as for
14093 When this option is not used, GCC will optimize for the processor
14094 specified by @option{-march}. By using @option{-march} and
14095 @option{-mtune} together, it is possible to generate code that will
14096 run on a family of processors, but optimize the code for one
14097 particular member of that family.
14099 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
14100 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
14101 @samp{-march} ones described above.
14105 Equivalent to @samp{-march=mips1}.
14109 Equivalent to @samp{-march=mips2}.
14113 Equivalent to @samp{-march=mips3}.
14117 Equivalent to @samp{-march=mips4}.
14121 Equivalent to @samp{-march=mips32}.
14125 Equivalent to @samp{-march=mips32r2}.
14129 Equivalent to @samp{-march=mips64}.
14133 Equivalent to @samp{-march=mips64r2}.
14138 @opindex mno-mips16
14139 Generate (do not generate) MIPS16 code. If GCC is targetting a
14140 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
14142 MIPS16 code generation can also be controlled on a per-function basis
14143 by means of @code{mips16} and @code{nomips16} attributes.
14144 @xref{Function Attributes}, for more information.
14146 @item -mflip-mips16
14147 @opindex mflip-mips16
14148 Generate MIPS16 code on alternating functions. This option is provided
14149 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
14150 not intended for ordinary use in compiling user code.
14152 @item -minterlink-mips16
14153 @itemx -mno-interlink-mips16
14154 @opindex minterlink-mips16
14155 @opindex mno-interlink-mips16
14156 Require (do not require) that non-MIPS16 code be link-compatible with
14159 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
14160 it must either use a call or an indirect jump. @option{-minterlink-mips16}
14161 therefore disables direct jumps unless GCC knows that the target of the
14162 jump is not MIPS16.
14174 Generate code for the given ABI@.
14176 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
14177 generates 64-bit code when you select a 64-bit architecture, but you
14178 can use @option{-mgp32} to get 32-bit code instead.
14180 For information about the O64 ABI, see
14181 @uref{http://gcc.gnu.org/@/projects/@/mipso64-abi.html}.
14183 GCC supports a variant of the o32 ABI in which floating-point registers
14184 are 64 rather than 32 bits wide. You can select this combination with
14185 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
14186 and @samp{mfhc1} instructions and is therefore only supported for
14187 MIPS32R2 processors.
14189 The register assignments for arguments and return values remain the
14190 same, but each scalar value is passed in a single 64-bit register
14191 rather than a pair of 32-bit registers. For example, scalar
14192 floating-point values are returned in @samp{$f0} only, not a
14193 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
14194 remains the same, but all 64 bits are saved.
14197 @itemx -mno-abicalls
14199 @opindex mno-abicalls
14200 Generate (do not generate) code that is suitable for SVR4-style
14201 dynamic objects. @option{-mabicalls} is the default for SVR4-based
14206 Generate (do not generate) code that is fully position-independent,
14207 and that can therefore be linked into shared libraries. This option
14208 only affects @option{-mabicalls}.
14210 All @option{-mabicalls} code has traditionally been position-independent,
14211 regardless of options like @option{-fPIC} and @option{-fpic}. However,
14212 as an extension, the GNU toolchain allows executables to use absolute
14213 accesses for locally-binding symbols. It can also use shorter GP
14214 initialization sequences and generate direct calls to locally-defined
14215 functions. This mode is selected by @option{-mno-shared}.
14217 @option{-mno-shared} depends on binutils 2.16 or higher and generates
14218 objects that can only be linked by the GNU linker. However, the option
14219 does not affect the ABI of the final executable; it only affects the ABI
14220 of relocatable objects. Using @option{-mno-shared} will generally make
14221 executables both smaller and quicker.
14223 @option{-mshared} is the default.
14229 Assume (do not assume) that the static and dynamic linkers
14230 support PLTs and copy relocations. This option only affects
14231 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
14232 has no effect without @samp{-msym32}.
14234 You can make @option{-mplt} the default by configuring
14235 GCC with @option{--with-mips-plt}. The default is
14236 @option{-mno-plt} otherwise.
14242 Lift (do not lift) the usual restrictions on the size of the global
14245 GCC normally uses a single instruction to load values from the GOT@.
14246 While this is relatively efficient, it will only work if the GOT
14247 is smaller than about 64k. Anything larger will cause the linker
14248 to report an error such as:
14250 @cindex relocation truncated to fit (MIPS)
14252 relocation truncated to fit: R_MIPS_GOT16 foobar
14255 If this happens, you should recompile your code with @option{-mxgot}.
14256 It should then work with very large GOTs, although it will also be
14257 less efficient, since it will take three instructions to fetch the
14258 value of a global symbol.
14260 Note that some linkers can create multiple GOTs. If you have such a
14261 linker, you should only need to use @option{-mxgot} when a single object
14262 file accesses more than 64k's worth of GOT entries. Very few do.
14264 These options have no effect unless GCC is generating position
14269 Assume that general-purpose registers are 32 bits wide.
14273 Assume that general-purpose registers are 64 bits wide.
14277 Assume that floating-point registers are 32 bits wide.
14281 Assume that floating-point registers are 64 bits wide.
14284 @opindex mhard-float
14285 Use floating-point coprocessor instructions.
14288 @opindex msoft-float
14289 Do not use floating-point coprocessor instructions. Implement
14290 floating-point calculations using library calls instead.
14292 @item -msingle-float
14293 @opindex msingle-float
14294 Assume that the floating-point coprocessor only supports single-precision
14297 @item -mdouble-float
14298 @opindex mdouble-float
14299 Assume that the floating-point coprocessor supports double-precision
14300 operations. This is the default.
14306 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
14307 implement atomic memory built-in functions. When neither option is
14308 specified, GCC will use the instructions if the target architecture
14311 @option{-mllsc} is useful if the runtime environment can emulate the
14312 instructions and @option{-mno-llsc} can be useful when compiling for
14313 nonstandard ISAs. You can make either option the default by
14314 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
14315 respectively. @option{--with-llsc} is the default for some
14316 configurations; see the installation documentation for details.
14322 Use (do not use) revision 1 of the MIPS DSP ASE@.
14323 @xref{MIPS DSP Built-in Functions}. This option defines the
14324 preprocessor macro @samp{__mips_dsp}. It also defines
14325 @samp{__mips_dsp_rev} to 1.
14331 Use (do not use) revision 2 of the MIPS DSP ASE@.
14332 @xref{MIPS DSP Built-in Functions}. This option defines the
14333 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
14334 It also defines @samp{__mips_dsp_rev} to 2.
14337 @itemx -mno-smartmips
14338 @opindex msmartmips
14339 @opindex mno-smartmips
14340 Use (do not use) the MIPS SmartMIPS ASE.
14342 @item -mpaired-single
14343 @itemx -mno-paired-single
14344 @opindex mpaired-single
14345 @opindex mno-paired-single
14346 Use (do not use) paired-single floating-point instructions.
14347 @xref{MIPS Paired-Single Support}. This option requires
14348 hardware floating-point support to be enabled.
14354 Use (do not use) MIPS Digital Media Extension instructions.
14355 This option can only be used when generating 64-bit code and requires
14356 hardware floating-point support to be enabled.
14361 @opindex mno-mips3d
14362 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
14363 The option @option{-mips3d} implies @option{-mpaired-single}.
14369 Use (do not use) MT Multithreading instructions.
14373 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
14374 an explanation of the default and the way that the pointer size is
14379 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
14381 The default size of @code{int}s, @code{long}s and pointers depends on
14382 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
14383 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
14384 32-bit @code{long}s. Pointers are the same size as @code{long}s,
14385 or the same size as integer registers, whichever is smaller.
14391 Assume (do not assume) that all symbols have 32-bit values, regardless
14392 of the selected ABI@. This option is useful in combination with
14393 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
14394 to generate shorter and faster references to symbolic addresses.
14398 Put definitions of externally-visible data in a small data section
14399 if that data is no bigger than @var{num} bytes. GCC can then access
14400 the data more efficiently; see @option{-mgpopt} for details.
14402 The default @option{-G} option depends on the configuration.
14404 @item -mlocal-sdata
14405 @itemx -mno-local-sdata
14406 @opindex mlocal-sdata
14407 @opindex mno-local-sdata
14408 Extend (do not extend) the @option{-G} behavior to local data too,
14409 such as to static variables in C@. @option{-mlocal-sdata} is the
14410 default for all configurations.
14412 If the linker complains that an application is using too much small data,
14413 you might want to try rebuilding the less performance-critical parts with
14414 @option{-mno-local-sdata}. You might also want to build large
14415 libraries with @option{-mno-local-sdata}, so that the libraries leave
14416 more room for the main program.
14418 @item -mextern-sdata
14419 @itemx -mno-extern-sdata
14420 @opindex mextern-sdata
14421 @opindex mno-extern-sdata
14422 Assume (do not assume) that externally-defined data will be in
14423 a small data section if that data is within the @option{-G} limit.
14424 @option{-mextern-sdata} is the default for all configurations.
14426 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
14427 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
14428 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
14429 is placed in a small data section. If @var{Var} is defined by another
14430 module, you must either compile that module with a high-enough
14431 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
14432 definition. If @var{Var} is common, you must link the application
14433 with a high-enough @option{-G} setting.
14435 The easiest way of satisfying these restrictions is to compile
14436 and link every module with the same @option{-G} option. However,
14437 you may wish to build a library that supports several different
14438 small data limits. You can do this by compiling the library with
14439 the highest supported @option{-G} setting and additionally using
14440 @option{-mno-extern-sdata} to stop the library from making assumptions
14441 about externally-defined data.
14447 Use (do not use) GP-relative accesses for symbols that are known to be
14448 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
14449 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
14452 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
14453 might not hold the value of @code{_gp}. For example, if the code is
14454 part of a library that might be used in a boot monitor, programs that
14455 call boot monitor routines will pass an unknown value in @code{$gp}.
14456 (In such situations, the boot monitor itself would usually be compiled
14457 with @option{-G0}.)
14459 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
14460 @option{-mno-extern-sdata}.
14462 @item -membedded-data
14463 @itemx -mno-embedded-data
14464 @opindex membedded-data
14465 @opindex mno-embedded-data
14466 Allocate variables to the read-only data section first if possible, then
14467 next in the small data section if possible, otherwise in data. This gives
14468 slightly slower code than the default, but reduces the amount of RAM required
14469 when executing, and thus may be preferred for some embedded systems.
14471 @item -muninit-const-in-rodata
14472 @itemx -mno-uninit-const-in-rodata
14473 @opindex muninit-const-in-rodata
14474 @opindex mno-uninit-const-in-rodata
14475 Put uninitialized @code{const} variables in the read-only data section.
14476 This option is only meaningful in conjunction with @option{-membedded-data}.
14478 @item -mcode-readable=@var{setting}
14479 @opindex mcode-readable
14480 Specify whether GCC may generate code that reads from executable sections.
14481 There are three possible settings:
14484 @item -mcode-readable=yes
14485 Instructions may freely access executable sections. This is the
14488 @item -mcode-readable=pcrel
14489 MIPS16 PC-relative load instructions can access executable sections,
14490 but other instructions must not do so. This option is useful on 4KSc
14491 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
14492 It is also useful on processors that can be configured to have a dual
14493 instruction/data SRAM interface and that, like the M4K, automatically
14494 redirect PC-relative loads to the instruction RAM.
14496 @item -mcode-readable=no
14497 Instructions must not access executable sections. This option can be
14498 useful on targets that are configured to have a dual instruction/data
14499 SRAM interface but that (unlike the M4K) do not automatically redirect
14500 PC-relative loads to the instruction RAM.
14503 @item -msplit-addresses
14504 @itemx -mno-split-addresses
14505 @opindex msplit-addresses
14506 @opindex mno-split-addresses
14507 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
14508 relocation operators. This option has been superseded by
14509 @option{-mexplicit-relocs} but is retained for backwards compatibility.
14511 @item -mexplicit-relocs
14512 @itemx -mno-explicit-relocs
14513 @opindex mexplicit-relocs
14514 @opindex mno-explicit-relocs
14515 Use (do not use) assembler relocation operators when dealing with symbolic
14516 addresses. The alternative, selected by @option{-mno-explicit-relocs},
14517 is to use assembler macros instead.
14519 @option{-mexplicit-relocs} is the default if GCC was configured
14520 to use an assembler that supports relocation operators.
14522 @item -mcheck-zero-division
14523 @itemx -mno-check-zero-division
14524 @opindex mcheck-zero-division
14525 @opindex mno-check-zero-division
14526 Trap (do not trap) on integer division by zero.
14528 The default is @option{-mcheck-zero-division}.
14530 @item -mdivide-traps
14531 @itemx -mdivide-breaks
14532 @opindex mdivide-traps
14533 @opindex mdivide-breaks
14534 MIPS systems check for division by zero by generating either a
14535 conditional trap or a break instruction. Using traps results in
14536 smaller code, but is only supported on MIPS II and later. Also, some
14537 versions of the Linux kernel have a bug that prevents trap from
14538 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
14539 allow conditional traps on architectures that support them and
14540 @option{-mdivide-breaks} to force the use of breaks.
14542 The default is usually @option{-mdivide-traps}, but this can be
14543 overridden at configure time using @option{--with-divide=breaks}.
14544 Divide-by-zero checks can be completely disabled using
14545 @option{-mno-check-zero-division}.
14550 @opindex mno-memcpy
14551 Force (do not force) the use of @code{memcpy()} for non-trivial block
14552 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
14553 most constant-sized copies.
14556 @itemx -mno-long-calls
14557 @opindex mlong-calls
14558 @opindex mno-long-calls
14559 Disable (do not disable) use of the @code{jal} instruction. Calling
14560 functions using @code{jal} is more efficient but requires the caller
14561 and callee to be in the same 256 megabyte segment.
14563 This option has no effect on abicalls code. The default is
14564 @option{-mno-long-calls}.
14570 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
14571 instructions, as provided by the R4650 ISA@.
14574 @itemx -mno-fused-madd
14575 @opindex mfused-madd
14576 @opindex mno-fused-madd
14577 Enable (disable) use of the floating point multiply-accumulate
14578 instructions, when they are available. The default is
14579 @option{-mfused-madd}.
14581 When multiply-accumulate instructions are used, the intermediate
14582 product is calculated to infinite precision and is not subject to
14583 the FCSR Flush to Zero bit. This may be undesirable in some
14588 Tell the MIPS assembler to not run its preprocessor over user
14589 assembler files (with a @samp{.s} suffix) when assembling them.
14592 @itemx -mno-fix-r4000
14593 @opindex mfix-r4000
14594 @opindex mno-fix-r4000
14595 Work around certain R4000 CPU errata:
14598 A double-word or a variable shift may give an incorrect result if executed
14599 immediately after starting an integer division.
14601 A double-word or a variable shift may give an incorrect result if executed
14602 while an integer multiplication is in progress.
14604 An integer division may give an incorrect result if started in a delay slot
14605 of a taken branch or a jump.
14609 @itemx -mno-fix-r4400
14610 @opindex mfix-r4400
14611 @opindex mno-fix-r4400
14612 Work around certain R4400 CPU errata:
14615 A double-word or a variable shift may give an incorrect result if executed
14616 immediately after starting an integer division.
14620 @itemx -mno-fix-r10000
14621 @opindex mfix-r10000
14622 @opindex mno-fix-r10000
14623 Work around certain R10000 errata:
14626 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14627 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14630 This option can only be used if the target architecture supports
14631 branch-likely instructions. @option{-mfix-r10000} is the default when
14632 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14636 @itemx -mno-fix-vr4120
14637 @opindex mfix-vr4120
14638 Work around certain VR4120 errata:
14641 @code{dmultu} does not always produce the correct result.
14643 @code{div} and @code{ddiv} do not always produce the correct result if one
14644 of the operands is negative.
14646 The workarounds for the division errata rely on special functions in
14647 @file{libgcc.a}. At present, these functions are only provided by
14648 the @code{mips64vr*-elf} configurations.
14650 Other VR4120 errata require a nop to be inserted between certain pairs of
14651 instructions. These errata are handled by the assembler, not by GCC itself.
14654 @opindex mfix-vr4130
14655 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14656 workarounds are implemented by the assembler rather than by GCC,
14657 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14658 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14659 instructions are available instead.
14662 @itemx -mno-fix-sb1
14664 Work around certain SB-1 CPU core errata.
14665 (This flag currently works around the SB-1 revision 2
14666 ``F1'' and ``F2'' floating point errata.)
14668 @item -mr10k-cache-barrier=@var{setting}
14669 @opindex mr10k-cache-barrier
14670 Specify whether GCC should insert cache barriers to avoid the
14671 side-effects of speculation on R10K processors.
14673 In common with many processors, the R10K tries to predict the outcome
14674 of a conditional branch and speculatively executes instructions from
14675 the ``taken'' branch. It later aborts these instructions if the
14676 predicted outcome was wrong. However, on the R10K, even aborted
14677 instructions can have side effects.
14679 This problem only affects kernel stores and, depending on the system,
14680 kernel loads. As an example, a speculatively-executed store may load
14681 the target memory into cache and mark the cache line as dirty, even if
14682 the store itself is later aborted. If a DMA operation writes to the
14683 same area of memory before the ``dirty'' line is flushed, the cached
14684 data will overwrite the DMA-ed data. See the R10K processor manual
14685 for a full description, including other potential problems.
14687 One workaround is to insert cache barrier instructions before every memory
14688 access that might be speculatively executed and that might have side
14689 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14690 controls GCC's implementation of this workaround. It assumes that
14691 aborted accesses to any byte in the following regions will not have
14696 the memory occupied by the current function's stack frame;
14699 the memory occupied by an incoming stack argument;
14702 the memory occupied by an object with a link-time-constant address.
14705 It is the kernel's responsibility to ensure that speculative
14706 accesses to these regions are indeed safe.
14708 If the input program contains a function declaration such as:
14714 then the implementation of @code{foo} must allow @code{j foo} and
14715 @code{jal foo} to be executed speculatively. GCC honors this
14716 restriction for functions it compiles itself. It expects non-GCC
14717 functions (such as hand-written assembly code) to do the same.
14719 The option has three forms:
14722 @item -mr10k-cache-barrier=load-store
14723 Insert a cache barrier before a load or store that might be
14724 speculatively executed and that might have side effects even
14727 @item -mr10k-cache-barrier=store
14728 Insert a cache barrier before a store that might be speculatively
14729 executed and that might have side effects even if aborted.
14731 @item -mr10k-cache-barrier=none
14732 Disable the insertion of cache barriers. This is the default setting.
14735 @item -mflush-func=@var{func}
14736 @itemx -mno-flush-func
14737 @opindex mflush-func
14738 Specifies the function to call to flush the I and D caches, or to not
14739 call any such function. If called, the function must take the same
14740 arguments as the common @code{_flush_func()}, that is, the address of the
14741 memory range for which the cache is being flushed, the size of the
14742 memory range, and the number 3 (to flush both caches). The default
14743 depends on the target GCC was configured for, but commonly is either
14744 @samp{_flush_func} or @samp{__cpu_flush}.
14746 @item mbranch-cost=@var{num}
14747 @opindex mbranch-cost
14748 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14749 This cost is only a heuristic and is not guaranteed to produce
14750 consistent results across releases. A zero cost redundantly selects
14751 the default, which is based on the @option{-mtune} setting.
14753 @item -mbranch-likely
14754 @itemx -mno-branch-likely
14755 @opindex mbranch-likely
14756 @opindex mno-branch-likely
14757 Enable or disable use of Branch Likely instructions, regardless of the
14758 default for the selected architecture. By default, Branch Likely
14759 instructions may be generated if they are supported by the selected
14760 architecture. An exception is for the MIPS32 and MIPS64 architectures
14761 and processors which implement those architectures; for those, Branch
14762 Likely instructions will not be generated by default because the MIPS32
14763 and MIPS64 architectures specifically deprecate their use.
14765 @item -mfp-exceptions
14766 @itemx -mno-fp-exceptions
14767 @opindex mfp-exceptions
14768 Specifies whether FP exceptions are enabled. This affects how we schedule
14769 FP instructions for some processors. The default is that FP exceptions are
14772 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14773 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14776 @item -mvr4130-align
14777 @itemx -mno-vr4130-align
14778 @opindex mvr4130-align
14779 The VR4130 pipeline is two-way superscalar, but can only issue two
14780 instructions together if the first one is 8-byte aligned. When this
14781 option is enabled, GCC will align pairs of instructions that it
14782 thinks should execute in parallel.
14784 This option only has an effect when optimizing for the VR4130.
14785 It normally makes code faster, but at the expense of making it bigger.
14786 It is enabled by default at optimization level @option{-O3}.
14791 Enable (disable) generation of @code{synci} instructions on
14792 architectures that support it. The @code{synci} instructions (if
14793 enabled) will be generated when @code{__builtin___clear_cache()} is
14796 This option defaults to @code{-mno-synci}, but the default can be
14797 overridden by configuring with @code{--with-synci}.
14799 When compiling code for single processor systems, it is generally safe
14800 to use @code{synci}. However, on many multi-core (SMP) systems, it
14801 will not invalidate the instruction caches on all cores and may lead
14802 to undefined behavior.
14804 @item -mrelax-pic-calls
14805 @itemx -mno-relax-pic-calls
14806 @opindex mrelax-pic-calls
14807 Try to turn PIC calls that are normally dispatched via register
14808 @code{$25} into direct calls. This is only possible if the linker can
14809 resolve the destination at link-time and if the destination is within
14810 range for a direct call.
14812 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14813 an assembler and a linker that supports the @code{.reloc} assembly
14814 directive and @code{-mexplicit-relocs} is in effect. With
14815 @code{-mno-explicit-relocs}, this optimization can be performed by the
14816 assembler and the linker alone without help from the compiler.
14818 @item -mmcount-ra-address
14819 @itemx -mno-mcount-ra-address
14820 @opindex mmcount-ra-address
14821 @opindex mno-mcount-ra-address
14822 Emit (do not emit) code that allows @code{_mcount} to modify the
14823 calling function's return address. When enabled, this option extends
14824 the usual @code{_mcount} interface with a new @var{ra-address}
14825 parameter, which has type @code{intptr_t *} and is passed in register
14826 @code{$12}. @code{_mcount} can then modify the return address by
14827 doing both of the following:
14830 Returning the new address in register @code{$31}.
14832 Storing the new address in @code{*@var{ra-address}},
14833 if @var{ra-address} is nonnull.
14836 The default is @option{-mno-mcount-ra-address}.
14841 @subsection MMIX Options
14842 @cindex MMIX Options
14844 These options are defined for the MMIX:
14848 @itemx -mno-libfuncs
14850 @opindex mno-libfuncs
14851 Specify that intrinsic library functions are being compiled, passing all
14852 values in registers, no matter the size.
14855 @itemx -mno-epsilon
14857 @opindex mno-epsilon
14858 Generate floating-point comparison instructions that compare with respect
14859 to the @code{rE} epsilon register.
14861 @item -mabi=mmixware
14863 @opindex mabi=mmixware
14865 Generate code that passes function parameters and return values that (in
14866 the called function) are seen as registers @code{$0} and up, as opposed to
14867 the GNU ABI which uses global registers @code{$231} and up.
14869 @item -mzero-extend
14870 @itemx -mno-zero-extend
14871 @opindex mzero-extend
14872 @opindex mno-zero-extend
14873 When reading data from memory in sizes shorter than 64 bits, use (do not
14874 use) zero-extending load instructions by default, rather than
14875 sign-extending ones.
14878 @itemx -mno-knuthdiv
14880 @opindex mno-knuthdiv
14881 Make the result of a division yielding a remainder have the same sign as
14882 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14883 remainder follows the sign of the dividend. Both methods are
14884 arithmetically valid, the latter being almost exclusively used.
14886 @item -mtoplevel-symbols
14887 @itemx -mno-toplevel-symbols
14888 @opindex mtoplevel-symbols
14889 @opindex mno-toplevel-symbols
14890 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14891 code can be used with the @code{PREFIX} assembly directive.
14895 Generate an executable in the ELF format, rather than the default
14896 @samp{mmo} format used by the @command{mmix} simulator.
14898 @item -mbranch-predict
14899 @itemx -mno-branch-predict
14900 @opindex mbranch-predict
14901 @opindex mno-branch-predict
14902 Use (do not use) the probable-branch instructions, when static branch
14903 prediction indicates a probable branch.
14905 @item -mbase-addresses
14906 @itemx -mno-base-addresses
14907 @opindex mbase-addresses
14908 @opindex mno-base-addresses
14909 Generate (do not generate) code that uses @emph{base addresses}. Using a
14910 base address automatically generates a request (handled by the assembler
14911 and the linker) for a constant to be set up in a global register. The
14912 register is used for one or more base address requests within the range 0
14913 to 255 from the value held in the register. The generally leads to short
14914 and fast code, but the number of different data items that can be
14915 addressed is limited. This means that a program that uses lots of static
14916 data may require @option{-mno-base-addresses}.
14918 @item -msingle-exit
14919 @itemx -mno-single-exit
14920 @opindex msingle-exit
14921 @opindex mno-single-exit
14922 Force (do not force) generated code to have a single exit point in each
14926 @node MN10300 Options
14927 @subsection MN10300 Options
14928 @cindex MN10300 options
14930 These @option{-m} options are defined for Matsushita MN10300 architectures:
14935 Generate code to avoid bugs in the multiply instructions for the MN10300
14936 processors. This is the default.
14938 @item -mno-mult-bug
14939 @opindex mno-mult-bug
14940 Do not generate code to avoid bugs in the multiply instructions for the
14941 MN10300 processors.
14945 Generate code which uses features specific to the AM33 processor.
14949 Do not generate code which uses features specific to the AM33 processor. This
14954 Generate code which uses features specific to the AM33/2.0 processor.
14958 Generate code which uses features specific to the AM34 processor.
14960 @item -mtune=@var{cpu-type}
14962 Use the timing characteristics of the indicated CPU type when
14963 scheduling instructions. This does not change the targeted processor
14964 type. The CPU type must be one of @samp{mn10300}, @samp{am33},
14965 @samp{am33-2} or @samp{am34}.
14967 @item -mreturn-pointer-on-d0
14968 @opindex mreturn-pointer-on-d0
14969 When generating a function which returns a pointer, return the pointer
14970 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14971 only in a0, and attempts to call such functions without a prototype
14972 would result in errors. Note that this option is on by default; use
14973 @option{-mno-return-pointer-on-d0} to disable it.
14977 Do not link in the C run-time initialization object file.
14981 Indicate to the linker that it should perform a relaxation optimization pass
14982 to shorten branches, calls and absolute memory addresses. This option only
14983 has an effect when used on the command line for the final link step.
14985 This option makes symbolic debugging impossible.
14988 @node PDP-11 Options
14989 @subsection PDP-11 Options
14990 @cindex PDP-11 Options
14992 These options are defined for the PDP-11:
14997 Use hardware FPP floating point. This is the default. (FIS floating
14998 point on the PDP-11/40 is not supported.)
15001 @opindex msoft-float
15002 Do not use hardware floating point.
15006 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
15010 Return floating-point results in memory. This is the default.
15014 Generate code for a PDP-11/40.
15018 Generate code for a PDP-11/45. This is the default.
15022 Generate code for a PDP-11/10.
15024 @item -mbcopy-builtin
15025 @opindex mbcopy-builtin
15026 Use inline @code{movmemhi} patterns for copying memory. This is the
15031 Do not use inline @code{movmemhi} patterns for copying memory.
15037 Use 16-bit @code{int}. This is the default.
15043 Use 32-bit @code{int}.
15046 @itemx -mno-float32
15048 @opindex mno-float32
15049 Use 64-bit @code{float}. This is the default.
15052 @itemx -mno-float64
15054 @opindex mno-float64
15055 Use 32-bit @code{float}.
15059 Use @code{abshi2} pattern. This is the default.
15063 Do not use @code{abshi2} pattern.
15065 @item -mbranch-expensive
15066 @opindex mbranch-expensive
15067 Pretend that branches are expensive. This is for experimenting with
15068 code generation only.
15070 @item -mbranch-cheap
15071 @opindex mbranch-cheap
15072 Do not pretend that branches are expensive. This is the default.
15076 Use Unix assembler syntax. This is the default when configured for
15077 @samp{pdp11-*-bsd}.
15081 Use DEC assembler syntax. This is the default when configured for any
15082 PDP-11 target other than @samp{pdp11-*-bsd}.
15085 @node picoChip Options
15086 @subsection picoChip Options
15087 @cindex picoChip options
15089 These @samp{-m} options are defined for picoChip implementations:
15093 @item -mae=@var{ae_type}
15095 Set the instruction set, register set, and instruction scheduling
15096 parameters for array element type @var{ae_type}. Supported values
15097 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
15099 @option{-mae=ANY} selects a completely generic AE type. Code
15100 generated with this option will run on any of the other AE types. The
15101 code will not be as efficient as it would be if compiled for a specific
15102 AE type, and some types of operation (e.g., multiplication) will not
15103 work properly on all types of AE.
15105 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
15106 for compiled code, and is the default.
15108 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
15109 option may suffer from poor performance of byte (char) manipulation,
15110 since the DSP AE does not provide hardware support for byte load/stores.
15112 @item -msymbol-as-address
15113 Enable the compiler to directly use a symbol name as an address in a
15114 load/store instruction, without first loading it into a
15115 register. Typically, the use of this option will generate larger
15116 programs, which run faster than when the option isn't used. However, the
15117 results vary from program to program, so it is left as a user option,
15118 rather than being permanently enabled.
15120 @item -mno-inefficient-warnings
15121 Disables warnings about the generation of inefficient code. These
15122 warnings can be generated, for example, when compiling code which
15123 performs byte-level memory operations on the MAC AE type. The MAC AE has
15124 no hardware support for byte-level memory operations, so all byte
15125 load/stores must be synthesized from word load/store operations. This is
15126 inefficient and a warning will be generated indicating to the programmer
15127 that they should rewrite the code to avoid byte operations, or to target
15128 an AE type which has the necessary hardware support. This option enables
15129 the warning to be turned off.
15133 @node PowerPC Options
15134 @subsection PowerPC Options
15135 @cindex PowerPC options
15137 These are listed under @xref{RS/6000 and PowerPC Options}.
15139 @node RS/6000 and PowerPC Options
15140 @subsection IBM RS/6000 and PowerPC Options
15141 @cindex RS/6000 and PowerPC Options
15142 @cindex IBM RS/6000 and PowerPC Options
15144 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
15151 @itemx -mno-powerpc
15152 @itemx -mpowerpc-gpopt
15153 @itemx -mno-powerpc-gpopt
15154 @itemx -mpowerpc-gfxopt
15155 @itemx -mno-powerpc-gfxopt
15158 @itemx -mno-powerpc64
15162 @itemx -mno-popcntb
15164 @itemx -mno-popcntd
15173 @itemx -mno-hard-dfp
15177 @opindex mno-power2
15179 @opindex mno-powerpc
15180 @opindex mpowerpc-gpopt
15181 @opindex mno-powerpc-gpopt
15182 @opindex mpowerpc-gfxopt
15183 @opindex mno-powerpc-gfxopt
15184 @opindex mpowerpc64
15185 @opindex mno-powerpc64
15189 @opindex mno-popcntb
15191 @opindex mno-popcntd
15197 @opindex mno-mfpgpr
15199 @opindex mno-hard-dfp
15200 GCC supports two related instruction set architectures for the
15201 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
15202 instructions supported by the @samp{rios} chip set used in the original
15203 RS/6000 systems and the @dfn{PowerPC} instruction set is the
15204 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
15205 the IBM 4xx, 6xx, and follow-on microprocessors.
15207 Neither architecture is a subset of the other. However there is a
15208 large common subset of instructions supported by both. An MQ
15209 register is included in processors supporting the POWER architecture.
15211 You use these options to specify which instructions are available on the
15212 processor you are using. The default value of these options is
15213 determined when configuring GCC@. Specifying the
15214 @option{-mcpu=@var{cpu_type}} overrides the specification of these
15215 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
15216 rather than the options listed above.
15218 The @option{-mpower} option allows GCC to generate instructions that
15219 are found only in the POWER architecture and to use the MQ register.
15220 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
15221 to generate instructions that are present in the POWER2 architecture but
15222 not the original POWER architecture.
15224 The @option{-mpowerpc} option allows GCC to generate instructions that
15225 are found only in the 32-bit subset of the PowerPC architecture.
15226 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
15227 GCC to use the optional PowerPC architecture instructions in the
15228 General Purpose group, including floating-point square root. Specifying
15229 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
15230 use the optional PowerPC architecture instructions in the Graphics
15231 group, including floating-point select.
15233 The @option{-mmfcrf} option allows GCC to generate the move from
15234 condition register field instruction implemented on the POWER4
15235 processor and other processors that support the PowerPC V2.01
15237 The @option{-mpopcntb} option allows GCC to generate the popcount and
15238 double precision FP reciprocal estimate instruction implemented on the
15239 POWER5 processor and other processors that support the PowerPC V2.02
15241 The @option{-mpopcntd} option allows GCC to generate the popcount
15242 instruction implemented on the POWER7 processor and other processors
15243 that support the PowerPC V2.06 architecture.
15244 The @option{-mfprnd} option allows GCC to generate the FP round to
15245 integer instructions implemented on the POWER5+ processor and other
15246 processors that support the PowerPC V2.03 architecture.
15247 The @option{-mcmpb} option allows GCC to generate the compare bytes
15248 instruction implemented on the POWER6 processor and other processors
15249 that support the PowerPC V2.05 architecture.
15250 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
15251 general purpose register instructions implemented on the POWER6X
15252 processor and other processors that support the extended PowerPC V2.05
15254 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
15255 point instructions implemented on some POWER processors.
15257 The @option{-mpowerpc64} option allows GCC to generate the additional
15258 64-bit instructions that are found in the full PowerPC64 architecture
15259 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
15260 @option{-mno-powerpc64}.
15262 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
15263 will use only the instructions in the common subset of both
15264 architectures plus some special AIX common-mode calls, and will not use
15265 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
15266 permits GCC to use any instruction from either architecture and to
15267 allow use of the MQ register; specify this for the Motorola MPC601.
15269 @item -mnew-mnemonics
15270 @itemx -mold-mnemonics
15271 @opindex mnew-mnemonics
15272 @opindex mold-mnemonics
15273 Select which mnemonics to use in the generated assembler code. With
15274 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
15275 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
15276 assembler mnemonics defined for the POWER architecture. Instructions
15277 defined in only one architecture have only one mnemonic; GCC uses that
15278 mnemonic irrespective of which of these options is specified.
15280 GCC defaults to the mnemonics appropriate for the architecture in
15281 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
15282 value of these option. Unless you are building a cross-compiler, you
15283 should normally not specify either @option{-mnew-mnemonics} or
15284 @option{-mold-mnemonics}, but should instead accept the default.
15286 @item -mcpu=@var{cpu_type}
15288 Set architecture type, register usage, choice of mnemonics, and
15289 instruction scheduling parameters for machine type @var{cpu_type}.
15290 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
15291 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
15292 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
15293 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
15294 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
15295 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
15296 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
15297 @samp{G4}, @samp{G5}, @samp{titan}, @samp{power}, @samp{power2}, @samp{power3},
15298 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
15299 @samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
15300 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
15302 @option{-mcpu=common} selects a completely generic processor. Code
15303 generated under this option will run on any POWER or PowerPC processor.
15304 GCC will use only the instructions in the common subset of both
15305 architectures, and will not use the MQ register. GCC assumes a generic
15306 processor model for scheduling purposes.
15308 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
15309 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
15310 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
15311 types, with an appropriate, generic processor model assumed for
15312 scheduling purposes.
15314 The other options specify a specific processor. Code generated under
15315 those options will run best on that processor, and may not run at all on
15318 The @option{-mcpu} options automatically enable or disable the
15321 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
15322 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
15323 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
15324 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
15326 The particular options set for any particular CPU will vary between
15327 compiler versions, depending on what setting seems to produce optimal
15328 code for that CPU; it doesn't necessarily reflect the actual hardware's
15329 capabilities. If you wish to set an individual option to a particular
15330 value, you may specify it after the @option{-mcpu} option, like
15331 @samp{-mcpu=970 -mno-altivec}.
15333 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
15334 not enabled or disabled by the @option{-mcpu} option at present because
15335 AIX does not have full support for these options. You may still
15336 enable or disable them individually if you're sure it'll work in your
15339 @item -mtune=@var{cpu_type}
15341 Set the instruction scheduling parameters for machine type
15342 @var{cpu_type}, but do not set the architecture type, register usage, or
15343 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
15344 values for @var{cpu_type} are used for @option{-mtune} as for
15345 @option{-mcpu}. If both are specified, the code generated will use the
15346 architecture, registers, and mnemonics set by @option{-mcpu}, but the
15347 scheduling parameters set by @option{-mtune}.
15349 @item -mcmodel=small
15350 @opindex mcmodel=small
15351 Generate PowerPC64 code for the small model: The TOC is limited to
15354 @item -mcmodel=medium
15355 @opindex mcmodel=medium
15356 Generate PowerPC64 code for the medium model: The TOC and other static
15357 data may be up to a total of 4G in size.
15359 @item -mcmodel=large
15360 @opindex mcmodel=large
15361 Generate PowerPC64 code for the large model: The TOC may be up to 4G
15362 in size. Other data and code is only limited by the 64-bit address
15366 @itemx -mno-altivec
15368 @opindex mno-altivec
15369 Generate code that uses (does not use) AltiVec instructions, and also
15370 enable the use of built-in functions that allow more direct access to
15371 the AltiVec instruction set. You may also need to set
15372 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
15378 @opindex mno-vrsave
15379 Generate VRSAVE instructions when generating AltiVec code.
15381 @item -mgen-cell-microcode
15382 @opindex mgen-cell-microcode
15383 Generate Cell microcode instructions
15385 @item -mwarn-cell-microcode
15386 @opindex mwarn-cell-microcode
15387 Warning when a Cell microcode instruction is going to emitted. An example
15388 of a Cell microcode instruction is a variable shift.
15391 @opindex msecure-plt
15392 Generate code that allows ld and ld.so to build executables and shared
15393 libraries with non-exec .plt and .got sections. This is a PowerPC
15394 32-bit SYSV ABI option.
15398 Generate code that uses a BSS .plt section that ld.so fills in, and
15399 requires .plt and .got sections that are both writable and executable.
15400 This is a PowerPC 32-bit SYSV ABI option.
15406 This switch enables or disables the generation of ISEL instructions.
15408 @item -misel=@var{yes/no}
15409 This switch has been deprecated. Use @option{-misel} and
15410 @option{-mno-isel} instead.
15416 This switch enables or disables the generation of SPE simd
15422 @opindex mno-paired
15423 This switch enables or disables the generation of PAIRED simd
15426 @item -mspe=@var{yes/no}
15427 This option has been deprecated. Use @option{-mspe} and
15428 @option{-mno-spe} instead.
15434 Generate code that uses (does not use) vector/scalar (VSX)
15435 instructions, and also enable the use of built-in functions that allow
15436 more direct access to the VSX instruction set.
15438 @item -mfloat-gprs=@var{yes/single/double/no}
15439 @itemx -mfloat-gprs
15440 @opindex mfloat-gprs
15441 This switch enables or disables the generation of floating point
15442 operations on the general purpose registers for architectures that
15445 The argument @var{yes} or @var{single} enables the use of
15446 single-precision floating point operations.
15448 The argument @var{double} enables the use of single and
15449 double-precision floating point operations.
15451 The argument @var{no} disables floating point operations on the
15452 general purpose registers.
15454 This option is currently only available on the MPC854x.
15460 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
15461 targets (including GNU/Linux). The 32-bit environment sets int, long
15462 and pointer to 32 bits and generates code that runs on any PowerPC
15463 variant. The 64-bit environment sets int to 32 bits and long and
15464 pointer to 64 bits, and generates code for PowerPC64, as for
15465 @option{-mpowerpc64}.
15468 @itemx -mno-fp-in-toc
15469 @itemx -mno-sum-in-toc
15470 @itemx -mminimal-toc
15472 @opindex mno-fp-in-toc
15473 @opindex mno-sum-in-toc
15474 @opindex mminimal-toc
15475 Modify generation of the TOC (Table Of Contents), which is created for
15476 every executable file. The @option{-mfull-toc} option is selected by
15477 default. In that case, GCC will allocate at least one TOC entry for
15478 each unique non-automatic variable reference in your program. GCC
15479 will also place floating-point constants in the TOC@. However, only
15480 16,384 entries are available in the TOC@.
15482 If you receive a linker error message that saying you have overflowed
15483 the available TOC space, you can reduce the amount of TOC space used
15484 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
15485 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
15486 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
15487 generate code to calculate the sum of an address and a constant at
15488 run-time instead of putting that sum into the TOC@. You may specify one
15489 or both of these options. Each causes GCC to produce very slightly
15490 slower and larger code at the expense of conserving TOC space.
15492 If you still run out of space in the TOC even when you specify both of
15493 these options, specify @option{-mminimal-toc} instead. This option causes
15494 GCC to make only one TOC entry for every file. When you specify this
15495 option, GCC will produce code that is slower and larger but which
15496 uses extremely little TOC space. You may wish to use this option
15497 only on files that contain less frequently executed code.
15503 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
15504 @code{long} type, and the infrastructure needed to support them.
15505 Specifying @option{-maix64} implies @option{-mpowerpc64} and
15506 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
15507 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
15510 @itemx -mno-xl-compat
15511 @opindex mxl-compat
15512 @opindex mno-xl-compat
15513 Produce code that conforms more closely to IBM XL compiler semantics
15514 when using AIX-compatible ABI@. Pass floating-point arguments to
15515 prototyped functions beyond the register save area (RSA) on the stack
15516 in addition to argument FPRs. Do not assume that most significant
15517 double in 128-bit long double value is properly rounded when comparing
15518 values and converting to double. Use XL symbol names for long double
15521 The AIX calling convention was extended but not initially documented to
15522 handle an obscure K&R C case of calling a function that takes the
15523 address of its arguments with fewer arguments than declared. IBM XL
15524 compilers access floating point arguments which do not fit in the
15525 RSA from the stack when a subroutine is compiled without
15526 optimization. Because always storing floating-point arguments on the
15527 stack is inefficient and rarely needed, this option is not enabled by
15528 default and only is necessary when calling subroutines compiled by IBM
15529 XL compilers without optimization.
15533 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
15534 application written to use message passing with special startup code to
15535 enable the application to run. The system must have PE installed in the
15536 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
15537 must be overridden with the @option{-specs=} option to specify the
15538 appropriate directory location. The Parallel Environment does not
15539 support threads, so the @option{-mpe} option and the @option{-pthread}
15540 option are incompatible.
15542 @item -malign-natural
15543 @itemx -malign-power
15544 @opindex malign-natural
15545 @opindex malign-power
15546 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
15547 @option{-malign-natural} overrides the ABI-defined alignment of larger
15548 types, such as floating-point doubles, on their natural size-based boundary.
15549 The option @option{-malign-power} instructs GCC to follow the ABI-specified
15550 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
15552 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
15556 @itemx -mhard-float
15557 @opindex msoft-float
15558 @opindex mhard-float
15559 Generate code that does not use (uses) the floating-point register set.
15560 Software floating point emulation is provided if you use the
15561 @option{-msoft-float} option, and pass the option to GCC when linking.
15563 @item -msingle-float
15564 @itemx -mdouble-float
15565 @opindex msingle-float
15566 @opindex mdouble-float
15567 Generate code for single or double-precision floating point operations.
15568 @option{-mdouble-float} implies @option{-msingle-float}.
15571 @opindex msimple-fpu
15572 Do not generate sqrt and div instructions for hardware floating point unit.
15576 Specify type of floating point unit. Valid values are @var{sp_lite}
15577 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
15578 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
15579 and @var{dp_full} (equivalent to -mdouble-float).
15582 @opindex mxilinx-fpu
15583 Perform optimizations for floating point unit on Xilinx PPC 405/440.
15586 @itemx -mno-multiple
15588 @opindex mno-multiple
15589 Generate code that uses (does not use) the load multiple word
15590 instructions and the store multiple word instructions. These
15591 instructions are generated by default on POWER systems, and not
15592 generated on PowerPC systems. Do not use @option{-mmultiple} on little
15593 endian PowerPC systems, since those instructions do not work when the
15594 processor is in little endian mode. The exceptions are PPC740 and
15595 PPC750 which permit the instructions usage in little endian mode.
15600 @opindex mno-string
15601 Generate code that uses (does not use) the load string instructions
15602 and the store string word instructions to save multiple registers and
15603 do small block moves. These instructions are generated by default on
15604 POWER systems, and not generated on PowerPC systems. Do not use
15605 @option{-mstring} on little endian PowerPC systems, since those
15606 instructions do not work when the processor is in little endian mode.
15607 The exceptions are PPC740 and PPC750 which permit the instructions
15608 usage in little endian mode.
15613 @opindex mno-update
15614 Generate code that uses (does not use) the load or store instructions
15615 that update the base register to the address of the calculated memory
15616 location. These instructions are generated by default. If you use
15617 @option{-mno-update}, there is a small window between the time that the
15618 stack pointer is updated and the address of the previous frame is
15619 stored, which means code that walks the stack frame across interrupts or
15620 signals may get corrupted data.
15622 @item -mavoid-indexed-addresses
15623 @itemx -mno-avoid-indexed-addresses
15624 @opindex mavoid-indexed-addresses
15625 @opindex mno-avoid-indexed-addresses
15626 Generate code that tries to avoid (not avoid) the use of indexed load
15627 or store instructions. These instructions can incur a performance
15628 penalty on Power6 processors in certain situations, such as when
15629 stepping through large arrays that cross a 16M boundary. This option
15630 is enabled by default when targetting Power6 and disabled otherwise.
15633 @itemx -mno-fused-madd
15634 @opindex mfused-madd
15635 @opindex mno-fused-madd
15636 Generate code that uses (does not use) the floating point multiply and
15637 accumulate instructions. These instructions are generated by default
15638 if hardware floating point is used. The machine dependent
15639 @option{-mfused-madd} option is now mapped to the machine independent
15640 @option{-ffp-contract=fast} option, and @option{-mno-fused-madd} is
15641 mapped to @option{-ffp-contract=off}.
15647 Generate code that uses (does not use) the half-word multiply and
15648 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15649 These instructions are generated by default when targetting those
15656 Generate code that uses (does not use) the string-search @samp{dlmzb}
15657 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15658 generated by default when targetting those processors.
15660 @item -mno-bit-align
15662 @opindex mno-bit-align
15663 @opindex mbit-align
15664 On System V.4 and embedded PowerPC systems do not (do) force structures
15665 and unions that contain bit-fields to be aligned to the base type of the
15668 For example, by default a structure containing nothing but 8
15669 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15670 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15671 the structure would be aligned to a 1 byte boundary and be one byte in
15674 @item -mno-strict-align
15675 @itemx -mstrict-align
15676 @opindex mno-strict-align
15677 @opindex mstrict-align
15678 On System V.4 and embedded PowerPC systems do not (do) assume that
15679 unaligned memory references will be handled by the system.
15681 @item -mrelocatable
15682 @itemx -mno-relocatable
15683 @opindex mrelocatable
15684 @opindex mno-relocatable
15685 Generate code that allows (does not allow) a static executable to be
15686 relocated to a different address at runtime. A simple embedded
15687 PowerPC system loader should relocate the entire contents of
15688 @code{.got2} and 4-byte locations listed in the @code{.fixup} section,
15689 a table of 32-bit addresses generated by this option. For this to
15690 work, all objects linked together must be compiled with
15691 @option{-mrelocatable} or @option{-mrelocatable-lib}.
15692 @option{-mrelocatable} code aligns the stack to an 8 byte boundary.
15694 @item -mrelocatable-lib
15695 @itemx -mno-relocatable-lib
15696 @opindex mrelocatable-lib
15697 @opindex mno-relocatable-lib
15698 Like @option{-mrelocatable}, @option{-mrelocatable-lib} generates a
15699 @code{.fixup} section to allow static executables to be relocated at
15700 runtime, but @option{-mrelocatable-lib} does not use the smaller stack
15701 alignment of @option{-mrelocatable}. Objects compiled with
15702 @option{-mrelocatable-lib} may be linked with objects compiled with
15703 any combination of the @option{-mrelocatable} options.
15709 On System V.4 and embedded PowerPC systems do not (do) assume that
15710 register 2 contains a pointer to a global area pointing to the addresses
15711 used in the program.
15714 @itemx -mlittle-endian
15716 @opindex mlittle-endian
15717 On System V.4 and embedded PowerPC systems compile code for the
15718 processor in little endian mode. The @option{-mlittle-endian} option is
15719 the same as @option{-mlittle}.
15722 @itemx -mbig-endian
15724 @opindex mbig-endian
15725 On System V.4 and embedded PowerPC systems compile code for the
15726 processor in big endian mode. The @option{-mbig-endian} option is
15727 the same as @option{-mbig}.
15729 @item -mdynamic-no-pic
15730 @opindex mdynamic-no-pic
15731 On Darwin and Mac OS X systems, compile code so that it is not
15732 relocatable, but that its external references are relocatable. The
15733 resulting code is suitable for applications, but not shared
15736 @item -msingle-pic-base
15737 @opindex msingle-pic-base
15738 Treat the register used for PIC addressing as read-only, rather than
15739 loading it in the prologue for each function. The run-time system is
15740 responsible for initializing this register with an appropriate value
15741 before execution begins.
15743 @item -mprioritize-restricted-insns=@var{priority}
15744 @opindex mprioritize-restricted-insns
15745 This option controls the priority that is assigned to
15746 dispatch-slot restricted instructions during the second scheduling
15747 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15748 @var{no/highest/second-highest} priority to dispatch slot restricted
15751 @item -msched-costly-dep=@var{dependence_type}
15752 @opindex msched-costly-dep
15753 This option controls which dependences are considered costly
15754 by the target during instruction scheduling. The argument
15755 @var{dependence_type} takes one of the following values:
15756 @var{no}: no dependence is costly,
15757 @var{all}: all dependences are costly,
15758 @var{true_store_to_load}: a true dependence from store to load is costly,
15759 @var{store_to_load}: any dependence from store to load is costly,
15760 @var{number}: any dependence which latency >= @var{number} is costly.
15762 @item -minsert-sched-nops=@var{scheme}
15763 @opindex minsert-sched-nops
15764 This option controls which nop insertion scheme will be used during
15765 the second scheduling pass. The argument @var{scheme} takes one of the
15767 @var{no}: Don't insert nops.
15768 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15769 according to the scheduler's grouping.
15770 @var{regroup_exact}: Insert nops to force costly dependent insns into
15771 separate groups. Insert exactly as many nops as needed to force an insn
15772 to a new group, according to the estimated processor grouping.
15773 @var{number}: Insert nops to force costly dependent insns into
15774 separate groups. Insert @var{number} nops to force an insn to a new group.
15777 @opindex mcall-sysv
15778 On System V.4 and embedded PowerPC systems compile code using calling
15779 conventions that adheres to the March 1995 draft of the System V
15780 Application Binary Interface, PowerPC processor supplement. This is the
15781 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15783 @item -mcall-sysv-eabi
15785 @opindex mcall-sysv-eabi
15786 @opindex mcall-eabi
15787 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15789 @item -mcall-sysv-noeabi
15790 @opindex mcall-sysv-noeabi
15791 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15793 @item -mcall-aixdesc
15795 On System V.4 and embedded PowerPC systems compile code for the AIX
15799 @opindex mcall-linux
15800 On System V.4 and embedded PowerPC systems compile code for the
15801 Linux-based GNU system.
15805 On System V.4 and embedded PowerPC systems compile code for the
15806 Hurd-based GNU system.
15808 @item -mcall-freebsd
15809 @opindex mcall-freebsd
15810 On System V.4 and embedded PowerPC systems compile code for the
15811 FreeBSD operating system.
15813 @item -mcall-netbsd
15814 @opindex mcall-netbsd
15815 On System V.4 and embedded PowerPC systems compile code for the
15816 NetBSD operating system.
15818 @item -mcall-openbsd
15819 @opindex mcall-netbsd
15820 On System V.4 and embedded PowerPC systems compile code for the
15821 OpenBSD operating system.
15823 @item -maix-struct-return
15824 @opindex maix-struct-return
15825 Return all structures in memory (as specified by the AIX ABI)@.
15827 @item -msvr4-struct-return
15828 @opindex msvr4-struct-return
15829 Return structures smaller than 8 bytes in registers (as specified by the
15832 @item -mabi=@var{abi-type}
15834 Extend the current ABI with a particular extension, or remove such extension.
15835 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15836 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15840 Extend the current ABI with SPE ABI extensions. This does not change
15841 the default ABI, instead it adds the SPE ABI extensions to the current
15845 @opindex mabi=no-spe
15846 Disable Booke SPE ABI extensions for the current ABI@.
15848 @item -mabi=ibmlongdouble
15849 @opindex mabi=ibmlongdouble
15850 Change the current ABI to use IBM extended precision long double.
15851 This is a PowerPC 32-bit SYSV ABI option.
15853 @item -mabi=ieeelongdouble
15854 @opindex mabi=ieeelongdouble
15855 Change the current ABI to use IEEE extended precision long double.
15856 This is a PowerPC 32-bit Linux ABI option.
15859 @itemx -mno-prototype
15860 @opindex mprototype
15861 @opindex mno-prototype
15862 On System V.4 and embedded PowerPC systems assume that all calls to
15863 variable argument functions are properly prototyped. Otherwise, the
15864 compiler must insert an instruction before every non prototyped call to
15865 set or clear bit 6 of the condition code register (@var{CR}) to
15866 indicate whether floating point values were passed in the floating point
15867 registers in case the function takes a variable arguments. With
15868 @option{-mprototype}, only calls to prototyped variable argument functions
15869 will set or clear the bit.
15873 On embedded PowerPC systems, assume that the startup module is called
15874 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15875 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15880 On embedded PowerPC systems, assume that the startup module is called
15881 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15886 On embedded PowerPC systems, assume that the startup module is called
15887 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15890 @item -myellowknife
15891 @opindex myellowknife
15892 On embedded PowerPC systems, assume that the startup module is called
15893 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15898 On System V.4 and embedded PowerPC systems, specify that you are
15899 compiling for a VxWorks system.
15903 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15904 header to indicate that @samp{eabi} extended relocations are used.
15910 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15911 Embedded Applications Binary Interface (eabi) which is a set of
15912 modifications to the System V.4 specifications. Selecting @option{-meabi}
15913 means that the stack is aligned to an 8 byte boundary, a function
15914 @code{__eabi} is called to from @code{main} to set up the eabi
15915 environment, and the @option{-msdata} option can use both @code{r2} and
15916 @code{r13} to point to two separate small data areas. Selecting
15917 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15918 do not call an initialization function from @code{main}, and the
15919 @option{-msdata} option will only use @code{r13} to point to a single
15920 small data area. The @option{-meabi} option is on by default if you
15921 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15924 @opindex msdata=eabi
15925 On System V.4 and embedded PowerPC systems, put small initialized
15926 @code{const} global and static data in the @samp{.sdata2} section, which
15927 is pointed to by register @code{r2}. Put small initialized
15928 non-@code{const} global and static data in the @samp{.sdata} section,
15929 which is pointed to by register @code{r13}. Put small uninitialized
15930 global and static data in the @samp{.sbss} section, which is adjacent to
15931 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15932 incompatible with the @option{-mrelocatable} option. The
15933 @option{-msdata=eabi} option also sets the @option{-memb} option.
15936 @opindex msdata=sysv
15937 On System V.4 and embedded PowerPC systems, put small global and static
15938 data in the @samp{.sdata} section, which is pointed to by register
15939 @code{r13}. Put small uninitialized global and static data in the
15940 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15941 The @option{-msdata=sysv} option is incompatible with the
15942 @option{-mrelocatable} option.
15944 @item -msdata=default
15946 @opindex msdata=default
15948 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15949 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15950 same as @option{-msdata=sysv}.
15953 @opindex msdata=data
15954 On System V.4 and embedded PowerPC systems, put small global
15955 data in the @samp{.sdata} section. Put small uninitialized global
15956 data in the @samp{.sbss} section. Do not use register @code{r13}
15957 to address small data however. This is the default behavior unless
15958 other @option{-msdata} options are used.
15962 @opindex msdata=none
15964 On embedded PowerPC systems, put all initialized global and static data
15965 in the @samp{.data} section, and all uninitialized data in the
15966 @samp{.bss} section.
15968 @item -mblock-move-inline-limit=@var{num}
15969 @opindex mblock-move-inline-limit
15970 Inline all block moves (such as calls to @code{memcpy} or structure
15971 copies) less than or equal to @var{num} bytes. The minimum value for
15972 @var{num} is 32 bytes on 32-bit targets and 64 bytes on 64-bit
15973 targets. The default value is target-specific.
15977 @cindex smaller data references (PowerPC)
15978 @cindex .sdata/.sdata2 references (PowerPC)
15979 On embedded PowerPC systems, put global and static items less than or
15980 equal to @var{num} bytes into the small data or bss sections instead of
15981 the normal data or bss section. By default, @var{num} is 8. The
15982 @option{-G @var{num}} switch is also passed to the linker.
15983 All modules should be compiled with the same @option{-G @var{num}} value.
15986 @itemx -mno-regnames
15988 @opindex mno-regnames
15989 On System V.4 and embedded PowerPC systems do (do not) emit register
15990 names in the assembly language output using symbolic forms.
15993 @itemx -mno-longcall
15995 @opindex mno-longcall
15996 By default assume that all calls are far away so that a longer more
15997 expensive calling sequence is required. This is required for calls
15998 further than 32 megabytes (33,554,432 bytes) from the current location.
15999 A short call will be generated if the compiler knows
16000 the call cannot be that far away. This setting can be overridden by
16001 the @code{shortcall} function attribute, or by @code{#pragma
16004 Some linkers are capable of detecting out-of-range calls and generating
16005 glue code on the fly. On these systems, long calls are unnecessary and
16006 generate slower code. As of this writing, the AIX linker can do this,
16007 as can the GNU linker for PowerPC/64. It is planned to add this feature
16008 to the GNU linker for 32-bit PowerPC systems as well.
16010 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
16011 callee, L42'', plus a ``branch island'' (glue code). The two target
16012 addresses represent the callee and the ``branch island''. The
16013 Darwin/PPC linker will prefer the first address and generate a ``bl
16014 callee'' if the PPC ``bl'' instruction will reach the callee directly;
16015 otherwise, the linker will generate ``bl L42'' to call the ``branch
16016 island''. The ``branch island'' is appended to the body of the
16017 calling function; it computes the full 32-bit address of the callee
16020 On Mach-O (Darwin) systems, this option directs the compiler emit to
16021 the glue for every direct call, and the Darwin linker decides whether
16022 to use or discard it.
16024 In the future, we may cause GCC to ignore all longcall specifications
16025 when the linker is known to generate glue.
16027 @item -mtls-markers
16028 @itemx -mno-tls-markers
16029 @opindex mtls-markers
16030 @opindex mno-tls-markers
16031 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
16032 specifying the function argument. The relocation allows ld to
16033 reliably associate function call with argument setup instructions for
16034 TLS optimization, which in turn allows gcc to better schedule the
16039 Adds support for multithreading with the @dfn{pthreads} library.
16040 This option sets flags for both the preprocessor and linker.
16045 This option will enable GCC to use the reciprocal estimate and
16046 reciprocal square root estimate instructions with additional
16047 Newton-Raphson steps to increase precision instead of doing a divide or
16048 square root and divide for floating point arguments. You should use
16049 the @option{-ffast-math} option when using @option{-mrecip} (or at
16050 least @option{-funsafe-math-optimizations},
16051 @option{-finite-math-only}, @option{-freciprocal-math} and
16052 @option{-fno-trapping-math}). Note that while the throughput of the
16053 sequence is generally higher than the throughput of the non-reciprocal
16054 instruction, the precision of the sequence can be decreased by up to 2
16055 ulp (i.e. the inverse of 1.0 equals 0.99999994) for reciprocal square
16058 @item -mrecip=@var{opt}
16059 @opindex mrecip=opt
16060 This option allows to control which reciprocal estimate instructions
16061 may be used. @var{opt} is a comma separated list of options, that may
16062 be preceeded by a @code{!} to invert the option:
16063 @code{all}: enable all estimate instructions,
16064 @code{default}: enable the default instructions, equvalent to @option{-mrecip},
16065 @code{none}: disable all estimate instructions, equivalent to @option{-mno-recip};
16066 @code{div}: enable the reciprocal approximation instructions for both single and double precision;
16067 @code{divf}: enable the single precision reciprocal approximation instructions;
16068 @code{divd}: enable the double precision reciprocal approximation instructions;
16069 @code{rsqrt}: enable the reciprocal square root approximation instructions for both single and double precision;
16070 @code{rsqrtf}: enable the single precision reciprocal square root approximation instructions;
16071 @code{rsqrtd}: enable the double precision reciprocal square root approximation instructions;
16073 So for example, @option{-mrecip=all,!rsqrtd} would enable the
16074 all of the reciprocal estimate instructions, except for the
16075 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
16076 which handle the double precision reciprocal square root calculations.
16078 @item -mrecip-precision
16079 @itemx -mno-recip-precision
16080 @opindex mrecip-precision
16081 Assume (do not assume) that the reciprocal estimate instructions
16082 provide higher precision estimates than is mandated by the powerpc
16083 ABI. Selecting @option{-mcpu=power6} or @option{-mcpu=power7}
16084 automatically selects @option{-mrecip-precision}. The double
16085 precision square root estimate instructions are not generated by
16086 default on low precision machines, since they do not provide an
16087 estimate that converges after three steps.
16089 @item -mveclibabi=@var{type}
16090 @opindex mveclibabi
16091 Specifies the ABI type to use for vectorizing intrinsics using an
16092 external library. The only type supported at present is @code{mass},
16093 which specifies to use IBM's Mathematical Acceleration Subsystem
16094 (MASS) libraries for vectorizing intrinsics using external libraries.
16095 GCC will currently emit calls to @code{acosd2}, @code{acosf4},
16096 @code{acoshd2}, @code{acoshf4}, @code{asind2}, @code{asinf4},
16097 @code{asinhd2}, @code{asinhf4}, @code{atan2d2}, @code{atan2f4},
16098 @code{atand2}, @code{atanf4}, @code{atanhd2}, @code{atanhf4},
16099 @code{cbrtd2}, @code{cbrtf4}, @code{cosd2}, @code{cosf4},
16100 @code{coshd2}, @code{coshf4}, @code{erfcd2}, @code{erfcf4},
16101 @code{erfd2}, @code{erff4}, @code{exp2d2}, @code{exp2f4},
16102 @code{expd2}, @code{expf4}, @code{expm1d2}, @code{expm1f4},
16103 @code{hypotd2}, @code{hypotf4}, @code{lgammad2}, @code{lgammaf4},
16104 @code{log10d2}, @code{log10f4}, @code{log1pd2}, @code{log1pf4},
16105 @code{log2d2}, @code{log2f4}, @code{logd2}, @code{logf4},
16106 @code{powd2}, @code{powf4}, @code{sind2}, @code{sinf4}, @code{sinhd2},
16107 @code{sinhf4}, @code{sqrtd2}, @code{sqrtf4}, @code{tand2},
16108 @code{tanf4}, @code{tanhd2}, and @code{tanhf4} when generating code
16109 for power7. Both @option{-ftree-vectorize} and
16110 @option{-funsafe-math-optimizations} have to be enabled. The MASS
16111 libraries will have to be specified at link time.
16116 Generate (do not generate) the @code{friz} instruction when the
16117 @option{-funsafe-math-optimizations} option is used to optimize
16118 rounding a floating point value to 64-bit integer and back to floating
16119 point. The @code{friz} instruction does not return the same value if
16120 the floating point number is too large to fit in an integer.
16124 @subsection RX Options
16127 These command line options are defined for RX targets:
16130 @item -m64bit-doubles
16131 @itemx -m32bit-doubles
16132 @opindex m64bit-doubles
16133 @opindex m32bit-doubles
16134 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
16135 or 32-bits (@option{-m32bit-doubles}) in size. The default is
16136 @option{-m32bit-doubles}. @emph{Note} RX floating point hardware only
16137 works on 32-bit values, which is why the default is
16138 @option{-m32bit-doubles}.
16144 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
16145 floating point hardware. The default is enabled for the @var{RX600}
16146 series and disabled for the @var{RX200} series.
16148 Floating point instructions will only be generated for 32-bit floating
16149 point values however, so if the @option{-m64bit-doubles} option is in
16150 use then the FPU hardware will not be used for doubles.
16152 @emph{Note} If the @option{-fpu} option is enabled then
16153 @option{-funsafe-math-optimizations} is also enabled automatically.
16154 This is because the RX FPU instructions are themselves unsafe.
16156 @item -mcpu=@var{name}
16158 Selects the type of RX CPU to be targeted. Currently three types are
16159 supported, the generic @var{RX600} and @var{RX200} series hardware and
16160 the specific @var{RX610} cpu. The default is @var{RX600}.
16162 The only difference between @var{RX600} and @var{RX610} is that the
16163 @var{RX610} does not support the @code{MVTIPL} instruction.
16165 The @var{RX200} series does not have a hardware floating point unit
16166 and so @option{-nofpu} is enabled by default when this type is
16169 @item -mbig-endian-data
16170 @itemx -mlittle-endian-data
16171 @opindex mbig-endian-data
16172 @opindex mlittle-endian-data
16173 Store data (but not code) in the big-endian format. The default is
16174 @option{-mlittle-endian-data}, ie to store data in the little endian
16177 @item -msmall-data-limit=@var{N}
16178 @opindex msmall-data-limit
16179 Specifies the maximum size in bytes of global and static variables
16180 which can be placed into the small data area. Using the small data
16181 area can lead to smaller and faster code, but the size of area is
16182 limited and it is up to the programmer to ensure that the area does
16183 not overflow. Also when the small data area is used one of the RX's
16184 registers (@code{r13}) is reserved for use pointing to this area, so
16185 it is no longer available for use by the compiler. This could result
16186 in slower and/or larger code if variables which once could have been
16187 held in @code{r13} are now pushed onto the stack.
16189 Note, common variables (variables which have not been initialised) and
16190 constants are not placed into the small data area as they are assigned
16191 to other sections in the output executable.
16193 The default value is zero, which disables this feature. Note, this
16194 feature is not enabled by default with higher optimization levels
16195 (@option{-O2} etc) because of the potentially detrimental effects of
16196 reserving register @code{r13}. It is up to the programmer to
16197 experiment and discover whether this feature is of benefit to their
16204 Use the simulator runtime. The default is to use the libgloss board
16207 @item -mas100-syntax
16208 @itemx -mno-as100-syntax
16209 @opindex mas100-syntax
16210 @opindex mno-as100-syntax
16211 When generating assembler output use a syntax that is compatible with
16212 Renesas's AS100 assembler. This syntax can also be handled by the GAS
16213 assembler but it has some restrictions so generating it is not the
16216 @item -mmax-constant-size=@var{N}
16217 @opindex mmax-constant-size
16218 Specifies the maximum size, in bytes, of a constant that can be used as
16219 an operand in a RX instruction. Although the RX instruction set does
16220 allow constants of up to 4 bytes in length to be used in instructions,
16221 a longer value equates to a longer instruction. Thus in some
16222 circumstances it can be beneficial to restrict the size of constants
16223 that are used in instructions. Constants that are too big are instead
16224 placed into a constant pool and referenced via register indirection.
16226 The value @var{N} can be between 0 and 4. A value of 0 (the default)
16227 or 4 means that constants of any size are allowed.
16231 Enable linker relaxation. Linker relaxation is a process whereby the
16232 linker will attempt to reduce the size of a program by finding shorter
16233 versions of various instructions. Disabled by default.
16235 @item -mint-register=@var{N}
16236 @opindex mint-register
16237 Specify the number of registers to reserve for fast interrupt handler
16238 functions. The value @var{N} can be between 0 and 4. A value of 1
16239 means that register @code{r13} will be reserved for the exclusive use
16240 of fast interrupt handlers. A value of 2 reserves @code{r13} and
16241 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
16242 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
16243 A value of 0, the default, does not reserve any registers.
16245 @item -msave-acc-in-interrupts
16246 @opindex msave-acc-in-interrupts
16247 Specifies that interrupt handler functions should preserve the
16248 accumulator register. This is only necessary if normal code might use
16249 the accumulator register, for example because it performs 64-bit
16250 multiplications. The default is to ignore the accumulator as this
16251 makes the interrupt handlers faster.
16255 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
16256 has special significance to the RX port when used with the
16257 @code{interrupt} function attribute. This attribute indicates a
16258 function intended to process fast interrupts. GCC will will ensure
16259 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
16260 and/or @code{r13} and only provided that the normal use of the
16261 corresponding registers have been restricted via the
16262 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
16265 @node S/390 and zSeries Options
16266 @subsection S/390 and zSeries Options
16267 @cindex S/390 and zSeries Options
16269 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
16273 @itemx -msoft-float
16274 @opindex mhard-float
16275 @opindex msoft-float
16276 Use (do not use) the hardware floating-point instructions and registers
16277 for floating-point operations. When @option{-msoft-float} is specified,
16278 functions in @file{libgcc.a} will be used to perform floating-point
16279 operations. When @option{-mhard-float} is specified, the compiler
16280 generates IEEE floating-point instructions. This is the default.
16283 @itemx -mno-hard-dfp
16285 @opindex mno-hard-dfp
16286 Use (do not use) the hardware decimal-floating-point instructions for
16287 decimal-floating-point operations. When @option{-mno-hard-dfp} is
16288 specified, functions in @file{libgcc.a} will be used to perform
16289 decimal-floating-point operations. When @option{-mhard-dfp} is
16290 specified, the compiler generates decimal-floating-point hardware
16291 instructions. This is the default for @option{-march=z9-ec} or higher.
16293 @item -mlong-double-64
16294 @itemx -mlong-double-128
16295 @opindex mlong-double-64
16296 @opindex mlong-double-128
16297 These switches control the size of @code{long double} type. A size
16298 of 64bit makes the @code{long double} type equivalent to the @code{double}
16299 type. This is the default.
16302 @itemx -mno-backchain
16303 @opindex mbackchain
16304 @opindex mno-backchain
16305 Store (do not store) the address of the caller's frame as backchain pointer
16306 into the callee's stack frame.
16307 A backchain may be needed to allow debugging using tools that do not understand
16308 DWARF-2 call frame information.
16309 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
16310 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
16311 the backchain is placed into the topmost word of the 96/160 byte register
16314 In general, code compiled with @option{-mbackchain} is call-compatible with
16315 code compiled with @option{-mmo-backchain}; however, use of the backchain
16316 for debugging purposes usually requires that the whole binary is built with
16317 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
16318 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
16319 to build a linux kernel use @option{-msoft-float}.
16321 The default is to not maintain the backchain.
16323 @item -mpacked-stack
16324 @itemx -mno-packed-stack
16325 @opindex mpacked-stack
16326 @opindex mno-packed-stack
16327 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
16328 specified, the compiler uses the all fields of the 96/160 byte register save
16329 area only for their default purpose; unused fields still take up stack space.
16330 When @option{-mpacked-stack} is specified, register save slots are densely
16331 packed at the top of the register save area; unused space is reused for other
16332 purposes, allowing for more efficient use of the available stack space.
16333 However, when @option{-mbackchain} is also in effect, the topmost word of
16334 the save area is always used to store the backchain, and the return address
16335 register is always saved two words below the backchain.
16337 As long as the stack frame backchain is not used, code generated with
16338 @option{-mpacked-stack} is call-compatible with code generated with
16339 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
16340 S/390 or zSeries generated code that uses the stack frame backchain at run
16341 time, not just for debugging purposes. Such code is not call-compatible
16342 with code compiled with @option{-mpacked-stack}. Also, note that the
16343 combination of @option{-mbackchain},
16344 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
16345 to build a linux kernel use @option{-msoft-float}.
16347 The default is to not use the packed stack layout.
16350 @itemx -mno-small-exec
16351 @opindex msmall-exec
16352 @opindex mno-small-exec
16353 Generate (or do not generate) code using the @code{bras} instruction
16354 to do subroutine calls.
16355 This only works reliably if the total executable size does not
16356 exceed 64k. The default is to use the @code{basr} instruction instead,
16357 which does not have this limitation.
16363 When @option{-m31} is specified, generate code compliant to the
16364 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
16365 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
16366 particular to generate 64-bit instructions. For the @samp{s390}
16367 targets, the default is @option{-m31}, while the @samp{s390x}
16368 targets default to @option{-m64}.
16374 When @option{-mzarch} is specified, generate code using the
16375 instructions available on z/Architecture.
16376 When @option{-mesa} is specified, generate code using the
16377 instructions available on ESA/390. Note that @option{-mesa} is
16378 not possible with @option{-m64}.
16379 When generating code compliant to the GNU/Linux for S/390 ABI,
16380 the default is @option{-mesa}. When generating code compliant
16381 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
16387 Generate (or do not generate) code using the @code{mvcle} instruction
16388 to perform block moves. When @option{-mno-mvcle} is specified,
16389 use a @code{mvc} loop instead. This is the default unless optimizing for
16396 Print (or do not print) additional debug information when compiling.
16397 The default is to not print debug information.
16399 @item -march=@var{cpu-type}
16401 Generate code that will run on @var{cpu-type}, which is the name of a system
16402 representing a certain processor type. Possible values for
16403 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
16404 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
16405 When generating code using the instructions available on z/Architecture,
16406 the default is @option{-march=z900}. Otherwise, the default is
16407 @option{-march=g5}.
16409 @item -mtune=@var{cpu-type}
16411 Tune to @var{cpu-type} everything applicable about the generated code,
16412 except for the ABI and the set of available instructions.
16413 The list of @var{cpu-type} values is the same as for @option{-march}.
16414 The default is the value used for @option{-march}.
16417 @itemx -mno-tpf-trace
16418 @opindex mtpf-trace
16419 @opindex mno-tpf-trace
16420 Generate code that adds (does not add) in TPF OS specific branches to trace
16421 routines in the operating system. This option is off by default, even
16422 when compiling for the TPF OS@.
16425 @itemx -mno-fused-madd
16426 @opindex mfused-madd
16427 @opindex mno-fused-madd
16428 Generate code that uses (does not use) the floating point multiply and
16429 accumulate instructions. These instructions are generated by default if
16430 hardware floating point is used.
16432 @item -mwarn-framesize=@var{framesize}
16433 @opindex mwarn-framesize
16434 Emit a warning if the current function exceeds the given frame size. Because
16435 this is a compile time check it doesn't need to be a real problem when the program
16436 runs. It is intended to identify functions which most probably cause
16437 a stack overflow. It is useful to be used in an environment with limited stack
16438 size e.g.@: the linux kernel.
16440 @item -mwarn-dynamicstack
16441 @opindex mwarn-dynamicstack
16442 Emit a warning if the function calls alloca or uses dynamically
16443 sized arrays. This is generally a bad idea with a limited stack size.
16445 @item -mstack-guard=@var{stack-guard}
16446 @itemx -mstack-size=@var{stack-size}
16447 @opindex mstack-guard
16448 @opindex mstack-size
16449 If these options are provided the s390 back end emits additional instructions in
16450 the function prologue which trigger a trap if the stack size is @var{stack-guard}
16451 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
16452 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
16453 the frame size of the compiled function is chosen.
16454 These options are intended to be used to help debugging stack overflow problems.
16455 The additionally emitted code causes only little overhead and hence can also be
16456 used in production like systems without greater performance degradation. The given
16457 values have to be exact powers of 2 and @var{stack-size} has to be greater than
16458 @var{stack-guard} without exceeding 64k.
16459 In order to be efficient the extra code makes the assumption that the stack starts
16460 at an address aligned to the value given by @var{stack-size}.
16461 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
16464 @node Score Options
16465 @subsection Score Options
16466 @cindex Score Options
16468 These options are defined for Score implementations:
16473 Compile code for big endian mode. This is the default.
16477 Compile code for little endian mode.
16481 Disable generate bcnz instruction.
16485 Enable generate unaligned load and store instruction.
16489 Enable the use of multiply-accumulate instructions. Disabled by default.
16493 Specify the SCORE5 as the target architecture.
16497 Specify the SCORE5U of the target architecture.
16501 Specify the SCORE7 as the target architecture. This is the default.
16505 Specify the SCORE7D as the target architecture.
16509 @subsection SH Options
16511 These @samp{-m} options are defined for the SH implementations:
16516 Generate code for the SH1.
16520 Generate code for the SH2.
16523 Generate code for the SH2e.
16527 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
16528 that the floating-point unit is not used.
16530 @item -m2a-single-only
16531 @opindex m2a-single-only
16532 Generate code for the SH2a-FPU, in such a way that no double-precision
16533 floating point operations are used.
16536 @opindex m2a-single
16537 Generate code for the SH2a-FPU assuming the floating-point unit is in
16538 single-precision mode by default.
16542 Generate code for the SH2a-FPU assuming the floating-point unit is in
16543 double-precision mode by default.
16547 Generate code for the SH3.
16551 Generate code for the SH3e.
16555 Generate code for the SH4 without a floating-point unit.
16557 @item -m4-single-only
16558 @opindex m4-single-only
16559 Generate code for the SH4 with a floating-point unit that only
16560 supports single-precision arithmetic.
16564 Generate code for the SH4 assuming the floating-point unit is in
16565 single-precision mode by default.
16569 Generate code for the SH4.
16573 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
16574 floating-point unit is not used.
16576 @item -m4a-single-only
16577 @opindex m4a-single-only
16578 Generate code for the SH4a, in such a way that no double-precision
16579 floating point operations are used.
16582 @opindex m4a-single
16583 Generate code for the SH4a assuming the floating-point unit is in
16584 single-precision mode by default.
16588 Generate code for the SH4a.
16592 Same as @option{-m4a-nofpu}, except that it implicitly passes
16593 @option{-dsp} to the assembler. GCC doesn't generate any DSP
16594 instructions at the moment.
16598 Compile code for the processor in big endian mode.
16602 Compile code for the processor in little endian mode.
16606 Align doubles at 64-bit boundaries. Note that this changes the calling
16607 conventions, and thus some functions from the standard C library will
16608 not work unless you recompile it first with @option{-mdalign}.
16612 Shorten some address references at link time, when possible; uses the
16613 linker option @option{-relax}.
16617 Use 32-bit offsets in @code{switch} tables. The default is to use
16622 Enable the use of bit manipulation instructions on SH2A.
16626 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
16627 alignment constraints.
16631 Comply with the calling conventions defined by Renesas.
16635 Comply with the calling conventions defined by Renesas.
16639 Comply with the calling conventions defined for GCC before the Renesas
16640 conventions were available. This option is the default for all
16641 targets of the SH toolchain except for @samp{sh-symbianelf}.
16644 @opindex mnomacsave
16645 Mark the @code{MAC} register as call-clobbered, even if
16646 @option{-mhitachi} is given.
16650 Increase IEEE-compliance of floating-point code.
16651 At the moment, this is equivalent to @option{-fno-finite-math-only}.
16652 When generating 16 bit SH opcodes, getting IEEE-conforming results for
16653 comparisons of NANs / infinities incurs extra overhead in every
16654 floating point comparison, therefore the default is set to
16655 @option{-ffinite-math-only}.
16657 @item -minline-ic_invalidate
16658 @opindex minline-ic_invalidate
16659 Inline code to invalidate instruction cache entries after setting up
16660 nested function trampolines.
16661 This option has no effect if -musermode is in effect and the selected
16662 code generation option (e.g. -m4) does not allow the use of the icbi
16664 If the selected code generation option does not allow the use of the icbi
16665 instruction, and -musermode is not in effect, the inlined code will
16666 manipulate the instruction cache address array directly with an associative
16667 write. This not only requires privileged mode, but it will also
16668 fail if the cache line had been mapped via the TLB and has become unmapped.
16672 Dump instruction size and location in the assembly code.
16675 @opindex mpadstruct
16676 This option is deprecated. It pads structures to multiple of 4 bytes,
16677 which is incompatible with the SH ABI@.
16681 Optimize for space instead of speed. Implied by @option{-Os}.
16684 @opindex mprefergot
16685 When generating position-independent code, emit function calls using
16686 the Global Offset Table instead of the Procedure Linkage Table.
16690 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
16691 if the inlined code would not work in user mode.
16692 This is the default when the target is @code{sh-*-linux*}.
16694 @item -multcost=@var{number}
16695 @opindex multcost=@var{number}
16696 Set the cost to assume for a multiply insn.
16698 @item -mdiv=@var{strategy}
16699 @opindex mdiv=@var{strategy}
16700 Set the division strategy to use for SHmedia code. @var{strategy} must be
16701 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
16702 inv:call2, inv:fp .
16703 "fp" performs the operation in floating point. This has a very high latency,
16704 but needs only a few instructions, so it might be a good choice if
16705 your code has enough easily exploitable ILP to allow the compiler to
16706 schedule the floating point instructions together with other instructions.
16707 Division by zero causes a floating point exception.
16708 "inv" uses integer operations to calculate the inverse of the divisor,
16709 and then multiplies the dividend with the inverse. This strategy allows
16710 cse and hoisting of the inverse calculation. Division by zero calculates
16711 an unspecified result, but does not trap.
16712 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
16713 have been found, or if the entire operation has been hoisted to the same
16714 place, the last stages of the inverse calculation are intertwined with the
16715 final multiply to reduce the overall latency, at the expense of using a few
16716 more instructions, and thus offering fewer scheduling opportunities with
16718 "call" calls a library function that usually implements the inv:minlat
16720 This gives high code density for m5-*media-nofpu compilations.
16721 "call2" uses a different entry point of the same library function, where it
16722 assumes that a pointer to a lookup table has already been set up, which
16723 exposes the pointer load to cse / code hoisting optimizations.
16724 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
16725 code generation, but if the code stays unoptimized, revert to the "call",
16726 "call2", or "fp" strategies, respectively. Note that the
16727 potentially-trapping side effect of division by zero is carried by a
16728 separate instruction, so it is possible that all the integer instructions
16729 are hoisted out, but the marker for the side effect stays where it is.
16730 A recombination to fp operations or a call is not possible in that case.
16731 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
16732 that the inverse calculation was nor separated from the multiply, they speed
16733 up division where the dividend fits into 20 bits (plus sign where applicable),
16734 by inserting a test to skip a number of operations in this case; this test
16735 slows down the case of larger dividends. inv20u assumes the case of a such
16736 a small dividend to be unlikely, and inv20l assumes it to be likely.
16738 @item -maccumulate-outgoing-args
16739 @opindex maccumulate-outgoing-args
16740 Reserve space once for outgoing arguments in the function prologue rather
16741 than around each call. Generally beneficial for performance and size. Also
16742 needed for unwinding to avoid changing the stack frame around conditional code.
16744 @item -mdivsi3_libfunc=@var{name}
16745 @opindex mdivsi3_libfunc=@var{name}
16746 Set the name of the library function used for 32 bit signed division to
16747 @var{name}. This only affect the name used in the call and inv:call
16748 division strategies, and the compiler will still expect the same
16749 sets of input/output/clobbered registers as if this option was not present.
16751 @item -mfixed-range=@var{register-range}
16752 @opindex mfixed-range
16753 Generate code treating the given register range as fixed registers.
16754 A fixed register is one that the register allocator can not use. This is
16755 useful when compiling kernel code. A register range is specified as
16756 two registers separated by a dash. Multiple register ranges can be
16757 specified separated by a comma.
16759 @item -madjust-unroll
16760 @opindex madjust-unroll
16761 Throttle unrolling to avoid thrashing target registers.
16762 This option only has an effect if the gcc code base supports the
16763 TARGET_ADJUST_UNROLL_MAX target hook.
16765 @item -mindexed-addressing
16766 @opindex mindexed-addressing
16767 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16768 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16769 semantics for the indexed addressing mode. The architecture allows the
16770 implementation of processors with 64 bit MMU, which the OS could use to
16771 get 32 bit addressing, but since no current hardware implementation supports
16772 this or any other way to make the indexed addressing mode safe to use in
16773 the 32 bit ABI, the default is -mno-indexed-addressing.
16775 @item -mgettrcost=@var{number}
16776 @opindex mgettrcost=@var{number}
16777 Set the cost assumed for the gettr instruction to @var{number}.
16778 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16782 Assume pt* instructions won't trap. This will generally generate better
16783 scheduled code, but is unsafe on current hardware. The current architecture
16784 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16785 This has the unintentional effect of making it unsafe to schedule ptabs /
16786 ptrel before a branch, or hoist it out of a loop. For example,
16787 __do_global_ctors, a part of libgcc that runs constructors at program
16788 startup, calls functions in a list which is delimited by @minus{}1. With the
16789 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16790 That means that all the constructors will be run a bit quicker, but when
16791 the loop comes to the end of the list, the program crashes because ptabs
16792 loads @minus{}1 into a target register. Since this option is unsafe for any
16793 hardware implementing the current architecture specification, the default
16794 is -mno-pt-fixed. Unless the user specifies a specific cost with
16795 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16796 this deters register allocation using target registers for storing
16799 @item -minvalid-symbols
16800 @opindex minvalid-symbols
16801 Assume symbols might be invalid. Ordinary function symbols generated by
16802 the compiler will always be valid to load with movi/shori/ptabs or
16803 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16804 to generate symbols that will cause ptabs / ptrel to trap.
16805 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16806 It will then prevent cross-basic-block cse, hoisting and most scheduling
16807 of symbol loads. The default is @option{-mno-invalid-symbols}.
16810 @node Solaris 2 Options
16811 @subsection Solaris 2 Options
16812 @cindex Solaris 2 options
16814 These @samp{-m} options are supported on Solaris 2:
16817 @item -mimpure-text
16818 @opindex mimpure-text
16819 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16820 the compiler to not pass @option{-z text} to the linker when linking a
16821 shared object. Using this option, you can link position-dependent
16822 code into a shared object.
16824 @option{-mimpure-text} suppresses the ``relocations remain against
16825 allocatable but non-writable sections'' linker error message.
16826 However, the necessary relocations will trigger copy-on-write, and the
16827 shared object is not actually shared across processes. Instead of
16828 using @option{-mimpure-text}, you should compile all source code with
16829 @option{-fpic} or @option{-fPIC}.
16833 These switches are supported in addition to the above on Solaris 2:
16838 Add support for multithreading using the Solaris threads library. This
16839 option sets flags for both the preprocessor and linker. This option does
16840 not affect the thread safety of object code produced by the compiler or
16841 that of libraries supplied with it.
16845 Add support for multithreading using the POSIX threads library. This
16846 option sets flags for both the preprocessor and linker. This option does
16847 not affect the thread safety of object code produced by the compiler or
16848 that of libraries supplied with it.
16852 This is a synonym for @option{-pthreads}.
16855 @node SPARC Options
16856 @subsection SPARC Options
16857 @cindex SPARC options
16859 These @samp{-m} options are supported on the SPARC:
16862 @item -mno-app-regs
16864 @opindex mno-app-regs
16866 Specify @option{-mapp-regs} to generate output using the global registers
16867 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16870 To be fully SVR4 ABI compliant at the cost of some performance loss,
16871 specify @option{-mno-app-regs}. You should compile libraries and system
16872 software with this option.
16875 @itemx -mhard-float
16877 @opindex mhard-float
16878 Generate output containing floating point instructions. This is the
16882 @itemx -msoft-float
16884 @opindex msoft-float
16885 Generate output containing library calls for floating point.
16886 @strong{Warning:} the requisite libraries are not available for all SPARC
16887 targets. Normally the facilities of the machine's usual C compiler are
16888 used, but this cannot be done directly in cross-compilation. You must make
16889 your own arrangements to provide suitable library functions for
16890 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16891 @samp{sparclite-*-*} do provide software floating point support.
16893 @option{-msoft-float} changes the calling convention in the output file;
16894 therefore, it is only useful if you compile @emph{all} of a program with
16895 this option. In particular, you need to compile @file{libgcc.a}, the
16896 library that comes with GCC, with @option{-msoft-float} in order for
16899 @item -mhard-quad-float
16900 @opindex mhard-quad-float
16901 Generate output containing quad-word (long double) floating point
16904 @item -msoft-quad-float
16905 @opindex msoft-quad-float
16906 Generate output containing library calls for quad-word (long double)
16907 floating point instructions. The functions called are those specified
16908 in the SPARC ABI@. This is the default.
16910 As of this writing, there are no SPARC implementations that have hardware
16911 support for the quad-word floating point instructions. They all invoke
16912 a trap handler for one of these instructions, and then the trap handler
16913 emulates the effect of the instruction. Because of the trap handler overhead,
16914 this is much slower than calling the ABI library routines. Thus the
16915 @option{-msoft-quad-float} option is the default.
16917 @item -mno-unaligned-doubles
16918 @itemx -munaligned-doubles
16919 @opindex mno-unaligned-doubles
16920 @opindex munaligned-doubles
16921 Assume that doubles have 8 byte alignment. This is the default.
16923 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16924 alignment only if they are contained in another type, or if they have an
16925 absolute address. Otherwise, it assumes they have 4 byte alignment.
16926 Specifying this option avoids some rare compatibility problems with code
16927 generated by other compilers. It is not the default because it results
16928 in a performance loss, especially for floating point code.
16930 @item -mno-faster-structs
16931 @itemx -mfaster-structs
16932 @opindex mno-faster-structs
16933 @opindex mfaster-structs
16934 With @option{-mfaster-structs}, the compiler assumes that structures
16935 should have 8 byte alignment. This enables the use of pairs of
16936 @code{ldd} and @code{std} instructions for copies in structure
16937 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16938 However, the use of this changed alignment directly violates the SPARC
16939 ABI@. Thus, it's intended only for use on targets where the developer
16940 acknowledges that their resulting code will not be directly in line with
16941 the rules of the ABI@.
16943 @item -mcpu=@var{cpu_type}
16945 Set the instruction set, register set, and instruction scheduling parameters
16946 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16947 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16948 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16949 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16950 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16952 Default instruction scheduling parameters are used for values that select
16953 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16954 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16956 Here is a list of each supported architecture and their supported
16961 v8: supersparc, hypersparc
16962 sparclite: f930, f934, sparclite86x
16964 v9: ultrasparc, ultrasparc3, niagara, niagara2
16967 By default (unless configured otherwise), GCC generates code for the V7
16968 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16969 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16970 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16971 SPARCStation 1, 2, IPX etc.
16973 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16974 architecture. The only difference from V7 code is that the compiler emits
16975 the integer multiply and integer divide instructions which exist in SPARC-V8
16976 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16977 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16980 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16981 the SPARC architecture. This adds the integer multiply, integer divide step
16982 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16983 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16984 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16985 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16986 MB86934 chip, which is the more recent SPARClite with FPU@.
16988 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16989 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16990 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16991 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16992 optimizes it for the TEMIC SPARClet chip.
16994 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16995 architecture. This adds 64-bit integer and floating-point move instructions,
16996 3 additional floating-point condition code registers and conditional move
16997 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16998 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16999 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
17000 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
17001 @option{-mcpu=niagara}, the compiler additionally optimizes it for
17002 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
17003 additionally optimizes it for Sun UltraSPARC T2 chips.
17005 @item -mtune=@var{cpu_type}
17007 Set the instruction scheduling parameters for machine type
17008 @var{cpu_type}, but do not set the instruction set or register set that the
17009 option @option{-mcpu=@var{cpu_type}} would.
17011 The same values for @option{-mcpu=@var{cpu_type}} can be used for
17012 @option{-mtune=@var{cpu_type}}, but the only useful values are those
17013 that select a particular cpu implementation. Those are @samp{cypress},
17014 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
17015 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
17016 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
17021 @opindex mno-v8plus
17022 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
17023 difference from the V8 ABI is that the global and out registers are
17024 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
17025 mode for all SPARC-V9 processors.
17031 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
17032 Visual Instruction Set extensions. The default is @option{-mno-vis}.
17035 These @samp{-m} options are supported in addition to the above
17036 on SPARC-V9 processors in 64-bit environments:
17039 @item -mlittle-endian
17040 @opindex mlittle-endian
17041 Generate code for a processor running in little-endian mode. It is only
17042 available for a few configurations and most notably not on Solaris and Linux.
17048 Generate code for a 32-bit or 64-bit environment.
17049 The 32-bit environment sets int, long and pointer to 32 bits.
17050 The 64-bit environment sets int to 32 bits and long and pointer
17053 @item -mcmodel=medlow
17054 @opindex mcmodel=medlow
17055 Generate code for the Medium/Low code model: 64-bit addresses, programs
17056 must be linked in the low 32 bits of memory. Programs can be statically
17057 or dynamically linked.
17059 @item -mcmodel=medmid
17060 @opindex mcmodel=medmid
17061 Generate code for the Medium/Middle code model: 64-bit addresses, programs
17062 must be linked in the low 44 bits of memory, the text and data segments must
17063 be less than 2GB in size and the data segment must be located within 2GB of
17066 @item -mcmodel=medany
17067 @opindex mcmodel=medany
17068 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
17069 may be linked anywhere in memory, the text and data segments must be less
17070 than 2GB in size and the data segment must be located within 2GB of the
17073 @item -mcmodel=embmedany
17074 @opindex mcmodel=embmedany
17075 Generate code for the Medium/Anywhere code model for embedded systems:
17076 64-bit addresses, the text and data segments must be less than 2GB in
17077 size, both starting anywhere in memory (determined at link time). The
17078 global register %g4 points to the base of the data segment. Programs
17079 are statically linked and PIC is not supported.
17082 @itemx -mno-stack-bias
17083 @opindex mstack-bias
17084 @opindex mno-stack-bias
17085 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
17086 frame pointer if present, are offset by @minus{}2047 which must be added back
17087 when making stack frame references. This is the default in 64-bit mode.
17088 Otherwise, assume no such offset is present.
17092 @subsection SPU Options
17093 @cindex SPU options
17095 These @samp{-m} options are supported on the SPU:
17099 @itemx -merror-reloc
17100 @opindex mwarn-reloc
17101 @opindex merror-reloc
17103 The loader for SPU does not handle dynamic relocations. By default, GCC
17104 will give an error when it generates code that requires a dynamic
17105 relocation. @option{-mno-error-reloc} disables the error,
17106 @option{-mwarn-reloc} will generate a warning instead.
17109 @itemx -munsafe-dma
17111 @opindex munsafe-dma
17113 Instructions which initiate or test completion of DMA must not be
17114 reordered with respect to loads and stores of the memory which is being
17115 accessed. Users typically address this problem using the volatile
17116 keyword, but that can lead to inefficient code in places where the
17117 memory is known to not change. Rather than mark the memory as volatile
17118 we treat the DMA instructions as potentially effecting all memory. With
17119 @option{-munsafe-dma} users must use the volatile keyword to protect
17122 @item -mbranch-hints
17123 @opindex mbranch-hints
17125 By default, GCC will generate a branch hint instruction to avoid
17126 pipeline stalls for always taken or probably taken branches. A hint
17127 will not be generated closer than 8 instructions away from its branch.
17128 There is little reason to disable them, except for debugging purposes,
17129 or to make an object a little bit smaller.
17133 @opindex msmall-mem
17134 @opindex mlarge-mem
17136 By default, GCC generates code assuming that addresses are never larger
17137 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
17138 a full 32 bit address.
17143 By default, GCC links against startup code that assumes the SPU-style
17144 main function interface (which has an unconventional parameter list).
17145 With @option{-mstdmain}, GCC will link your program against startup
17146 code that assumes a C99-style interface to @code{main}, including a
17147 local copy of @code{argv} strings.
17149 @item -mfixed-range=@var{register-range}
17150 @opindex mfixed-range
17151 Generate code treating the given register range as fixed registers.
17152 A fixed register is one that the register allocator can not use. This is
17153 useful when compiling kernel code. A register range is specified as
17154 two registers separated by a dash. Multiple register ranges can be
17155 specified separated by a comma.
17161 Compile code assuming that pointers to the PPU address space accessed
17162 via the @code{__ea} named address space qualifier are either 32 or 64
17163 bits wide. The default is 32 bits. As this is an ABI changing option,
17164 all object code in an executable must be compiled with the same setting.
17166 @item -maddress-space-conversion
17167 @itemx -mno-address-space-conversion
17168 @opindex maddress-space-conversion
17169 @opindex mno-address-space-conversion
17170 Allow/disallow treating the @code{__ea} address space as superset
17171 of the generic address space. This enables explicit type casts
17172 between @code{__ea} and generic pointer as well as implicit
17173 conversions of generic pointers to @code{__ea} pointers. The
17174 default is to allow address space pointer conversions.
17176 @item -mcache-size=@var{cache-size}
17177 @opindex mcache-size
17178 This option controls the version of libgcc that the compiler links to an
17179 executable and selects a software-managed cache for accessing variables
17180 in the @code{__ea} address space with a particular cache size. Possible
17181 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
17182 and @samp{128}. The default cache size is 64KB.
17184 @item -matomic-updates
17185 @itemx -mno-atomic-updates
17186 @opindex matomic-updates
17187 @opindex mno-atomic-updates
17188 This option controls the version of libgcc that the compiler links to an
17189 executable and selects whether atomic updates to the software-managed
17190 cache of PPU-side variables are used. If you use atomic updates, changes
17191 to a PPU variable from SPU code using the @code{__ea} named address space
17192 qualifier will not interfere with changes to other PPU variables residing
17193 in the same cache line from PPU code. If you do not use atomic updates,
17194 such interference may occur; however, writing back cache lines will be
17195 more efficient. The default behavior is to use atomic updates.
17198 @itemx -mdual-nops=@var{n}
17199 @opindex mdual-nops
17200 By default, GCC will insert nops to increase dual issue when it expects
17201 it to increase performance. @var{n} can be a value from 0 to 10. A
17202 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
17203 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
17205 @item -mhint-max-nops=@var{n}
17206 @opindex mhint-max-nops
17207 Maximum number of nops to insert for a branch hint. A branch hint must
17208 be at least 8 instructions away from the branch it is effecting. GCC
17209 will insert up to @var{n} nops to enforce this, otherwise it will not
17210 generate the branch hint.
17212 @item -mhint-max-distance=@var{n}
17213 @opindex mhint-max-distance
17214 The encoding of the branch hint instruction limits the hint to be within
17215 256 instructions of the branch it is effecting. By default, GCC makes
17216 sure it is within 125.
17219 @opindex msafe-hints
17220 Work around a hardware bug which causes the SPU to stall indefinitely.
17221 By default, GCC will insert the @code{hbrp} instruction to make sure
17222 this stall won't happen.
17226 @node System V Options
17227 @subsection Options for System V
17229 These additional options are available on System V Release 4 for
17230 compatibility with other compilers on those systems:
17235 Create a shared object.
17236 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
17240 Identify the versions of each tool used by the compiler, in a
17241 @code{.ident} assembler directive in the output.
17245 Refrain from adding @code{.ident} directives to the output file (this is
17248 @item -YP,@var{dirs}
17250 Search the directories @var{dirs}, and no others, for libraries
17251 specified with @option{-l}.
17253 @item -Ym,@var{dir}
17255 Look in the directory @var{dir} to find the M4 preprocessor.
17256 The assembler uses this option.
17257 @c This is supposed to go with a -Yd for predefined M4 macro files, but
17258 @c the generic assembler that comes with Solaris takes just -Ym.
17262 @subsection V850 Options
17263 @cindex V850 Options
17265 These @samp{-m} options are defined for V850 implementations:
17269 @itemx -mno-long-calls
17270 @opindex mlong-calls
17271 @opindex mno-long-calls
17272 Treat all calls as being far away (near). If calls are assumed to be
17273 far away, the compiler will always load the functions address up into a
17274 register, and call indirect through the pointer.
17280 Do not optimize (do optimize) basic blocks that use the same index
17281 pointer 4 or more times to copy pointer into the @code{ep} register, and
17282 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
17283 option is on by default if you optimize.
17285 @item -mno-prolog-function
17286 @itemx -mprolog-function
17287 @opindex mno-prolog-function
17288 @opindex mprolog-function
17289 Do not use (do use) external functions to save and restore registers
17290 at the prologue and epilogue of a function. The external functions
17291 are slower, but use less code space if more than one function saves
17292 the same number of registers. The @option{-mprolog-function} option
17293 is on by default if you optimize.
17297 Try to make the code as small as possible. At present, this just turns
17298 on the @option{-mep} and @option{-mprolog-function} options.
17300 @item -mtda=@var{n}
17302 Put static or global variables whose size is @var{n} bytes or less into
17303 the tiny data area that register @code{ep} points to. The tiny data
17304 area can hold up to 256 bytes in total (128 bytes for byte references).
17306 @item -msda=@var{n}
17308 Put static or global variables whose size is @var{n} bytes or less into
17309 the small data area that register @code{gp} points to. The small data
17310 area can hold up to 64 kilobytes.
17312 @item -mzda=@var{n}
17314 Put static or global variables whose size is @var{n} bytes or less into
17315 the first 32 kilobytes of memory.
17319 Specify that the target processor is the V850.
17322 @opindex mbig-switch
17323 Generate code suitable for big switch tables. Use this option only if
17324 the assembler/linker complain about out of range branches within a switch
17329 This option will cause r2 and r5 to be used in the code generated by
17330 the compiler. This setting is the default.
17332 @item -mno-app-regs
17333 @opindex mno-app-regs
17334 This option will cause r2 and r5 to be treated as fixed registers.
17338 Specify that the target processor is the V850E2V3. The preprocessor
17339 constants @samp{__v850e2v3__} will be defined if
17340 this option is used.
17344 Specify that the target processor is the V850E2. The preprocessor
17345 constants @samp{__v850e2__} will be defined if
17349 Specify that the target processor is the V850E1. The preprocessor
17350 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
17354 Specify that the target processor is the V850E@. The preprocessor
17355 constant @samp{__v850e__} will be defined if this option is used.
17357 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
17358 nor @option{-mv850e2} nor @option{-mv850e2v3}
17359 are defined then a default target processor will be chosen and the
17360 relevant @samp{__v850*__} preprocessor constant will be defined.
17362 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
17363 defined, regardless of which processor variant is the target.
17365 @item -mdisable-callt
17366 @opindex mdisable-callt
17367 This option will suppress generation of the CALLT instruction for the
17368 v850e, v850e1, v850e2 and v850e2v3 flavors of the v850 architecture. The default is
17369 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
17374 @subsection VAX Options
17375 @cindex VAX options
17377 These @samp{-m} options are defined for the VAX:
17382 Do not output certain jump instructions (@code{aobleq} and so on)
17383 that the Unix assembler for the VAX cannot handle across long
17388 Do output those jump instructions, on the assumption that you
17389 will assemble with the GNU assembler.
17393 Output code for g-format floating point numbers instead of d-format.
17396 @node VxWorks Options
17397 @subsection VxWorks Options
17398 @cindex VxWorks Options
17400 The options in this section are defined for all VxWorks targets.
17401 Options specific to the target hardware are listed with the other
17402 options for that target.
17407 GCC can generate code for both VxWorks kernels and real time processes
17408 (RTPs). This option switches from the former to the latter. It also
17409 defines the preprocessor macro @code{__RTP__}.
17412 @opindex non-static
17413 Link an RTP executable against shared libraries rather than static
17414 libraries. The options @option{-static} and @option{-shared} can
17415 also be used for RTPs (@pxref{Link Options}); @option{-static}
17422 These options are passed down to the linker. They are defined for
17423 compatibility with Diab.
17426 @opindex Xbind-lazy
17427 Enable lazy binding of function calls. This option is equivalent to
17428 @option{-Wl,-z,now} and is defined for compatibility with Diab.
17432 Disable lazy binding of function calls. This option is the default and
17433 is defined for compatibility with Diab.
17436 @node x86-64 Options
17437 @subsection x86-64 Options
17438 @cindex x86-64 options
17440 These are listed under @xref{i386 and x86-64 Options}.
17442 @node i386 and x86-64 Windows Options
17443 @subsection i386 and x86-64 Windows Options
17444 @cindex i386 and x86-64 Windows Options
17446 These additional options are available for Windows targets:
17451 This option is available for Cygwin and MinGW targets. It
17452 specifies that a console application is to be generated, by
17453 instructing the linker to set the PE header subsystem type
17454 required for console applications.
17455 This is the default behavior for Cygwin and MinGW targets.
17459 This option is available for Cygwin and MinGW targets. It
17460 specifies that a DLL - a dynamic link library - is to be
17461 generated, enabling the selection of the required runtime
17462 startup object and entry point.
17464 @item -mnop-fun-dllimport
17465 @opindex mnop-fun-dllimport
17466 This option is available for Cygwin and MinGW targets. It
17467 specifies that the dllimport attribute should be ignored.
17471 This option is available for MinGW targets. It specifies
17472 that MinGW-specific thread support is to be used.
17476 This option is available for mingw-w64 targets. It specifies
17477 that the UNICODE macro is getting pre-defined and that the
17478 unicode capable runtime startup code is chosen.
17482 This option is available for Cygwin and MinGW targets. It
17483 specifies that the typical Windows pre-defined macros are to
17484 be set in the pre-processor, but does not influence the choice
17485 of runtime library/startup code.
17489 This option is available for Cygwin and MinGW targets. It
17490 specifies that a GUI application is to be generated by
17491 instructing the linker to set the PE header subsystem type
17494 @item -fno-set-stack-executable
17495 @opindex fno-set-stack-executable
17496 This option is available for MinGW targets. It specifies that
17497 the executable flag for stack used by nested functions isn't
17498 set. This is necessary for binaries running in kernel mode of
17499 Windows, as there the user32 API, which is used to set executable
17500 privileges, isn't available.
17502 @item -mpe-aligned-commons
17503 @opindex mpe-aligned-commons
17504 This option is available for Cygwin and MinGW targets. It
17505 specifies that the GNU extension to the PE file format that
17506 permits the correct alignment of COMMON variables should be
17507 used when generating code. It will be enabled by default if
17508 GCC detects that the target assembler found during configuration
17509 supports the feature.
17512 See also under @ref{i386 and x86-64 Options} for standard options.
17514 @node Xstormy16 Options
17515 @subsection Xstormy16 Options
17516 @cindex Xstormy16 Options
17518 These options are defined for Xstormy16:
17523 Choose startup files and linker script suitable for the simulator.
17526 @node Xtensa Options
17527 @subsection Xtensa Options
17528 @cindex Xtensa Options
17530 These options are supported for Xtensa targets:
17534 @itemx -mno-const16
17536 @opindex mno-const16
17537 Enable or disable use of @code{CONST16} instructions for loading
17538 constant values. The @code{CONST16} instruction is currently not a
17539 standard option from Tensilica. When enabled, @code{CONST16}
17540 instructions are always used in place of the standard @code{L32R}
17541 instructions. The use of @code{CONST16} is enabled by default only if
17542 the @code{L32R} instruction is not available.
17545 @itemx -mno-fused-madd
17546 @opindex mfused-madd
17547 @opindex mno-fused-madd
17548 Enable or disable use of fused multiply/add and multiply/subtract
17549 instructions in the floating-point option. This has no effect if the
17550 floating-point option is not also enabled. Disabling fused multiply/add
17551 and multiply/subtract instructions forces the compiler to use separate
17552 instructions for the multiply and add/subtract operations. This may be
17553 desirable in some cases where strict IEEE 754-compliant results are
17554 required: the fused multiply add/subtract instructions do not round the
17555 intermediate result, thereby producing results with @emph{more} bits of
17556 precision than specified by the IEEE standard. Disabling fused multiply
17557 add/subtract instructions also ensures that the program output is not
17558 sensitive to the compiler's ability to combine multiply and add/subtract
17561 @item -mserialize-volatile
17562 @itemx -mno-serialize-volatile
17563 @opindex mserialize-volatile
17564 @opindex mno-serialize-volatile
17565 When this option is enabled, GCC inserts @code{MEMW} instructions before
17566 @code{volatile} memory references to guarantee sequential consistency.
17567 The default is @option{-mserialize-volatile}. Use
17568 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
17570 @item -mforce-no-pic
17571 @opindex mforce-no-pic
17572 For targets, like GNU/Linux, where all user-mode Xtensa code must be
17573 position-independent code (PIC), this option disables PIC for compiling
17576 @item -mtext-section-literals
17577 @itemx -mno-text-section-literals
17578 @opindex mtext-section-literals
17579 @opindex mno-text-section-literals
17580 Control the treatment of literal pools. The default is
17581 @option{-mno-text-section-literals}, which places literals in a separate
17582 section in the output file. This allows the literal pool to be placed
17583 in a data RAM/ROM, and it also allows the linker to combine literal
17584 pools from separate object files to remove redundant literals and
17585 improve code size. With @option{-mtext-section-literals}, the literals
17586 are interspersed in the text section in order to keep them as close as
17587 possible to their references. This may be necessary for large assembly
17590 @item -mtarget-align
17591 @itemx -mno-target-align
17592 @opindex mtarget-align
17593 @opindex mno-target-align
17594 When this option is enabled, GCC instructs the assembler to
17595 automatically align instructions to reduce branch penalties at the
17596 expense of some code density. The assembler attempts to widen density
17597 instructions to align branch targets and the instructions following call
17598 instructions. If there are not enough preceding safe density
17599 instructions to align a target, no widening will be performed. The
17600 default is @option{-mtarget-align}. These options do not affect the
17601 treatment of auto-aligned instructions like @code{LOOP}, which the
17602 assembler will always align, either by widening density instructions or
17603 by inserting no-op instructions.
17606 @itemx -mno-longcalls
17607 @opindex mlongcalls
17608 @opindex mno-longcalls
17609 When this option is enabled, GCC instructs the assembler to translate
17610 direct calls to indirect calls unless it can determine that the target
17611 of a direct call is in the range allowed by the call instruction. This
17612 translation typically occurs for calls to functions in other source
17613 files. Specifically, the assembler translates a direct @code{CALL}
17614 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
17615 The default is @option{-mno-longcalls}. This option should be used in
17616 programs where the call target can potentially be out of range. This
17617 option is implemented in the assembler, not the compiler, so the
17618 assembly code generated by GCC will still show direct call
17619 instructions---look at the disassembled object code to see the actual
17620 instructions. Note that the assembler will use an indirect call for
17621 every cross-file call, not just those that really will be out of range.
17624 @node zSeries Options
17625 @subsection zSeries Options
17626 @cindex zSeries options
17628 These are listed under @xref{S/390 and zSeries Options}.
17630 @node Code Gen Options
17631 @section Options for Code Generation Conventions
17632 @cindex code generation conventions
17633 @cindex options, code generation
17634 @cindex run-time options
17636 These machine-independent options control the interface conventions
17637 used in code generation.
17639 Most of them have both positive and negative forms; the negative form
17640 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
17641 one of the forms is listed---the one which is not the default. You
17642 can figure out the other form by either removing @samp{no-} or adding
17646 @item -fbounds-check
17647 @opindex fbounds-check
17648 For front-ends that support it, generate additional code to check that
17649 indices used to access arrays are within the declared range. This is
17650 currently only supported by the Java and Fortran front-ends, where
17651 this option defaults to true and false respectively.
17655 This option generates traps for signed overflow on addition, subtraction,
17656 multiplication operations.
17660 This option instructs the compiler to assume that signed arithmetic
17661 overflow of addition, subtraction and multiplication wraps around
17662 using twos-complement representation. This flag enables some optimizations
17663 and disables others. This option is enabled by default for the Java
17664 front-end, as required by the Java language specification.
17667 @opindex fexceptions
17668 Enable exception handling. Generates extra code needed to propagate
17669 exceptions. For some targets, this implies GCC will generate frame
17670 unwind information for all functions, which can produce significant data
17671 size overhead, although it does not affect execution. If you do not
17672 specify this option, GCC will enable it by default for languages like
17673 C++ which normally require exception handling, and disable it for
17674 languages like C that do not normally require it. However, you may need
17675 to enable this option when compiling C code that needs to interoperate
17676 properly with exception handlers written in C++. You may also wish to
17677 disable this option if you are compiling older C++ programs that don't
17678 use exception handling.
17680 @item -fnon-call-exceptions
17681 @opindex fnon-call-exceptions
17682 Generate code that allows trapping instructions to throw exceptions.
17683 Note that this requires platform-specific runtime support that does
17684 not exist everywhere. Moreover, it only allows @emph{trapping}
17685 instructions to throw exceptions, i.e.@: memory references or floating
17686 point instructions. It does not allow exceptions to be thrown from
17687 arbitrary signal handlers such as @code{SIGALRM}.
17689 @item -funwind-tables
17690 @opindex funwind-tables
17691 Similar to @option{-fexceptions}, except that it will just generate any needed
17692 static data, but will not affect the generated code in any other way.
17693 You will normally not enable this option; instead, a language processor
17694 that needs this handling would enable it on your behalf.
17696 @item -fasynchronous-unwind-tables
17697 @opindex fasynchronous-unwind-tables
17698 Generate unwind table in dwarf2 format, if supported by target machine. The
17699 table is exact at each instruction boundary, so it can be used for stack
17700 unwinding from asynchronous events (such as debugger or garbage collector).
17702 @item -fpcc-struct-return
17703 @opindex fpcc-struct-return
17704 Return ``short'' @code{struct} and @code{union} values in memory like
17705 longer ones, rather than in registers. This convention is less
17706 efficient, but it has the advantage of allowing intercallability between
17707 GCC-compiled files and files compiled with other compilers, particularly
17708 the Portable C Compiler (pcc).
17710 The precise convention for returning structures in memory depends
17711 on the target configuration macros.
17713 Short structures and unions are those whose size and alignment match
17714 that of some integer type.
17716 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
17717 switch is not binary compatible with code compiled with the
17718 @option{-freg-struct-return} switch.
17719 Use it to conform to a non-default application binary interface.
17721 @item -freg-struct-return
17722 @opindex freg-struct-return
17723 Return @code{struct} and @code{union} values in registers when possible.
17724 This is more efficient for small structures than
17725 @option{-fpcc-struct-return}.
17727 If you specify neither @option{-fpcc-struct-return} nor
17728 @option{-freg-struct-return}, GCC defaults to whichever convention is
17729 standard for the target. If there is no standard convention, GCC
17730 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
17731 the principal compiler. In those cases, we can choose the standard, and
17732 we chose the more efficient register return alternative.
17734 @strong{Warning:} code compiled with the @option{-freg-struct-return}
17735 switch is not binary compatible with code compiled with the
17736 @option{-fpcc-struct-return} switch.
17737 Use it to conform to a non-default application binary interface.
17739 @item -fshort-enums
17740 @opindex fshort-enums
17741 Allocate to an @code{enum} type only as many bytes as it needs for the
17742 declared range of possible values. Specifically, the @code{enum} type
17743 will be equivalent to the smallest integer type which has enough room.
17745 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
17746 code that is not binary compatible with code generated without that switch.
17747 Use it to conform to a non-default application binary interface.
17749 @item -fshort-double
17750 @opindex fshort-double
17751 Use the same size for @code{double} as for @code{float}.
17753 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
17754 code that is not binary compatible with code generated without that switch.
17755 Use it to conform to a non-default application binary interface.
17757 @item -fshort-wchar
17758 @opindex fshort-wchar
17759 Override the underlying type for @samp{wchar_t} to be @samp{short
17760 unsigned int} instead of the default for the target. This option is
17761 useful for building programs to run under WINE@.
17763 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
17764 code that is not binary compatible with code generated without that switch.
17765 Use it to conform to a non-default application binary interface.
17768 @opindex fno-common
17769 In C code, controls the placement of uninitialized global variables.
17770 Unix C compilers have traditionally permitted multiple definitions of
17771 such variables in different compilation units by placing the variables
17773 This is the behavior specified by @option{-fcommon}, and is the default
17774 for GCC on most targets.
17775 On the other hand, this behavior is not required by ISO C, and on some
17776 targets may carry a speed or code size penalty on variable references.
17777 The @option{-fno-common} option specifies that the compiler should place
17778 uninitialized global variables in the data section of the object file,
17779 rather than generating them as common blocks.
17780 This has the effect that if the same variable is declared
17781 (without @code{extern}) in two different compilations,
17782 you will get a multiple-definition error when you link them.
17783 In this case, you must compile with @option{-fcommon} instead.
17784 Compiling with @option{-fno-common} is useful on targets for which
17785 it provides better performance, or if you wish to verify that the
17786 program will work on other systems which always treat uninitialized
17787 variable declarations this way.
17791 Ignore the @samp{#ident} directive.
17793 @item -finhibit-size-directive
17794 @opindex finhibit-size-directive
17795 Don't output a @code{.size} assembler directive, or anything else that
17796 would cause trouble if the function is split in the middle, and the
17797 two halves are placed at locations far apart in memory. This option is
17798 used when compiling @file{crtstuff.c}; you should not need to use it
17801 @item -fverbose-asm
17802 @opindex fverbose-asm
17803 Put extra commentary information in the generated assembly code to
17804 make it more readable. This option is generally only of use to those
17805 who actually need to read the generated assembly code (perhaps while
17806 debugging the compiler itself).
17808 @option{-fno-verbose-asm}, the default, causes the
17809 extra information to be omitted and is useful when comparing two assembler
17812 @item -frecord-gcc-switches
17813 @opindex frecord-gcc-switches
17814 This switch causes the command line that was used to invoke the
17815 compiler to be recorded into the object file that is being created.
17816 This switch is only implemented on some targets and the exact format
17817 of the recording is target and binary file format dependent, but it
17818 usually takes the form of a section containing ASCII text. This
17819 switch is related to the @option{-fverbose-asm} switch, but that
17820 switch only records information in the assembler output file as
17821 comments, so it never reaches the object file.
17825 @cindex global offset table
17827 Generate position-independent code (PIC) suitable for use in a shared
17828 library, if supported for the target machine. Such code accesses all
17829 constant addresses through a global offset table (GOT)@. The dynamic
17830 loader resolves the GOT entries when the program starts (the dynamic
17831 loader is not part of GCC; it is part of the operating system). If
17832 the GOT size for the linked executable exceeds a machine-specific
17833 maximum size, you get an error message from the linker indicating that
17834 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17835 instead. (These maximums are 8k on the SPARC and 32k
17836 on the m68k and RS/6000. The 386 has no such limit.)
17838 Position-independent code requires special support, and therefore works
17839 only on certain machines. For the 386, GCC supports PIC for System V
17840 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17841 position-independent.
17843 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17848 If supported for the target machine, emit position-independent code,
17849 suitable for dynamic linking and avoiding any limit on the size of the
17850 global offset table. This option makes a difference on the m68k,
17851 PowerPC and SPARC@.
17853 Position-independent code requires special support, and therefore works
17854 only on certain machines.
17856 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17863 These options are similar to @option{-fpic} and @option{-fPIC}, but
17864 generated position independent code can be only linked into executables.
17865 Usually these options are used when @option{-pie} GCC option will be
17866 used during linking.
17868 @option{-fpie} and @option{-fPIE} both define the macros
17869 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17870 for @option{-fpie} and 2 for @option{-fPIE}.
17872 @item -fno-jump-tables
17873 @opindex fno-jump-tables
17874 Do not use jump tables for switch statements even where it would be
17875 more efficient than other code generation strategies. This option is
17876 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17877 building code which forms part of a dynamic linker and cannot
17878 reference the address of a jump table. On some targets, jump tables
17879 do not require a GOT and this option is not needed.
17881 @item -ffixed-@var{reg}
17883 Treat the register named @var{reg} as a fixed register; generated code
17884 should never refer to it (except perhaps as a stack pointer, frame
17885 pointer or in some other fixed role).
17887 @var{reg} must be the name of a register. The register names accepted
17888 are machine-specific and are defined in the @code{REGISTER_NAMES}
17889 macro in the machine description macro file.
17891 This flag does not have a negative form, because it specifies a
17894 @item -fcall-used-@var{reg}
17895 @opindex fcall-used
17896 Treat the register named @var{reg} as an allocable register that is
17897 clobbered by function calls. It may be allocated for temporaries or
17898 variables that do not live across a call. Functions compiled this way
17899 will not save and restore the register @var{reg}.
17901 It is an error to used this flag with the frame pointer or stack pointer.
17902 Use of this flag for other registers that have fixed pervasive roles in
17903 the machine's execution model will produce disastrous results.
17905 This flag does not have a negative form, because it specifies a
17908 @item -fcall-saved-@var{reg}
17909 @opindex fcall-saved
17910 Treat the register named @var{reg} as an allocable register saved by
17911 functions. It may be allocated even for temporaries or variables that
17912 live across a call. Functions compiled this way will save and restore
17913 the register @var{reg} if they use it.
17915 It is an error to used this flag with the frame pointer or stack pointer.
17916 Use of this flag for other registers that have fixed pervasive roles in
17917 the machine's execution model will produce disastrous results.
17919 A different sort of disaster will result from the use of this flag for
17920 a register in which function values may be returned.
17922 This flag does not have a negative form, because it specifies a
17925 @item -fpack-struct[=@var{n}]
17926 @opindex fpack-struct
17927 Without a value specified, pack all structure members together without
17928 holes. When a value is specified (which must be a small power of two), pack
17929 structure members according to this value, representing the maximum
17930 alignment (that is, objects with default alignment requirements larger than
17931 this will be output potentially unaligned at the next fitting location.
17933 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17934 code that is not binary compatible with code generated without that switch.
17935 Additionally, it makes the code suboptimal.
17936 Use it to conform to a non-default application binary interface.
17938 @item -finstrument-functions
17939 @opindex finstrument-functions
17940 Generate instrumentation calls for entry and exit to functions. Just
17941 after function entry and just before function exit, the following
17942 profiling functions will be called with the address of the current
17943 function and its call site. (On some platforms,
17944 @code{__builtin_return_address} does not work beyond the current
17945 function, so the call site information may not be available to the
17946 profiling functions otherwise.)
17949 void __cyg_profile_func_enter (void *this_fn,
17951 void __cyg_profile_func_exit (void *this_fn,
17955 The first argument is the address of the start of the current function,
17956 which may be looked up exactly in the symbol table.
17958 This instrumentation is also done for functions expanded inline in other
17959 functions. The profiling calls will indicate where, conceptually, the
17960 inline function is entered and exited. This means that addressable
17961 versions of such functions must be available. If all your uses of a
17962 function are expanded inline, this may mean an additional expansion of
17963 code size. If you use @samp{extern inline} in your C code, an
17964 addressable version of such functions must be provided. (This is
17965 normally the case anyways, but if you get lucky and the optimizer always
17966 expands the functions inline, you might have gotten away without
17967 providing static copies.)
17969 A function may be given the attribute @code{no_instrument_function}, in
17970 which case this instrumentation will not be done. This can be used, for
17971 example, for the profiling functions listed above, high-priority
17972 interrupt routines, and any functions from which the profiling functions
17973 cannot safely be called (perhaps signal handlers, if the profiling
17974 routines generate output or allocate memory).
17976 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17977 @opindex finstrument-functions-exclude-file-list
17979 Set the list of functions that are excluded from instrumentation (see
17980 the description of @code{-finstrument-functions}). If the file that
17981 contains a function definition matches with one of @var{file}, then
17982 that function is not instrumented. The match is done on substrings:
17983 if the @var{file} parameter is a substring of the file name, it is
17984 considered to be a match.
17989 -finstrument-functions-exclude-file-list=/bits/stl,include/sys
17993 will exclude any inline function defined in files whose pathnames
17994 contain @code{/bits/stl} or @code{include/sys}.
17996 If, for some reason, you want to include letter @code{','} in one of
17997 @var{sym}, write @code{'\,'}. For example,
17998 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17999 (note the single quote surrounding the option).
18001 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
18002 @opindex finstrument-functions-exclude-function-list
18004 This is similar to @code{-finstrument-functions-exclude-file-list},
18005 but this option sets the list of function names to be excluded from
18006 instrumentation. The function name to be matched is its user-visible
18007 name, such as @code{vector<int> blah(const vector<int> &)}, not the
18008 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
18009 match is done on substrings: if the @var{sym} parameter is a substring
18010 of the function name, it is considered to be a match. For C99 and C++
18011 extended identifiers, the function name must be given in UTF-8, not
18012 using universal character names.
18014 @item -fstack-check
18015 @opindex fstack-check
18016 Generate code to verify that you do not go beyond the boundary of the
18017 stack. You should specify this flag if you are running in an
18018 environment with multiple threads, but only rarely need to specify it in
18019 a single-threaded environment since stack overflow is automatically
18020 detected on nearly all systems if there is only one stack.
18022 Note that this switch does not actually cause checking to be done; the
18023 operating system or the language runtime must do that. The switch causes
18024 generation of code to ensure that they see the stack being extended.
18026 You can additionally specify a string parameter: @code{no} means no
18027 checking, @code{generic} means force the use of old-style checking,
18028 @code{specific} means use the best checking method and is equivalent
18029 to bare @option{-fstack-check}.
18031 Old-style checking is a generic mechanism that requires no specific
18032 target support in the compiler but comes with the following drawbacks:
18036 Modified allocation strategy for large objects: they will always be
18037 allocated dynamically if their size exceeds a fixed threshold.
18040 Fixed limit on the size of the static frame of functions: when it is
18041 topped by a particular function, stack checking is not reliable and
18042 a warning is issued by the compiler.
18045 Inefficiency: because of both the modified allocation strategy and the
18046 generic implementation, the performances of the code are hampered.
18049 Note that old-style stack checking is also the fallback method for
18050 @code{specific} if no target support has been added in the compiler.
18052 @item -fstack-limit-register=@var{reg}
18053 @itemx -fstack-limit-symbol=@var{sym}
18054 @itemx -fno-stack-limit
18055 @opindex fstack-limit-register
18056 @opindex fstack-limit-symbol
18057 @opindex fno-stack-limit
18058 Generate code to ensure that the stack does not grow beyond a certain value,
18059 either the value of a register or the address of a symbol. If the stack
18060 would grow beyond the value, a signal is raised. For most targets,
18061 the signal is raised before the stack overruns the boundary, so
18062 it is possible to catch the signal without taking special precautions.
18064 For instance, if the stack starts at absolute address @samp{0x80000000}
18065 and grows downwards, you can use the flags
18066 @option{-fstack-limit-symbol=__stack_limit} and
18067 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
18068 of 128KB@. Note that this may only work with the GNU linker.
18070 @item -fsplit-stack
18071 @opindex fsplit-stack
18072 Generate code to automatically split the stack before it overflows.
18073 The resulting program has a discontiguous stack which can only
18074 overflow if the program is unable to allocate any more memory. This
18075 is most useful when running threaded programs, as it is no longer
18076 necessary to calculate a good stack size to use for each thread. This
18077 is currently only implemented for the i386 and x86_64 backends running
18080 When code compiled with @option{-fsplit-stack} calls code compiled
18081 without @option{-fsplit-stack}, there may not be much stack space
18082 available for the latter code to run. If compiling all code,
18083 including library code, with @option{-fsplit-stack} is not an option,
18084 then the linker can fix up these calls so that the code compiled
18085 without @option{-fsplit-stack} always has a large stack. Support for
18086 this is implemented in the gold linker in GNU binutils release 2.21
18089 @item -fleading-underscore
18090 @opindex fleading-underscore
18091 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
18092 change the way C symbols are represented in the object file. One use
18093 is to help link with legacy assembly code.
18095 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
18096 generate code that is not binary compatible with code generated without that
18097 switch. Use it to conform to a non-default application binary interface.
18098 Not all targets provide complete support for this switch.
18100 @item -ftls-model=@var{model}
18101 @opindex ftls-model
18102 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
18103 The @var{model} argument should be one of @code{global-dynamic},
18104 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
18106 The default without @option{-fpic} is @code{initial-exec}; with
18107 @option{-fpic} the default is @code{global-dynamic}.
18109 @item -fvisibility=@var{default|internal|hidden|protected}
18110 @opindex fvisibility
18111 Set the default ELF image symbol visibility to the specified option---all
18112 symbols will be marked with this unless overridden within the code.
18113 Using this feature can very substantially improve linking and
18114 load times of shared object libraries, produce more optimized
18115 code, provide near-perfect API export and prevent symbol clashes.
18116 It is @strong{strongly} recommended that you use this in any shared objects
18119 Despite the nomenclature, @code{default} always means public ie;
18120 available to be linked against from outside the shared object.
18121 @code{protected} and @code{internal} are pretty useless in real-world
18122 usage so the only other commonly used option will be @code{hidden}.
18123 The default if @option{-fvisibility} isn't specified is
18124 @code{default}, i.e., make every
18125 symbol public---this causes the same behavior as previous versions of
18128 A good explanation of the benefits offered by ensuring ELF
18129 symbols have the correct visibility is given by ``How To Write
18130 Shared Libraries'' by Ulrich Drepper (which can be found at
18131 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
18132 solution made possible by this option to marking things hidden when
18133 the default is public is to make the default hidden and mark things
18134 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
18135 and @code{__attribute__ ((visibility("default")))} instead of
18136 @code{__declspec(dllexport)} you get almost identical semantics with
18137 identical syntax. This is a great boon to those working with
18138 cross-platform projects.
18140 For those adding visibility support to existing code, you may find
18141 @samp{#pragma GCC visibility} of use. This works by you enclosing
18142 the declarations you wish to set visibility for with (for example)
18143 @samp{#pragma GCC visibility push(hidden)} and
18144 @samp{#pragma GCC visibility pop}.
18145 Bear in mind that symbol visibility should be viewed @strong{as
18146 part of the API interface contract} and thus all new code should
18147 always specify visibility when it is not the default ie; declarations
18148 only for use within the local DSO should @strong{always} be marked explicitly
18149 as hidden as so to avoid PLT indirection overheads---making this
18150 abundantly clear also aids readability and self-documentation of the code.
18151 Note that due to ISO C++ specification requirements, operator new and
18152 operator delete must always be of default visibility.
18154 Be aware that headers from outside your project, in particular system
18155 headers and headers from any other library you use, may not be
18156 expecting to be compiled with visibility other than the default. You
18157 may need to explicitly say @samp{#pragma GCC visibility push(default)}
18158 before including any such headers.
18160 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
18161 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
18162 no modifications. However, this means that calls to @samp{extern}
18163 functions with no explicit visibility will use the PLT, so it is more
18164 effective to use @samp{__attribute ((visibility))} and/or
18165 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
18166 declarations should be treated as hidden.
18168 Note that @samp{-fvisibility} does affect C++ vague linkage
18169 entities. This means that, for instance, an exception class that will
18170 be thrown between DSOs must be explicitly marked with default
18171 visibility so that the @samp{type_info} nodes will be unified between
18174 An overview of these techniques, their benefits and how to use them
18175 is at @uref{http://gcc.gnu.org/@/wiki/@/Visibility}.
18177 @item -fstrict-volatile-bitfields
18178 @opindex fstrict-volatile-bitfields
18179 This option should be used if accesses to volatile bitfields (or other
18180 structure fields, although the compiler usually honors those types
18181 anyway) should use a single access of the width of the
18182 field's type, aligned to a natural alignment if possible. For
18183 example, targets with memory-mapped peripheral registers might require
18184 all such accesses to be 16 bits wide; with this flag the user could
18185 declare all peripheral bitfields as ``unsigned short'' (assuming short
18186 is 16 bits on these targets) to force GCC to use 16 bit accesses
18187 instead of, perhaps, a more efficient 32 bit access.
18189 If this option is disabled, the compiler will use the most efficient
18190 instruction. In the previous example, that might be a 32-bit load
18191 instruction, even though that will access bytes that do not contain
18192 any portion of the bitfield, or memory-mapped registers unrelated to
18193 the one being updated.
18195 If the target requires strict alignment, and honoring the field
18196 type would require violating this alignment, a warning is issued.
18197 If the field has @code{packed} attribute, the access is done without
18198 honoring the field type. If the field doesn't have @code{packed}
18199 attribute, the access is done honoring the field type. In both cases,
18200 GCC assumes that the user knows something about the target hardware
18201 that it is unaware of.
18203 The default value of this option is determined by the application binary
18204 interface for the target processor.
18210 @node Environment Variables
18211 @section Environment Variables Affecting GCC
18212 @cindex environment variables
18214 @c man begin ENVIRONMENT
18215 This section describes several environment variables that affect how GCC
18216 operates. Some of them work by specifying directories or prefixes to use
18217 when searching for various kinds of files. Some are used to specify other
18218 aspects of the compilation environment.
18220 Note that you can also specify places to search using options such as
18221 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
18222 take precedence over places specified using environment variables, which
18223 in turn take precedence over those specified by the configuration of GCC@.
18224 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
18225 GNU Compiler Collection (GCC) Internals}.
18230 @c @itemx LC_COLLATE
18232 @c @itemx LC_MONETARY
18233 @c @itemx LC_NUMERIC
18238 @c @findex LC_COLLATE
18239 @findex LC_MESSAGES
18240 @c @findex LC_MONETARY
18241 @c @findex LC_NUMERIC
18245 These environment variables control the way that GCC uses
18246 localization information that allow GCC to work with different
18247 national conventions. GCC inspects the locale categories
18248 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
18249 so. These locale categories can be set to any value supported by your
18250 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
18251 Kingdom encoded in UTF-8.
18253 The @env{LC_CTYPE} environment variable specifies character
18254 classification. GCC uses it to determine the character boundaries in
18255 a string; this is needed for some multibyte encodings that contain quote
18256 and escape characters that would otherwise be interpreted as a string
18259 The @env{LC_MESSAGES} environment variable specifies the language to
18260 use in diagnostic messages.
18262 If the @env{LC_ALL} environment variable is set, it overrides the value
18263 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
18264 and @env{LC_MESSAGES} default to the value of the @env{LANG}
18265 environment variable. If none of these variables are set, GCC
18266 defaults to traditional C English behavior.
18270 If @env{TMPDIR} is set, it specifies the directory to use for temporary
18271 files. GCC uses temporary files to hold the output of one stage of
18272 compilation which is to be used as input to the next stage: for example,
18273 the output of the preprocessor, which is the input to the compiler
18276 @item GCC_EXEC_PREFIX
18277 @findex GCC_EXEC_PREFIX
18278 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
18279 names of the subprograms executed by the compiler. No slash is added
18280 when this prefix is combined with the name of a subprogram, but you can
18281 specify a prefix that ends with a slash if you wish.
18283 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
18284 an appropriate prefix to use based on the pathname it was invoked with.
18286 If GCC cannot find the subprogram using the specified prefix, it
18287 tries looking in the usual places for the subprogram.
18289 The default value of @env{GCC_EXEC_PREFIX} is
18290 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
18291 the installed compiler. In many cases @var{prefix} is the value
18292 of @code{prefix} when you ran the @file{configure} script.
18294 Other prefixes specified with @option{-B} take precedence over this prefix.
18296 This prefix is also used for finding files such as @file{crt0.o} that are
18299 In addition, the prefix is used in an unusual way in finding the
18300 directories to search for header files. For each of the standard
18301 directories whose name normally begins with @samp{/usr/local/lib/gcc}
18302 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
18303 replacing that beginning with the specified prefix to produce an
18304 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
18305 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
18306 These alternate directories are searched first; the standard directories
18307 come next. If a standard directory begins with the configured
18308 @var{prefix} then the value of @var{prefix} is replaced by
18309 @env{GCC_EXEC_PREFIX} when looking for header files.
18311 @item COMPILER_PATH
18312 @findex COMPILER_PATH
18313 The value of @env{COMPILER_PATH} is a colon-separated list of
18314 directories, much like @env{PATH}. GCC tries the directories thus
18315 specified when searching for subprograms, if it can't find the
18316 subprograms using @env{GCC_EXEC_PREFIX}.
18319 @findex LIBRARY_PATH
18320 The value of @env{LIBRARY_PATH} is a colon-separated list of
18321 directories, much like @env{PATH}. When configured as a native compiler,
18322 GCC tries the directories thus specified when searching for special
18323 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
18324 using GCC also uses these directories when searching for ordinary
18325 libraries for the @option{-l} option (but directories specified with
18326 @option{-L} come first).
18330 @cindex locale definition
18331 This variable is used to pass locale information to the compiler. One way in
18332 which this information is used is to determine the character set to be used
18333 when character literals, string literals and comments are parsed in C and C++.
18334 When the compiler is configured to allow multibyte characters,
18335 the following values for @env{LANG} are recognized:
18339 Recognize JIS characters.
18341 Recognize SJIS characters.
18343 Recognize EUCJP characters.
18346 If @env{LANG} is not defined, or if it has some other value, then the
18347 compiler will use mblen and mbtowc as defined by the default locale to
18348 recognize and translate multibyte characters.
18352 Some additional environments variables affect the behavior of the
18355 @include cppenv.texi
18359 @node Precompiled Headers
18360 @section Using Precompiled Headers
18361 @cindex precompiled headers
18362 @cindex speed of compilation
18364 Often large projects have many header files that are included in every
18365 source file. The time the compiler takes to process these header files
18366 over and over again can account for nearly all of the time required to
18367 build the project. To make builds faster, GCC allows users to
18368 `precompile' a header file; then, if builds can use the precompiled
18369 header file they will be much faster.
18371 To create a precompiled header file, simply compile it as you would any
18372 other file, if necessary using the @option{-x} option to make the driver
18373 treat it as a C or C++ header file. You will probably want to use a
18374 tool like @command{make} to keep the precompiled header up-to-date when
18375 the headers it contains change.
18377 A precompiled header file will be searched for when @code{#include} is
18378 seen in the compilation. As it searches for the included file
18379 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
18380 compiler looks for a precompiled header in each directory just before it
18381 looks for the include file in that directory. The name searched for is
18382 the name specified in the @code{#include} with @samp{.gch} appended. If
18383 the precompiled header file can't be used, it is ignored.
18385 For instance, if you have @code{#include "all.h"}, and you have
18386 @file{all.h.gch} in the same directory as @file{all.h}, then the
18387 precompiled header file will be used if possible, and the original
18388 header will be used otherwise.
18390 Alternatively, you might decide to put the precompiled header file in a
18391 directory and use @option{-I} to ensure that directory is searched
18392 before (or instead of) the directory containing the original header.
18393 Then, if you want to check that the precompiled header file is always
18394 used, you can put a file of the same name as the original header in this
18395 directory containing an @code{#error} command.
18397 This also works with @option{-include}. So yet another way to use
18398 precompiled headers, good for projects not designed with precompiled
18399 header files in mind, is to simply take most of the header files used by
18400 a project, include them from another header file, precompile that header
18401 file, and @option{-include} the precompiled header. If the header files
18402 have guards against multiple inclusion, they will be skipped because
18403 they've already been included (in the precompiled header).
18405 If you need to precompile the same header file for different
18406 languages, targets, or compiler options, you can instead make a
18407 @emph{directory} named like @file{all.h.gch}, and put each precompiled
18408 header in the directory, perhaps using @option{-o}. It doesn't matter
18409 what you call the files in the directory, every precompiled header in
18410 the directory will be considered. The first precompiled header
18411 encountered in the directory that is valid for this compilation will
18412 be used; they're searched in no particular order.
18414 There are many other possibilities, limited only by your imagination,
18415 good sense, and the constraints of your build system.
18417 A precompiled header file can be used only when these conditions apply:
18421 Only one precompiled header can be used in a particular compilation.
18424 A precompiled header can't be used once the first C token is seen. You
18425 can have preprocessor directives before a precompiled header; you can
18426 even include a precompiled header from inside another header, so long as
18427 there are no C tokens before the @code{#include}.
18430 The precompiled header file must be produced for the same language as
18431 the current compilation. You can't use a C precompiled header for a C++
18435 The precompiled header file must have been produced by the same compiler
18436 binary as the current compilation is using.
18439 Any macros defined before the precompiled header is included must
18440 either be defined in the same way as when the precompiled header was
18441 generated, or must not affect the precompiled header, which usually
18442 means that they don't appear in the precompiled header at all.
18444 The @option{-D} option is one way to define a macro before a
18445 precompiled header is included; using a @code{#define} can also do it.
18446 There are also some options that define macros implicitly, like
18447 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
18450 @item If debugging information is output when using the precompiled
18451 header, using @option{-g} or similar, the same kind of debugging information
18452 must have been output when building the precompiled header. However,
18453 a precompiled header built using @option{-g} can be used in a compilation
18454 when no debugging information is being output.
18456 @item The same @option{-m} options must generally be used when building
18457 and using the precompiled header. @xref{Submodel Options},
18458 for any cases where this rule is relaxed.
18460 @item Each of the following options must be the same when building and using
18461 the precompiled header:
18463 @gccoptlist{-fexceptions}
18466 Some other command-line options starting with @option{-f},
18467 @option{-p}, or @option{-O} must be defined in the same way as when
18468 the precompiled header was generated. At present, it's not clear
18469 which options are safe to change and which are not; the safest choice
18470 is to use exactly the same options when generating and using the
18471 precompiled header. The following are known to be safe:
18473 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
18474 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
18475 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
18480 For all of these except the last, the compiler will automatically
18481 ignore the precompiled header if the conditions aren't met. If you
18482 find an option combination that doesn't work and doesn't cause the
18483 precompiled header to be ignored, please consider filing a bug report,
18486 If you do use differing options when generating and using the
18487 precompiled header, the actual behavior will be a mixture of the
18488 behavior for the options. For instance, if you use @option{-g} to
18489 generate the precompiled header but not when using it, you may or may
18490 not get debugging information for routines in the precompiled header.