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 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
156 @section Option Summary
158 Here is a summary of all the options, grouped by type. Explanations are
159 in the following sections.
162 @item Overall Options
163 @xref{Overall Options,,Options Controlling the Kind of Output}.
164 @gccoptlist{-c -S -E -o @var{file} -combine -no-canonical-prefixes @gol
165 -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
167 --version -wrapper@@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -fno-pretty-templates @gol
193 -frepo -fno-rtti -fstats -ftemplate-depth=@var{n} @gol
194 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
195 -fno-default-inline -fvisibility-inlines-hidden @gol
196 -fvisibility-ms-compat @gol
197 -Wabi -Wconversion-null -Wctor-dtor-privacy @gol
198 -Wnon-virtual-dtor -Wreorder @gol
199 -Weffc++ -Wstrict-null-sentinel @gol
200 -Wno-non-template-friend -Wold-style-cast @gol
201 -Woverloaded-virtual -Wno-pmf-conversions @gol
204 @item Objective-C and Objective-C++ Language Options
205 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
206 Objective-C and Objective-C++ Dialects}.
207 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
208 -fgnu-runtime -fnext-runtime @gol
209 -fno-nil-receivers @gol
210 -fobjc-call-cxx-cdtors @gol
211 -fobjc-direct-dispatch @gol
212 -fobjc-exceptions @gol
214 -freplace-objc-classes @gol
217 -Wassign-intercept @gol
218 -Wno-protocol -Wselector @gol
219 -Wstrict-selector-match @gol
220 -Wundeclared-selector}
222 @item Language Independent Options
223 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
224 @gccoptlist{-fmessage-length=@var{n} @gol
225 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
226 -fdiagnostics-show-option}
228 @item Warning Options
229 @xref{Warning Options,,Options to Request or Suppress Warnings}.
230 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
231 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
232 -Wno-attributes -Wno-builtin-macro-redefined @gol
233 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
234 -Wchar-subscripts -Wclobbered -Wcomment @gol
235 -Wconversion -Wcoverage-mismatch -Wno-deprecated @gol
236 -Wno-deprecated-declarations -Wdisabled-optimization @gol
237 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
238 -Werror -Werror=* @gol
239 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
240 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
241 -Wformat-security -Wformat-y2k @gol
242 -Wframe-larger-than=@var{len} -Wjump-misses-init -Wignored-qualifiers @gol
243 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
244 -Winit-self -Winline @gol
245 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
246 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
247 -Wlogical-op -Wlong-long @gol
248 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
249 -Wmissing-format-attribute -Wmissing-include-dirs @gol
250 -Wmissing-noreturn -Wno-mudflap @gol
251 -Wno-multichar -Wnonnull -Wno-overflow @gol
252 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
253 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
254 -Wpointer-arith -Wno-pointer-to-int-cast @gol
255 -Wredundant-decls @gol
256 -Wreturn-type -Wsequence-point -Wshadow @gol
257 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
258 -Wstrict-aliasing -Wstrict-aliasing=n @gol
259 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
260 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
261 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
262 -Wunknown-pragmas -Wno-pragmas @gol
263 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
264 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value -Wunused-variable @gol
265 -Wvariadic-macros -Wvla @gol
266 -Wvolatile-register-var -Wwrite-strings}
268 @item C and Objective-C-only Warning Options
269 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
270 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
271 -Wold-style-declaration -Wold-style-definition @gol
272 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
273 -Wdeclaration-after-statement -Wpointer-sign}
275 @item Debugging Options
276 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
277 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
278 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
279 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
280 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
281 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
282 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
283 -fdump-statistics @gol
285 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
289 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-nrv -fdump-tree-vect @gol
299 -fdump-tree-sink @gol
300 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
301 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
302 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
303 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
304 -ftree-vectorizer-verbose=@var{n} @gol
305 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
306 -fdump-final-insns=@var{file} @gol
307 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
308 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
309 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
310 -fenable-icf-debug @gol
311 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
312 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
313 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
314 -ftest-coverage -ftime-report -fvar-tracking @gol
315 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
316 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
317 -ggdb -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
318 -gvms -gxcoff -gxcoff+ @gol
319 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
320 -fdebug-prefix-map=@var{old}=@var{new} @gol
321 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
322 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
323 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
324 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
325 -print-prog-name=@var{program} -print-search-dirs -Q @gol
326 -print-sysroot -print-sysroot-headers-suffix @gol
327 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
329 @item Optimization Options
330 @xref{Optimize Options,,Options that Control Optimization}.
332 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
333 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
334 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
335 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
336 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
337 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
338 -fdata-sections -fdce -fdce @gol
339 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
340 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffast-math @gol
341 -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
342 -fforward-propagate -ffunction-sections @gol
343 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
344 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
345 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
346 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
347 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
348 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
349 -fira-region=@var{region} -fira-coalesce @gol
350 -fira-loop-pressure -fno-ira-share-save-slots @gol
351 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
352 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
353 -floop-block -floop-interchange -floop-strip-mine -fgraphite-identity @gol
354 -floop-parallelize-all -flto -flto-compression-level -flto-report -fltrans @gol
355 -fltrans-output-list -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
356 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
357 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
358 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
359 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
360 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
361 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
362 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
363 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
364 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
365 -fprofile-generate=@var{path} @gol
366 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
367 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
368 -freorder-blocks-and-partition -freorder-functions @gol
369 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
370 -frounding-math -fsched2-use-superblocks -fsched-pressure @gol
371 -fsched-spec-load -fsched-spec-load-dangerous @gol
372 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
373 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
374 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
375 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
376 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
377 -fselective-scheduling -fselective-scheduling2 @gol
378 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
379 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
380 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
381 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
382 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
383 -ftree-copyrename -ftree-dce @gol
384 -ftree-dominator-opts -ftree-dse -ftree-forwprop -ftree-fre -ftree-loop-im @gol
385 -ftree-phiprop -ftree-loop-distribution @gol
386 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
387 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
388 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
389 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
390 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
391 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
392 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
393 -fwhole-program -fwhopr -fwpa -fuse-linker-plugin @gol
394 --param @var{name}=@var{value}
395 -O -O0 -O1 -O2 -O3 -Os}
397 @item Preprocessor Options
398 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
399 @gccoptlist{-A@var{question}=@var{answer} @gol
400 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
401 -C -dD -dI -dM -dN @gol
402 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
403 -idirafter @var{dir} @gol
404 -include @var{file} -imacros @var{file} @gol
405 -iprefix @var{file} -iwithprefix @var{dir} @gol
406 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
407 -imultilib @var{dir} -isysroot @var{dir} @gol
408 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
409 -P -fworking-directory -remap @gol
410 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
411 -Xpreprocessor @var{option}}
413 @item Assembler Option
414 @xref{Assembler Options,,Passing Options to the Assembler}.
415 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
418 @xref{Link Options,,Options for Linking}.
419 @gccoptlist{@var{object-file-name} -l@var{library} @gol
420 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
421 -s -static -static-libgcc -static-libstdc++ -shared @gol
422 -shared-libgcc -symbolic @gol
423 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
426 @item Directory Options
427 @xref{Directory Options,,Options for Directory Search}.
428 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
429 -specs=@var{file} -I- --sysroot=@var{dir}}
432 @c I wrote this xref this way to avoid overfull hbox. -- rms
433 @xref{Target Options}.
434 @gccoptlist{-V @var{version} -b @var{machine}}
436 @item Machine Dependent Options
437 @xref{Submodel Options,,Hardware Models and Configurations}.
438 @c This list is ordered alphanumerically by subsection name.
439 @c Try and put the significant identifier (CPU or system) first,
440 @c so users have a clue at guessing where the ones they want will be.
443 @gccoptlist{-EB -EL @gol
444 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
445 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
448 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
449 -mabi=@var{name} @gol
450 -mapcs-stack-check -mno-apcs-stack-check @gol
451 -mapcs-float -mno-apcs-float @gol
452 -mapcs-reentrant -mno-apcs-reentrant @gol
453 -msched-prolog -mno-sched-prolog @gol
454 -mlittle-endian -mbig-endian -mwords-little-endian @gol
455 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
456 -mfp16-format=@var{name}
457 -mthumb-interwork -mno-thumb-interwork @gol
458 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
459 -mstructure-size-boundary=@var{n} @gol
460 -mabort-on-noreturn @gol
461 -mlong-calls -mno-long-calls @gol
462 -msingle-pic-base -mno-single-pic-base @gol
463 -mpic-register=@var{reg} @gol
464 -mnop-fun-dllimport @gol
465 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
466 -mpoke-function-name @gol
468 -mtpcs-frame -mtpcs-leaf-frame @gol
469 -mcaller-super-interworking -mcallee-super-interworking @gol
471 -mword-relocations @gol
472 -mfix-cortex-m3-ldrd}
475 @gccoptlist{-mmcu=@var{mcu} -mno-interrupts @gol
476 -mcall-prologues -mtiny-stack -mint8}
478 @emph{Blackfin Options}
479 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
480 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
481 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
482 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
483 -mno-id-shared-library -mshared-library-id=@var{n} @gol
484 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
485 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
486 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
490 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
491 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
492 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
493 -mstack-align -mdata-align -mconst-align @gol
494 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
495 -melf -maout -melinux -mlinux -sim -sim2 @gol
496 -mmul-bug-workaround -mno-mul-bug-workaround}
499 @gccoptlist{-mmac -mpush-args}
501 @emph{Darwin Options}
502 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
503 -arch_only -bind_at_load -bundle -bundle_loader @gol
504 -client_name -compatibility_version -current_version @gol
506 -dependency-file -dylib_file -dylinker_install_name @gol
507 -dynamic -dynamiclib -exported_symbols_list @gol
508 -filelist -flat_namespace -force_cpusubtype_ALL @gol
509 -force_flat_namespace -headerpad_max_install_names @gol
511 -image_base -init -install_name -keep_private_externs @gol
512 -multi_module -multiply_defined -multiply_defined_unused @gol
513 -noall_load -no_dead_strip_inits_and_terms @gol
514 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
515 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
516 -private_bundle -read_only_relocs -sectalign @gol
517 -sectobjectsymbols -whyload -seg1addr @gol
518 -sectcreate -sectobjectsymbols -sectorder @gol
519 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
520 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
521 -segprot -segs_read_only_addr -segs_read_write_addr @gol
522 -single_module -static -sub_library -sub_umbrella @gol
523 -twolevel_namespace -umbrella -undefined @gol
524 -unexported_symbols_list -weak_reference_mismatches @gol
525 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
526 -mkernel -mone-byte-bool}
528 @emph{DEC Alpha Options}
529 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
530 -mieee -mieee-with-inexact -mieee-conformant @gol
531 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
532 -mtrap-precision=@var{mode} -mbuild-constants @gol
533 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
534 -mbwx -mmax -mfix -mcix @gol
535 -mfloat-vax -mfloat-ieee @gol
536 -mexplicit-relocs -msmall-data -mlarge-data @gol
537 -msmall-text -mlarge-text @gol
538 -mmemory-latency=@var{time}}
540 @emph{DEC Alpha/VMS Options}
541 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
544 @gccoptlist{-msmall-model -mno-lsim}
547 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
548 -mhard-float -msoft-float @gol
549 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
550 -mdouble -mno-double @gol
551 -mmedia -mno-media -mmuladd -mno-muladd @gol
552 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
553 -mlinked-fp -mlong-calls -malign-labels @gol
554 -mlibrary-pic -macc-4 -macc-8 @gol
555 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
556 -moptimize-membar -mno-optimize-membar @gol
557 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
558 -mvliw-branch -mno-vliw-branch @gol
559 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
560 -mno-nested-cond-exec -mtomcat-stats @gol
564 @emph{GNU/Linux Options}
565 @gccoptlist{-muclibc}
567 @emph{H8/300 Options}
568 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
571 @gccoptlist{-march=@var{architecture-type} @gol
572 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
573 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
574 -mfixed-range=@var{register-range} @gol
575 -mjump-in-delay -mlinker-opt -mlong-calls @gol
576 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
577 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
578 -mno-jump-in-delay -mno-long-load-store @gol
579 -mno-portable-runtime -mno-soft-float @gol
580 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
581 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
582 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
583 -munix=@var{unix-std} -nolibdld -static -threads}
585 @emph{i386 and x86-64 Options}
586 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
587 -mfpmath=@var{unit} @gol
588 -masm=@var{dialect} -mno-fancy-math-387 @gol
589 -mno-fp-ret-in-387 -msoft-float @gol
590 -mno-wide-multiply -mrtd -malign-double @gol
591 -mpreferred-stack-boundary=@var{num}
592 -mincoming-stack-boundary=@var{num}
593 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip @gol
594 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
595 -maes -mpclmul -mfused-madd @gol
596 -msse4a -m3dnow -mpopcnt -mabm -mfma4 -mxop -mlwp @gol
597 -mthreads -mno-align-stringops -minline-all-stringops @gol
598 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
599 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
600 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
601 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
602 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
603 -mcmodel=@var{code-model} -mabi=@var{name} @gol
604 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
608 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
609 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
610 -mconstant-gp -mauto-pic -mfused-madd @gol
611 -minline-float-divide-min-latency @gol
612 -minline-float-divide-max-throughput @gol
613 -mno-inline-float-divide @gol
614 -minline-int-divide-min-latency @gol
615 -minline-int-divide-max-throughput @gol
616 -mno-inline-int-divide @gol
617 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
618 -mno-inline-sqrt @gol
619 -mdwarf2-asm -mearly-stop-bits @gol
620 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
621 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
622 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
623 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
624 -msched-spec-ldc -msched-spec-control-ldc @gol
625 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
626 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
627 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
628 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
630 @emph{IA-64/VMS Options}
631 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
634 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
635 -msign-extend-enabled -muser-enabled}
637 @emph{M32R/D Options}
638 @gccoptlist{-m32r2 -m32rx -m32r @gol
640 -malign-loops -mno-align-loops @gol
641 -missue-rate=@var{number} @gol
642 -mbranch-cost=@var{number} @gol
643 -mmodel=@var{code-size-model-type} @gol
644 -msdata=@var{sdata-type} @gol
645 -mno-flush-func -mflush-func=@var{name} @gol
646 -mno-flush-trap -mflush-trap=@var{number} @gol
650 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
652 @emph{M680x0 Options}
653 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
654 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
655 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
656 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
657 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
658 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
659 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
660 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
663 @emph{M68hc1x Options}
664 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
665 -mauto-incdec -minmax -mlong-calls -mshort @gol
666 -msoft-reg-count=@var{count}}
669 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
670 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
671 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
672 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
673 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
676 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
677 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
678 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
679 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
683 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
684 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
685 -mips64 -mips64r2 @gol
686 -mips16 -mno-mips16 -mflip-mips16 @gol
687 -minterlink-mips16 -mno-interlink-mips16 @gol
688 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
689 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
690 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
691 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
692 -mfpu=@var{fpu-type} @gol
693 -msmartmips -mno-smartmips @gol
694 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
695 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
696 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
697 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
698 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
699 -membedded-data -mno-embedded-data @gol
700 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
701 -mcode-readable=@var{setting} @gol
702 -msplit-addresses -mno-split-addresses @gol
703 -mexplicit-relocs -mno-explicit-relocs @gol
704 -mcheck-zero-division -mno-check-zero-division @gol
705 -mdivide-traps -mdivide-breaks @gol
706 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
707 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
708 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
709 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
710 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
711 -mflush-func=@var{func} -mno-flush-func @gol
712 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
713 -mfp-exceptions -mno-fp-exceptions @gol
714 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
715 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
718 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
719 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
720 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
721 -mno-base-addresses -msingle-exit -mno-single-exit}
723 @emph{MN10300 Options}
724 @gccoptlist{-mmult-bug -mno-mult-bug @gol
725 -mam33 -mno-am33 @gol
726 -mam33-2 -mno-am33-2 @gol
727 -mreturn-pointer-on-d0 @gol
730 @emph{PDP-11 Options}
731 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
732 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
733 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
734 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
735 -mbranch-expensive -mbranch-cheap @gol
736 -msplit -mno-split -munix-asm -mdec-asm}
738 @emph{picoChip Options}
739 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
740 -msymbol-as-address -mno-inefficient-warnings}
742 @emph{PowerPC Options}
743 See RS/6000 and PowerPC Options.
745 @emph{RS/6000 and PowerPC Options}
746 @gccoptlist{-mcpu=@var{cpu-type} @gol
747 -mtune=@var{cpu-type} @gol
748 -mpower -mno-power -mpower2 -mno-power2 @gol
749 -mpowerpc -mpowerpc64 -mno-powerpc @gol
750 -maltivec -mno-altivec @gol
751 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
752 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
753 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
754 -mfprnd -mno-fprnd @gol
755 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
756 -mnew-mnemonics -mold-mnemonics @gol
757 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
758 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
759 -malign-power -malign-natural @gol
760 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
761 -msingle-float -mdouble-float -msimple-fpu @gol
762 -mstring -mno-string -mupdate -mno-update @gol
763 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
764 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
765 -mstrict-align -mno-strict-align -mrelocatable @gol
766 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
767 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
768 -mdynamic-no-pic -maltivec -mswdiv @gol
769 -mprioritize-restricted-insns=@var{priority} @gol
770 -msched-costly-dep=@var{dependence_type} @gol
771 -minsert-sched-nops=@var{scheme} @gol
772 -mcall-sysv -mcall-netbsd @gol
773 -maix-struct-return -msvr4-struct-return @gol
774 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
775 -misel -mno-isel @gol
776 -misel=yes -misel=no @gol
778 -mspe=yes -mspe=no @gol
780 -mgen-cell-microcode -mwarn-cell-microcode @gol
781 -mvrsave -mno-vrsave @gol
782 -mmulhw -mno-mulhw @gol
783 -mdlmzb -mno-dlmzb @gol
784 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
785 -mprototype -mno-prototype @gol
786 -msim -mmvme -mads -myellowknife -memb -msdata @gol
787 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
790 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
792 -mbig-endian-data -mlittle-endian-data @gol
795 -mas100-syntax -mno-as100-syntax@gol
797 -mmax-constant-size=@gol
799 -msave-acc-in-interrupts}
801 @emph{S/390 and zSeries Options}
802 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
803 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
804 -mlong-double-64 -mlong-double-128 @gol
805 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
806 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
807 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
808 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
809 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
812 @gccoptlist{-meb -mel @gol
816 -mscore5 -mscore5u -mscore7 -mscore7d}
819 @gccoptlist{-m1 -m2 -m2e @gol
820 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
822 -m4-nofpu -m4-single-only -m4-single -m4 @gol
823 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
824 -m5-64media -m5-64media-nofpu @gol
825 -m5-32media -m5-32media-nofpu @gol
826 -m5-compact -m5-compact-nofpu @gol
827 -mb -ml -mdalign -mrelax @gol
828 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
829 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
830 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
831 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
832 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
836 @gccoptlist{-mcpu=@var{cpu-type} @gol
837 -mtune=@var{cpu-type} @gol
838 -mcmodel=@var{code-model} @gol
839 -m32 -m64 -mapp-regs -mno-app-regs @gol
840 -mfaster-structs -mno-faster-structs @gol
841 -mfpu -mno-fpu -mhard-float -msoft-float @gol
842 -mhard-quad-float -msoft-quad-float @gol
843 -mimpure-text -mno-impure-text -mlittle-endian @gol
844 -mstack-bias -mno-stack-bias @gol
845 -munaligned-doubles -mno-unaligned-doubles @gol
846 -mv8plus -mno-v8plus -mvis -mno-vis
847 -threads -pthreads -pthread}
850 @gccoptlist{-mwarn-reloc -merror-reloc @gol
851 -msafe-dma -munsafe-dma @gol
853 -msmall-mem -mlarge-mem -mstdmain @gol
854 -mfixed-range=@var{register-range} @gol
856 -maddress-space-conversion -mno-address-space-conversion @gol
857 -mcache-size=@var{cache-size} @gol
858 -matomic-updates -mno-atomic-updates}
860 @emph{System V Options}
861 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
864 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
865 -mprolog-function -mno-prolog-function -mspace @gol
866 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
867 -mapp-regs -mno-app-regs @gol
868 -mdisable-callt -mno-disable-callt @gol
874 @gccoptlist{-mg -mgnu -munix}
876 @emph{VxWorks Options}
877 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
878 -Xbind-lazy -Xbind-now}
880 @emph{x86-64 Options}
881 See i386 and x86-64 Options.
883 @emph{i386 and x86-64 Windows Options}
884 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
885 -mnop-fun-dllimport -mthread -municode -mwin32 -mwindows
886 -fno-set-stack-executable}
888 @emph{Xstormy16 Options}
891 @emph{Xtensa Options}
892 @gccoptlist{-mconst16 -mno-const16 @gol
893 -mfused-madd -mno-fused-madd @gol
894 -mserialize-volatile -mno-serialize-volatile @gol
895 -mtext-section-literals -mno-text-section-literals @gol
896 -mtarget-align -mno-target-align @gol
897 -mlongcalls -mno-longcalls}
899 @emph{zSeries Options}
900 See S/390 and zSeries Options.
902 @item Code Generation Options
903 @xref{Code Gen Options,,Options for Code Generation Conventions}.
904 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
905 -ffixed-@var{reg} -fexceptions @gol
906 -fnon-call-exceptions -funwind-tables @gol
907 -fasynchronous-unwind-tables @gol
908 -finhibit-size-directive -finstrument-functions @gol
909 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
910 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
911 -fno-common -fno-ident @gol
912 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
913 -fno-jump-tables @gol
914 -frecord-gcc-switches @gol
915 -freg-struct-return -fshort-enums @gol
916 -fshort-double -fshort-wchar @gol
917 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
918 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
919 -fno-stack-limit -fargument-alias -fargument-noalias @gol
920 -fargument-noalias-global -fargument-noalias-anything @gol
921 -fleading-underscore -ftls-model=@var{model} @gol
922 -ftrapv -fwrapv -fbounds-check @gol
927 * Overall Options:: Controlling the kind of output:
928 an executable, object files, assembler files,
929 or preprocessed source.
930 * C Dialect Options:: Controlling the variant of C language compiled.
931 * C++ Dialect Options:: Variations on C++.
932 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
934 * Language Independent Options:: Controlling how diagnostics should be
936 * Warning Options:: How picky should the compiler be?
937 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
938 * Optimize Options:: How much optimization?
939 * Preprocessor Options:: Controlling header files and macro definitions.
940 Also, getting dependency information for Make.
941 * Assembler Options:: Passing options to the assembler.
942 * Link Options:: Specifying libraries and so on.
943 * Directory Options:: Where to find header files and libraries.
944 Where to find the compiler executable files.
945 * Spec Files:: How to pass switches to sub-processes.
946 * Target Options:: Running a cross-compiler, or an old version of GCC.
949 @node Overall Options
950 @section Options Controlling the Kind of Output
952 Compilation can involve up to four stages: preprocessing, compilation
953 proper, assembly and linking, always in that order. GCC is capable of
954 preprocessing and compiling several files either into several
955 assembler input files, or into one assembler input file; then each
956 assembler input file produces an object file, and linking combines all
957 the object files (those newly compiled, and those specified as input)
958 into an executable file.
960 @cindex file name suffix
961 For any given input file, the file name suffix determines what kind of
966 C source code which must be preprocessed.
969 C source code which should not be preprocessed.
972 C++ source code which should not be preprocessed.
975 Objective-C source code. Note that you must link with the @file{libobjc}
976 library to make an Objective-C program work.
979 Objective-C source code which should not be preprocessed.
983 Objective-C++ source code. Note that you must link with the @file{libobjc}
984 library to make an Objective-C++ program work. Note that @samp{.M} refers
985 to a literal capital M@.
988 Objective-C++ source code which should not be preprocessed.
991 C, C++, Objective-C or Objective-C++ header file to be turned into a
996 @itemx @var{file}.cxx
997 @itemx @var{file}.cpp
998 @itemx @var{file}.CPP
999 @itemx @var{file}.c++
1001 C++ source code which must be preprocessed. Note that in @samp{.cxx},
1002 the last two letters must both be literally @samp{x}. Likewise,
1003 @samp{.C} refers to a literal capital C@.
1007 Objective-C++ source code which must be preprocessed.
1009 @item @var{file}.mii
1010 Objective-C++ source code which should not be preprocessed.
1014 @itemx @var{file}.hp
1015 @itemx @var{file}.hxx
1016 @itemx @var{file}.hpp
1017 @itemx @var{file}.HPP
1018 @itemx @var{file}.h++
1019 @itemx @var{file}.tcc
1020 C++ header file to be turned into a precompiled header.
1023 @itemx @var{file}.for
1024 @itemx @var{file}.ftn
1025 Fixed form Fortran source code which should not be preprocessed.
1028 @itemx @var{file}.FOR
1029 @itemx @var{file}.fpp
1030 @itemx @var{file}.FPP
1031 @itemx @var{file}.FTN
1032 Fixed form Fortran source code which must be preprocessed (with the traditional
1035 @item @var{file}.f90
1036 @itemx @var{file}.f95
1037 @itemx @var{file}.f03
1038 @itemx @var{file}.f08
1039 Free form Fortran source code which should not be preprocessed.
1041 @item @var{file}.F90
1042 @itemx @var{file}.F95
1043 @itemx @var{file}.F03
1044 @itemx @var{file}.F08
1045 Free form Fortran source code which must be preprocessed (with the
1046 traditional preprocessor).
1048 @c FIXME: Descriptions of Java file types.
1054 @item @var{file}.ads
1055 Ada source code file which contains a library unit declaration (a
1056 declaration of a package, subprogram, or generic, or a generic
1057 instantiation), or a library unit renaming declaration (a package,
1058 generic, or subprogram renaming declaration). Such files are also
1061 @item @var{file}.adb
1062 Ada source code file containing a library unit body (a subprogram or
1063 package body). Such files are also called @dfn{bodies}.
1065 @c GCC also knows about some suffixes for languages not yet included:
1076 @itemx @var{file}.sx
1077 Assembler code which must be preprocessed.
1080 An object file to be fed straight into linking.
1081 Any file name with no recognized suffix is treated this way.
1085 You can specify the input language explicitly with the @option{-x} option:
1088 @item -x @var{language}
1089 Specify explicitly the @var{language} for the following input files
1090 (rather than letting the compiler choose a default based on the file
1091 name suffix). This option applies to all following input files until
1092 the next @option{-x} option. Possible values for @var{language} are:
1094 c c-header c-cpp-output
1095 c++ c++-header c++-cpp-output
1096 objective-c objective-c-header objective-c-cpp-output
1097 objective-c++ objective-c++-header objective-c++-cpp-output
1098 assembler assembler-with-cpp
1100 f77 f77-cpp-input f95 f95-cpp-input
1105 Turn off any specification of a language, so that subsequent files are
1106 handled according to their file name suffixes (as they are if @option{-x}
1107 has not been used at all).
1109 @item -pass-exit-codes
1110 @opindex pass-exit-codes
1111 Normally the @command{gcc} program will exit with the code of 1 if any
1112 phase of the compiler returns a non-success return code. If you specify
1113 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1114 numerically highest error produced by any phase that returned an error
1115 indication. The C, C++, and Fortran frontends return 4, if an internal
1116 compiler error is encountered.
1119 If you only want some of the stages of compilation, you can use
1120 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1121 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1122 @command{gcc} is to stop. Note that some combinations (for example,
1123 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1128 Compile or assemble the source files, but do not link. The linking
1129 stage simply is not done. The ultimate output is in the form of an
1130 object file for each source file.
1132 By default, the object file name for a source file is made by replacing
1133 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1135 Unrecognized input files, not requiring compilation or assembly, are
1140 Stop after the stage of compilation proper; do not assemble. The output
1141 is in the form of an assembler code file for each non-assembler input
1144 By default, the assembler file name for a source file is made by
1145 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1147 Input files that don't require compilation are ignored.
1151 Stop after the preprocessing stage; do not run the compiler proper. The
1152 output is in the form of preprocessed source code, which is sent to the
1155 Input files which don't require preprocessing are ignored.
1157 @cindex output file option
1160 Place output in file @var{file}. This applies regardless to whatever
1161 sort of output is being produced, whether it be an executable file,
1162 an object file, an assembler file or preprocessed C code.
1164 If @option{-o} is not specified, the default is to put an executable
1165 file in @file{a.out}, the object file for
1166 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1167 assembler file in @file{@var{source}.s}, a precompiled header file in
1168 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1173 Print (on standard error output) the commands executed to run the stages
1174 of compilation. Also print the version number of the compiler driver
1175 program and of the preprocessor and the compiler proper.
1179 Like @option{-v} except the commands are not executed and all command
1180 arguments are quoted. This is useful for shell scripts to capture the
1181 driver-generated command lines.
1185 Use pipes rather than temporary files for communication between the
1186 various stages of compilation. This fails to work on some systems where
1187 the assembler is unable to read from a pipe; but the GNU assembler has
1192 If you are compiling multiple source files, this option tells the driver
1193 to pass all the source files to the compiler at once (for those
1194 languages for which the compiler can handle this). This will allow
1195 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1196 language for which this is supported is C@. If you pass source files for
1197 multiple languages to the driver, using this option, the driver will invoke
1198 the compiler(s) that support IMA once each, passing each compiler all the
1199 source files appropriate for it. For those languages that do not support
1200 IMA this option will be ignored, and the compiler will be invoked once for
1201 each source file in that language. If you use this option in conjunction
1202 with @option{-save-temps}, the compiler will generate multiple
1204 (one for each source file), but only one (combined) @file{.o} or
1209 Print (on the standard output) a description of the command line options
1210 understood by @command{gcc}. If the @option{-v} option is also specified
1211 then @option{--help} will also be passed on to the various processes
1212 invoked by @command{gcc}, so that they can display the command line options
1213 they accept. If the @option{-Wextra} option has also been specified
1214 (prior to the @option{--help} option), then command line options which
1215 have no documentation associated with them will also be displayed.
1218 @opindex target-help
1219 Print (on the standard output) a description of target-specific command
1220 line options for each tool. For some targets extra target-specific
1221 information may also be printed.
1223 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1224 Print (on the standard output) a description of the command line
1225 options understood by the compiler that fit into all specified classes
1226 and qualifiers. These are the supported classes:
1229 @item @samp{optimizers}
1230 This will display all of the optimization options supported by the
1233 @item @samp{warnings}
1234 This will display all of the options controlling warning messages
1235 produced by the compiler.
1238 This will display target-specific options. Unlike the
1239 @option{--target-help} option however, target-specific options of the
1240 linker and assembler will not be displayed. This is because those
1241 tools do not currently support the extended @option{--help=} syntax.
1244 This will display the values recognized by the @option{--param}
1247 @item @var{language}
1248 This will display the options supported for @var{language}, where
1249 @var{language} is the name of one of the languages supported in this
1253 This will display the options that are common to all languages.
1256 These are the supported qualifiers:
1259 @item @samp{undocumented}
1260 Display only those options which are undocumented.
1263 Display options which take an argument that appears after an equal
1264 sign in the same continuous piece of text, such as:
1265 @samp{--help=target}.
1267 @item @samp{separate}
1268 Display options which take an argument that appears as a separate word
1269 following the original option, such as: @samp{-o output-file}.
1272 Thus for example to display all the undocumented target-specific
1273 switches supported by the compiler the following can be used:
1276 --help=target,undocumented
1279 The sense of a qualifier can be inverted by prefixing it with the
1280 @samp{^} character, so for example to display all binary warning
1281 options (i.e., ones that are either on or off and that do not take an
1282 argument), which have a description the following can be used:
1285 --help=warnings,^joined,^undocumented
1288 The argument to @option{--help=} should not consist solely of inverted
1291 Combining several classes is possible, although this usually
1292 restricts the output by so much that there is nothing to display. One
1293 case where it does work however is when one of the classes is
1294 @var{target}. So for example to display all the target-specific
1295 optimization options the following can be used:
1298 --help=target,optimizers
1301 The @option{--help=} option can be repeated on the command line. Each
1302 successive use will display its requested class of options, skipping
1303 those that have already been displayed.
1305 If the @option{-Q} option appears on the command line before the
1306 @option{--help=} option, then the descriptive text displayed by
1307 @option{--help=} is changed. Instead of describing the displayed
1308 options, an indication is given as to whether the option is enabled,
1309 disabled or set to a specific value (assuming that the compiler
1310 knows this at the point where the @option{--help=} option is used).
1312 Here is a truncated example from the ARM port of @command{gcc}:
1315 % gcc -Q -mabi=2 --help=target -c
1316 The following options are target specific:
1318 -mabort-on-noreturn [disabled]
1322 The output is sensitive to the effects of previous command line
1323 options, so for example it is possible to find out which optimizations
1324 are enabled at @option{-O2} by using:
1327 -Q -O2 --help=optimizers
1330 Alternatively you can discover which binary optimizations are enabled
1331 by @option{-O3} by using:
1334 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1335 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1336 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1339 @item -no-canonical-prefixes
1340 @opindex no-canonical-prefixes
1341 Do not expand any symbolic links, resolve references to @samp{/../}
1342 or @samp{/./}, or make the path absolute when generating a relative
1347 Display the version number and copyrights of the invoked GCC@.
1351 Invoke all subcommands under a wrapper program. It takes a single
1352 comma separated list as an argument, which will be used to invoke
1356 gcc -c t.c -wrapper gdb,--args
1359 This will invoke all subprograms of gcc under "gdb --args",
1360 thus cc1 invocation will be "gdb --args cc1 ...".
1362 @item -fplugin=@var{name}.so
1363 Load the plugin code in file @var{name}.so, assumed to be a
1364 shared object to be dlopen'd by the compiler. The base name of
1365 the shared object file is used to identify the plugin for the
1366 purposes of argument parsing (See
1367 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1368 Each plugin should define the callback functions specified in the
1371 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1372 Define an argument called @var{key} with a value of @var{value}
1373 for the plugin called @var{name}.
1375 @include @value{srcdir}/../libiberty/at-file.texi
1379 @section Compiling C++ Programs
1381 @cindex suffixes for C++ source
1382 @cindex C++ source file suffixes
1383 C++ source files conventionally use one of the suffixes @samp{.C},
1384 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1385 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1386 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1387 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1388 files with these names and compiles them as C++ programs even if you
1389 call the compiler the same way as for compiling C programs (usually
1390 with the name @command{gcc}).
1394 However, the use of @command{gcc} does not add the C++ library.
1395 @command{g++} is a program that calls GCC and treats @samp{.c},
1396 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1397 files unless @option{-x} is used, and automatically specifies linking
1398 against the C++ library. This program is also useful when
1399 precompiling a C header file with a @samp{.h} extension for use in C++
1400 compilations. On many systems, @command{g++} is also installed with
1401 the name @command{c++}.
1403 @cindex invoking @command{g++}
1404 When you compile C++ programs, you may specify many of the same
1405 command-line options that you use for compiling programs in any
1406 language; or command-line options meaningful for C and related
1407 languages; or options that are meaningful only for C++ programs.
1408 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1409 explanations of options for languages related to C@.
1410 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1411 explanations of options that are meaningful only for C++ programs.
1413 @node C Dialect Options
1414 @section Options Controlling C Dialect
1415 @cindex dialect options
1416 @cindex language dialect options
1417 @cindex options, dialect
1419 The following options control the dialect of C (or languages derived
1420 from C, such as C++, Objective-C and Objective-C++) that the compiler
1424 @cindex ANSI support
1428 In C mode, this is equivalent to @samp{-std=c90}. In C++ mode, it is
1429 equivalent to @samp{-std=c++98}.
1431 This turns off certain features of GCC that are incompatible with ISO
1432 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1433 such as the @code{asm} and @code{typeof} keywords, and
1434 predefined macros such as @code{unix} and @code{vax} that identify the
1435 type of system you are using. It also enables the undesirable and
1436 rarely used ISO trigraph feature. For the C compiler,
1437 it disables recognition of C++ style @samp{//} comments as well as
1438 the @code{inline} keyword.
1440 The alternate keywords @code{__asm__}, @code{__extension__},
1441 @code{__inline__} and @code{__typeof__} continue to work despite
1442 @option{-ansi}. You would not want to use them in an ISO C program, of
1443 course, but it is useful to put them in header files that might be included
1444 in compilations done with @option{-ansi}. Alternate predefined macros
1445 such as @code{__unix__} and @code{__vax__} are also available, with or
1446 without @option{-ansi}.
1448 The @option{-ansi} option does not cause non-ISO programs to be
1449 rejected gratuitously. For that, @option{-pedantic} is required in
1450 addition to @option{-ansi}. @xref{Warning Options}.
1452 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1453 option is used. Some header files may notice this macro and refrain
1454 from declaring certain functions or defining certain macros that the
1455 ISO standard doesn't call for; this is to avoid interfering with any
1456 programs that might use these names for other things.
1458 Functions that would normally be built in but do not have semantics
1459 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1460 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1461 built-in functions provided by GCC}, for details of the functions
1466 Determine the language standard. @xref{Standards,,Language Standards
1467 Supported by GCC}, for details of these standard versions. This option
1468 is currently only supported when compiling C or C++.
1470 The compiler can accept several base standards, such as @samp{c90} or
1471 @samp{c++98}, and GNU dialects of those standards, such as
1472 @samp{gnu90} or @samp{gnu++98}. By specifying a base standard, the
1473 compiler will accept all programs following that standard and those
1474 using GNU extensions that do not contradict it. For example,
1475 @samp{-std=c90} turns off certain features of GCC that are
1476 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1477 keywords, but not other GNU extensions that do not have a meaning in
1478 ISO C90, such as omitting the middle term of a @code{?:}
1479 expression. On the other hand, by specifying a GNU dialect of a
1480 standard, all features the compiler support are enabled, even when
1481 those features change the meaning of the base standard and some
1482 strict-conforming programs may be rejected. The particular standard
1483 is used by @option{-pedantic} to identify which features are GNU
1484 extensions given that version of the standard. For example
1485 @samp{-std=gnu90 -pedantic} would warn about C++ style @samp{//}
1486 comments, while @samp{-std=gnu99 -pedantic} would not.
1488 A value for this option must be provided; possible values are
1494 Support all ISO C90 programs (certain GNU extensions that conflict
1495 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1497 @item iso9899:199409
1498 ISO C90 as modified in amendment 1.
1504 ISO C99. Note that this standard is not yet fully supported; see
1505 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1506 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1510 GNU dialect of ISO C90 (including some C99 features). This
1511 is the default for C code.
1515 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1516 this will become the default. The name @samp{gnu9x} is deprecated.
1519 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1523 GNU dialect of @option{-std=c++98}. This is the default for
1527 The working draft of the upcoming ISO C++0x standard. This option
1528 enables experimental features that are likely to be included in
1529 C++0x. The working draft is constantly changing, and any feature that is
1530 enabled by this flag may be removed from future versions of GCC if it is
1531 not part of the C++0x standard.
1534 GNU dialect of @option{-std=c++0x}. This option enables
1535 experimental features that may be removed in future versions of GCC.
1538 @item -fgnu89-inline
1539 @opindex fgnu89-inline
1540 The option @option{-fgnu89-inline} tells GCC to use the traditional
1541 GNU semantics for @code{inline} functions when in C99 mode.
1542 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1543 is accepted and ignored by GCC versions 4.1.3 up to but not including
1544 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1545 C99 mode. Using this option is roughly equivalent to adding the
1546 @code{gnu_inline} function attribute to all inline functions
1547 (@pxref{Function Attributes}).
1549 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1550 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1551 specifies the default behavior). This option was first supported in
1552 GCC 4.3. This option is not supported in @option{-std=c90} or
1553 @option{-std=gnu90} mode.
1555 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1556 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1557 in effect for @code{inline} functions. @xref{Common Predefined
1558 Macros,,,cpp,The C Preprocessor}.
1560 @item -aux-info @var{filename}
1562 Output to the given filename prototyped declarations for all functions
1563 declared and/or defined in a translation unit, including those in header
1564 files. This option is silently ignored in any language other than C@.
1566 Besides declarations, the file indicates, in comments, the origin of
1567 each declaration (source file and line), whether the declaration was
1568 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1569 @samp{O} for old, respectively, in the first character after the line
1570 number and the colon), and whether it came from a declaration or a
1571 definition (@samp{C} or @samp{F}, respectively, in the following
1572 character). In the case of function definitions, a K&R-style list of
1573 arguments followed by their declarations is also provided, inside
1574 comments, after the declaration.
1578 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1579 keyword, so that code can use these words as identifiers. You can use
1580 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1581 instead. @option{-ansi} implies @option{-fno-asm}.
1583 In C++, this switch only affects the @code{typeof} keyword, since
1584 @code{asm} and @code{inline} are standard keywords. You may want to
1585 use the @option{-fno-gnu-keywords} flag instead, which has the same
1586 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1587 switch only affects the @code{asm} and @code{typeof} keywords, since
1588 @code{inline} is a standard keyword in ISO C99.
1591 @itemx -fno-builtin-@var{function}
1592 @opindex fno-builtin
1593 @cindex built-in functions
1594 Don't recognize built-in functions that do not begin with
1595 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1596 functions provided by GCC}, for details of the functions affected,
1597 including those which are not built-in functions when @option{-ansi} or
1598 @option{-std} options for strict ISO C conformance are used because they
1599 do not have an ISO standard meaning.
1601 GCC normally generates special code to handle certain built-in functions
1602 more efficiently; for instance, calls to @code{alloca} may become single
1603 instructions that adjust the stack directly, and calls to @code{memcpy}
1604 may become inline copy loops. The resulting code is often both smaller
1605 and faster, but since the function calls no longer appear as such, you
1606 cannot set a breakpoint on those calls, nor can you change the behavior
1607 of the functions by linking with a different library. In addition,
1608 when a function is recognized as a built-in function, GCC may use
1609 information about that function to warn about problems with calls to
1610 that function, or to generate more efficient code, even if the
1611 resulting code still contains calls to that function. For example,
1612 warnings are given with @option{-Wformat} for bad calls to
1613 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1614 known not to modify global memory.
1616 With the @option{-fno-builtin-@var{function}} option
1617 only the built-in function @var{function} is
1618 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1619 function is named that is not built-in in this version of GCC, this
1620 option is ignored. There is no corresponding
1621 @option{-fbuiltin-@var{function}} option; if you wish to enable
1622 built-in functions selectively when using @option{-fno-builtin} or
1623 @option{-ffreestanding}, you may define macros such as:
1626 #define abs(n) __builtin_abs ((n))
1627 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1632 @cindex hosted environment
1634 Assert that compilation takes place in a hosted environment. This implies
1635 @option{-fbuiltin}. A hosted environment is one in which the
1636 entire standard library is available, and in which @code{main} has a return
1637 type of @code{int}. Examples are nearly everything except a kernel.
1638 This is equivalent to @option{-fno-freestanding}.
1640 @item -ffreestanding
1641 @opindex ffreestanding
1642 @cindex hosted environment
1644 Assert that compilation takes place in a freestanding environment. This
1645 implies @option{-fno-builtin}. A freestanding environment
1646 is one in which the standard library may not exist, and program startup may
1647 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1648 This is equivalent to @option{-fno-hosted}.
1650 @xref{Standards,,Language Standards Supported by GCC}, for details of
1651 freestanding and hosted environments.
1655 @cindex openmp parallel
1656 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1657 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1658 compiler generates parallel code according to the OpenMP Application
1659 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1660 implies @option{-pthread}, and thus is only supported on targets that
1661 have support for @option{-pthread}.
1663 @item -fms-extensions
1664 @opindex fms-extensions
1665 Accept some non-standard constructs used in Microsoft header files.
1667 Some cases of unnamed fields in structures and unions are only
1668 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1669 fields within structs/unions}, for details.
1673 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1674 options for strict ISO C conformance) implies @option{-trigraphs}.
1676 @item -no-integrated-cpp
1677 @opindex no-integrated-cpp
1678 Performs a compilation in two passes: preprocessing and compiling. This
1679 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1680 @option{-B} option. The user supplied compilation step can then add in
1681 an additional preprocessing step after normal preprocessing but before
1682 compiling. The default is to use the integrated cpp (internal cpp)
1684 The semantics of this option will change if "cc1", "cc1plus", and
1685 "cc1obj" are merged.
1687 @cindex traditional C language
1688 @cindex C language, traditional
1690 @itemx -traditional-cpp
1691 @opindex traditional-cpp
1692 @opindex traditional
1693 Formerly, these options caused GCC to attempt to emulate a pre-standard
1694 C compiler. They are now only supported with the @option{-E} switch.
1695 The preprocessor continues to support a pre-standard mode. See the GNU
1696 CPP manual for details.
1698 @item -fcond-mismatch
1699 @opindex fcond-mismatch
1700 Allow conditional expressions with mismatched types in the second and
1701 third arguments. The value of such an expression is void. This option
1702 is not supported for C++.
1704 @item -flax-vector-conversions
1705 @opindex flax-vector-conversions
1706 Allow implicit conversions between vectors with differing numbers of
1707 elements and/or incompatible element types. This option should not be
1710 @item -funsigned-char
1711 @opindex funsigned-char
1712 Let the type @code{char} be unsigned, like @code{unsigned char}.
1714 Each kind of machine has a default for what @code{char} should
1715 be. It is either like @code{unsigned char} by default or like
1716 @code{signed char} by default.
1718 Ideally, a portable program should always use @code{signed char} or
1719 @code{unsigned char} when it depends on the signedness of an object.
1720 But many programs have been written to use plain @code{char} and
1721 expect it to be signed, or expect it to be unsigned, depending on the
1722 machines they were written for. This option, and its inverse, let you
1723 make such a program work with the opposite default.
1725 The type @code{char} is always a distinct type from each of
1726 @code{signed char} or @code{unsigned char}, even though its behavior
1727 is always just like one of those two.
1730 @opindex fsigned-char
1731 Let the type @code{char} be signed, like @code{signed char}.
1733 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1734 the negative form of @option{-funsigned-char}. Likewise, the option
1735 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1737 @item -fsigned-bitfields
1738 @itemx -funsigned-bitfields
1739 @itemx -fno-signed-bitfields
1740 @itemx -fno-unsigned-bitfields
1741 @opindex fsigned-bitfields
1742 @opindex funsigned-bitfields
1743 @opindex fno-signed-bitfields
1744 @opindex fno-unsigned-bitfields
1745 These options control whether a bit-field is signed or unsigned, when the
1746 declaration does not use either @code{signed} or @code{unsigned}. By
1747 default, such a bit-field is signed, because this is consistent: the
1748 basic integer types such as @code{int} are signed types.
1751 @node C++ Dialect Options
1752 @section Options Controlling C++ Dialect
1754 @cindex compiler options, C++
1755 @cindex C++ options, command line
1756 @cindex options, C++
1757 This section describes the command-line options that are only meaningful
1758 for C++ programs; but you can also use most of the GNU compiler options
1759 regardless of what language your program is in. For example, you
1760 might compile a file @code{firstClass.C} like this:
1763 g++ -g -frepo -O -c firstClass.C
1767 In this example, only @option{-frepo} is an option meant
1768 only for C++ programs; you can use the other options with any
1769 language supported by GCC@.
1771 Here is a list of options that are @emph{only} for compiling C++ programs:
1775 @item -fabi-version=@var{n}
1776 @opindex fabi-version
1777 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1778 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1779 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1780 the version that conforms most closely to the C++ ABI specification.
1781 Therefore, the ABI obtained using version 0 will change as ABI bugs
1784 The default is version 2.
1786 Version 3 corrects an error in mangling a constant address as a
1789 Version 4 implements a standard mangling for vector types.
1791 See also @option{-Wabi}.
1793 @item -fno-access-control
1794 @opindex fno-access-control
1795 Turn off all access checking. This switch is mainly useful for working
1796 around bugs in the access control code.
1800 Check that the pointer returned by @code{operator new} is non-null
1801 before attempting to modify the storage allocated. This check is
1802 normally unnecessary because the C++ standard specifies that
1803 @code{operator new} will only return @code{0} if it is declared
1804 @samp{throw()}, in which case the compiler will always check the
1805 return value even without this option. In all other cases, when
1806 @code{operator new} has a non-empty exception specification, memory
1807 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1808 @samp{new (nothrow)}.
1810 @item -fconserve-space
1811 @opindex fconserve-space
1812 Put uninitialized or runtime-initialized global variables into the
1813 common segment, as C does. This saves space in the executable at the
1814 cost of not diagnosing duplicate definitions. If you compile with this
1815 flag and your program mysteriously crashes after @code{main()} has
1816 completed, you may have an object that is being destroyed twice because
1817 two definitions were merged.
1819 This option is no longer useful on most targets, now that support has
1820 been added for putting variables into BSS without making them common.
1822 @item -fno-deduce-init-list
1823 @opindex fno-deduce-init-list
1824 Disable deduction of a template type parameter as
1825 std::initializer_list from a brace-enclosed initializer list, i.e.
1828 template <class T> auto forward(T t) -> decltype (realfn (t))
1835 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1839 This option is present because this deduction is an extension to the
1840 current specification in the C++0x working draft, and there was
1841 some concern about potential overload resolution problems.
1843 @item -ffriend-injection
1844 @opindex ffriend-injection
1845 Inject friend functions into the enclosing namespace, so that they are
1846 visible outside the scope of the class in which they are declared.
1847 Friend functions were documented to work this way in the old Annotated
1848 C++ Reference Manual, and versions of G++ before 4.1 always worked
1849 that way. However, in ISO C++ a friend function which is not declared
1850 in an enclosing scope can only be found using argument dependent
1851 lookup. This option causes friends to be injected as they were in
1854 This option is for compatibility, and may be removed in a future
1857 @item -fno-elide-constructors
1858 @opindex fno-elide-constructors
1859 The C++ standard allows an implementation to omit creating a temporary
1860 which is only used to initialize another object of the same type.
1861 Specifying this option disables that optimization, and forces G++ to
1862 call the copy constructor in all cases.
1864 @item -fno-enforce-eh-specs
1865 @opindex fno-enforce-eh-specs
1866 Don't generate code to check for violation of exception specifications
1867 at runtime. This option violates the C++ standard, but may be useful
1868 for reducing code size in production builds, much like defining
1869 @samp{NDEBUG}. This does not give user code permission to throw
1870 exceptions in violation of the exception specifications; the compiler
1871 will still optimize based on the specifications, so throwing an
1872 unexpected exception will result in undefined behavior.
1875 @itemx -fno-for-scope
1877 @opindex fno-for-scope
1878 If @option{-ffor-scope} is specified, the scope of variables declared in
1879 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1880 as specified by the C++ standard.
1881 If @option{-fno-for-scope} is specified, the scope of variables declared in
1882 a @i{for-init-statement} extends to the end of the enclosing scope,
1883 as was the case in old versions of G++, and other (traditional)
1884 implementations of C++.
1886 The default if neither flag is given to follow the standard,
1887 but to allow and give a warning for old-style code that would
1888 otherwise be invalid, or have different behavior.
1890 @item -fno-gnu-keywords
1891 @opindex fno-gnu-keywords
1892 Do not recognize @code{typeof} as a keyword, so that code can use this
1893 word as an identifier. You can use the keyword @code{__typeof__} instead.
1894 @option{-ansi} implies @option{-fno-gnu-keywords}.
1896 @item -fno-implicit-templates
1897 @opindex fno-implicit-templates
1898 Never emit code for non-inline templates which are instantiated
1899 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1900 @xref{Template Instantiation}, for more information.
1902 @item -fno-implicit-inline-templates
1903 @opindex fno-implicit-inline-templates
1904 Don't emit code for implicit instantiations of inline templates, either.
1905 The default is to handle inlines differently so that compiles with and
1906 without optimization will need the same set of explicit instantiations.
1908 @item -fno-implement-inlines
1909 @opindex fno-implement-inlines
1910 To save space, do not emit out-of-line copies of inline functions
1911 controlled by @samp{#pragma implementation}. This will cause linker
1912 errors if these functions are not inlined everywhere they are called.
1914 @item -fms-extensions
1915 @opindex fms-extensions
1916 Disable pedantic warnings about constructs used in MFC, such as implicit
1917 int and getting a pointer to member function via non-standard syntax.
1919 @item -fno-nonansi-builtins
1920 @opindex fno-nonansi-builtins
1921 Disable built-in declarations of functions that are not mandated by
1922 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1923 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1925 @item -fno-operator-names
1926 @opindex fno-operator-names
1927 Do not treat the operator name keywords @code{and}, @code{bitand},
1928 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1929 synonyms as keywords.
1931 @item -fno-optional-diags
1932 @opindex fno-optional-diags
1933 Disable diagnostics that the standard says a compiler does not need to
1934 issue. Currently, the only such diagnostic issued by G++ is the one for
1935 a name having multiple meanings within a class.
1938 @opindex fpermissive
1939 Downgrade some diagnostics about nonconformant code from errors to
1940 warnings. Thus, using @option{-fpermissive} will allow some
1941 nonconforming code to compile.
1943 @item -fno-pretty-templates
1944 @opindex fno-pretty-templates
1945 When an error message refers to a specialization of a function
1946 template, the compiler will normally print the signature of the
1947 template followed by the template arguments and any typedefs or
1948 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1949 rather than @code{void f(int)}) so that it's clear which template is
1950 involved. When an error message refers to a specialization of a class
1951 template, the compiler will omit any template arguments which match
1952 the default template arguments for that template. If either of these
1953 behaviors make it harder to understand the error message rather than
1954 easier, using @option{-fno-pretty-templates} will disable them.
1958 Enable automatic template instantiation at link time. This option also
1959 implies @option{-fno-implicit-templates}. @xref{Template
1960 Instantiation}, for more information.
1964 Disable generation of information about every class with virtual
1965 functions for use by the C++ runtime type identification features
1966 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1967 of the language, you can save some space by using this flag. Note that
1968 exception handling uses the same information, but it will generate it as
1969 needed. The @samp{dynamic_cast} operator can still be used for casts that
1970 do not require runtime type information, i.e.@: casts to @code{void *} or to
1971 unambiguous base classes.
1975 Emit statistics about front-end processing at the end of the compilation.
1976 This information is generally only useful to the G++ development team.
1978 @item -ftemplate-depth=@var{n}
1979 @opindex ftemplate-depth
1980 Set the maximum instantiation depth for template classes to @var{n}.
1981 A limit on the template instantiation depth is needed to detect
1982 endless recursions during template class instantiation. ANSI/ISO C++
1983 conforming programs must not rely on a maximum depth greater than 17
1984 (changed to 1024 in C++0x).
1986 @item -fno-threadsafe-statics
1987 @opindex fno-threadsafe-statics
1988 Do not emit the extra code to use the routines specified in the C++
1989 ABI for thread-safe initialization of local statics. You can use this
1990 option to reduce code size slightly in code that doesn't need to be
1993 @item -fuse-cxa-atexit
1994 @opindex fuse-cxa-atexit
1995 Register destructors for objects with static storage duration with the
1996 @code{__cxa_atexit} function rather than the @code{atexit} function.
1997 This option is required for fully standards-compliant handling of static
1998 destructors, but will only work if your C library supports
1999 @code{__cxa_atexit}.
2001 @item -fno-use-cxa-get-exception-ptr
2002 @opindex fno-use-cxa-get-exception-ptr
2003 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2004 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2005 if the runtime routine is not available.
2007 @item -fvisibility-inlines-hidden
2008 @opindex fvisibility-inlines-hidden
2009 This switch declares that the user does not attempt to compare
2010 pointers to inline methods where the addresses of the two functions
2011 were taken in different shared objects.
2013 The effect of this is that GCC may, effectively, mark inline methods with
2014 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2015 appear in the export table of a DSO and do not require a PLT indirection
2016 when used within the DSO@. Enabling this option can have a dramatic effect
2017 on load and link times of a DSO as it massively reduces the size of the
2018 dynamic export table when the library makes heavy use of templates.
2020 The behavior of this switch is not quite the same as marking the
2021 methods as hidden directly, because it does not affect static variables
2022 local to the function or cause the compiler to deduce that
2023 the function is defined in only one shared object.
2025 You may mark a method as having a visibility explicitly to negate the
2026 effect of the switch for that method. For example, if you do want to
2027 compare pointers to a particular inline method, you might mark it as
2028 having default visibility. Marking the enclosing class with explicit
2029 visibility will have no effect.
2031 Explicitly instantiated inline methods are unaffected by this option
2032 as their linkage might otherwise cross a shared library boundary.
2033 @xref{Template Instantiation}.
2035 @item -fvisibility-ms-compat
2036 @opindex fvisibility-ms-compat
2037 This flag attempts to use visibility settings to make GCC's C++
2038 linkage model compatible with that of Microsoft Visual Studio.
2040 The flag makes these changes to GCC's linkage model:
2044 It sets the default visibility to @code{hidden}, like
2045 @option{-fvisibility=hidden}.
2048 Types, but not their members, are not hidden by default.
2051 The One Definition Rule is relaxed for types without explicit
2052 visibility specifications which are defined in more than one different
2053 shared object: those declarations are permitted if they would have
2054 been permitted when this option was not used.
2057 In new code it is better to use @option{-fvisibility=hidden} and
2058 export those classes which are intended to be externally visible.
2059 Unfortunately it is possible for code to rely, perhaps accidentally,
2060 on the Visual Studio behavior.
2062 Among the consequences of these changes are that static data members
2063 of the same type with the same name but defined in different shared
2064 objects will be different, so changing one will not change the other;
2065 and that pointers to function members defined in different shared
2066 objects may not compare equal. When this flag is given, it is a
2067 violation of the ODR to define types with the same name differently.
2071 Do not use weak symbol support, even if it is provided by the linker.
2072 By default, G++ will use weak symbols if they are available. This
2073 option exists only for testing, and should not be used by end-users;
2074 it will result in inferior code and has no benefits. This option may
2075 be removed in a future release of G++.
2079 Do not search for header files in the standard directories specific to
2080 C++, but do still search the other standard directories. (This option
2081 is used when building the C++ library.)
2084 In addition, these optimization, warning, and code generation options
2085 have meanings only for C++ programs:
2088 @item -fno-default-inline
2089 @opindex fno-default-inline
2090 Do not assume @samp{inline} for functions defined inside a class scope.
2091 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2092 functions will have linkage like inline functions; they just won't be
2095 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2098 Warn when G++ generates code that is probably not compatible with the
2099 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2100 all such cases, there are probably some cases that are not warned about,
2101 even though G++ is generating incompatible code. There may also be
2102 cases where warnings are emitted even though the code that is generated
2105 You should rewrite your code to avoid these warnings if you are
2106 concerned about the fact that code generated by G++ may not be binary
2107 compatible with code generated by other compilers.
2109 The known incompatibilities in @option{-fabi-version=2} (the default) include:
2114 A template with a non-type template parameter of reference type is
2115 mangled incorrectly:
2118 template <int &> struct S @{@};
2122 This is fixed in @option{-fabi-version=3}.
2125 SIMD vector types declared using @code{__attribute ((vector_size))} are
2126 mangled in a non-standard way that does not allow for overloading of
2127 functions taking vectors of different sizes.
2129 The mangling is changed in @option{-fabi-version=4}.
2132 The known incompatibilities in @option{-fabi-version=1} include:
2137 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2138 pack data into the same byte as a base class. For example:
2141 struct A @{ virtual void f(); int f1 : 1; @};
2142 struct B : public A @{ int f2 : 1; @};
2146 In this case, G++ will place @code{B::f2} into the same byte
2147 as@code{A::f1}; other compilers will not. You can avoid this problem
2148 by explicitly padding @code{A} so that its size is a multiple of the
2149 byte size on your platform; that will cause G++ and other compilers to
2150 layout @code{B} identically.
2153 Incorrect handling of tail-padding for virtual bases. G++ does not use
2154 tail padding when laying out virtual bases. For example:
2157 struct A @{ virtual void f(); char c1; @};
2158 struct B @{ B(); char c2; @};
2159 struct C : public A, public virtual B @{@};
2163 In this case, G++ will not place @code{B} into the tail-padding for
2164 @code{A}; other compilers will. You can avoid this problem by
2165 explicitly padding @code{A} so that its size is a multiple of its
2166 alignment (ignoring virtual base classes); that will cause G++ and other
2167 compilers to layout @code{C} identically.
2170 Incorrect handling of bit-fields with declared widths greater than that
2171 of their underlying types, when the bit-fields appear in a union. For
2175 union U @{ int i : 4096; @};
2179 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2180 union too small by the number of bits in an @code{int}.
2183 Empty classes can be placed at incorrect offsets. For example:
2193 struct C : public B, public A @{@};
2197 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2198 it should be placed at offset zero. G++ mistakenly believes that the
2199 @code{A} data member of @code{B} is already at offset zero.
2202 Names of template functions whose types involve @code{typename} or
2203 template template parameters can be mangled incorrectly.
2206 template <typename Q>
2207 void f(typename Q::X) @{@}
2209 template <template <typename> class Q>
2210 void f(typename Q<int>::X) @{@}
2214 Instantiations of these templates may be mangled incorrectly.
2218 It also warns psABI related changes. The known psABI changes at this
2224 For SYSV/x86-64, when passing union with long double, it is changed to
2225 pass in memory as specified in psABI. For example:
2235 @code{union U} will always be passed in memory.
2239 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2240 @opindex Wctor-dtor-privacy
2241 @opindex Wno-ctor-dtor-privacy
2242 Warn when a class seems unusable because all the constructors or
2243 destructors in that class are private, and it has neither friends nor
2244 public static member functions.
2246 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2247 @opindex Wnon-virtual-dtor
2248 @opindex Wno-non-virtual-dtor
2249 Warn when a class has virtual functions and accessible non-virtual
2250 destructor, in which case it would be possible but unsafe to delete
2251 an instance of a derived class through a pointer to the base class.
2252 This warning is also enabled if -Weffc++ is specified.
2254 @item -Wreorder @r{(C++ and Objective-C++ only)}
2256 @opindex Wno-reorder
2257 @cindex reordering, warning
2258 @cindex warning for reordering of member initializers
2259 Warn when the order of member initializers given in the code does not
2260 match the order in which they must be executed. For instance:
2266 A(): j (0), i (1) @{ @}
2270 The compiler will rearrange the member initializers for @samp{i}
2271 and @samp{j} to match the declaration order of the members, emitting
2272 a warning to that effect. This warning is enabled by @option{-Wall}.
2275 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2278 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2281 Warn about violations of the following style guidelines from Scott Meyers'
2282 @cite{Effective C++} book:
2286 Item 11: Define a copy constructor and an assignment operator for classes
2287 with dynamically allocated memory.
2290 Item 12: Prefer initialization to assignment in constructors.
2293 Item 14: Make destructors virtual in base classes.
2296 Item 15: Have @code{operator=} return a reference to @code{*this}.
2299 Item 23: Don't try to return a reference when you must return an object.
2303 Also warn about violations of the following style guidelines from
2304 Scott Meyers' @cite{More Effective C++} book:
2308 Item 6: Distinguish between prefix and postfix forms of increment and
2309 decrement operators.
2312 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2316 When selecting this option, be aware that the standard library
2317 headers do not obey all of these guidelines; use @samp{grep -v}
2318 to filter out those warnings.
2320 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2321 @opindex Wstrict-null-sentinel
2322 @opindex Wno-strict-null-sentinel
2323 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2324 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2325 to @code{__null}. Although it is a null pointer constant not a null pointer,
2326 it is guaranteed to be of the same size as a pointer. But this use is
2327 not portable across different compilers.
2329 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2330 @opindex Wno-non-template-friend
2331 @opindex Wnon-template-friend
2332 Disable warnings when non-templatized friend functions are declared
2333 within a template. Since the advent of explicit template specification
2334 support in G++, if the name of the friend is an unqualified-id (i.e.,
2335 @samp{friend foo(int)}), the C++ language specification demands that the
2336 friend declare or define an ordinary, nontemplate function. (Section
2337 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2338 could be interpreted as a particular specialization of a templatized
2339 function. Because this non-conforming behavior is no longer the default
2340 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2341 check existing code for potential trouble spots and is on by default.
2342 This new compiler behavior can be turned off with
2343 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2344 but disables the helpful warning.
2346 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2347 @opindex Wold-style-cast
2348 @opindex Wno-old-style-cast
2349 Warn if an old-style (C-style) cast to a non-void type is used within
2350 a C++ program. The new-style casts (@samp{dynamic_cast},
2351 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2352 less vulnerable to unintended effects and much easier to search for.
2354 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2355 @opindex Woverloaded-virtual
2356 @opindex Wno-overloaded-virtual
2357 @cindex overloaded virtual fn, warning
2358 @cindex warning for overloaded virtual fn
2359 Warn when a function declaration hides virtual functions from a
2360 base class. For example, in:
2367 struct B: public A @{
2372 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2380 will fail to compile.
2382 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2383 @opindex Wno-pmf-conversions
2384 @opindex Wpmf-conversions
2385 Disable the diagnostic for converting a bound pointer to member function
2388 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2389 @opindex Wsign-promo
2390 @opindex Wno-sign-promo
2391 Warn when overload resolution chooses a promotion from unsigned or
2392 enumerated type to a signed type, over a conversion to an unsigned type of
2393 the same size. Previous versions of G++ would try to preserve
2394 unsignedness, but the standard mandates the current behavior.
2399 A& operator = (int);
2409 In this example, G++ will synthesize a default @samp{A& operator =
2410 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2413 @node Objective-C and Objective-C++ Dialect Options
2414 @section Options Controlling Objective-C and Objective-C++ Dialects
2416 @cindex compiler options, Objective-C and Objective-C++
2417 @cindex Objective-C and Objective-C++ options, command line
2418 @cindex options, Objective-C and Objective-C++
2419 (NOTE: This manual does not describe the Objective-C and Objective-C++
2420 languages themselves. See @xref{Standards,,Language Standards
2421 Supported by GCC}, for references.)
2423 This section describes the command-line options that are only meaningful
2424 for Objective-C and Objective-C++ programs, but you can also use most of
2425 the language-independent GNU compiler options.
2426 For example, you might compile a file @code{some_class.m} like this:
2429 gcc -g -fgnu-runtime -O -c some_class.m
2433 In this example, @option{-fgnu-runtime} is an option meant only for
2434 Objective-C and Objective-C++ programs; you can use the other options with
2435 any language supported by GCC@.
2437 Note that since Objective-C is an extension of the C language, Objective-C
2438 compilations may also use options specific to the C front-end (e.g.,
2439 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2440 C++-specific options (e.g., @option{-Wabi}).
2442 Here is a list of options that are @emph{only} for compiling Objective-C
2443 and Objective-C++ programs:
2446 @item -fconstant-string-class=@var{class-name}
2447 @opindex fconstant-string-class
2448 Use @var{class-name} as the name of the class to instantiate for each
2449 literal string specified with the syntax @code{@@"@dots{}"}. The default
2450 class name is @code{NXConstantString} if the GNU runtime is being used, and
2451 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2452 @option{-fconstant-cfstrings} option, if also present, will override the
2453 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2454 to be laid out as constant CoreFoundation strings.
2457 @opindex fgnu-runtime
2458 Generate object code compatible with the standard GNU Objective-C
2459 runtime. This is the default for most types of systems.
2461 @item -fnext-runtime
2462 @opindex fnext-runtime
2463 Generate output compatible with the NeXT runtime. This is the default
2464 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2465 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2468 @item -fno-nil-receivers
2469 @opindex fno-nil-receivers
2470 Assume that all Objective-C message dispatches (e.g.,
2471 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2472 is not @code{nil}. This allows for more efficient entry points in the runtime
2473 to be used. Currently, this option is only available in conjunction with
2474 the NeXT runtime on Mac OS X 10.3 and later.
2476 @item -fobjc-call-cxx-cdtors
2477 @opindex fobjc-call-cxx-cdtors
2478 For each Objective-C class, check if any of its instance variables is a
2479 C++ object with a non-trivial default constructor. If so, synthesize a
2480 special @code{- (id) .cxx_construct} instance method that will run
2481 non-trivial default constructors on any such instance variables, in order,
2482 and then return @code{self}. Similarly, check if any instance variable
2483 is a C++ object with a non-trivial destructor, and if so, synthesize a
2484 special @code{- (void) .cxx_destruct} method that will run
2485 all such default destructors, in reverse order.
2487 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2488 thusly generated will only operate on instance variables declared in the
2489 current Objective-C class, and not those inherited from superclasses. It
2490 is the responsibility of the Objective-C runtime to invoke all such methods
2491 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2492 will be invoked by the runtime immediately after a new object
2493 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2494 be invoked immediately before the runtime deallocates an object instance.
2496 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2497 support for invoking the @code{- (id) .cxx_construct} and
2498 @code{- (void) .cxx_destruct} methods.
2500 @item -fobjc-direct-dispatch
2501 @opindex fobjc-direct-dispatch
2502 Allow fast jumps to the message dispatcher. On Darwin this is
2503 accomplished via the comm page.
2505 @item -fobjc-exceptions
2506 @opindex fobjc-exceptions
2507 Enable syntactic support for structured exception handling in Objective-C,
2508 similar to what is offered by C++ and Java. This option is
2509 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2518 @@catch (AnObjCClass *exc) @{
2525 @@catch (AnotherClass *exc) @{
2528 @@catch (id allOthers) @{
2538 The @code{@@throw} statement may appear anywhere in an Objective-C or
2539 Objective-C++ program; when used inside of a @code{@@catch} block, the
2540 @code{@@throw} may appear without an argument (as shown above), in which case
2541 the object caught by the @code{@@catch} will be rethrown.
2543 Note that only (pointers to) Objective-C objects may be thrown and
2544 caught using this scheme. When an object is thrown, it will be caught
2545 by the nearest @code{@@catch} clause capable of handling objects of that type,
2546 analogously to how @code{catch} blocks work in C++ and Java. A
2547 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2548 any and all Objective-C exceptions not caught by previous @code{@@catch}
2551 The @code{@@finally} clause, if present, will be executed upon exit from the
2552 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2553 regardless of whether any exceptions are thrown, caught or rethrown
2554 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2555 of the @code{finally} clause in Java.
2557 There are several caveats to using the new exception mechanism:
2561 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2562 idioms provided by the @code{NSException} class, the new
2563 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2564 systems, due to additional functionality needed in the (NeXT) Objective-C
2568 As mentioned above, the new exceptions do not support handling
2569 types other than Objective-C objects. Furthermore, when used from
2570 Objective-C++, the Objective-C exception model does not interoperate with C++
2571 exceptions at this time. This means you cannot @code{@@throw} an exception
2572 from Objective-C and @code{catch} it in C++, or vice versa
2573 (i.e., @code{throw @dots{} @@catch}).
2576 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2577 blocks for thread-safe execution:
2580 @@synchronized (ObjCClass *guard) @{
2585 Upon entering the @code{@@synchronized} block, a thread of execution shall
2586 first check whether a lock has been placed on the corresponding @code{guard}
2587 object by another thread. If it has, the current thread shall wait until
2588 the other thread relinquishes its lock. Once @code{guard} becomes available,
2589 the current thread will place its own lock on it, execute the code contained in
2590 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2591 making @code{guard} available to other threads).
2593 Unlike Java, Objective-C does not allow for entire methods to be marked
2594 @code{@@synchronized}. Note that throwing exceptions out of
2595 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2596 to be unlocked properly.
2600 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2602 @item -freplace-objc-classes
2603 @opindex freplace-objc-classes
2604 Emit a special marker instructing @command{ld(1)} not to statically link in
2605 the resulting object file, and allow @command{dyld(1)} to load it in at
2606 run time instead. This is used in conjunction with the Fix-and-Continue
2607 debugging mode, where the object file in question may be recompiled and
2608 dynamically reloaded in the course of program execution, without the need
2609 to restart the program itself. Currently, Fix-and-Continue functionality
2610 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2615 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2616 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2617 compile time) with static class references that get initialized at load time,
2618 which improves run-time performance. Specifying the @option{-fzero-link} flag
2619 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2620 to be retained. This is useful in Zero-Link debugging mode, since it allows
2621 for individual class implementations to be modified during program execution.
2625 Dump interface declarations for all classes seen in the source file to a
2626 file named @file{@var{sourcename}.decl}.
2628 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2629 @opindex Wassign-intercept
2630 @opindex Wno-assign-intercept
2631 Warn whenever an Objective-C assignment is being intercepted by the
2634 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2635 @opindex Wno-protocol
2637 If a class is declared to implement a protocol, a warning is issued for
2638 every method in the protocol that is not implemented by the class. The
2639 default behavior is to issue a warning for every method not explicitly
2640 implemented in the class, even if a method implementation is inherited
2641 from the superclass. If you use the @option{-Wno-protocol} option, then
2642 methods inherited from the superclass are considered to be implemented,
2643 and no warning is issued for them.
2645 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2647 @opindex Wno-selector
2648 Warn if multiple methods of different types for the same selector are
2649 found during compilation. The check is performed on the list of methods
2650 in the final stage of compilation. Additionally, a check is performed
2651 for each selector appearing in a @code{@@selector(@dots{})}
2652 expression, and a corresponding method for that selector has been found
2653 during compilation. Because these checks scan the method table only at
2654 the end of compilation, these warnings are not produced if the final
2655 stage of compilation is not reached, for example because an error is
2656 found during compilation, or because the @option{-fsyntax-only} option is
2659 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2660 @opindex Wstrict-selector-match
2661 @opindex Wno-strict-selector-match
2662 Warn if multiple methods with differing argument and/or return types are
2663 found for a given selector when attempting to send a message using this
2664 selector to a receiver of type @code{id} or @code{Class}. When this flag
2665 is off (which is the default behavior), the compiler will omit such warnings
2666 if any differences found are confined to types which share the same size
2669 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2670 @opindex Wundeclared-selector
2671 @opindex Wno-undeclared-selector
2672 Warn if a @code{@@selector(@dots{})} expression referring to an
2673 undeclared selector is found. A selector is considered undeclared if no
2674 method with that name has been declared before the
2675 @code{@@selector(@dots{})} expression, either explicitly in an
2676 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2677 an @code{@@implementation} section. This option always performs its
2678 checks as soon as a @code{@@selector(@dots{})} expression is found,
2679 while @option{-Wselector} only performs its checks in the final stage of
2680 compilation. This also enforces the coding style convention
2681 that methods and selectors must be declared before being used.
2683 @item -print-objc-runtime-info
2684 @opindex print-objc-runtime-info
2685 Generate C header describing the largest structure that is passed by
2690 @node Language Independent Options
2691 @section Options to Control Diagnostic Messages Formatting
2692 @cindex options to control diagnostics formatting
2693 @cindex diagnostic messages
2694 @cindex message formatting
2696 Traditionally, diagnostic messages have been formatted irrespective of
2697 the output device's aspect (e.g.@: its width, @dots{}). The options described
2698 below can be used to control the diagnostic messages formatting
2699 algorithm, e.g.@: how many characters per line, how often source location
2700 information should be reported. Right now, only the C++ front end can
2701 honor these options. However it is expected, in the near future, that
2702 the remaining front ends would be able to digest them correctly.
2705 @item -fmessage-length=@var{n}
2706 @opindex fmessage-length
2707 Try to format error messages so that they fit on lines of about @var{n}
2708 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2709 the front ends supported by GCC@. If @var{n} is zero, then no
2710 line-wrapping will be done; each error message will appear on a single
2713 @opindex fdiagnostics-show-location
2714 @item -fdiagnostics-show-location=once
2715 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2716 reporter to emit @emph{once} source location information; that is, in
2717 case the message is too long to fit on a single physical line and has to
2718 be wrapped, the source location won't be emitted (as prefix) again,
2719 over and over, in subsequent continuation lines. This is the default
2722 @item -fdiagnostics-show-location=every-line
2723 Only meaningful in line-wrapping mode. Instructs the diagnostic
2724 messages reporter to emit the same source location information (as
2725 prefix) for physical lines that result from the process of breaking
2726 a message which is too long to fit on a single line.
2728 @item -fdiagnostics-show-option
2729 @opindex fdiagnostics-show-option
2730 This option instructs the diagnostic machinery to add text to each
2731 diagnostic emitted, which indicates which command line option directly
2732 controls that diagnostic, when such an option is known to the
2733 diagnostic machinery.
2735 @item -Wcoverage-mismatch
2736 @opindex Wcoverage-mismatch
2737 Warn if feedback profiles do not match when using the
2738 @option{-fprofile-use} option.
2739 If a source file was changed between @option{-fprofile-gen} and
2740 @option{-fprofile-use}, the files with the profile feedback can fail
2741 to match the source file and GCC can not use the profile feedback
2742 information. By default, GCC emits an error message in this case.
2743 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2744 error. GCC does not use appropriate feedback profiles, so using this
2745 option can result in poorly optimized code. This option is useful
2746 only in the case of very minor changes such as bug fixes to an
2751 @node Warning Options
2752 @section Options to Request or Suppress Warnings
2753 @cindex options to control warnings
2754 @cindex warning messages
2755 @cindex messages, warning
2756 @cindex suppressing warnings
2758 Warnings are diagnostic messages that report constructions which
2759 are not inherently erroneous but which are risky or suggest there
2760 may have been an error.
2762 The following language-independent options do not enable specific
2763 warnings but control the kinds of diagnostics produced by GCC.
2766 @cindex syntax checking
2768 @opindex fsyntax-only
2769 Check the code for syntax errors, but don't do anything beyond that.
2773 Inhibit all warning messages.
2778 Make all warnings into errors.
2783 Make the specified warning into an error. The specifier for a warning
2784 is appended, for example @option{-Werror=switch} turns the warnings
2785 controlled by @option{-Wswitch} into errors. This switch takes a
2786 negative form, to be used to negate @option{-Werror} for specific
2787 warnings, for example @option{-Wno-error=switch} makes
2788 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2789 is in effect. You can use the @option{-fdiagnostics-show-option}
2790 option to have each controllable warning amended with the option which
2791 controls it, to determine what to use with this option.
2793 Note that specifying @option{-Werror=}@var{foo} automatically implies
2794 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2797 @item -Wfatal-errors
2798 @opindex Wfatal-errors
2799 @opindex Wno-fatal-errors
2800 This option causes the compiler to abort compilation on the first error
2801 occurred rather than trying to keep going and printing further error
2806 You can request many specific warnings with options beginning
2807 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2808 implicit declarations. Each of these specific warning options also
2809 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2810 example, @option{-Wno-implicit}. This manual lists only one of the
2811 two forms, whichever is not the default. For further,
2812 language-specific options also refer to @ref{C++ Dialect Options} and
2813 @ref{Objective-C and Objective-C++ Dialect Options}.
2818 Issue all the warnings demanded by strict ISO C and ISO C++;
2819 reject all programs that use forbidden extensions, and some other
2820 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2821 version of the ISO C standard specified by any @option{-std} option used.
2823 Valid ISO C and ISO C++ programs should compile properly with or without
2824 this option (though a rare few will require @option{-ansi} or a
2825 @option{-std} option specifying the required version of ISO C)@. However,
2826 without this option, certain GNU extensions and traditional C and C++
2827 features are supported as well. With this option, they are rejected.
2829 @option{-pedantic} does not cause warning messages for use of the
2830 alternate keywords whose names begin and end with @samp{__}. Pedantic
2831 warnings are also disabled in the expression that follows
2832 @code{__extension__}. However, only system header files should use
2833 these escape routes; application programs should avoid them.
2834 @xref{Alternate Keywords}.
2836 Some users try to use @option{-pedantic} to check programs for strict ISO
2837 C conformance. They soon find that it does not do quite what they want:
2838 it finds some non-ISO practices, but not all---only those for which
2839 ISO C @emph{requires} a diagnostic, and some others for which
2840 diagnostics have been added.
2842 A feature to report any failure to conform to ISO C might be useful in
2843 some instances, but would require considerable additional work and would
2844 be quite different from @option{-pedantic}. We don't have plans to
2845 support such a feature in the near future.
2847 Where the standard specified with @option{-std} represents a GNU
2848 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
2849 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2850 extended dialect is based. Warnings from @option{-pedantic} are given
2851 where they are required by the base standard. (It would not make sense
2852 for such warnings to be given only for features not in the specified GNU
2853 C dialect, since by definition the GNU dialects of C include all
2854 features the compiler supports with the given option, and there would be
2855 nothing to warn about.)
2857 @item -pedantic-errors
2858 @opindex pedantic-errors
2859 Like @option{-pedantic}, except that errors are produced rather than
2865 This enables all the warnings about constructions that some users
2866 consider questionable, and that are easy to avoid (or modify to
2867 prevent the warning), even in conjunction with macros. This also
2868 enables some language-specific warnings described in @ref{C++ Dialect
2869 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2871 @option{-Wall} turns on the following warning flags:
2873 @gccoptlist{-Waddress @gol
2874 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2876 -Wchar-subscripts @gol
2877 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2879 -Wimplicit-function-declaration @gol
2882 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2883 -Wmissing-braces @gol
2889 -Wsequence-point @gol
2890 -Wsign-compare @r{(only in C++)} @gol
2891 -Wstrict-aliasing @gol
2892 -Wstrict-overflow=1 @gol
2895 -Wuninitialized @gol
2896 -Wunknown-pragmas @gol
2897 -Wunused-function @gol
2900 -Wunused-variable @gol
2901 -Wvolatile-register-var @gol
2904 Note that some warning flags are not implied by @option{-Wall}. Some of
2905 them warn about constructions that users generally do not consider
2906 questionable, but which occasionally you might wish to check for;
2907 others warn about constructions that are necessary or hard to avoid in
2908 some cases, and there is no simple way to modify the code to suppress
2909 the warning. Some of them are enabled by @option{-Wextra} but many of
2910 them must be enabled individually.
2916 This enables some extra warning flags that are not enabled by
2917 @option{-Wall}. (This option used to be called @option{-W}. The older
2918 name is still supported, but the newer name is more descriptive.)
2920 @gccoptlist{-Wclobbered @gol
2922 -Wignored-qualifiers @gol
2923 -Wmissing-field-initializers @gol
2924 -Wmissing-parameter-type @r{(C only)} @gol
2925 -Wold-style-declaration @r{(C only)} @gol
2926 -Woverride-init @gol
2929 -Wuninitialized @gol
2930 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2933 The option @option{-Wextra} also prints warning messages for the
2939 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2940 @samp{>}, or @samp{>=}.
2943 (C++ only) An enumerator and a non-enumerator both appear in a
2944 conditional expression.
2947 (C++ only) Ambiguous virtual bases.
2950 (C++ only) Subscripting an array which has been declared @samp{register}.
2953 (C++ only) Taking the address of a variable which has been declared
2957 (C++ only) A base class is not initialized in a derived class' copy
2962 @item -Wchar-subscripts
2963 @opindex Wchar-subscripts
2964 @opindex Wno-char-subscripts
2965 Warn if an array subscript has type @code{char}. This is a common cause
2966 of error, as programmers often forget that this type is signed on some
2968 This warning is enabled by @option{-Wall}.
2972 @opindex Wno-comment
2973 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2974 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2975 This warning is enabled by @option{-Wall}.
2980 @opindex ffreestanding
2981 @opindex fno-builtin
2982 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2983 the arguments supplied have types appropriate to the format string
2984 specified, and that the conversions specified in the format string make
2985 sense. This includes standard functions, and others specified by format
2986 attributes (@pxref{Function Attributes}), in the @code{printf},
2987 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2988 not in the C standard) families (or other target-specific families).
2989 Which functions are checked without format attributes having been
2990 specified depends on the standard version selected, and such checks of
2991 functions without the attribute specified are disabled by
2992 @option{-ffreestanding} or @option{-fno-builtin}.
2994 The formats are checked against the format features supported by GNU
2995 libc version 2.2. These include all ISO C90 and C99 features, as well
2996 as features from the Single Unix Specification and some BSD and GNU
2997 extensions. Other library implementations may not support all these
2998 features; GCC does not support warning about features that go beyond a
2999 particular library's limitations. However, if @option{-pedantic} is used
3000 with @option{-Wformat}, warnings will be given about format features not
3001 in the selected standard version (but not for @code{strfmon} formats,
3002 since those are not in any version of the C standard). @xref{C Dialect
3003 Options,,Options Controlling C Dialect}.
3005 Since @option{-Wformat} also checks for null format arguments for
3006 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3008 @option{-Wformat} is included in @option{-Wall}. For more control over some
3009 aspects of format checking, the options @option{-Wformat-y2k},
3010 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3011 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3012 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
3015 @opindex Wformat-y2k
3016 @opindex Wno-format-y2k
3017 If @option{-Wformat} is specified, also warn about @code{strftime}
3018 formats which may yield only a two-digit year.
3020 @item -Wno-format-contains-nul
3021 @opindex Wno-format-contains-nul
3022 @opindex Wformat-contains-nul
3023 If @option{-Wformat} is specified, do not warn about format strings that
3026 @item -Wno-format-extra-args
3027 @opindex Wno-format-extra-args
3028 @opindex Wformat-extra-args
3029 If @option{-Wformat} is specified, do not warn about excess arguments to a
3030 @code{printf} or @code{scanf} format function. The C standard specifies
3031 that such arguments are ignored.
3033 Where the unused arguments lie between used arguments that are
3034 specified with @samp{$} operand number specifications, normally
3035 warnings are still given, since the implementation could not know what
3036 type to pass to @code{va_arg} to skip the unused arguments. However,
3037 in the case of @code{scanf} formats, this option will suppress the
3038 warning if the unused arguments are all pointers, since the Single
3039 Unix Specification says that such unused arguments are allowed.
3041 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3042 @opindex Wno-format-zero-length
3043 @opindex Wformat-zero-length
3044 If @option{-Wformat} is specified, do not warn about zero-length formats.
3045 The C standard specifies that zero-length formats are allowed.
3047 @item -Wformat-nonliteral
3048 @opindex Wformat-nonliteral
3049 @opindex Wno-format-nonliteral
3050 If @option{-Wformat} is specified, also warn if the format string is not a
3051 string literal and so cannot be checked, unless the format function
3052 takes its format arguments as a @code{va_list}.
3054 @item -Wformat-security
3055 @opindex Wformat-security
3056 @opindex Wno-format-security
3057 If @option{-Wformat} is specified, also warn about uses of format
3058 functions that represent possible security problems. At present, this
3059 warns about calls to @code{printf} and @code{scanf} functions where the
3060 format string is not a string literal and there are no format arguments,
3061 as in @code{printf (foo);}. This may be a security hole if the format
3062 string came from untrusted input and contains @samp{%n}. (This is
3063 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3064 in future warnings may be added to @option{-Wformat-security} that are not
3065 included in @option{-Wformat-nonliteral}.)
3069 @opindex Wno-format=2
3070 Enable @option{-Wformat} plus format checks not included in
3071 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3072 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3074 @item -Wnonnull @r{(C and Objective-C only)}
3076 @opindex Wno-nonnull
3077 Warn about passing a null pointer for arguments marked as
3078 requiring a non-null value by the @code{nonnull} function attribute.
3080 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3081 can be disabled with the @option{-Wno-nonnull} option.
3083 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3085 @opindex Wno-init-self
3086 Warn about uninitialized variables which are initialized with themselves.
3087 Note this option can only be used with the @option{-Wuninitialized} option.
3089 For example, GCC will warn about @code{i} being uninitialized in the
3090 following snippet only when @option{-Winit-self} has been specified:
3101 @item -Wimplicit-int @r{(C and Objective-C only)}
3102 @opindex Wimplicit-int
3103 @opindex Wno-implicit-int
3104 Warn when a declaration does not specify a type.
3105 This warning is enabled by @option{-Wall}.
3107 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3108 @opindex Wimplicit-function-declaration
3109 @opindex Wno-implicit-function-declaration
3110 Give a warning whenever a function is used before being declared. In
3111 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3112 enabled by default and it is made into an error by
3113 @option{-pedantic-errors}. This warning is also enabled by
3118 @opindex Wno-implicit
3119 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3120 This warning is enabled by @option{-Wall}.
3122 @item -Wignored-qualifiers @r{(C and C++ only)}
3123 @opindex Wignored-qualifiers
3124 @opindex Wno-ignored-qualifiers
3125 Warn if the return type of a function has a type qualifier
3126 such as @code{const}. For ISO C such a type qualifier has no effect,
3127 since the value returned by a function is not an lvalue.
3128 For C++, the warning is only emitted for scalar types or @code{void}.
3129 ISO C prohibits qualified @code{void} return types on function
3130 definitions, so such return types always receive a warning
3131 even without this option.
3133 This warning is also enabled by @option{-Wextra}.
3138 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3139 a function with external linkage, returning int, taking either zero
3140 arguments, two, or three arguments of appropriate types. This warning
3141 is enabled by default in C++ and is enabled by either @option{-Wall}
3142 or @option{-pedantic}.
3144 @item -Wmissing-braces
3145 @opindex Wmissing-braces
3146 @opindex Wno-missing-braces
3147 Warn if an aggregate or union initializer is not fully bracketed. In
3148 the following example, the initializer for @samp{a} is not fully
3149 bracketed, but that for @samp{b} is fully bracketed.
3152 int a[2][2] = @{ 0, 1, 2, 3 @};
3153 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3156 This warning is enabled by @option{-Wall}.
3158 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3159 @opindex Wmissing-include-dirs
3160 @opindex Wno-missing-include-dirs
3161 Warn if a user-supplied include directory does not exist.
3164 @opindex Wparentheses
3165 @opindex Wno-parentheses
3166 Warn if parentheses are omitted in certain contexts, such
3167 as when there is an assignment in a context where a truth value
3168 is expected, or when operators are nested whose precedence people
3169 often get confused about.
3171 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3172 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3173 interpretation from that of ordinary mathematical notation.
3175 Also warn about constructions where there may be confusion to which
3176 @code{if} statement an @code{else} branch belongs. Here is an example of
3191 In C/C++, every @code{else} branch belongs to the innermost possible
3192 @code{if} statement, which in this example is @code{if (b)}. This is
3193 often not what the programmer expected, as illustrated in the above
3194 example by indentation the programmer chose. When there is the
3195 potential for this confusion, GCC will issue a warning when this flag
3196 is specified. To eliminate the warning, add explicit braces around
3197 the innermost @code{if} statement so there is no way the @code{else}
3198 could belong to the enclosing @code{if}. The resulting code would
3215 This warning is enabled by @option{-Wall}.
3217 @item -Wsequence-point
3218 @opindex Wsequence-point
3219 @opindex Wno-sequence-point
3220 Warn about code that may have undefined semantics because of violations
3221 of sequence point rules in the C and C++ standards.
3223 The C and C++ standards defines the order in which expressions in a C/C++
3224 program are evaluated in terms of @dfn{sequence points}, which represent
3225 a partial ordering between the execution of parts of the program: those
3226 executed before the sequence point, and those executed after it. These
3227 occur after the evaluation of a full expression (one which is not part
3228 of a larger expression), after the evaluation of the first operand of a
3229 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3230 function is called (but after the evaluation of its arguments and the
3231 expression denoting the called function), and in certain other places.
3232 Other than as expressed by the sequence point rules, the order of
3233 evaluation of subexpressions of an expression is not specified. All
3234 these rules describe only a partial order rather than a total order,
3235 since, for example, if two functions are called within one expression
3236 with no sequence point between them, the order in which the functions
3237 are called is not specified. However, the standards committee have
3238 ruled that function calls do not overlap.
3240 It is not specified when between sequence points modifications to the
3241 values of objects take effect. Programs whose behavior depends on this
3242 have undefined behavior; the C and C++ standards specify that ``Between
3243 the previous and next sequence point an object shall have its stored
3244 value modified at most once by the evaluation of an expression.
3245 Furthermore, the prior value shall be read only to determine the value
3246 to be stored.''. If a program breaks these rules, the results on any
3247 particular implementation are entirely unpredictable.
3249 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3250 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3251 diagnosed by this option, and it may give an occasional false positive
3252 result, but in general it has been found fairly effective at detecting
3253 this sort of problem in programs.
3255 The standard is worded confusingly, therefore there is some debate
3256 over the precise meaning of the sequence point rules in subtle cases.
3257 Links to discussions of the problem, including proposed formal
3258 definitions, may be found on the GCC readings page, at
3259 @w{@uref{http://gcc.gnu.org/readings.html}}.
3261 This warning is enabled by @option{-Wall} for C and C++.
3264 @opindex Wreturn-type
3265 @opindex Wno-return-type
3266 Warn whenever a function is defined with a return-type that defaults
3267 to @code{int}. Also warn about any @code{return} statement with no
3268 return-value in a function whose return-type is not @code{void}
3269 (falling off the end of the function body is considered returning
3270 without a value), and about a @code{return} statement with an
3271 expression in a function whose return-type is @code{void}.
3273 For C++, a function without return type always produces a diagnostic
3274 message, even when @option{-Wno-return-type} is specified. The only
3275 exceptions are @samp{main} and functions defined in system headers.
3277 This warning is enabled by @option{-Wall}.
3282 Warn whenever a @code{switch} statement has an index of enumerated type
3283 and lacks a @code{case} for one or more of the named codes of that
3284 enumeration. (The presence of a @code{default} label prevents this
3285 warning.) @code{case} labels outside the enumeration range also
3286 provoke warnings when this option is used (even if there is a
3287 @code{default} label).
3288 This warning is enabled by @option{-Wall}.
3290 @item -Wswitch-default
3291 @opindex Wswitch-default
3292 @opindex Wno-switch-default
3293 Warn whenever a @code{switch} statement does not have a @code{default}
3297 @opindex Wswitch-enum
3298 @opindex Wno-switch-enum
3299 Warn whenever a @code{switch} statement has an index of enumerated type
3300 and lacks a @code{case} for one or more of the named codes of that
3301 enumeration. @code{case} labels outside the enumeration range also
3302 provoke warnings when this option is used. The only difference
3303 between @option{-Wswitch} and this option is that this option gives a
3304 warning about an omitted enumeration code even if there is a
3305 @code{default} label.
3307 @item -Wsync-nand @r{(C and C++ only)}
3309 @opindex Wno-sync-nand
3310 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3311 built-in functions are used. These functions changed semantics in GCC 4.4.
3315 @opindex Wno-trigraphs
3316 Warn if any trigraphs are encountered that might change the meaning of
3317 the program (trigraphs within comments are not warned about).
3318 This warning is enabled by @option{-Wall}.
3320 @item -Wunused-function
3321 @opindex Wunused-function
3322 @opindex Wno-unused-function
3323 Warn whenever a static function is declared but not defined or a
3324 non-inline static function is unused.
3325 This warning is enabled by @option{-Wall}.
3327 @item -Wunused-label
3328 @opindex Wunused-label
3329 @opindex Wno-unused-label
3330 Warn whenever a label is declared but not used.
3331 This warning is enabled by @option{-Wall}.
3333 To suppress this warning use the @samp{unused} attribute
3334 (@pxref{Variable Attributes}).
3336 @item -Wunused-parameter
3337 @opindex Wunused-parameter
3338 @opindex Wno-unused-parameter
3339 Warn whenever a function parameter is unused aside from its declaration.
3341 To suppress this warning use the @samp{unused} attribute
3342 (@pxref{Variable Attributes}).
3344 @item -Wno-unused-result
3345 @opindex Wunused-result
3346 @opindex Wno-unused-result
3347 Do not warn if a caller of a function marked with attribute
3348 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3349 its return value. The default is @option{-Wunused-result}.
3351 @item -Wunused-variable
3352 @opindex Wunused-variable
3353 @opindex Wno-unused-variable
3354 Warn whenever a local variable or non-constant static variable is unused
3355 aside from its declaration.
3356 This warning is enabled by @option{-Wall}.
3358 To suppress this warning use the @samp{unused} attribute
3359 (@pxref{Variable Attributes}).
3361 @item -Wunused-value
3362 @opindex Wunused-value
3363 @opindex Wno-unused-value
3364 Warn whenever a statement computes a result that is explicitly not
3365 used. To suppress this warning cast the unused expression to
3366 @samp{void}. This includes an expression-statement or the left-hand
3367 side of a comma expression that contains no side effects. For example,
3368 an expression such as @samp{x[i,j]} will cause a warning, while
3369 @samp{x[(void)i,j]} will not.
3371 This warning is enabled by @option{-Wall}.
3376 All the above @option{-Wunused} options combined.
3378 In order to get a warning about an unused function parameter, you must
3379 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3380 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3382 @item -Wuninitialized
3383 @opindex Wuninitialized
3384 @opindex Wno-uninitialized
3385 Warn if an automatic variable is used without first being initialized
3386 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3387 warn if a non-static reference or non-static @samp{const} member
3388 appears in a class without constructors.
3390 If you want to warn about code which uses the uninitialized value of the
3391 variable in its own initializer, use the @option{-Winit-self} option.
3393 These warnings occur for individual uninitialized or clobbered
3394 elements of structure, union or array variables as well as for
3395 variables which are uninitialized or clobbered as a whole. They do
3396 not occur for variables or elements declared @code{volatile}. Because
3397 these warnings depend on optimization, the exact variables or elements
3398 for which there are warnings will depend on the precise optimization
3399 options and version of GCC used.
3401 Note that there may be no warning about a variable that is used only
3402 to compute a value that itself is never used, because such
3403 computations may be deleted by data flow analysis before the warnings
3406 These warnings are made optional because GCC is not smart
3407 enough to see all the reasons why the code might be correct
3408 despite appearing to have an error. Here is one example of how
3429 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3430 always initialized, but GCC doesn't know this. Here is
3431 another common case:
3436 if (change_y) save_y = y, y = new_y;
3438 if (change_y) y = save_y;
3443 This has no bug because @code{save_y} is used only if it is set.
3445 @cindex @code{longjmp} warnings
3446 This option also warns when a non-volatile automatic variable might be
3447 changed by a call to @code{longjmp}. These warnings as well are possible
3448 only in optimizing compilation.
3450 The compiler sees only the calls to @code{setjmp}. It cannot know
3451 where @code{longjmp} will be called; in fact, a signal handler could
3452 call it at any point in the code. As a result, you may get a warning
3453 even when there is in fact no problem because @code{longjmp} cannot
3454 in fact be called at the place which would cause a problem.
3456 Some spurious warnings can be avoided if you declare all the functions
3457 you use that never return as @code{noreturn}. @xref{Function
3460 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3462 @item -Wunknown-pragmas
3463 @opindex Wunknown-pragmas
3464 @opindex Wno-unknown-pragmas
3465 @cindex warning for unknown pragmas
3466 @cindex unknown pragmas, warning
3467 @cindex pragmas, warning of unknown
3468 Warn when a #pragma directive is encountered which is not understood by
3469 GCC@. If this command line option is used, warnings will even be issued
3470 for unknown pragmas in system header files. This is not the case if
3471 the warnings were only enabled by the @option{-Wall} command line option.
3474 @opindex Wno-pragmas
3476 Do not warn about misuses of pragmas, such as incorrect parameters,
3477 invalid syntax, or conflicts between pragmas. See also
3478 @samp{-Wunknown-pragmas}.
3480 @item -Wstrict-aliasing
3481 @opindex Wstrict-aliasing
3482 @opindex Wno-strict-aliasing
3483 This option is only active when @option{-fstrict-aliasing} is active.
3484 It warns about code which might break the strict aliasing rules that the
3485 compiler is using for optimization. The warning does not catch all
3486 cases, but does attempt to catch the more common pitfalls. It is
3487 included in @option{-Wall}.
3488 It is equivalent to @option{-Wstrict-aliasing=3}
3490 @item -Wstrict-aliasing=n
3491 @opindex Wstrict-aliasing=n
3492 @opindex Wno-strict-aliasing=n
3493 This option is only active when @option{-fstrict-aliasing} is active.
3494 It warns about code which might break the strict aliasing rules that the
3495 compiler is using for optimization.
3496 Higher levels correspond to higher accuracy (fewer false positives).
3497 Higher levels also correspond to more effort, similar to the way -O works.
3498 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3501 Level 1: Most aggressive, quick, least accurate.
3502 Possibly useful when higher levels
3503 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3504 false negatives. However, it has many false positives.
3505 Warns for all pointer conversions between possibly incompatible types,
3506 even if never dereferenced. Runs in the frontend only.
3508 Level 2: Aggressive, quick, not too precise.
3509 May still have many false positives (not as many as level 1 though),
3510 and few false negatives (but possibly more than level 1).
3511 Unlike level 1, it only warns when an address is taken. Warns about
3512 incomplete types. Runs in the frontend only.
3514 Level 3 (default for @option{-Wstrict-aliasing}):
3515 Should have very few false positives and few false
3516 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3517 Takes care of the common pun+dereference pattern in the frontend:
3518 @code{*(int*)&some_float}.
3519 If optimization is enabled, it also runs in the backend, where it deals
3520 with multiple statement cases using flow-sensitive points-to information.
3521 Only warns when the converted pointer is dereferenced.
3522 Does not warn about incomplete types.
3524 @item -Wstrict-overflow
3525 @itemx -Wstrict-overflow=@var{n}
3526 @opindex Wstrict-overflow
3527 @opindex Wno-strict-overflow
3528 This option is only active when @option{-fstrict-overflow} is active.
3529 It warns about cases where the compiler optimizes based on the
3530 assumption that signed overflow does not occur. Note that it does not
3531 warn about all cases where the code might overflow: it only warns
3532 about cases where the compiler implements some optimization. Thus
3533 this warning depends on the optimization level.
3535 An optimization which assumes that signed overflow does not occur is
3536 perfectly safe if the values of the variables involved are such that
3537 overflow never does, in fact, occur. Therefore this warning can
3538 easily give a false positive: a warning about code which is not
3539 actually a problem. To help focus on important issues, several
3540 warning levels are defined. No warnings are issued for the use of
3541 undefined signed overflow when estimating how many iterations a loop
3542 will require, in particular when determining whether a loop will be
3546 @item -Wstrict-overflow=1
3547 Warn about cases which are both questionable and easy to avoid. For
3548 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3549 compiler will simplify this to @code{1}. This level of
3550 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3551 are not, and must be explicitly requested.
3553 @item -Wstrict-overflow=2
3554 Also warn about other cases where a comparison is simplified to a
3555 constant. For example: @code{abs (x) >= 0}. This can only be
3556 simplified when @option{-fstrict-overflow} is in effect, because
3557 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3558 zero. @option{-Wstrict-overflow} (with no level) is the same as
3559 @option{-Wstrict-overflow=2}.
3561 @item -Wstrict-overflow=3
3562 Also warn about other cases where a comparison is simplified. For
3563 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3565 @item -Wstrict-overflow=4
3566 Also warn about other simplifications not covered by the above cases.
3567 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3569 @item -Wstrict-overflow=5
3570 Also warn about cases where the compiler reduces the magnitude of a
3571 constant involved in a comparison. For example: @code{x + 2 > y} will
3572 be simplified to @code{x + 1 >= y}. This is reported only at the
3573 highest warning level because this simplification applies to many
3574 comparisons, so this warning level will give a very large number of
3578 @item -Warray-bounds
3579 @opindex Wno-array-bounds
3580 @opindex Warray-bounds
3581 This option is only active when @option{-ftree-vrp} is active
3582 (default for -O2 and above). It warns about subscripts to arrays
3583 that are always out of bounds. This warning is enabled by @option{-Wall}.
3585 @item -Wno-div-by-zero
3586 @opindex Wno-div-by-zero
3587 @opindex Wdiv-by-zero
3588 Do not warn about compile-time integer division by zero. Floating point
3589 division by zero is not warned about, as it can be a legitimate way of
3590 obtaining infinities and NaNs.
3592 @item -Wsystem-headers
3593 @opindex Wsystem-headers
3594 @opindex Wno-system-headers
3595 @cindex warnings from system headers
3596 @cindex system headers, warnings from
3597 Print warning messages for constructs found in system header files.
3598 Warnings from system headers are normally suppressed, on the assumption
3599 that they usually do not indicate real problems and would only make the
3600 compiler output harder to read. Using this command line option tells
3601 GCC to emit warnings from system headers as if they occurred in user
3602 code. However, note that using @option{-Wall} in conjunction with this
3603 option will @emph{not} warn about unknown pragmas in system
3604 headers---for that, @option{-Wunknown-pragmas} must also be used.
3607 @opindex Wfloat-equal
3608 @opindex Wno-float-equal
3609 Warn if floating point values are used in equality comparisons.
3611 The idea behind this is that sometimes it is convenient (for the
3612 programmer) to consider floating-point values as approximations to
3613 infinitely precise real numbers. If you are doing this, then you need
3614 to compute (by analyzing the code, or in some other way) the maximum or
3615 likely maximum error that the computation introduces, and allow for it
3616 when performing comparisons (and when producing output, but that's a
3617 different problem). In particular, instead of testing for equality, you
3618 would check to see whether the two values have ranges that overlap; and
3619 this is done with the relational operators, so equality comparisons are
3622 @item -Wtraditional @r{(C and Objective-C only)}
3623 @opindex Wtraditional
3624 @opindex Wno-traditional
3625 Warn about certain constructs that behave differently in traditional and
3626 ISO C@. Also warn about ISO C constructs that have no traditional C
3627 equivalent, and/or problematic constructs which should be avoided.
3631 Macro parameters that appear within string literals in the macro body.
3632 In traditional C macro replacement takes place within string literals,
3633 but does not in ISO C@.
3636 In traditional C, some preprocessor directives did not exist.
3637 Traditional preprocessors would only consider a line to be a directive
3638 if the @samp{#} appeared in column 1 on the line. Therefore
3639 @option{-Wtraditional} warns about directives that traditional C
3640 understands but would ignore because the @samp{#} does not appear as the
3641 first character on the line. It also suggests you hide directives like
3642 @samp{#pragma} not understood by traditional C by indenting them. Some
3643 traditional implementations would not recognize @samp{#elif}, so it
3644 suggests avoiding it altogether.
3647 A function-like macro that appears without arguments.
3650 The unary plus operator.
3653 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3654 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3655 constants.) Note, these suffixes appear in macros defined in the system
3656 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3657 Use of these macros in user code might normally lead to spurious
3658 warnings, however GCC's integrated preprocessor has enough context to
3659 avoid warning in these cases.
3662 A function declared external in one block and then used after the end of
3666 A @code{switch} statement has an operand of type @code{long}.
3669 A non-@code{static} function declaration follows a @code{static} one.
3670 This construct is not accepted by some traditional C compilers.
3673 The ISO type of an integer constant has a different width or
3674 signedness from its traditional type. This warning is only issued if
3675 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3676 typically represent bit patterns, are not warned about.
3679 Usage of ISO string concatenation is detected.
3682 Initialization of automatic aggregates.
3685 Identifier conflicts with labels. Traditional C lacks a separate
3686 namespace for labels.
3689 Initialization of unions. If the initializer is zero, the warning is
3690 omitted. This is done under the assumption that the zero initializer in
3691 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3692 initializer warnings and relies on default initialization to zero in the
3696 Conversions by prototypes between fixed/floating point values and vice
3697 versa. The absence of these prototypes when compiling with traditional
3698 C would cause serious problems. This is a subset of the possible
3699 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3702 Use of ISO C style function definitions. This warning intentionally is
3703 @emph{not} issued for prototype declarations or variadic functions
3704 because these ISO C features will appear in your code when using
3705 libiberty's traditional C compatibility macros, @code{PARAMS} and
3706 @code{VPARAMS}. This warning is also bypassed for nested functions
3707 because that feature is already a GCC extension and thus not relevant to
3708 traditional C compatibility.
3711 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3712 @opindex Wtraditional-conversion
3713 @opindex Wno-traditional-conversion
3714 Warn if a prototype causes a type conversion that is different from what
3715 would happen to the same argument in the absence of a prototype. This
3716 includes conversions of fixed point to floating and vice versa, and
3717 conversions changing the width or signedness of a fixed point argument
3718 except when the same as the default promotion.
3720 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3721 @opindex Wdeclaration-after-statement
3722 @opindex Wno-declaration-after-statement
3723 Warn when a declaration is found after a statement in a block. This
3724 construct, known from C++, was introduced with ISO C99 and is by default
3725 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3726 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3731 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3733 @item -Wno-endif-labels
3734 @opindex Wno-endif-labels
3735 @opindex Wendif-labels
3736 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3741 Warn whenever a local variable shadows another local variable, parameter or
3742 global variable or whenever a built-in function is shadowed.
3744 @item -Wlarger-than=@var{len}
3745 @opindex Wlarger-than=@var{len}
3746 @opindex Wlarger-than-@var{len}
3747 Warn whenever an object of larger than @var{len} bytes is defined.
3749 @item -Wframe-larger-than=@var{len}
3750 @opindex Wframe-larger-than
3751 Warn if the size of a function frame is larger than @var{len} bytes.
3752 The computation done to determine the stack frame size is approximate
3753 and not conservative.
3754 The actual requirements may be somewhat greater than @var{len}
3755 even if you do not get a warning. In addition, any space allocated
3756 via @code{alloca}, variable-length arrays, or related constructs
3757 is not included by the compiler when determining
3758 whether or not to issue a warning.
3760 @item -Wunsafe-loop-optimizations
3761 @opindex Wunsafe-loop-optimizations
3762 @opindex Wno-unsafe-loop-optimizations
3763 Warn if the loop cannot be optimized because the compiler could not
3764 assume anything on the bounds of the loop indices. With
3765 @option{-funsafe-loop-optimizations} warn if the compiler made
3768 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3769 @opindex Wno-pedantic-ms-format
3770 @opindex Wpedantic-ms-format
3771 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3772 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3773 depending on the MS runtime, when you are using the options @option{-Wformat}
3774 and @option{-pedantic} without gnu-extensions.
3776 @item -Wpointer-arith
3777 @opindex Wpointer-arith
3778 @opindex Wno-pointer-arith
3779 Warn about anything that depends on the ``size of'' a function type or
3780 of @code{void}. GNU C assigns these types a size of 1, for
3781 convenience in calculations with @code{void *} pointers and pointers
3782 to functions. In C++, warn also when an arithmetic operation involves
3783 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3786 @opindex Wtype-limits
3787 @opindex Wno-type-limits
3788 Warn if a comparison is always true or always false due to the limited
3789 range of the data type, but do not warn for constant expressions. For
3790 example, warn if an unsigned variable is compared against zero with
3791 @samp{<} or @samp{>=}. This warning is also enabled by
3794 @item -Wbad-function-cast @r{(C and Objective-C only)}
3795 @opindex Wbad-function-cast
3796 @opindex Wno-bad-function-cast
3797 Warn whenever a function call is cast to a non-matching type.
3798 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3800 @item -Wc++-compat @r{(C and Objective-C only)}
3801 Warn about ISO C constructs that are outside of the common subset of
3802 ISO C and ISO C++, e.g.@: request for implicit conversion from
3803 @code{void *} to a pointer to non-@code{void} type.
3805 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3806 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3807 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3808 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3812 @opindex Wno-cast-qual
3813 Warn whenever a pointer is cast so as to remove a type qualifier from
3814 the target type. For example, warn if a @code{const char *} is cast
3815 to an ordinary @code{char *}.
3817 Also warn when making a cast which introduces a type qualifier in an
3818 unsafe way. For example, casting @code{char **} to @code{const char **}
3819 is unsafe, as in this example:
3822 /* p is char ** value. */
3823 const char **q = (const char **) p;
3824 /* Assignment of readonly string to const char * is OK. */
3826 /* Now char** pointer points to read-only memory. */
3831 @opindex Wcast-align
3832 @opindex Wno-cast-align
3833 Warn whenever a pointer is cast such that the required alignment of the
3834 target is increased. For example, warn if a @code{char *} is cast to
3835 an @code{int *} on machines where integers can only be accessed at
3836 two- or four-byte boundaries.
3838 @item -Wwrite-strings
3839 @opindex Wwrite-strings
3840 @opindex Wno-write-strings
3841 When compiling C, give string constants the type @code{const
3842 char[@var{length}]} so that copying the address of one into a
3843 non-@code{const} @code{char *} pointer will get a warning. These
3844 warnings will help you find at compile time code that can try to write
3845 into a string constant, but only if you have been very careful about
3846 using @code{const} in declarations and prototypes. Otherwise, it will
3847 just be a nuisance. This is why we did not make @option{-Wall} request
3850 When compiling C++, warn about the deprecated conversion from string
3851 literals to @code{char *}. This warning is enabled by default for C++
3856 @opindex Wno-clobbered
3857 Warn for variables that might be changed by @samp{longjmp} or
3858 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3861 @opindex Wconversion
3862 @opindex Wno-conversion
3863 Warn for implicit conversions that may alter a value. This includes
3864 conversions between real and integer, like @code{abs (x)} when
3865 @code{x} is @code{double}; conversions between signed and unsigned,
3866 like @code{unsigned ui = -1}; and conversions to smaller types, like
3867 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3868 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3869 changed by the conversion like in @code{abs (2.0)}. Warnings about
3870 conversions between signed and unsigned integers can be disabled by
3871 using @option{-Wno-sign-conversion}.
3873 For C++, also warn for confusing overload resolution for user-defined
3874 conversions; and conversions that will never use a type conversion
3875 operator: conversions to @code{void}, the same type, a base class or a
3876 reference to them. Warnings about conversions between signed and
3877 unsigned integers are disabled by default in C++ unless
3878 @option{-Wsign-conversion} is explicitly enabled.
3880 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
3881 @opindex Wconversion-null
3882 @opindex Wno-conversion-null
3883 Do not warn for conversions between @code{NULL} and non-pointer
3884 types. @option{-Wconversion-null} is enabled by default.
3887 @opindex Wempty-body
3888 @opindex Wno-empty-body
3889 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3890 while} statement. This warning is also enabled by @option{-Wextra}.
3892 @item -Wenum-compare
3893 @opindex Wenum-compare
3894 @opindex Wno-enum-compare
3895 Warn about a comparison between values of different enum types. In C++
3896 this warning is enabled by default. In C this warning is enabled by
3899 @item -Wjump-misses-init @r{(C, Objective-C only)}
3900 @opindex Wjump-misses-init
3901 @opindex Wno-jump-misses-init
3902 Warn if a @code{goto} statement or a @code{switch} statement jumps
3903 forward across the initialization of a variable, or jumps backward to a
3904 label after the variable has been initialized. This only warns about
3905 variables which are initialized when they are declared. This warning is
3906 only supported for C and Objective C; in C++ this sort of branch is an
3909 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
3910 can be disabled with the @option{-Wno-jump-misses-init} option.
3912 @item -Wsign-compare
3913 @opindex Wsign-compare
3914 @opindex Wno-sign-compare
3915 @cindex warning for comparison of signed and unsigned values
3916 @cindex comparison of signed and unsigned values, warning
3917 @cindex signed and unsigned values, comparison warning
3918 Warn when a comparison between signed and unsigned values could produce
3919 an incorrect result when the signed value is converted to unsigned.
3920 This warning is also enabled by @option{-Wextra}; to get the other warnings
3921 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3923 @item -Wsign-conversion
3924 @opindex Wsign-conversion
3925 @opindex Wno-sign-conversion
3926 Warn for implicit conversions that may change the sign of an integer
3927 value, like assigning a signed integer expression to an unsigned
3928 integer variable. An explicit cast silences the warning. In C, this
3929 option is enabled also by @option{-Wconversion}.
3933 @opindex Wno-address
3934 Warn about suspicious uses of memory addresses. These include using
3935 the address of a function in a conditional expression, such as
3936 @code{void func(void); if (func)}, and comparisons against the memory
3937 address of a string literal, such as @code{if (x == "abc")}. Such
3938 uses typically indicate a programmer error: the address of a function
3939 always evaluates to true, so their use in a conditional usually
3940 indicate that the programmer forgot the parentheses in a function
3941 call; and comparisons against string literals result in unspecified
3942 behavior and are not portable in C, so they usually indicate that the
3943 programmer intended to use @code{strcmp}. This warning is enabled by
3947 @opindex Wlogical-op
3948 @opindex Wno-logical-op
3949 Warn about suspicious uses of logical operators in expressions.
3950 This includes using logical operators in contexts where a
3951 bit-wise operator is likely to be expected.
3953 @item -Waggregate-return
3954 @opindex Waggregate-return
3955 @opindex Wno-aggregate-return
3956 Warn if any functions that return structures or unions are defined or
3957 called. (In languages where you can return an array, this also elicits
3960 @item -Wno-attributes
3961 @opindex Wno-attributes
3962 @opindex Wattributes
3963 Do not warn if an unexpected @code{__attribute__} is used, such as
3964 unrecognized attributes, function attributes applied to variables,
3965 etc. This will not stop errors for incorrect use of supported
3968 @item -Wno-builtin-macro-redefined
3969 @opindex Wno-builtin-macro-redefined
3970 @opindex Wbuiltin-macro-redefined
3971 Do not warn if certain built-in macros are redefined. This suppresses
3972 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3973 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3975 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3976 @opindex Wstrict-prototypes
3977 @opindex Wno-strict-prototypes
3978 Warn if a function is declared or defined without specifying the
3979 argument types. (An old-style function definition is permitted without
3980 a warning if preceded by a declaration which specifies the argument
3983 @item -Wold-style-declaration @r{(C and Objective-C only)}
3984 @opindex Wold-style-declaration
3985 @opindex Wno-old-style-declaration
3986 Warn for obsolescent usages, according to the C Standard, in a
3987 declaration. For example, warn if storage-class specifiers like
3988 @code{static} are not the first things in a declaration. This warning
3989 is also enabled by @option{-Wextra}.
3991 @item -Wold-style-definition @r{(C and Objective-C only)}
3992 @opindex Wold-style-definition
3993 @opindex Wno-old-style-definition
3994 Warn if an old-style function definition is used. A warning is given
3995 even if there is a previous prototype.
3997 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3998 @opindex Wmissing-parameter-type
3999 @opindex Wno-missing-parameter-type
4000 A function parameter is declared without a type specifier in K&R-style
4007 This warning is also enabled by @option{-Wextra}.
4009 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4010 @opindex Wmissing-prototypes
4011 @opindex Wno-missing-prototypes
4012 Warn if a global function is defined without a previous prototype
4013 declaration. This warning is issued even if the definition itself
4014 provides a prototype. The aim is to detect global functions that fail
4015 to be declared in header files.
4017 @item -Wmissing-declarations
4018 @opindex Wmissing-declarations
4019 @opindex Wno-missing-declarations
4020 Warn if a global function is defined without a previous declaration.
4021 Do so even if the definition itself provides a prototype.
4022 Use this option to detect global functions that are not declared in
4023 header files. In C++, no warnings are issued for function templates,
4024 or for inline functions, or for functions in anonymous namespaces.
4026 @item -Wmissing-field-initializers
4027 @opindex Wmissing-field-initializers
4028 @opindex Wno-missing-field-initializers
4032 Warn if a structure's initializer has some fields missing. For
4033 example, the following code would cause such a warning, because
4034 @code{x.h} is implicitly zero:
4037 struct s @{ int f, g, h; @};
4038 struct s x = @{ 3, 4 @};
4041 This option does not warn about designated initializers, so the following
4042 modification would not trigger a warning:
4045 struct s @{ int f, g, h; @};
4046 struct s x = @{ .f = 3, .g = 4 @};
4049 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4050 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4052 @item -Wmissing-noreturn
4053 @opindex Wmissing-noreturn
4054 @opindex Wno-missing-noreturn
4055 Warn about functions which might be candidates for attribute @code{noreturn}.
4056 Note these are only possible candidates, not absolute ones. Care should
4057 be taken to manually verify functions actually do not ever return before
4058 adding the @code{noreturn} attribute, otherwise subtle code generation
4059 bugs could be introduced. You will not get a warning for @code{main} in
4060 hosted C environments.
4062 @item -Wmissing-format-attribute
4063 @opindex Wmissing-format-attribute
4064 @opindex Wno-missing-format-attribute
4067 Warn about function pointers which might be candidates for @code{format}
4068 attributes. Note these are only possible candidates, not absolute ones.
4069 GCC will guess that function pointers with @code{format} attributes that
4070 are used in assignment, initialization, parameter passing or return
4071 statements should have a corresponding @code{format} attribute in the
4072 resulting type. I.e.@: the left-hand side of the assignment or
4073 initialization, the type of the parameter variable, or the return type
4074 of the containing function respectively should also have a @code{format}
4075 attribute to avoid the warning.
4077 GCC will also warn about function definitions which might be
4078 candidates for @code{format} attributes. Again, these are only
4079 possible candidates. GCC will guess that @code{format} attributes
4080 might be appropriate for any function that calls a function like
4081 @code{vprintf} or @code{vscanf}, but this might not always be the
4082 case, and some functions for which @code{format} attributes are
4083 appropriate may not be detected.
4085 @item -Wno-multichar
4086 @opindex Wno-multichar
4088 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4089 Usually they indicate a typo in the user's code, as they have
4090 implementation-defined values, and should not be used in portable code.
4092 @item -Wnormalized=<none|id|nfc|nfkc>
4093 @opindex Wnormalized=
4096 @cindex character set, input normalization
4097 In ISO C and ISO C++, two identifiers are different if they are
4098 different sequences of characters. However, sometimes when characters
4099 outside the basic ASCII character set are used, you can have two
4100 different character sequences that look the same. To avoid confusion,
4101 the ISO 10646 standard sets out some @dfn{normalization rules} which
4102 when applied ensure that two sequences that look the same are turned into
4103 the same sequence. GCC can warn you if you are using identifiers which
4104 have not been normalized; this option controls that warning.
4106 There are four levels of warning that GCC supports. The default is
4107 @option{-Wnormalized=nfc}, which warns about any identifier which is
4108 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4109 recommended form for most uses.
4111 Unfortunately, there are some characters which ISO C and ISO C++ allow
4112 in identifiers that when turned into NFC aren't allowable as
4113 identifiers. That is, there's no way to use these symbols in portable
4114 ISO C or C++ and have all your identifiers in NFC@.
4115 @option{-Wnormalized=id} suppresses the warning for these characters.
4116 It is hoped that future versions of the standards involved will correct
4117 this, which is why this option is not the default.
4119 You can switch the warning off for all characters by writing
4120 @option{-Wnormalized=none}. You would only want to do this if you
4121 were using some other normalization scheme (like ``D''), because
4122 otherwise you can easily create bugs that are literally impossible to see.
4124 Some characters in ISO 10646 have distinct meanings but look identical
4125 in some fonts or display methodologies, especially once formatting has
4126 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4127 LETTER N'', will display just like a regular @code{n} which has been
4128 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4129 normalization scheme to convert all these into a standard form as
4130 well, and GCC will warn if your code is not in NFKC if you use
4131 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4132 about every identifier that contains the letter O because it might be
4133 confused with the digit 0, and so is not the default, but may be
4134 useful as a local coding convention if the programming environment is
4135 unable to be fixed to display these characters distinctly.
4137 @item -Wno-deprecated
4138 @opindex Wno-deprecated
4139 @opindex Wdeprecated
4140 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4142 @item -Wno-deprecated-declarations
4143 @opindex Wno-deprecated-declarations
4144 @opindex Wdeprecated-declarations
4145 Do not warn about uses of functions (@pxref{Function Attributes}),
4146 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4147 Attributes}) marked as deprecated by using the @code{deprecated}
4151 @opindex Wno-overflow
4153 Do not warn about compile-time overflow in constant expressions.
4155 @item -Woverride-init @r{(C and Objective-C only)}
4156 @opindex Woverride-init
4157 @opindex Wno-override-init
4161 Warn if an initialized field without side effects is overridden when
4162 using designated initializers (@pxref{Designated Inits, , Designated
4165 This warning is included in @option{-Wextra}. To get other
4166 @option{-Wextra} warnings without this one, use @samp{-Wextra
4167 -Wno-override-init}.
4172 Warn if a structure is given the packed attribute, but the packed
4173 attribute has no effect on the layout or size of the structure.
4174 Such structures may be mis-aligned for little benefit. For
4175 instance, in this code, the variable @code{f.x} in @code{struct bar}
4176 will be misaligned even though @code{struct bar} does not itself
4177 have the packed attribute:
4184 @} __attribute__((packed));
4192 @item -Wpacked-bitfield-compat
4193 @opindex Wpacked-bitfield-compat
4194 @opindex Wno-packed-bitfield-compat
4195 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4196 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4197 the change can lead to differences in the structure layout. GCC
4198 informs you when the offset of such a field has changed in GCC 4.4.
4199 For example there is no longer a 4-bit padding between field @code{a}
4200 and @code{b} in this structure:
4207 @} __attribute__ ((packed));
4210 This warning is enabled by default. Use
4211 @option{-Wno-packed-bitfield-compat} to disable this warning.
4216 Warn if padding is included in a structure, either to align an element
4217 of the structure or to align the whole structure. Sometimes when this
4218 happens it is possible to rearrange the fields of the structure to
4219 reduce the padding and so make the structure smaller.
4221 @item -Wredundant-decls
4222 @opindex Wredundant-decls
4223 @opindex Wno-redundant-decls
4224 Warn if anything is declared more than once in the same scope, even in
4225 cases where multiple declaration is valid and changes nothing.
4227 @item -Wnested-externs @r{(C and Objective-C only)}
4228 @opindex Wnested-externs
4229 @opindex Wno-nested-externs
4230 Warn if an @code{extern} declaration is encountered within a function.
4235 Warn if a function can not be inlined and it was declared as inline.
4236 Even with this option, the compiler will not warn about failures to
4237 inline functions declared in system headers.
4239 The compiler uses a variety of heuristics to determine whether or not
4240 to inline a function. For example, the compiler takes into account
4241 the size of the function being inlined and the amount of inlining
4242 that has already been done in the current function. Therefore,
4243 seemingly insignificant changes in the source program can cause the
4244 warnings produced by @option{-Winline} to appear or disappear.
4246 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4247 @opindex Wno-invalid-offsetof
4248 @opindex Winvalid-offsetof
4249 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4250 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4251 to a non-POD type is undefined. In existing C++ implementations,
4252 however, @samp{offsetof} typically gives meaningful results even when
4253 applied to certain kinds of non-POD types. (Such as a simple
4254 @samp{struct} that fails to be a POD type only by virtue of having a
4255 constructor.) This flag is for users who are aware that they are
4256 writing nonportable code and who have deliberately chosen to ignore the
4259 The restrictions on @samp{offsetof} may be relaxed in a future version
4260 of the C++ standard.
4262 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4263 @opindex Wno-int-to-pointer-cast
4264 @opindex Wint-to-pointer-cast
4265 Suppress warnings from casts to pointer type of an integer of a
4268 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4269 @opindex Wno-pointer-to-int-cast
4270 @opindex Wpointer-to-int-cast
4271 Suppress warnings from casts from a pointer to an integer type of a
4275 @opindex Winvalid-pch
4276 @opindex Wno-invalid-pch
4277 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4278 the search path but can't be used.
4282 @opindex Wno-long-long
4283 Warn if @samp{long long} type is used. This is enabled by either
4284 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4285 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4287 @item -Wvariadic-macros
4288 @opindex Wvariadic-macros
4289 @opindex Wno-variadic-macros
4290 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4291 alternate syntax when in pedantic ISO C99 mode. This is default.
4292 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4297 Warn if variable length array is used in the code.
4298 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4299 the variable length array.
4301 @item -Wvolatile-register-var
4302 @opindex Wvolatile-register-var
4303 @opindex Wno-volatile-register-var
4304 Warn if a register variable is declared volatile. The volatile
4305 modifier does not inhibit all optimizations that may eliminate reads
4306 and/or writes to register variables. This warning is enabled by
4309 @item -Wdisabled-optimization
4310 @opindex Wdisabled-optimization
4311 @opindex Wno-disabled-optimization
4312 Warn if a requested optimization pass is disabled. This warning does
4313 not generally indicate that there is anything wrong with your code; it
4314 merely indicates that GCC's optimizers were unable to handle the code
4315 effectively. Often, the problem is that your code is too big or too
4316 complex; GCC will refuse to optimize programs when the optimization
4317 itself is likely to take inordinate amounts of time.
4319 @item -Wpointer-sign @r{(C and Objective-C only)}
4320 @opindex Wpointer-sign
4321 @opindex Wno-pointer-sign
4322 Warn for pointer argument passing or assignment with different signedness.
4323 This option is only supported for C and Objective-C@. It is implied by
4324 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4325 @option{-Wno-pointer-sign}.
4327 @item -Wstack-protector
4328 @opindex Wstack-protector
4329 @opindex Wno-stack-protector
4330 This option is only active when @option{-fstack-protector} is active. It
4331 warns about functions that will not be protected against stack smashing.
4334 @opindex Wno-mudflap
4335 Suppress warnings about constructs that cannot be instrumented by
4338 @item -Woverlength-strings
4339 @opindex Woverlength-strings
4340 @opindex Wno-overlength-strings
4341 Warn about string constants which are longer than the ``minimum
4342 maximum'' length specified in the C standard. Modern compilers
4343 generally allow string constants which are much longer than the
4344 standard's minimum limit, but very portable programs should avoid
4345 using longer strings.
4347 The limit applies @emph{after} string constant concatenation, and does
4348 not count the trailing NUL@. In C90, the limit was 509 characters; in
4349 C99, it was raised to 4095. C++98 does not specify a normative
4350 minimum maximum, so we do not diagnose overlength strings in C++@.
4352 This option is implied by @option{-pedantic}, and can be disabled with
4353 @option{-Wno-overlength-strings}.
4355 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4356 @opindex Wunsuffixed-float-constants
4358 GCC will issue a warning for any floating constant that does not have
4359 a suffix. When used together with @option{-Wsystem-headers} it will
4360 warn about such constants in system header files. This can be useful
4361 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4362 from the decimal floating-point extension to C99.
4365 @node Debugging Options
4366 @section Options for Debugging Your Program or GCC
4367 @cindex options, debugging
4368 @cindex debugging information options
4370 GCC has various special options that are used for debugging
4371 either your program or GCC:
4376 Produce debugging information in the operating system's native format
4377 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4380 On most systems that use stabs format, @option{-g} enables use of extra
4381 debugging information that only GDB can use; this extra information
4382 makes debugging work better in GDB but will probably make other debuggers
4384 refuse to read the program. If you want to control for certain whether
4385 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4386 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4388 GCC allows you to use @option{-g} with
4389 @option{-O}. The shortcuts taken by optimized code may occasionally
4390 produce surprising results: some variables you declared may not exist
4391 at all; flow of control may briefly move where you did not expect it;
4392 some statements may not be executed because they compute constant
4393 results or their values were already at hand; some statements may
4394 execute in different places because they were moved out of loops.
4396 Nevertheless it proves possible to debug optimized output. This makes
4397 it reasonable to use the optimizer for programs that might have bugs.
4399 The following options are useful when GCC is generated with the
4400 capability for more than one debugging format.
4404 Produce debugging information for use by GDB@. This means to use the
4405 most expressive format available (DWARF 2, stabs, or the native format
4406 if neither of those are supported), including GDB extensions if at all
4411 Produce debugging information in stabs format (if that is supported),
4412 without GDB extensions. This is the format used by DBX on most BSD
4413 systems. On MIPS, Alpha and System V Release 4 systems this option
4414 produces stabs debugging output which is not understood by DBX or SDB@.
4415 On System V Release 4 systems this option requires the GNU assembler.
4417 @item -feliminate-unused-debug-symbols
4418 @opindex feliminate-unused-debug-symbols
4419 Produce debugging information in stabs format (if that is supported),
4420 for only symbols that are actually used.
4422 @item -femit-class-debug-always
4423 Instead of emitting debugging information for a C++ class in only one
4424 object file, emit it in all object files using the class. This option
4425 should be used only with debuggers that are unable to handle the way GCC
4426 normally emits debugging information for classes because using this
4427 option will increase the size of debugging information by as much as a
4432 Produce debugging information in stabs format (if that is supported),
4433 using GNU extensions understood only by the GNU debugger (GDB)@. The
4434 use of these extensions is likely to make other debuggers crash or
4435 refuse to read the program.
4439 Produce debugging information in COFF format (if that is supported).
4440 This is the format used by SDB on most System V systems prior to
4445 Produce debugging information in XCOFF format (if that is supported).
4446 This is the format used by the DBX debugger on IBM RS/6000 systems.
4450 Produce debugging information in XCOFF format (if that is supported),
4451 using GNU extensions understood only by the GNU debugger (GDB)@. The
4452 use of these extensions is likely to make other debuggers crash or
4453 refuse to read the program, and may cause assemblers other than the GNU
4454 assembler (GAS) to fail with an error.
4456 @item -gdwarf-@var{version}
4457 @opindex gdwarf-@var{version}
4458 Produce debugging information in DWARF format (if that is
4459 supported). This is the format used by DBX on IRIX 6. The value
4460 of @var{version} may be either 2, 3 or 4; the default version is 2.
4462 Note that with DWARF version 2 some ports require, and will always
4463 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4465 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4466 for maximum benefit.
4468 @item -gstrict-dwarf
4469 @opindex gstrict-dwarf
4470 Disallow using extensions of later DWARF standard version than selected
4471 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4472 DWARF extensions from later standard versions is allowed.
4474 @item -gno-strict-dwarf
4475 @opindex gno-strict-dwarf
4476 Allow using extensions of later DWARF standard version than selected with
4477 @option{-gdwarf-@var{version}}.
4481 Produce debugging information in VMS debug format (if that is
4482 supported). This is the format used by DEBUG on VMS systems.
4485 @itemx -ggdb@var{level}
4486 @itemx -gstabs@var{level}
4487 @itemx -gcoff@var{level}
4488 @itemx -gxcoff@var{level}
4489 @itemx -gvms@var{level}
4490 Request debugging information and also use @var{level} to specify how
4491 much information. The default level is 2.
4493 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4496 Level 1 produces minimal information, enough for making backtraces in
4497 parts of the program that you don't plan to debug. This includes
4498 descriptions of functions and external variables, but no information
4499 about local variables and no line numbers.
4501 Level 3 includes extra information, such as all the macro definitions
4502 present in the program. Some debuggers support macro expansion when
4503 you use @option{-g3}.
4505 @option{-gdwarf-2} does not accept a concatenated debug level, because
4506 GCC used to support an option @option{-gdwarf} that meant to generate
4507 debug information in version 1 of the DWARF format (which is very
4508 different from version 2), and it would have been too confusing. That
4509 debug format is long obsolete, but the option cannot be changed now.
4510 Instead use an additional @option{-g@var{level}} option to change the
4511 debug level for DWARF.
4515 Turn off generation of debug info, if leaving out this option would have
4516 generated it, or turn it on at level 2 otherwise. The position of this
4517 argument in the command line does not matter, it takes effect after all
4518 other options are processed, and it does so only once, no matter how
4519 many times it is given. This is mainly intended to be used with
4520 @option{-fcompare-debug}.
4522 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4523 @opindex fdump-final-insns
4524 Dump the final internal representation (RTL) to @var{file}. If the
4525 optional argument is omitted (or if @var{file} is @code{.}), the name
4526 of the dump file will be determined by appending @code{.gkd} to the
4527 compilation output file name.
4529 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4530 @opindex fcompare-debug
4531 @opindex fno-compare-debug
4532 If no error occurs during compilation, run the compiler a second time,
4533 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4534 passed to the second compilation. Dump the final internal
4535 representation in both compilations, and print an error if they differ.
4537 If the equal sign is omitted, the default @option{-gtoggle} is used.
4539 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4540 and nonzero, implicitly enables @option{-fcompare-debug}. If
4541 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4542 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4545 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4546 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4547 of the final representation and the second compilation, preventing even
4548 @env{GCC_COMPARE_DEBUG} from taking effect.
4550 To verify full coverage during @option{-fcompare-debug} testing, set
4551 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4552 which GCC will reject as an invalid option in any actual compilation
4553 (rather than preprocessing, assembly or linking). To get just a
4554 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4555 not overridden} will do.
4557 @item -fcompare-debug-second
4558 @opindex fcompare-debug-second
4559 This option is implicitly passed to the compiler for the second
4560 compilation requested by @option{-fcompare-debug}, along with options to
4561 silence warnings, and omitting other options that would cause
4562 side-effect compiler outputs to files or to the standard output. Dump
4563 files and preserved temporary files are renamed so as to contain the
4564 @code{.gk} additional extension during the second compilation, to avoid
4565 overwriting those generated by the first.
4567 When this option is passed to the compiler driver, it causes the
4568 @emph{first} compilation to be skipped, which makes it useful for little
4569 other than debugging the compiler proper.
4571 @item -feliminate-dwarf2-dups
4572 @opindex feliminate-dwarf2-dups
4573 Compress DWARF2 debugging information by eliminating duplicated
4574 information about each symbol. This option only makes sense when
4575 generating DWARF2 debugging information with @option{-gdwarf-2}.
4577 @item -femit-struct-debug-baseonly
4578 Emit debug information for struct-like types
4579 only when the base name of the compilation source file
4580 matches the base name of file in which the struct was defined.
4582 This option substantially reduces the size of debugging information,
4583 but at significant potential loss in type information to the debugger.
4584 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4585 See @option{-femit-struct-debug-detailed} for more detailed control.
4587 This option works only with DWARF 2.
4589 @item -femit-struct-debug-reduced
4590 Emit debug information for struct-like types
4591 only when the base name of the compilation source file
4592 matches the base name of file in which the type was defined,
4593 unless the struct is a template or defined in a system header.
4595 This option significantly reduces the size of debugging information,
4596 with some potential loss in type information to the debugger.
4597 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4598 See @option{-femit-struct-debug-detailed} for more detailed control.
4600 This option works only with DWARF 2.
4602 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4603 Specify the struct-like types
4604 for which the compiler will generate debug information.
4605 The intent is to reduce duplicate struct debug information
4606 between different object files within the same program.
4608 This option is a detailed version of
4609 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4610 which will serve for most needs.
4612 A specification has the syntax
4613 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4615 The optional first word limits the specification to
4616 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4617 A struct type is used directly when it is the type of a variable, member.
4618 Indirect uses arise through pointers to structs.
4619 That is, when use of an incomplete struct would be legal, the use is indirect.
4621 @samp{struct one direct; struct two * indirect;}.
4623 The optional second word limits the specification to
4624 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4625 Generic structs are a bit complicated to explain.
4626 For C++, these are non-explicit specializations of template classes,
4627 or non-template classes within the above.
4628 Other programming languages have generics,
4629 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4631 The third word specifies the source files for those
4632 structs for which the compiler will emit debug information.
4633 The values @samp{none} and @samp{any} have the normal meaning.
4634 The value @samp{base} means that
4635 the base of name of the file in which the type declaration appears
4636 must match the base of the name of the main compilation file.
4637 In practice, this means that
4638 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4639 but types declared in other header will not.
4640 The value @samp{sys} means those types satisfying @samp{base}
4641 or declared in system or compiler headers.
4643 You may need to experiment to determine the best settings for your application.
4645 The default is @samp{-femit-struct-debug-detailed=all}.
4647 This option works only with DWARF 2.
4649 @item -fenable-icf-debug
4650 @opindex fenable-icf-debug
4651 Generate additional debug information to support identical code folding (ICF).
4652 This option only works with DWARF version 2 or higher.
4654 @item -fno-merge-debug-strings
4655 @opindex fmerge-debug-strings
4656 @opindex fno-merge-debug-strings
4657 Direct the linker to not merge together strings in the debugging
4658 information which are identical in different object files. Merging is
4659 not supported by all assemblers or linkers. Merging decreases the size
4660 of the debug information in the output file at the cost of increasing
4661 link processing time. Merging is enabled by default.
4663 @item -fdebug-prefix-map=@var{old}=@var{new}
4664 @opindex fdebug-prefix-map
4665 When compiling files in directory @file{@var{old}}, record debugging
4666 information describing them as in @file{@var{new}} instead.
4668 @item -fno-dwarf2-cfi-asm
4669 @opindex fdwarf2-cfi-asm
4670 @opindex fno-dwarf2-cfi-asm
4671 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4672 instead of using GAS @code{.cfi_*} directives.
4674 @cindex @command{prof}
4677 Generate extra code to write profile information suitable for the
4678 analysis program @command{prof}. You must use this option when compiling
4679 the source files you want data about, and you must also use it when
4682 @cindex @command{gprof}
4685 Generate extra code to write profile information suitable for the
4686 analysis program @command{gprof}. You must use this option when compiling
4687 the source files you want data about, and you must also use it when
4692 Makes the compiler print out each function name as it is compiled, and
4693 print some statistics about each pass when it finishes.
4696 @opindex ftime-report
4697 Makes the compiler print some statistics about the time consumed by each
4698 pass when it finishes.
4701 @opindex fmem-report
4702 Makes the compiler print some statistics about permanent memory
4703 allocation when it finishes.
4705 @item -fpre-ipa-mem-report
4706 @opindex fpre-ipa-mem-report
4707 @item -fpost-ipa-mem-report
4708 @opindex fpost-ipa-mem-report
4709 Makes the compiler print some statistics about permanent memory
4710 allocation before or after interprocedural optimization.
4712 @item -fprofile-arcs
4713 @opindex fprofile-arcs
4714 Add code so that program flow @dfn{arcs} are instrumented. During
4715 execution the program records how many times each branch and call is
4716 executed and how many times it is taken or returns. When the compiled
4717 program exits it saves this data to a file called
4718 @file{@var{auxname}.gcda} for each source file. The data may be used for
4719 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4720 test coverage analysis (@option{-ftest-coverage}). Each object file's
4721 @var{auxname} is generated from the name of the output file, if
4722 explicitly specified and it is not the final executable, otherwise it is
4723 the basename of the source file. In both cases any suffix is removed
4724 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4725 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4726 @xref{Cross-profiling}.
4728 @cindex @command{gcov}
4732 This option is used to compile and link code instrumented for coverage
4733 analysis. The option is a synonym for @option{-fprofile-arcs}
4734 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4735 linking). See the documentation for those options for more details.
4740 Compile the source files with @option{-fprofile-arcs} plus optimization
4741 and code generation options. For test coverage analysis, use the
4742 additional @option{-ftest-coverage} option. You do not need to profile
4743 every source file in a program.
4746 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4747 (the latter implies the former).
4750 Run the program on a representative workload to generate the arc profile
4751 information. This may be repeated any number of times. You can run
4752 concurrent instances of your program, and provided that the file system
4753 supports locking, the data files will be correctly updated. Also
4754 @code{fork} calls are detected and correctly handled (double counting
4758 For profile-directed optimizations, compile the source files again with
4759 the same optimization and code generation options plus
4760 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4761 Control Optimization}).
4764 For test coverage analysis, use @command{gcov} to produce human readable
4765 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4766 @command{gcov} documentation for further information.
4770 With @option{-fprofile-arcs}, for each function of your program GCC
4771 creates a program flow graph, then finds a spanning tree for the graph.
4772 Only arcs that are not on the spanning tree have to be instrumented: the
4773 compiler adds code to count the number of times that these arcs are
4774 executed. When an arc is the only exit or only entrance to a block, the
4775 instrumentation code can be added to the block; otherwise, a new basic
4776 block must be created to hold the instrumentation code.
4779 @item -ftest-coverage
4780 @opindex ftest-coverage
4781 Produce a notes file that the @command{gcov} code-coverage utility
4782 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4783 show program coverage. Each source file's note file is called
4784 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4785 above for a description of @var{auxname} and instructions on how to
4786 generate test coverage data. Coverage data will match the source files
4787 more closely, if you do not optimize.
4789 @item -fdbg-cnt-list
4790 @opindex fdbg-cnt-list
4791 Print the name and the counter upperbound for all debug counters.
4793 @item -fdbg-cnt=@var{counter-value-list}
4795 Set the internal debug counter upperbound. @var{counter-value-list}
4796 is a comma-separated list of @var{name}:@var{value} pairs
4797 which sets the upperbound of each debug counter @var{name} to @var{value}.
4798 All debug counters have the initial upperbound of @var{UINT_MAX},
4799 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4800 e.g. With -fdbg-cnt=dce:10,tail_call:0
4801 dbg_cnt(dce) will return true only for first 10 invocations
4802 and dbg_cnt(tail_call) will return false always.
4804 @item -d@var{letters}
4805 @itemx -fdump-rtl-@var{pass}
4807 Says to make debugging dumps during compilation at times specified by
4808 @var{letters}. This is used for debugging the RTL-based passes of the
4809 compiler. The file names for most of the dumps are made by appending
4810 a pass number and a word to the @var{dumpname}, and the files are
4811 created in the directory of the output file. @var{dumpname} is
4812 generated from the name of the output file, if explicitly specified
4813 and it is not an executable, otherwise it is the basename of the
4814 source file. These switches may have different effects when
4815 @option{-E} is used for preprocessing.
4817 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4818 @option{-d} option @var{letters}. Here are the possible
4819 letters for use in @var{pass} and @var{letters}, and their meanings:
4823 @item -fdump-rtl-alignments
4824 @opindex fdump-rtl-alignments
4825 Dump after branch alignments have been computed.
4827 @item -fdump-rtl-asmcons
4828 @opindex fdump-rtl-asmcons
4829 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4831 @item -fdump-rtl-auto_inc_dec
4832 @opindex fdump-rtl-auto_inc_dec
4833 Dump after auto-inc-dec discovery. This pass is only run on
4834 architectures that have auto inc or auto dec instructions.
4836 @item -fdump-rtl-barriers
4837 @opindex fdump-rtl-barriers
4838 Dump after cleaning up the barrier instructions.
4840 @item -fdump-rtl-bbpart
4841 @opindex fdump-rtl-bbpart
4842 Dump after partitioning hot and cold basic blocks.
4844 @item -fdump-rtl-bbro
4845 @opindex fdump-rtl-bbro
4846 Dump after block reordering.
4848 @item -fdump-rtl-btl1
4849 @itemx -fdump-rtl-btl2
4850 @opindex fdump-rtl-btl2
4851 @opindex fdump-rtl-btl2
4852 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4853 after the two branch
4854 target load optimization passes.
4856 @item -fdump-rtl-bypass
4857 @opindex fdump-rtl-bypass
4858 Dump after jump bypassing and control flow optimizations.
4860 @item -fdump-rtl-combine
4861 @opindex fdump-rtl-combine
4862 Dump after the RTL instruction combination pass.
4864 @item -fdump-rtl-compgotos
4865 @opindex fdump-rtl-compgotos
4866 Dump after duplicating the computed gotos.
4868 @item -fdump-rtl-ce1
4869 @itemx -fdump-rtl-ce2
4870 @itemx -fdump-rtl-ce3
4871 @opindex fdump-rtl-ce1
4872 @opindex fdump-rtl-ce2
4873 @opindex fdump-rtl-ce3
4874 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4875 @option{-fdump-rtl-ce3} enable dumping after the three
4876 if conversion passes.
4878 @itemx -fdump-rtl-cprop_hardreg
4879 @opindex fdump-rtl-cprop_hardreg
4880 Dump after hard register copy propagation.
4882 @itemx -fdump-rtl-csa
4883 @opindex fdump-rtl-csa
4884 Dump after combining stack adjustments.
4886 @item -fdump-rtl-cse1
4887 @itemx -fdump-rtl-cse2
4888 @opindex fdump-rtl-cse1
4889 @opindex fdump-rtl-cse2
4890 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4891 the two common sub-expression elimination passes.
4893 @itemx -fdump-rtl-dce
4894 @opindex fdump-rtl-dce
4895 Dump after the standalone dead code elimination passes.
4897 @itemx -fdump-rtl-dbr
4898 @opindex fdump-rtl-dbr
4899 Dump after delayed branch scheduling.
4901 @item -fdump-rtl-dce1
4902 @itemx -fdump-rtl-dce2
4903 @opindex fdump-rtl-dce1
4904 @opindex fdump-rtl-dce2
4905 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4906 the two dead store elimination passes.
4909 @opindex fdump-rtl-eh
4910 Dump after finalization of EH handling code.
4912 @item -fdump-rtl-eh_ranges
4913 @opindex fdump-rtl-eh_ranges
4914 Dump after conversion of EH handling range regions.
4916 @item -fdump-rtl-expand
4917 @opindex fdump-rtl-expand
4918 Dump after RTL generation.
4920 @item -fdump-rtl-fwprop1
4921 @itemx -fdump-rtl-fwprop2
4922 @opindex fdump-rtl-fwprop1
4923 @opindex fdump-rtl-fwprop2
4924 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4925 dumping after the two forward propagation passes.
4927 @item -fdump-rtl-gcse1
4928 @itemx -fdump-rtl-gcse2
4929 @opindex fdump-rtl-gcse1
4930 @opindex fdump-rtl-gcse2
4931 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4932 after global common subexpression elimination.
4934 @item -fdump-rtl-init-regs
4935 @opindex fdump-rtl-init-regs
4936 Dump after the initialization of the registers.
4938 @item -fdump-rtl-initvals
4939 @opindex fdump-rtl-initvals
4940 Dump after the computation of the initial value sets.
4942 @itemx -fdump-rtl-into_cfglayout
4943 @opindex fdump-rtl-into_cfglayout
4944 Dump after converting to cfglayout mode.
4946 @item -fdump-rtl-ira
4947 @opindex fdump-rtl-ira
4948 Dump after iterated register allocation.
4950 @item -fdump-rtl-jump
4951 @opindex fdump-rtl-jump
4952 Dump after the second jump optimization.
4954 @item -fdump-rtl-loop2
4955 @opindex fdump-rtl-loop2
4956 @option{-fdump-rtl-loop2} enables dumping after the rtl
4957 loop optimization passes.
4959 @item -fdump-rtl-mach
4960 @opindex fdump-rtl-mach
4961 Dump after performing the machine dependent reorganization pass, if that
4964 @item -fdump-rtl-mode_sw
4965 @opindex fdump-rtl-mode_sw
4966 Dump after removing redundant mode switches.
4968 @item -fdump-rtl-rnreg
4969 @opindex fdump-rtl-rnreg
4970 Dump after register renumbering.
4972 @itemx -fdump-rtl-outof_cfglayout
4973 @opindex fdump-rtl-outof_cfglayout
4974 Dump after converting from cfglayout mode.
4976 @item -fdump-rtl-peephole2
4977 @opindex fdump-rtl-peephole2
4978 Dump after the peephole pass.
4980 @item -fdump-rtl-postreload
4981 @opindex fdump-rtl-postreload
4982 Dump after post-reload optimizations.
4984 @itemx -fdump-rtl-pro_and_epilogue
4985 @opindex fdump-rtl-pro_and_epilogue
4986 Dump after generating the function pro and epilogues.
4988 @item -fdump-rtl-regmove
4989 @opindex fdump-rtl-regmove
4990 Dump after the register move pass.
4992 @item -fdump-rtl-sched1
4993 @itemx -fdump-rtl-sched2
4994 @opindex fdump-rtl-sched1
4995 @opindex fdump-rtl-sched2
4996 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4997 after the basic block scheduling passes.
4999 @item -fdump-rtl-see
5000 @opindex fdump-rtl-see
5001 Dump after sign extension elimination.
5003 @item -fdump-rtl-seqabstr
5004 @opindex fdump-rtl-seqabstr
5005 Dump after common sequence discovery.
5007 @item -fdump-rtl-shorten
5008 @opindex fdump-rtl-shorten
5009 Dump after shortening branches.
5011 @item -fdump-rtl-sibling
5012 @opindex fdump-rtl-sibling
5013 Dump after sibling call optimizations.
5015 @item -fdump-rtl-split1
5016 @itemx -fdump-rtl-split2
5017 @itemx -fdump-rtl-split3
5018 @itemx -fdump-rtl-split4
5019 @itemx -fdump-rtl-split5
5020 @opindex fdump-rtl-split1
5021 @opindex fdump-rtl-split2
5022 @opindex fdump-rtl-split3
5023 @opindex fdump-rtl-split4
5024 @opindex fdump-rtl-split5
5025 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5026 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5027 @option{-fdump-rtl-split5} enable dumping after five rounds of
5028 instruction splitting.
5030 @item -fdump-rtl-sms
5031 @opindex fdump-rtl-sms
5032 Dump after modulo scheduling. This pass is only run on some
5035 @item -fdump-rtl-stack
5036 @opindex fdump-rtl-stack
5037 Dump after conversion from GCC's "flat register file" registers to the
5038 x87's stack-like registers. This pass is only run on x86 variants.
5040 @item -fdump-rtl-subreg1
5041 @itemx -fdump-rtl-subreg2
5042 @opindex fdump-rtl-subreg1
5043 @opindex fdump-rtl-subreg2
5044 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5045 the two subreg expansion passes.
5047 @item -fdump-rtl-unshare
5048 @opindex fdump-rtl-unshare
5049 Dump after all rtl has been unshared.
5051 @item -fdump-rtl-vartrack
5052 @opindex fdump-rtl-vartrack
5053 Dump after variable tracking.
5055 @item -fdump-rtl-vregs
5056 @opindex fdump-rtl-vregs
5057 Dump after converting virtual registers to hard registers.
5059 @item -fdump-rtl-web
5060 @opindex fdump-rtl-web
5061 Dump after live range splitting.
5063 @item -fdump-rtl-regclass
5064 @itemx -fdump-rtl-subregs_of_mode_init
5065 @itemx -fdump-rtl-subregs_of_mode_finish
5066 @itemx -fdump-rtl-dfinit
5067 @itemx -fdump-rtl-dfinish
5068 @opindex fdump-rtl-regclass
5069 @opindex fdump-rtl-subregs_of_mode_init
5070 @opindex fdump-rtl-subregs_of_mode_finish
5071 @opindex fdump-rtl-dfinit
5072 @opindex fdump-rtl-dfinish
5073 These dumps are defined but always produce empty files.
5075 @item -fdump-rtl-all
5076 @opindex fdump-rtl-all
5077 Produce all the dumps listed above.
5081 Annotate the assembler output with miscellaneous debugging information.
5085 Dump all macro definitions, at the end of preprocessing, in addition to
5090 Produce a core dump whenever an error occurs.
5094 Print statistics on memory usage, at the end of the run, to
5099 Annotate the assembler output with a comment indicating which
5100 pattern and alternative was used. The length of each instruction is
5105 Dump the RTL in the assembler output as a comment before each instruction.
5106 Also turns on @option{-dp} annotation.
5110 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5111 dump a representation of the control flow graph suitable for viewing with VCG
5112 to @file{@var{file}.@var{pass}.vcg}.
5116 Just generate RTL for a function instead of compiling it. Usually used
5117 with @option{-fdump-rtl-expand}.
5121 Dump debugging information during parsing, to standard error.
5125 @opindex fdump-noaddr
5126 When doing debugging dumps, suppress address output. This makes it more
5127 feasible to use diff on debugging dumps for compiler invocations with
5128 different compiler binaries and/or different
5129 text / bss / data / heap / stack / dso start locations.
5131 @item -fdump-unnumbered
5132 @opindex fdump-unnumbered
5133 When doing debugging dumps, suppress instruction numbers and address output.
5134 This makes it more feasible to use diff on debugging dumps for compiler
5135 invocations with different options, in particular with and without
5138 @item -fdump-unnumbered-links
5139 @opindex fdump-unnumbered-links
5140 When doing debugging dumps (see @option{-d} option above), suppress
5141 instruction numbers for the links to the previous and next instructions
5144 @item -fdump-translation-unit @r{(C++ only)}
5145 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5146 @opindex fdump-translation-unit
5147 Dump a representation of the tree structure for the entire translation
5148 unit to a file. The file name is made by appending @file{.tu} to the
5149 source file name, and the file is created in the same directory as the
5150 output file. If the @samp{-@var{options}} form is used, @var{options}
5151 controls the details of the dump as described for the
5152 @option{-fdump-tree} options.
5154 @item -fdump-class-hierarchy @r{(C++ only)}
5155 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5156 @opindex fdump-class-hierarchy
5157 Dump a representation of each class's hierarchy and virtual function
5158 table layout to a file. The file name is made by appending
5159 @file{.class} to the source file name, and the file is created in the
5160 same directory as the output file. If the @samp{-@var{options}} form
5161 is used, @var{options} controls the details of the dump as described
5162 for the @option{-fdump-tree} options.
5164 @item -fdump-ipa-@var{switch}
5166 Control the dumping at various stages of inter-procedural analysis
5167 language tree to a file. The file name is generated by appending a
5168 switch specific suffix to the source file name, and the file is created
5169 in the same directory as the output file. The following dumps are
5174 Enables all inter-procedural analysis dumps.
5177 Dumps information about call-graph optimization, unused function removal,
5178 and inlining decisions.
5181 Dump after function inlining.
5185 @item -fdump-statistics-@var{option}
5186 @opindex fdump-statistics
5187 Enable and control dumping of pass statistics in a separate file. The
5188 file name is generated by appending a suffix ending in
5189 @samp{.statistics} to the source file name, and the file is created in
5190 the same directory as the output file. If the @samp{-@var{option}}
5191 form is used, @samp{-stats} will cause counters to be summed over the
5192 whole compilation unit while @samp{-details} will dump every event as
5193 the passes generate them. The default with no option is to sum
5194 counters for each function compiled.
5196 @item -fdump-tree-@var{switch}
5197 @itemx -fdump-tree-@var{switch}-@var{options}
5199 Control the dumping at various stages of processing the intermediate
5200 language tree to a file. The file name is generated by appending a
5201 switch specific suffix to the source file name, and the file is
5202 created in the same directory as the output file. If the
5203 @samp{-@var{options}} form is used, @var{options} is a list of
5204 @samp{-} separated options that control the details of the dump. Not
5205 all options are applicable to all dumps, those which are not
5206 meaningful will be ignored. The following options are available
5210 Print the address of each node. Usually this is not meaningful as it
5211 changes according to the environment and source file. Its primary use
5212 is for tying up a dump file with a debug environment.
5214 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5215 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5216 use working backward from mangled names in the assembly file.
5218 Inhibit dumping of members of a scope or body of a function merely
5219 because that scope has been reached. Only dump such items when they
5220 are directly reachable by some other path. When dumping pretty-printed
5221 trees, this option inhibits dumping the bodies of control structures.
5223 Print a raw representation of the tree. By default, trees are
5224 pretty-printed into a C-like representation.
5226 Enable more detailed dumps (not honored by every dump option).
5228 Enable dumping various statistics about the pass (not honored by every dump
5231 Enable showing basic block boundaries (disabled in raw dumps).
5233 Enable showing virtual operands for every statement.
5235 Enable showing line numbers for statements.
5237 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5239 Enable showing the tree dump for each statement.
5241 Enable showing the EH region number holding each statement.
5243 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5244 and @option{lineno}.
5247 The following tree dumps are possible:
5251 @opindex fdump-tree-original
5252 Dump before any tree based optimization, to @file{@var{file}.original}.
5255 @opindex fdump-tree-optimized
5256 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5259 @opindex fdump-tree-gimple
5260 Dump each function before and after the gimplification pass to a file. The
5261 file name is made by appending @file{.gimple} to the source file name.
5264 @opindex fdump-tree-cfg
5265 Dump the control flow graph of each function to a file. The file name is
5266 made by appending @file{.cfg} to the source file name.
5269 @opindex fdump-tree-vcg
5270 Dump the control flow graph of each function to a file in VCG format. The
5271 file name is made by appending @file{.vcg} to the source file name. Note
5272 that if the file contains more than one function, the generated file cannot
5273 be used directly by VCG@. You will need to cut and paste each function's
5274 graph into its own separate file first.
5277 @opindex fdump-tree-ch
5278 Dump each function after copying loop headers. The file name is made by
5279 appending @file{.ch} to the source file name.
5282 @opindex fdump-tree-ssa
5283 Dump SSA related information to a file. The file name is made by appending
5284 @file{.ssa} to the source file name.
5287 @opindex fdump-tree-alias
5288 Dump aliasing information for each function. The file name is made by
5289 appending @file{.alias} to the source file name.
5292 @opindex fdump-tree-ccp
5293 Dump each function after CCP@. The file name is made by appending
5294 @file{.ccp} to the source file name.
5297 @opindex fdump-tree-storeccp
5298 Dump each function after STORE-CCP@. The file name is made by appending
5299 @file{.storeccp} to the source file name.
5302 @opindex fdump-tree-pre
5303 Dump trees after partial redundancy elimination. The file name is made
5304 by appending @file{.pre} to the source file name.
5307 @opindex fdump-tree-fre
5308 Dump trees after full redundancy elimination. The file name is made
5309 by appending @file{.fre} to the source file name.
5312 @opindex fdump-tree-copyprop
5313 Dump trees after copy propagation. The file name is made
5314 by appending @file{.copyprop} to the source file name.
5316 @item store_copyprop
5317 @opindex fdump-tree-store_copyprop
5318 Dump trees after store copy-propagation. The file name is made
5319 by appending @file{.store_copyprop} to the source file name.
5322 @opindex fdump-tree-dce
5323 Dump each function after dead code elimination. The file name is made by
5324 appending @file{.dce} to the source file name.
5327 @opindex fdump-tree-mudflap
5328 Dump each function after adding mudflap instrumentation. The file name is
5329 made by appending @file{.mudflap} to the source file name.
5332 @opindex fdump-tree-sra
5333 Dump each function after performing scalar replacement of aggregates. The
5334 file name is made by appending @file{.sra} to the source file name.
5337 @opindex fdump-tree-sink
5338 Dump each function after performing code sinking. The file name is made
5339 by appending @file{.sink} to the source file name.
5342 @opindex fdump-tree-dom
5343 Dump each function after applying dominator tree optimizations. The file
5344 name is made by appending @file{.dom} to the source file name.
5347 @opindex fdump-tree-dse
5348 Dump each function after applying dead store elimination. The file
5349 name is made by appending @file{.dse} to the source file name.
5352 @opindex fdump-tree-phiopt
5353 Dump each function after optimizing PHI nodes into straightline code. The file
5354 name is made by appending @file{.phiopt} to the source file name.
5357 @opindex fdump-tree-forwprop
5358 Dump each function after forward propagating single use variables. The file
5359 name is made by appending @file{.forwprop} to the source file name.
5362 @opindex fdump-tree-copyrename
5363 Dump each function after applying the copy rename optimization. The file
5364 name is made by appending @file{.copyrename} to the source file name.
5367 @opindex fdump-tree-nrv
5368 Dump each function after applying the named return value optimization on
5369 generic trees. The file name is made by appending @file{.nrv} to the source
5373 @opindex fdump-tree-vect
5374 Dump each function after applying vectorization of loops. The file name is
5375 made by appending @file{.vect} to the source file name.
5378 @opindex fdump-tree-slp
5379 Dump each function after applying vectorization of basic blocks. The file name
5380 is made by appending @file{.slp} to the source file name.
5383 @opindex fdump-tree-vrp
5384 Dump each function after Value Range Propagation (VRP). The file name
5385 is made by appending @file{.vrp} to the source file name.
5388 @opindex fdump-tree-all
5389 Enable all the available tree dumps with the flags provided in this option.
5392 @item -ftree-vectorizer-verbose=@var{n}
5393 @opindex ftree-vectorizer-verbose
5394 This option controls the amount of debugging output the vectorizer prints.
5395 This information is written to standard error, unless
5396 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5397 in which case it is output to the usual dump listing file, @file{.vect}.
5398 For @var{n}=0 no diagnostic information is reported.
5399 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5400 and the total number of loops that got vectorized.
5401 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5402 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5403 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5404 level that @option{-fdump-tree-vect-stats} uses.
5405 Higher verbosity levels mean either more information dumped for each
5406 reported loop, or same amount of information reported for more loops:
5407 if @var{n}=3, vectorizer cost model information is reported.
5408 If @var{n}=4, alignment related information is added to the reports.
5409 If @var{n}=5, data-references related information (e.g.@: memory dependences,
5410 memory access-patterns) is added to the reports.
5411 If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5412 that did not pass the first analysis phase (i.e., may not be countable, or
5413 may have complicated control-flow).
5414 If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5415 If @var{n}=8, SLP related information is added to the reports.
5416 For @var{n}=9, all the information the vectorizer generates during its
5417 analysis and transformation is reported. This is the same verbosity level
5418 that @option{-fdump-tree-vect-details} uses.
5420 @item -frandom-seed=@var{string}
5421 @opindex frandom-seed
5422 This option provides a seed that GCC uses when it would otherwise use
5423 random numbers. It is used to generate certain symbol names
5424 that have to be different in every compiled file. It is also used to
5425 place unique stamps in coverage data files and the object files that
5426 produce them. You can use the @option{-frandom-seed} option to produce
5427 reproducibly identical object files.
5429 The @var{string} should be different for every file you compile.
5431 @item -fsched-verbose=@var{n}
5432 @opindex fsched-verbose
5433 On targets that use instruction scheduling, this option controls the
5434 amount of debugging output the scheduler prints. This information is
5435 written to standard error, unless @option{-fdump-rtl-sched1} or
5436 @option{-fdump-rtl-sched2} is specified, in which case it is output
5437 to the usual dump listing file, @file{.sched1} or @file{.sched2}
5438 respectively. However for @var{n} greater than nine, the output is
5439 always printed to standard error.
5441 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5442 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5443 For @var{n} greater than one, it also output basic block probabilities,
5444 detailed ready list information and unit/insn info. For @var{n} greater
5445 than two, it includes RTL at abort point, control-flow and regions info.
5446 And for @var{n} over four, @option{-fsched-verbose} also includes
5450 @itemx -save-temps=cwd
5452 Store the usual ``temporary'' intermediate files permanently; place them
5453 in the current directory and name them based on the source file. Thus,
5454 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5455 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5456 preprocessed @file{foo.i} output file even though the compiler now
5457 normally uses an integrated preprocessor.
5459 When used in combination with the @option{-x} command line option,
5460 @option{-save-temps} is sensible enough to avoid over writing an
5461 input source file with the same extension as an intermediate file.
5462 The corresponding intermediate file may be obtained by renaming the
5463 source file before using @option{-save-temps}.
5465 If you invoke GCC in parallel, compiling several different source
5466 files that share a common base name in different subdirectories or the
5467 same source file compiled for multiple output destinations, it is
5468 likely that the different parallel compilers will interfere with each
5469 other, and overwrite the temporary files. For instance:
5472 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5473 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5476 may result in @file{foo.i} and @file{foo.o} being written to
5477 simultaneously by both compilers.
5479 @item -save-temps=obj
5480 @opindex save-temps=obj
5481 Store the usual ``temporary'' intermediate files permanently. If the
5482 @option{-o} option is used, the temporary files are based on the
5483 object file. If the @option{-o} option is not used, the
5484 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5489 gcc -save-temps=obj -c foo.c
5490 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5491 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5494 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5495 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5496 @file{dir2/yfoobar.o}.
5498 @item -time@r{[}=@var{file}@r{]}
5500 Report the CPU time taken by each subprocess in the compilation
5501 sequence. For C source files, this is the compiler proper and assembler
5502 (plus the linker if linking is done).
5504 Without the specification of an output file, the output looks like this:
5511 The first number on each line is the ``user time'', that is time spent
5512 executing the program itself. The second number is ``system time'',
5513 time spent executing operating system routines on behalf of the program.
5514 Both numbers are in seconds.
5516 With the specification of an output file, the output is appended to the
5517 named file, and it looks like this:
5520 0.12 0.01 cc1 @var{options}
5521 0.00 0.01 as @var{options}
5524 The ``user time'' and the ``system time'' are moved before the program
5525 name, and the options passed to the program are displayed, so that one
5526 can later tell what file was being compiled, and with which options.
5528 @item -fvar-tracking
5529 @opindex fvar-tracking
5530 Run variable tracking pass. It computes where variables are stored at each
5531 position in code. Better debugging information is then generated
5532 (if the debugging information format supports this information).
5534 It is enabled by default when compiling with optimization (@option{-Os},
5535 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5536 the debug info format supports it.
5538 @item -fvar-tracking-assignments
5539 @opindex fvar-tracking-assignments
5540 @opindex fno-var-tracking-assignments
5541 Annotate assignments to user variables early in the compilation and
5542 attempt to carry the annotations over throughout the compilation all the
5543 way to the end, in an attempt to improve debug information while
5544 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5546 It can be enabled even if var-tracking is disabled, in which case
5547 annotations will be created and maintained, but discarded at the end.
5549 @item -fvar-tracking-assignments-toggle
5550 @opindex fvar-tracking-assignments-toggle
5551 @opindex fno-var-tracking-assignments-toggle
5552 Toggle @option{-fvar-tracking-assignments}, in the same way that
5553 @option{-gtoggle} toggles @option{-g}.
5555 @item -print-file-name=@var{library}
5556 @opindex print-file-name
5557 Print the full absolute name of the library file @var{library} that
5558 would be used when linking---and don't do anything else. With this
5559 option, GCC does not compile or link anything; it just prints the
5562 @item -print-multi-directory
5563 @opindex print-multi-directory
5564 Print the directory name corresponding to the multilib selected by any
5565 other switches present in the command line. This directory is supposed
5566 to exist in @env{GCC_EXEC_PREFIX}.
5568 @item -print-multi-lib
5569 @opindex print-multi-lib
5570 Print the mapping from multilib directory names to compiler switches
5571 that enable them. The directory name is separated from the switches by
5572 @samp{;}, and each switch starts with an @samp{@@} instead of the
5573 @samp{-}, without spaces between multiple switches. This is supposed to
5574 ease shell-processing.
5576 @item -print-multi-os-directory
5577 @opindex print-multi-os-directory
5578 Print the path to OS libraries for the selected
5579 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5580 present in the @file{lib} subdirectory and no multilibs are used, this is
5581 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5582 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5583 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5584 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5586 @item -print-prog-name=@var{program}
5587 @opindex print-prog-name
5588 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5590 @item -print-libgcc-file-name
5591 @opindex print-libgcc-file-name
5592 Same as @option{-print-file-name=libgcc.a}.
5594 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5595 but you do want to link with @file{libgcc.a}. You can do
5598 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5601 @item -print-search-dirs
5602 @opindex print-search-dirs
5603 Print the name of the configured installation directory and a list of
5604 program and library directories @command{gcc} will search---and don't do anything else.
5606 This is useful when @command{gcc} prints the error message
5607 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5608 To resolve this you either need to put @file{cpp0} and the other compiler
5609 components where @command{gcc} expects to find them, or you can set the environment
5610 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5611 Don't forget the trailing @samp{/}.
5612 @xref{Environment Variables}.
5614 @item -print-sysroot
5615 @opindex print-sysroot
5616 Print the target sysroot directory that will be used during
5617 compilation. This is the target sysroot specified either at configure
5618 time or using the @option{--sysroot} option, possibly with an extra
5619 suffix that depends on compilation options. If no target sysroot is
5620 specified, the option prints nothing.
5622 @item -print-sysroot-headers-suffix
5623 @opindex print-sysroot-headers-suffix
5624 Print the suffix added to the target sysroot when searching for
5625 headers, or give an error if the compiler is not configured with such
5626 a suffix---and don't do anything else.
5629 @opindex dumpmachine
5630 Print the compiler's target machine (for example,
5631 @samp{i686-pc-linux-gnu})---and don't do anything else.
5634 @opindex dumpversion
5635 Print the compiler version (for example, @samp{3.0})---and don't do
5640 Print the compiler's built-in specs---and don't do anything else. (This
5641 is used when GCC itself is being built.) @xref{Spec Files}.
5643 @item -feliminate-unused-debug-types
5644 @opindex feliminate-unused-debug-types
5645 Normally, when producing DWARF2 output, GCC will emit debugging
5646 information for all types declared in a compilation
5647 unit, regardless of whether or not they are actually used
5648 in that compilation unit. Sometimes this is useful, such as
5649 if, in the debugger, you want to cast a value to a type that is
5650 not actually used in your program (but is declared). More often,
5651 however, this results in a significant amount of wasted space.
5652 With this option, GCC will avoid producing debug symbol output
5653 for types that are nowhere used in the source file being compiled.
5656 @node Optimize Options
5657 @section Options That Control Optimization
5658 @cindex optimize options
5659 @cindex options, optimization
5661 These options control various sorts of optimizations.
5663 Without any optimization option, the compiler's goal is to reduce the
5664 cost of compilation and to make debugging produce the expected
5665 results. Statements are independent: if you stop the program with a
5666 breakpoint between statements, you can then assign a new value to any
5667 variable or change the program counter to any other statement in the
5668 function and get exactly the results you would expect from the source
5671 Turning on optimization flags makes the compiler attempt to improve
5672 the performance and/or code size at the expense of compilation time
5673 and possibly the ability to debug the program.
5675 The compiler performs optimization based on the knowledge it has of the
5676 program. Compiling multiple files at once to a single output file mode allows
5677 the compiler to use information gained from all of the files when compiling
5680 Not all optimizations are controlled directly by a flag. Only
5681 optimizations that have a flag are listed in this section.
5683 Most optimizations are only enabled if an @option{-O} level is set on
5684 the command line. Otherwise they are disabled, even if individual
5685 optimization flags are specified.
5687 Depending on the target and how GCC was configured, a slightly different
5688 set of optimizations may be enabled at each @option{-O} level than
5689 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5690 to find out the exact set of optimizations that are enabled at each level.
5691 @xref{Overall Options}, for examples.
5698 Optimize. Optimizing compilation takes somewhat more time, and a lot
5699 more memory for a large function.
5701 With @option{-O}, the compiler tries to reduce code size and execution
5702 time, without performing any optimizations that take a great deal of
5705 @option{-O} turns on the following optimization flags:
5708 -fcprop-registers @gol
5711 -fdelayed-branch @gol
5713 -fguess-branch-probability @gol
5714 -fif-conversion2 @gol
5715 -fif-conversion @gol
5716 -fipa-pure-const @gol
5717 -fipa-reference @gol
5719 -fsplit-wide-types @gol
5720 -ftree-builtin-call-dce @gol
5723 -ftree-copyrename @gol
5725 -ftree-dominator-opts @gol
5727 -ftree-forwprop @gol
5735 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5736 where doing so does not interfere with debugging.
5740 Optimize even more. GCC performs nearly all supported optimizations
5741 that do not involve a space-speed tradeoff.
5742 As compared to @option{-O}, this option increases both compilation time
5743 and the performance of the generated code.
5745 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5746 also turns on the following optimization flags:
5747 @gccoptlist{-fthread-jumps @gol
5748 -falign-functions -falign-jumps @gol
5749 -falign-loops -falign-labels @gol
5752 -fcse-follow-jumps -fcse-skip-blocks @gol
5753 -fdelete-null-pointer-checks @gol
5754 -fexpensive-optimizations @gol
5755 -fgcse -fgcse-lm @gol
5756 -finline-small-functions @gol
5757 -findirect-inlining @gol
5759 -foptimize-sibling-calls @gol
5762 -freorder-blocks -freorder-functions @gol
5763 -frerun-cse-after-loop @gol
5764 -fsched-interblock -fsched-spec @gol
5765 -fschedule-insns -fschedule-insns2 @gol
5766 -fstrict-aliasing -fstrict-overflow @gol
5767 -ftree-switch-conversion @gol
5771 Please note the warning under @option{-fgcse} about
5772 invoking @option{-O2} on programs that use computed gotos.
5776 Optimize yet more. @option{-O3} turns on all optimizations specified
5777 by @option{-O2} and also turns on the @option{-finline-functions},
5778 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5779 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5783 Reduce compilation time and make debugging produce the expected
5784 results. This is the default.
5788 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5789 do not typically increase code size. It also performs further
5790 optimizations designed to reduce code size.
5792 @option{-Os} disables the following optimization flags:
5793 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5794 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5795 -fprefetch-loop-arrays -ftree-vect-loop-version}
5797 If you use multiple @option{-O} options, with or without level numbers,
5798 the last such option is the one that is effective.
5801 Options of the form @option{-f@var{flag}} specify machine-independent
5802 flags. Most flags have both positive and negative forms; the negative
5803 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5804 below, only one of the forms is listed---the one you typically will
5805 use. You can figure out the other form by either removing @samp{no-}
5808 The following options control specific optimizations. They are either
5809 activated by @option{-O} options or are related to ones that are. You
5810 can use the following flags in the rare cases when ``fine-tuning'' of
5811 optimizations to be performed is desired.
5814 @item -fno-default-inline
5815 @opindex fno-default-inline
5816 Do not make member functions inline by default merely because they are
5817 defined inside the class scope (C++ only). Otherwise, when you specify
5818 @w{@option{-O}}, member functions defined inside class scope are compiled
5819 inline by default; i.e., you don't need to add @samp{inline} in front of
5820 the member function name.
5822 @item -fno-defer-pop
5823 @opindex fno-defer-pop
5824 Always pop the arguments to each function call as soon as that function
5825 returns. For machines which must pop arguments after a function call,
5826 the compiler normally lets arguments accumulate on the stack for several
5827 function calls and pops them all at once.
5829 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5831 @item -fforward-propagate
5832 @opindex fforward-propagate
5833 Perform a forward propagation pass on RTL@. The pass tries to combine two
5834 instructions and checks if the result can be simplified. If loop unrolling
5835 is active, two passes are performed and the second is scheduled after
5838 This option is enabled by default at optimization levels @option{-O},
5839 @option{-O2}, @option{-O3}, @option{-Os}.
5841 @item -fomit-frame-pointer
5842 @opindex fomit-frame-pointer
5843 Don't keep the frame pointer in a register for functions that
5844 don't need one. This avoids the instructions to save, set up and
5845 restore frame pointers; it also makes an extra register available
5846 in many functions. @strong{It also makes debugging impossible on
5849 On some machines, such as the VAX, this flag has no effect, because
5850 the standard calling sequence automatically handles the frame pointer
5851 and nothing is saved by pretending it doesn't exist. The
5852 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5853 whether a target machine supports this flag. @xref{Registers,,Register
5854 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5856 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5858 @item -foptimize-sibling-calls
5859 @opindex foptimize-sibling-calls
5860 Optimize sibling and tail recursive calls.
5862 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5866 Don't pay attention to the @code{inline} keyword. Normally this option
5867 is used to keep the compiler from expanding any functions inline.
5868 Note that if you are not optimizing, no functions can be expanded inline.
5870 @item -finline-small-functions
5871 @opindex finline-small-functions
5872 Integrate functions into their callers when their body is smaller than expected
5873 function call code (so overall size of program gets smaller). The compiler
5874 heuristically decides which functions are simple enough to be worth integrating
5877 Enabled at level @option{-O2}.
5879 @item -findirect-inlining
5880 @opindex findirect-inlining
5881 Inline also indirect calls that are discovered to be known at compile
5882 time thanks to previous inlining. This option has any effect only
5883 when inlining itself is turned on by the @option{-finline-functions}
5884 or @option{-finline-small-functions} options.
5886 Enabled at level @option{-O2}.
5888 @item -finline-functions
5889 @opindex finline-functions
5890 Integrate all simple functions into their callers. The compiler
5891 heuristically decides which functions are simple enough to be worth
5892 integrating in this way.
5894 If all calls to a given function are integrated, and the function is
5895 declared @code{static}, then the function is normally not output as
5896 assembler code in its own right.
5898 Enabled at level @option{-O3}.
5900 @item -finline-functions-called-once
5901 @opindex finline-functions-called-once
5902 Consider all @code{static} functions called once for inlining into their
5903 caller even if they are not marked @code{inline}. If a call to a given
5904 function is integrated, then the function is not output as assembler code
5907 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5909 @item -fearly-inlining
5910 @opindex fearly-inlining
5911 Inline functions marked by @code{always_inline} and functions whose body seems
5912 smaller than the function call overhead early before doing
5913 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5914 makes profiling significantly cheaper and usually inlining faster on programs
5915 having large chains of nested wrapper functions.
5921 Perform interprocedural scalar replacement of aggregates, removal of
5922 unused parameters and replacement of parameters passed by reference
5923 by parameters passed by value.
5925 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
5927 @item -finline-limit=@var{n}
5928 @opindex finline-limit
5929 By default, GCC limits the size of functions that can be inlined. This flag
5930 allows coarse control of this limit. @var{n} is the size of functions that
5931 can be inlined in number of pseudo instructions.
5933 Inlining is actually controlled by a number of parameters, which may be
5934 specified individually by using @option{--param @var{name}=@var{value}}.
5935 The @option{-finline-limit=@var{n}} option sets some of these parameters
5939 @item max-inline-insns-single
5940 is set to @var{n}/2.
5941 @item max-inline-insns-auto
5942 is set to @var{n}/2.
5945 See below for a documentation of the individual
5946 parameters controlling inlining and for the defaults of these parameters.
5948 @emph{Note:} there may be no value to @option{-finline-limit} that results
5949 in default behavior.
5951 @emph{Note:} pseudo instruction represents, in this particular context, an
5952 abstract measurement of function's size. In no way does it represent a count
5953 of assembly instructions and as such its exact meaning might change from one
5954 release to an another.
5956 @item -fkeep-inline-functions
5957 @opindex fkeep-inline-functions
5958 In C, emit @code{static} functions that are declared @code{inline}
5959 into the object file, even if the function has been inlined into all
5960 of its callers. This switch does not affect functions using the
5961 @code{extern inline} extension in GNU C90@. In C++, emit any and all
5962 inline functions into the object file.
5964 @item -fkeep-static-consts
5965 @opindex fkeep-static-consts
5966 Emit variables declared @code{static const} when optimization isn't turned
5967 on, even if the variables aren't referenced.
5969 GCC enables this option by default. If you want to force the compiler to
5970 check if the variable was referenced, regardless of whether or not
5971 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5973 @item -fmerge-constants
5974 @opindex fmerge-constants
5975 Attempt to merge identical constants (string constants and floating point
5976 constants) across compilation units.
5978 This option is the default for optimized compilation if the assembler and
5979 linker support it. Use @option{-fno-merge-constants} to inhibit this
5982 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5984 @item -fmerge-all-constants
5985 @opindex fmerge-all-constants
5986 Attempt to merge identical constants and identical variables.
5988 This option implies @option{-fmerge-constants}. In addition to
5989 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5990 arrays or initialized constant variables with integral or floating point
5991 types. Languages like C or C++ require each variable, including multiple
5992 instances of the same variable in recursive calls, to have distinct locations,
5993 so using this option will result in non-conforming
5996 @item -fmodulo-sched
5997 @opindex fmodulo-sched
5998 Perform swing modulo scheduling immediately before the first scheduling
5999 pass. This pass looks at innermost loops and reorders their
6000 instructions by overlapping different iterations.
6002 @item -fmodulo-sched-allow-regmoves
6003 @opindex fmodulo-sched-allow-regmoves
6004 Perform more aggressive SMS based modulo scheduling with register moves
6005 allowed. By setting this flag certain anti-dependences edges will be
6006 deleted which will trigger the generation of reg-moves based on the
6007 life-range analysis. This option is effective only with
6008 @option{-fmodulo-sched} enabled.
6010 @item -fno-branch-count-reg
6011 @opindex fno-branch-count-reg
6012 Do not use ``decrement and branch'' instructions on a count register,
6013 but instead generate a sequence of instructions that decrement a
6014 register, compare it against zero, then branch based upon the result.
6015 This option is only meaningful on architectures that support such
6016 instructions, which include x86, PowerPC, IA-64 and S/390.
6018 The default is @option{-fbranch-count-reg}.
6020 @item -fno-function-cse
6021 @opindex fno-function-cse
6022 Do not put function addresses in registers; make each instruction that
6023 calls a constant function contain the function's address explicitly.
6025 This option results in less efficient code, but some strange hacks
6026 that alter the assembler output may be confused by the optimizations
6027 performed when this option is not used.
6029 The default is @option{-ffunction-cse}
6031 @item -fno-zero-initialized-in-bss
6032 @opindex fno-zero-initialized-in-bss
6033 If the target supports a BSS section, GCC by default puts variables that
6034 are initialized to zero into BSS@. This can save space in the resulting
6037 This option turns off this behavior because some programs explicitly
6038 rely on variables going to the data section. E.g., so that the
6039 resulting executable can find the beginning of that section and/or make
6040 assumptions based on that.
6042 The default is @option{-fzero-initialized-in-bss}.
6044 @item -fmudflap -fmudflapth -fmudflapir
6048 @cindex bounds checking
6050 For front-ends that support it (C and C++), instrument all risky
6051 pointer/array dereferencing operations, some standard library
6052 string/heap functions, and some other associated constructs with
6053 range/validity tests. Modules so instrumented should be immune to
6054 buffer overflows, invalid heap use, and some other classes of C/C++
6055 programming errors. The instrumentation relies on a separate runtime
6056 library (@file{libmudflap}), which will be linked into a program if
6057 @option{-fmudflap} is given at link time. Run-time behavior of the
6058 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6059 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6062 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6063 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6064 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6065 instrumentation should ignore pointer reads. This produces less
6066 instrumentation (and therefore faster execution) and still provides
6067 some protection against outright memory corrupting writes, but allows
6068 erroneously read data to propagate within a program.
6070 @item -fthread-jumps
6071 @opindex fthread-jumps
6072 Perform optimizations where we check to see if a jump branches to a
6073 location where another comparison subsumed by the first is found. If
6074 so, the first branch is redirected to either the destination of the
6075 second branch or a point immediately following it, depending on whether
6076 the condition is known to be true or false.
6078 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6080 @item -fsplit-wide-types
6081 @opindex fsplit-wide-types
6082 When using a type that occupies multiple registers, such as @code{long
6083 long} on a 32-bit system, split the registers apart and allocate them
6084 independently. This normally generates better code for those types,
6085 but may make debugging more difficult.
6087 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6090 @item -fcse-follow-jumps
6091 @opindex fcse-follow-jumps
6092 In common subexpression elimination (CSE), scan through jump instructions
6093 when the target of the jump is not reached by any other path. For
6094 example, when CSE encounters an @code{if} statement with an
6095 @code{else} clause, CSE will follow the jump when the condition
6098 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6100 @item -fcse-skip-blocks
6101 @opindex fcse-skip-blocks
6102 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6103 follow jumps which conditionally skip over blocks. When CSE
6104 encounters a simple @code{if} statement with no else clause,
6105 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6106 body of the @code{if}.
6108 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6110 @item -frerun-cse-after-loop
6111 @opindex frerun-cse-after-loop
6112 Re-run common subexpression elimination after loop optimizations has been
6115 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6119 Perform a global common subexpression elimination pass.
6120 This pass also performs global constant and copy propagation.
6122 @emph{Note:} When compiling a program using computed gotos, a GCC
6123 extension, you may get better runtime performance if you disable
6124 the global common subexpression elimination pass by adding
6125 @option{-fno-gcse} to the command line.
6127 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6131 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6132 attempt to move loads which are only killed by stores into themselves. This
6133 allows a loop containing a load/store sequence to be changed to a load outside
6134 the loop, and a copy/store within the loop.
6136 Enabled by default when gcse is enabled.
6140 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6141 global common subexpression elimination. This pass will attempt to move
6142 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6143 loops containing a load/store sequence can be changed to a load before
6144 the loop and a store after the loop.
6146 Not enabled at any optimization level.
6150 When @option{-fgcse-las} is enabled, the global common subexpression
6151 elimination pass eliminates redundant loads that come after stores to the
6152 same memory location (both partial and full redundancies).
6154 Not enabled at any optimization level.
6156 @item -fgcse-after-reload
6157 @opindex fgcse-after-reload
6158 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6159 pass is performed after reload. The purpose of this pass is to cleanup
6162 @item -funsafe-loop-optimizations
6163 @opindex funsafe-loop-optimizations
6164 If given, the loop optimizer will assume that loop indices do not
6165 overflow, and that the loops with nontrivial exit condition are not
6166 infinite. This enables a wider range of loop optimizations even if
6167 the loop optimizer itself cannot prove that these assumptions are valid.
6168 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6169 if it finds this kind of loop.
6171 @item -fcrossjumping
6172 @opindex fcrossjumping
6173 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6174 resulting code may or may not perform better than without cross-jumping.
6176 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6178 @item -fauto-inc-dec
6179 @opindex fauto-inc-dec
6180 Combine increments or decrements of addresses with memory accesses.
6181 This pass is always skipped on architectures that do not have
6182 instructions to support this. Enabled by default at @option{-O} and
6183 higher on architectures that support this.
6187 Perform dead code elimination (DCE) on RTL@.
6188 Enabled by default at @option{-O} and higher.
6192 Perform dead store elimination (DSE) on RTL@.
6193 Enabled by default at @option{-O} and higher.
6195 @item -fif-conversion
6196 @opindex fif-conversion
6197 Attempt to transform conditional jumps into branch-less equivalents. This
6198 include use of conditional moves, min, max, set flags and abs instructions, and
6199 some tricks doable by standard arithmetics. The use of conditional execution
6200 on chips where it is available is controlled by @code{if-conversion2}.
6202 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6204 @item -fif-conversion2
6205 @opindex fif-conversion2
6206 Use conditional execution (where available) to transform conditional jumps into
6207 branch-less equivalents.
6209 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6211 @item -fdelete-null-pointer-checks
6212 @opindex fdelete-null-pointer-checks
6213 Assume that programs cannot safely dereference null pointers, and that
6214 no code or data element resides there. This enables simple constant
6215 folding optimizations at all optimization levels. In addition, other
6216 optimization passes in GCC use this flag to control global dataflow
6217 analyses that eliminate useless checks for null pointers; these assume
6218 that if a pointer is checked after it has already been dereferenced,
6221 Note however that in some environments this assumption is not true.
6222 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6223 for programs which depend on that behavior.
6225 Some targets, especially embedded ones, disable this option at all levels.
6226 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6227 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6228 are enabled independently at different optimization levels.
6230 @item -fexpensive-optimizations
6231 @opindex fexpensive-optimizations
6232 Perform a number of minor optimizations that are relatively expensive.
6234 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6236 @item -foptimize-register-move
6238 @opindex foptimize-register-move
6240 Attempt to reassign register numbers in move instructions and as
6241 operands of other simple instructions in order to maximize the amount of
6242 register tying. This is especially helpful on machines with two-operand
6245 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6248 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6250 @item -fira-algorithm=@var{algorithm}
6251 Use specified coloring algorithm for the integrated register
6252 allocator. The @var{algorithm} argument should be @code{priority} or
6253 @code{CB}. The first algorithm specifies Chow's priority coloring,
6254 the second one specifies Chaitin-Briggs coloring. The second
6255 algorithm can be unimplemented for some architectures. If it is
6256 implemented, it is the default because Chaitin-Briggs coloring as a
6257 rule generates a better code.
6259 @item -fira-region=@var{region}
6260 Use specified regions for the integrated register allocator. The
6261 @var{region} argument should be one of @code{all}, @code{mixed}, or
6262 @code{one}. The first value means using all loops as register
6263 allocation regions, the second value which is the default means using
6264 all loops except for loops with small register pressure as the
6265 regions, and third one means using all function as a single region.
6266 The first value can give best result for machines with small size and
6267 irregular register set, the third one results in faster and generates
6268 decent code and the smallest size code, and the default value usually
6269 give the best results in most cases and for most architectures.
6271 @item -fira-coalesce
6272 @opindex fira-coalesce
6273 Do optimistic register coalescing. This option might be profitable for
6274 architectures with big regular register files.
6276 @item -fira-loop-pressure
6277 @opindex fira-loop-pressure
6278 Use IRA to evaluate register pressure in loops for decision to move
6279 loop invariants. Usage of this option usually results in generation
6280 of faster and smaller code on machines with big register files (>= 32
6281 registers) but it can slow compiler down.
6283 This option is enabled at level @option{-O3} for some targets.
6285 @item -fno-ira-share-save-slots
6286 @opindex fno-ira-share-save-slots
6287 Switch off sharing stack slots used for saving call used hard
6288 registers living through a call. Each hard register will get a
6289 separate stack slot and as a result function stack frame will be
6292 @item -fno-ira-share-spill-slots
6293 @opindex fno-ira-share-spill-slots
6294 Switch off sharing stack slots allocated for pseudo-registers. Each
6295 pseudo-register which did not get a hard register will get a separate
6296 stack slot and as a result function stack frame will be bigger.
6298 @item -fira-verbose=@var{n}
6299 @opindex fira-verbose
6300 Set up how verbose dump file for the integrated register allocator
6301 will be. Default value is 5. If the value is greater or equal to 10,
6302 the dump file will be stderr as if the value were @var{n} minus 10.
6304 @item -fdelayed-branch
6305 @opindex fdelayed-branch
6306 If supported for the target machine, attempt to reorder instructions
6307 to exploit instruction slots available after delayed branch
6310 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6312 @item -fschedule-insns
6313 @opindex fschedule-insns
6314 If supported for the target machine, attempt to reorder instructions to
6315 eliminate execution stalls due to required data being unavailable. This
6316 helps machines that have slow floating point or memory load instructions
6317 by allowing other instructions to be issued until the result of the load
6318 or floating point instruction is required.
6320 Enabled at levels @option{-O2}, @option{-O3}.
6322 @item -fschedule-insns2
6323 @opindex fschedule-insns2
6324 Similar to @option{-fschedule-insns}, but requests an additional pass of
6325 instruction scheduling after register allocation has been done. This is
6326 especially useful on machines with a relatively small number of
6327 registers and where memory load instructions take more than one cycle.
6329 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6331 @item -fno-sched-interblock
6332 @opindex fno-sched-interblock
6333 Don't schedule instructions across basic blocks. This is normally
6334 enabled by default when scheduling before register allocation, i.e.@:
6335 with @option{-fschedule-insns} or at @option{-O2} or higher.
6337 @item -fno-sched-spec
6338 @opindex fno-sched-spec
6339 Don't allow speculative motion of non-load instructions. This is normally
6340 enabled by default when scheduling before register allocation, i.e.@:
6341 with @option{-fschedule-insns} or at @option{-O2} or higher.
6343 @item -fsched-pressure
6344 @opindex fsched-pressure
6345 Enable register pressure sensitive insn scheduling before the register
6346 allocation. This only makes sense when scheduling before register
6347 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6348 @option{-O2} or higher. Usage of this option can improve the
6349 generated code and decrease its size by preventing register pressure
6350 increase above the number of available hard registers and as a
6351 consequence register spills in the register allocation.
6353 @item -fsched-spec-load
6354 @opindex fsched-spec-load
6355 Allow speculative motion of some load instructions. This only makes
6356 sense when scheduling before register allocation, i.e.@: with
6357 @option{-fschedule-insns} or at @option{-O2} or higher.
6359 @item -fsched-spec-load-dangerous
6360 @opindex fsched-spec-load-dangerous
6361 Allow speculative motion of more load instructions. This only makes
6362 sense when scheduling before register allocation, i.e.@: with
6363 @option{-fschedule-insns} or at @option{-O2} or higher.
6365 @item -fsched-stalled-insns
6366 @itemx -fsched-stalled-insns=@var{n}
6367 @opindex fsched-stalled-insns
6368 Define how many insns (if any) can be moved prematurely from the queue
6369 of stalled insns into the ready list, during the second scheduling pass.
6370 @option{-fno-sched-stalled-insns} means that no insns will be moved
6371 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6372 on how many queued insns can be moved prematurely.
6373 @option{-fsched-stalled-insns} without a value is equivalent to
6374 @option{-fsched-stalled-insns=1}.
6376 @item -fsched-stalled-insns-dep
6377 @itemx -fsched-stalled-insns-dep=@var{n}
6378 @opindex fsched-stalled-insns-dep
6379 Define how many insn groups (cycles) will be examined for a dependency
6380 on a stalled insn that is candidate for premature removal from the queue
6381 of stalled insns. This has an effect only during the second scheduling pass,
6382 and only if @option{-fsched-stalled-insns} is used.
6383 @option{-fno-sched-stalled-insns-dep} is equivalent to
6384 @option{-fsched-stalled-insns-dep=0}.
6385 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6386 @option{-fsched-stalled-insns-dep=1}.
6388 @item -fsched2-use-superblocks
6389 @opindex fsched2-use-superblocks
6390 When scheduling after register allocation, do use superblock scheduling
6391 algorithm. Superblock scheduling allows motion across basic block boundaries
6392 resulting on faster schedules. This option is experimental, as not all machine
6393 descriptions used by GCC model the CPU closely enough to avoid unreliable
6394 results from the algorithm.
6396 This only makes sense when scheduling after register allocation, i.e.@: with
6397 @option{-fschedule-insns2} or at @option{-O2} or higher.
6399 @item -fsched-group-heuristic
6400 @opindex fsched-group-heuristic
6401 Enable the group heuristic in the scheduler. This heuristic favors
6402 the instruction that belongs to a schedule group. This is enabled
6403 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6404 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6406 @item -fsched-critical-path-heuristic
6407 @opindex fsched-critical-path-heuristic
6408 Enable the critical-path heuristic in the scheduler. This heuristic favors
6409 instructions on the critical path. This is enabled by default when
6410 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6411 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6413 @item -fsched-spec-insn-heuristic
6414 @opindex fsched-spec-insn-heuristic
6415 Enable the speculative instruction heuristic in the scheduler. This
6416 heuristic favors speculative instructions with greater dependency weakness.
6417 This is enabled by default when scheduling is enabled, i.e.@:
6418 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6419 or at @option{-O2} or higher.
6421 @item -fsched-rank-heuristic
6422 @opindex fsched-rank-heuristic
6423 Enable the rank heuristic in the scheduler. This heuristic favors
6424 the instruction belonging to a basic block with greater size or frequency.
6425 This is enabled by default when scheduling is enabled, i.e.@:
6426 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6427 at @option{-O2} or higher.
6429 @item -fsched-last-insn-heuristic
6430 @opindex fsched-last-insn-heuristic
6431 Enable the last-instruction heuristic in the scheduler. This heuristic
6432 favors the instruction that is less dependent on the last instruction
6433 scheduled. This is enabled by default when scheduling is enabled,
6434 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6435 at @option{-O2} or higher.
6437 @item -fsched-dep-count-heuristic
6438 @opindex fsched-dep-count-heuristic
6439 Enable the dependent-count heuristic in the scheduler. This heuristic
6440 favors the instruction that has more instructions depending on it.
6441 This is enabled by default when scheduling is enabled, i.e.@:
6442 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6443 at @option{-O2} or higher.
6445 @item -freschedule-modulo-scheduled-loops
6446 @opindex freschedule-modulo-scheduled-loops
6447 The modulo scheduling comes before the traditional scheduling, if a loop
6448 was modulo scheduled we may want to prevent the later scheduling passes
6449 from changing its schedule, we use this option to control that.
6451 @item -fselective-scheduling
6452 @opindex fselective-scheduling
6453 Schedule instructions using selective scheduling algorithm. Selective
6454 scheduling runs instead of the first scheduler pass.
6456 @item -fselective-scheduling2
6457 @opindex fselective-scheduling2
6458 Schedule instructions using selective scheduling algorithm. Selective
6459 scheduling runs instead of the second scheduler pass.
6461 @item -fsel-sched-pipelining
6462 @opindex fsel-sched-pipelining
6463 Enable software pipelining of innermost loops during selective scheduling.
6464 This option has no effect until one of @option{-fselective-scheduling} or
6465 @option{-fselective-scheduling2} is turned on.
6467 @item -fsel-sched-pipelining-outer-loops
6468 @opindex fsel-sched-pipelining-outer-loops
6469 When pipelining loops during selective scheduling, also pipeline outer loops.
6470 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6472 @item -fcaller-saves
6473 @opindex fcaller-saves
6474 Enable values to be allocated in registers that will be clobbered by
6475 function calls, by emitting extra instructions to save and restore the
6476 registers around such calls. Such allocation is done only when it
6477 seems to result in better code than would otherwise be produced.
6479 This option is always enabled by default on certain machines, usually
6480 those which have no call-preserved registers to use instead.
6482 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6484 @item -fconserve-stack
6485 @opindex fconserve-stack
6486 Attempt to minimize stack usage. The compiler will attempt to use less
6487 stack space, even if that makes the program slower. This option
6488 implies setting the @option{large-stack-frame} parameter to 100
6489 and the @option{large-stack-frame-growth} parameter to 400.
6491 @item -ftree-reassoc
6492 @opindex ftree-reassoc
6493 Perform reassociation on trees. This flag is enabled by default
6494 at @option{-O} and higher.
6498 Perform partial redundancy elimination (PRE) on trees. This flag is
6499 enabled by default at @option{-O2} and @option{-O3}.
6501 @item -ftree-forwprop
6502 @opindex ftree-forwprop
6503 Perform forward propagation on trees. This flag is enabled by default
6504 at @option{-O} and higher.
6508 Perform full redundancy elimination (FRE) on trees. The difference
6509 between FRE and PRE is that FRE only considers expressions
6510 that are computed on all paths leading to the redundant computation.
6511 This analysis is faster than PRE, though it exposes fewer redundancies.
6512 This flag is enabled by default at @option{-O} and higher.
6514 @item -ftree-phiprop
6515 @opindex ftree-phiprop
6516 Perform hoisting of loads from conditional pointers on trees. This
6517 pass is enabled by default at @option{-O} and higher.
6519 @item -ftree-copy-prop
6520 @opindex ftree-copy-prop
6521 Perform copy propagation on trees. This pass eliminates unnecessary
6522 copy operations. This flag is enabled by default at @option{-O} and
6525 @item -fipa-pure-const
6526 @opindex fipa-pure-const
6527 Discover which functions are pure or constant.
6528 Enabled by default at @option{-O} and higher.
6530 @item -fipa-reference
6531 @opindex fipa-reference
6532 Discover which static variables do not escape cannot escape the
6534 Enabled by default at @option{-O} and higher.
6536 @item -fipa-struct-reorg
6537 @opindex fipa-struct-reorg
6538 Perform structure reorganization optimization, that change C-like structures
6539 layout in order to better utilize spatial locality. This transformation is
6540 affective for programs containing arrays of structures. Available in two
6541 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6542 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6543 to provide the safety of this transformation. It works only in whole program
6544 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6545 enabled. Structures considered @samp{cold} by this transformation are not
6546 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6548 With this flag, the program debug info reflects a new structure layout.
6552 Perform interprocedural pointer analysis. This option is experimental
6553 and does not affect generated code.
6557 Perform interprocedural constant propagation.
6558 This optimization analyzes the program to determine when values passed
6559 to functions are constants and then optimizes accordingly.
6560 This optimization can substantially increase performance
6561 if the application has constants passed to functions.
6562 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6564 @item -fipa-cp-clone
6565 @opindex fipa-cp-clone
6566 Perform function cloning to make interprocedural constant propagation stronger.
6567 When enabled, interprocedural constant propagation will perform function cloning
6568 when externally visible function can be called with constant arguments.
6569 Because this optimization can create multiple copies of functions,
6570 it may significantly increase code size
6571 (see @option{--param ipcp-unit-growth=@var{value}}).
6572 This flag is enabled by default at @option{-O3}.
6574 @item -fipa-matrix-reorg
6575 @opindex fipa-matrix-reorg
6576 Perform matrix flattening and transposing.
6577 Matrix flattening tries to replace an @math{m}-dimensional matrix
6578 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6579 This reduces the level of indirection needed for accessing the elements
6580 of the matrix. The second optimization is matrix transposing that
6581 attempts to change the order of the matrix's dimensions in order to
6582 improve cache locality.
6583 Both optimizations need the @option{-fwhole-program} flag.
6584 Transposing is enabled only if profiling information is available.
6588 Perform forward store motion on trees. This flag is
6589 enabled by default at @option{-O} and higher.
6593 Perform sparse conditional constant propagation (CCP) on trees. This
6594 pass only operates on local scalar variables and is enabled by default
6595 at @option{-O} and higher.
6597 @item -ftree-switch-conversion
6598 Perform conversion of simple initializations in a switch to
6599 initializations from a scalar array. This flag is enabled by default
6600 at @option{-O2} and higher.
6604 Perform dead code elimination (DCE) on trees. This flag is enabled by
6605 default at @option{-O} and higher.
6607 @item -ftree-builtin-call-dce
6608 @opindex ftree-builtin-call-dce
6609 Perform conditional dead code elimination (DCE) for calls to builtin functions
6610 that may set @code{errno} but are otherwise side-effect free. This flag is
6611 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6614 @item -ftree-dominator-opts
6615 @opindex ftree-dominator-opts
6616 Perform a variety of simple scalar cleanups (constant/copy
6617 propagation, redundancy elimination, range propagation and expression
6618 simplification) based on a dominator tree traversal. This also
6619 performs jump threading (to reduce jumps to jumps). This flag is
6620 enabled by default at @option{-O} and higher.
6624 Perform dead store elimination (DSE) on trees. A dead store is a store into
6625 a memory location which will later be overwritten by another store without
6626 any intervening loads. In this case the earlier store can be deleted. This
6627 flag is enabled by default at @option{-O} and higher.
6631 Perform loop header copying on trees. This is beneficial since it increases
6632 effectiveness of code motion optimizations. It also saves one jump. This flag
6633 is enabled by default at @option{-O} and higher. It is not enabled
6634 for @option{-Os}, since it usually increases code size.
6636 @item -ftree-loop-optimize
6637 @opindex ftree-loop-optimize
6638 Perform loop optimizations on trees. This flag is enabled by default
6639 at @option{-O} and higher.
6641 @item -ftree-loop-linear
6642 @opindex ftree-loop-linear
6643 Perform linear loop transformations on tree. This flag can improve cache
6644 performance and allow further loop optimizations to take place.
6646 @item -floop-interchange
6647 Perform loop interchange transformations on loops. Interchanging two
6648 nested loops switches the inner and outer loops. For example, given a
6653 A(J, I) = A(J, I) * C
6657 loop interchange will transform the loop as if the user had written:
6661 A(J, I) = A(J, I) * C
6665 which can be beneficial when @code{N} is larger than the caches,
6666 because in Fortran, the elements of an array are stored in memory
6667 contiguously by column, and the original loop iterates over rows,
6668 potentially creating at each access a cache miss. This optimization
6669 applies to all the languages supported by GCC and is not limited to
6670 Fortran. To use this code transformation, GCC has to be configured
6671 with @option{--with-ppl} and @option{--with-cloog} to enable the
6672 Graphite loop transformation infrastructure.
6674 @item -floop-strip-mine
6675 Perform loop strip mining transformations on loops. Strip mining
6676 splits a loop into two nested loops. The outer loop has strides
6677 equal to the strip size and the inner loop has strides of the
6678 original loop within a strip. The strip length can be changed
6679 using the @option{loop-block-tile-size} parameter. For example,
6686 loop strip mining will transform the loop as if the user had written:
6689 DO I = II, min (II + 50, N)
6694 This optimization applies to all the languages supported by GCC and is
6695 not limited to Fortran. To use this code transformation, GCC has to
6696 be configured with @option{--with-ppl} and @option{--with-cloog} to
6697 enable the Graphite loop transformation infrastructure.
6700 Perform loop blocking transformations on loops. Blocking strip mines
6701 each loop in the loop nest such that the memory accesses of the
6702 element loops fit inside caches. The strip length can be changed
6703 using the @option{loop-block-tile-size} parameter. For example, given
6708 A(J, I) = B(I) + C(J)
6712 loop blocking will transform the loop as if the user had written:
6716 DO I = II, min (II + 50, N)
6717 DO J = JJ, min (JJ + 50, M)
6718 A(J, I) = B(I) + C(J)
6724 which can be beneficial when @code{M} is larger than the caches,
6725 because the innermost loop will iterate over a smaller amount of data
6726 that can be kept in the caches. This optimization applies to all the
6727 languages supported by GCC and is not limited to Fortran. To use this
6728 code transformation, GCC has to be configured with @option{--with-ppl}
6729 and @option{--with-cloog} to enable the Graphite loop transformation
6732 @item -fgraphite-identity
6733 @opindex fgraphite-identity
6734 Enable the identity transformation for graphite. For every SCoP we generate
6735 the polyhedral representation and transform it back to gimple. Using
6736 @option{-fgraphite-identity} we can check the costs or benefits of the
6737 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6738 are also performed by the code generator CLooG, like index splitting and
6739 dead code elimination in loops.
6741 @item -floop-parallelize-all
6742 Use the Graphite data dependence analysis to identify loops that can
6743 be parallelized. Parallelize all the loops that can be analyzed to
6744 not contain loop carried dependences without checking that it is
6745 profitable to parallelize the loops.
6747 @item -fcheck-data-deps
6748 @opindex fcheck-data-deps
6749 Compare the results of several data dependence analyzers. This option
6750 is used for debugging the data dependence analyzers.
6752 @item -ftree-loop-distribution
6753 Perform loop distribution. This flag can improve cache performance on
6754 big loop bodies and allow further loop optimizations, like
6755 parallelization or vectorization, to take place. For example, the loop
6772 @item -ftree-loop-im
6773 @opindex ftree-loop-im
6774 Perform loop invariant motion on trees. This pass moves only invariants that
6775 would be hard to handle at RTL level (function calls, operations that expand to
6776 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6777 operands of conditions that are invariant out of the loop, so that we can use
6778 just trivial invariantness analysis in loop unswitching. The pass also includes
6781 @item -ftree-loop-ivcanon
6782 @opindex ftree-loop-ivcanon
6783 Create a canonical counter for number of iterations in the loop for that
6784 determining number of iterations requires complicated analysis. Later
6785 optimizations then may determine the number easily. Useful especially
6786 in connection with unrolling.
6790 Perform induction variable optimizations (strength reduction, induction
6791 variable merging and induction variable elimination) on trees.
6793 @item -ftree-parallelize-loops=n
6794 @opindex ftree-parallelize-loops
6795 Parallelize loops, i.e., split their iteration space to run in n threads.
6796 This is only possible for loops whose iterations are independent
6797 and can be arbitrarily reordered. The optimization is only
6798 profitable on multiprocessor machines, for loops that are CPU-intensive,
6799 rather than constrained e.g.@: by memory bandwidth. This option
6800 implies @option{-pthread}, and thus is only supported on targets
6801 that have support for @option{-pthread}.
6805 Perform function-local points-to analysis on trees. This flag is
6806 enabled by default at @option{-O} and higher.
6810 Perform scalar replacement of aggregates. This pass replaces structure
6811 references with scalars to prevent committing structures to memory too
6812 early. This flag is enabled by default at @option{-O} and higher.
6814 @item -ftree-copyrename
6815 @opindex ftree-copyrename
6816 Perform copy renaming on trees. This pass attempts to rename compiler
6817 temporaries to other variables at copy locations, usually resulting in
6818 variable names which more closely resemble the original variables. This flag
6819 is enabled by default at @option{-O} and higher.
6823 Perform temporary expression replacement during the SSA->normal phase. Single
6824 use/single def temporaries are replaced at their use location with their
6825 defining expression. This results in non-GIMPLE code, but gives the expanders
6826 much more complex trees to work on resulting in better RTL generation. This is
6827 enabled by default at @option{-O} and higher.
6829 @item -ftree-vectorize
6830 @opindex ftree-vectorize
6831 Perform loop vectorization on trees. This flag is enabled by default at
6834 @item -ftree-slp-vectorize
6835 @opindex ftree-slp-vectorize
6836 Perform basic block vectorization on trees. This flag is enabled by default at
6837 @option{-O3} and when @option{-ftree-vectorize} is enabled.
6839 @item -ftree-vect-loop-version
6840 @opindex ftree-vect-loop-version
6841 Perform loop versioning when doing loop vectorization on trees. When a loop
6842 appears to be vectorizable except that data alignment or data dependence cannot
6843 be determined at compile time then vectorized and non-vectorized versions of
6844 the loop are generated along with runtime checks for alignment or dependence
6845 to control which version is executed. This option is enabled by default
6846 except at level @option{-Os} where it is disabled.
6848 @item -fvect-cost-model
6849 @opindex fvect-cost-model
6850 Enable cost model for vectorization.
6854 Perform Value Range Propagation on trees. This is similar to the
6855 constant propagation pass, but instead of values, ranges of values are
6856 propagated. This allows the optimizers to remove unnecessary range
6857 checks like array bound checks and null pointer checks. This is
6858 enabled by default at @option{-O2} and higher. Null pointer check
6859 elimination is only done if @option{-fdelete-null-pointer-checks} is
6864 Perform tail duplication to enlarge superblock size. This transformation
6865 simplifies the control flow of the function allowing other optimizations to do
6868 @item -funroll-loops
6869 @opindex funroll-loops
6870 Unroll loops whose number of iterations can be determined at compile
6871 time or upon entry to the loop. @option{-funroll-loops} implies
6872 @option{-frerun-cse-after-loop}. This option makes code larger,
6873 and may or may not make it run faster.
6875 @item -funroll-all-loops
6876 @opindex funroll-all-loops
6877 Unroll all loops, even if their number of iterations is uncertain when
6878 the loop is entered. This usually makes programs run more slowly.
6879 @option{-funroll-all-loops} implies the same options as
6880 @option{-funroll-loops},
6882 @item -fsplit-ivs-in-unroller
6883 @opindex fsplit-ivs-in-unroller
6884 Enables expressing of values of induction variables in later iterations
6885 of the unrolled loop using the value in the first iteration. This breaks
6886 long dependency chains, thus improving efficiency of the scheduling passes.
6888 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6889 same effect. However in cases the loop body is more complicated than
6890 a single basic block, this is not reliable. It also does not work at all
6891 on some of the architectures due to restrictions in the CSE pass.
6893 This optimization is enabled by default.
6895 @item -fvariable-expansion-in-unroller
6896 @opindex fvariable-expansion-in-unroller
6897 With this option, the compiler will create multiple copies of some
6898 local variables when unrolling a loop which can result in superior code.
6900 @item -fpredictive-commoning
6901 @opindex fpredictive-commoning
6902 Perform predictive commoning optimization, i.e., reusing computations
6903 (especially memory loads and stores) performed in previous
6904 iterations of loops.
6906 This option is enabled at level @option{-O3}.
6908 @item -fprefetch-loop-arrays
6909 @opindex fprefetch-loop-arrays
6910 If supported by the target machine, generate instructions to prefetch
6911 memory to improve the performance of loops that access large arrays.
6913 This option may generate better or worse code; results are highly
6914 dependent on the structure of loops within the source code.
6916 Disabled at level @option{-Os}.
6919 @itemx -fno-peephole2
6920 @opindex fno-peephole
6921 @opindex fno-peephole2
6922 Disable any machine-specific peephole optimizations. The difference
6923 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6924 are implemented in the compiler; some targets use one, some use the
6925 other, a few use both.
6927 @option{-fpeephole} is enabled by default.
6928 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6930 @item -fno-guess-branch-probability
6931 @opindex fno-guess-branch-probability
6932 Do not guess branch probabilities using heuristics.
6934 GCC will use heuristics to guess branch probabilities if they are
6935 not provided by profiling feedback (@option{-fprofile-arcs}). These
6936 heuristics are based on the control flow graph. If some branch probabilities
6937 are specified by @samp{__builtin_expect}, then the heuristics will be
6938 used to guess branch probabilities for the rest of the control flow graph,
6939 taking the @samp{__builtin_expect} info into account. The interactions
6940 between the heuristics and @samp{__builtin_expect} can be complex, and in
6941 some cases, it may be useful to disable the heuristics so that the effects
6942 of @samp{__builtin_expect} are easier to understand.
6944 The default is @option{-fguess-branch-probability} at levels
6945 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6947 @item -freorder-blocks
6948 @opindex freorder-blocks
6949 Reorder basic blocks in the compiled function in order to reduce number of
6950 taken branches and improve code locality.
6952 Enabled at levels @option{-O2}, @option{-O3}.
6954 @item -freorder-blocks-and-partition
6955 @opindex freorder-blocks-and-partition
6956 In addition to reordering basic blocks in the compiled function, in order
6957 to reduce number of taken branches, partitions hot and cold basic blocks
6958 into separate sections of the assembly and .o files, to improve
6959 paging and cache locality performance.
6961 This optimization is automatically turned off in the presence of
6962 exception handling, for linkonce sections, for functions with a user-defined
6963 section attribute and on any architecture that does not support named
6966 @item -freorder-functions
6967 @opindex freorder-functions
6968 Reorder functions in the object file in order to
6969 improve code locality. This is implemented by using special
6970 subsections @code{.text.hot} for most frequently executed functions and
6971 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6972 the linker so object file format must support named sections and linker must
6973 place them in a reasonable way.
6975 Also profile feedback must be available in to make this option effective. See
6976 @option{-fprofile-arcs} for details.
6978 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6980 @item -fstrict-aliasing
6981 @opindex fstrict-aliasing
6982 Allow the compiler to assume the strictest aliasing rules applicable to
6983 the language being compiled. For C (and C++), this activates
6984 optimizations based on the type of expressions. In particular, an
6985 object of one type is assumed never to reside at the same address as an
6986 object of a different type, unless the types are almost the same. For
6987 example, an @code{unsigned int} can alias an @code{int}, but not a
6988 @code{void*} or a @code{double}. A character type may alias any other
6991 @anchor{Type-punning}Pay special attention to code like this:
7004 The practice of reading from a different union member than the one most
7005 recently written to (called ``type-punning'') is common. Even with
7006 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7007 is accessed through the union type. So, the code above will work as
7008 expected. @xref{Structures unions enumerations and bit-fields
7009 implementation}. However, this code might not:
7020 Similarly, access by taking the address, casting the resulting pointer
7021 and dereferencing the result has undefined behavior, even if the cast
7022 uses a union type, e.g.:
7026 return ((union a_union *) &d)->i;
7030 The @option{-fstrict-aliasing} option is enabled at levels
7031 @option{-O2}, @option{-O3}, @option{-Os}.
7033 @item -fstrict-overflow
7034 @opindex fstrict-overflow
7035 Allow the compiler to assume strict signed overflow rules, depending
7036 on the language being compiled. For C (and C++) this means that
7037 overflow when doing arithmetic with signed numbers is undefined, which
7038 means that the compiler may assume that it will not happen. This
7039 permits various optimizations. For example, the compiler will assume
7040 that an expression like @code{i + 10 > i} will always be true for
7041 signed @code{i}. This assumption is only valid if signed overflow is
7042 undefined, as the expression is false if @code{i + 10} overflows when
7043 using twos complement arithmetic. When this option is in effect any
7044 attempt to determine whether an operation on signed numbers will
7045 overflow must be written carefully to not actually involve overflow.
7047 This option also allows the compiler to assume strict pointer
7048 semantics: given a pointer to an object, if adding an offset to that
7049 pointer does not produce a pointer to the same object, the addition is
7050 undefined. This permits the compiler to conclude that @code{p + u >
7051 p} is always true for a pointer @code{p} and unsigned integer
7052 @code{u}. This assumption is only valid because pointer wraparound is
7053 undefined, as the expression is false if @code{p + u} overflows using
7054 twos complement arithmetic.
7056 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7057 that integer signed overflow is fully defined: it wraps. When
7058 @option{-fwrapv} is used, there is no difference between
7059 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7060 integers. With @option{-fwrapv} certain types of overflow are
7061 permitted. For example, if the compiler gets an overflow when doing
7062 arithmetic on constants, the overflowed value can still be used with
7063 @option{-fwrapv}, but not otherwise.
7065 The @option{-fstrict-overflow} option is enabled at levels
7066 @option{-O2}, @option{-O3}, @option{-Os}.
7068 @item -falign-functions
7069 @itemx -falign-functions=@var{n}
7070 @opindex falign-functions
7071 Align the start of functions to the next power-of-two greater than
7072 @var{n}, skipping up to @var{n} bytes. For instance,
7073 @option{-falign-functions=32} aligns functions to the next 32-byte
7074 boundary, but @option{-falign-functions=24} would align to the next
7075 32-byte boundary only if this can be done by skipping 23 bytes or less.
7077 @option{-fno-align-functions} and @option{-falign-functions=1} are
7078 equivalent and mean that functions will not be aligned.
7080 Some assemblers only support this flag when @var{n} is a power of two;
7081 in that case, it is rounded up.
7083 If @var{n} is not specified or is zero, use a machine-dependent default.
7085 Enabled at levels @option{-O2}, @option{-O3}.
7087 @item -falign-labels
7088 @itemx -falign-labels=@var{n}
7089 @opindex falign-labels
7090 Align all branch targets to a power-of-two boundary, skipping up to
7091 @var{n} bytes like @option{-falign-functions}. This option can easily
7092 make code slower, because it must insert dummy operations for when the
7093 branch target is reached in the usual flow of the code.
7095 @option{-fno-align-labels} and @option{-falign-labels=1} are
7096 equivalent and mean that labels will not be aligned.
7098 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7099 are greater than this value, then their values are used instead.
7101 If @var{n} is not specified or is zero, use a machine-dependent default
7102 which is very likely to be @samp{1}, meaning no alignment.
7104 Enabled at levels @option{-O2}, @option{-O3}.
7107 @itemx -falign-loops=@var{n}
7108 @opindex falign-loops
7109 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7110 like @option{-falign-functions}. The hope is that the loop will be
7111 executed many times, which will make up for any execution of the dummy
7114 @option{-fno-align-loops} and @option{-falign-loops=1} are
7115 equivalent and mean that loops will not be aligned.
7117 If @var{n} is not specified or is zero, use a machine-dependent default.
7119 Enabled at levels @option{-O2}, @option{-O3}.
7122 @itemx -falign-jumps=@var{n}
7123 @opindex falign-jumps
7124 Align branch targets to a power-of-two boundary, for branch targets
7125 where the targets can only be reached by jumping, skipping up to @var{n}
7126 bytes like @option{-falign-functions}. In this case, no dummy operations
7129 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7130 equivalent and mean that loops will not be aligned.
7132 If @var{n} is not specified or is zero, use a machine-dependent default.
7134 Enabled at levels @option{-O2}, @option{-O3}.
7136 @item -funit-at-a-time
7137 @opindex funit-at-a-time
7138 This option is left for compatibility reasons. @option{-funit-at-a-time}
7139 has no effect, while @option{-fno-unit-at-a-time} implies
7140 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7144 @item -fno-toplevel-reorder
7145 @opindex fno-toplevel-reorder
7146 Do not reorder top-level functions, variables, and @code{asm}
7147 statements. Output them in the same order that they appear in the
7148 input file. When this option is used, unreferenced static variables
7149 will not be removed. This option is intended to support existing code
7150 which relies on a particular ordering. For new code, it is better to
7153 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7154 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7159 Constructs webs as commonly used for register allocation purposes and assign
7160 each web individual pseudo register. This allows the register allocation pass
7161 to operate on pseudos directly, but also strengthens several other optimization
7162 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7163 however, make debugging impossible, since variables will no longer stay in a
7166 Enabled by default with @option{-funroll-loops}.
7168 @item -fwhole-program
7169 @opindex fwhole-program
7170 Assume that the current compilation unit represents the whole program being
7171 compiled. All public functions and variables with the exception of @code{main}
7172 and those merged by attribute @code{externally_visible} become static functions
7173 and in effect are optimized more aggressively by interprocedural optimizers.
7174 While this option is equivalent to proper use of the @code{static} keyword for
7175 programs consisting of a single file, in combination with option
7176 @option{-combine}, @option{-flto} or @option{-fwhopr} this flag can be used to
7177 compile many smaller scale programs since the functions and variables become
7178 local for the whole combined compilation unit, not for the single source file
7181 This option implies @option{-fwhole-file} for Fortran programs.
7185 This option runs the standard link-time optimizer. When invoked
7186 with source code, it generates GIMPLE (one of GCC's internal
7187 representations) and writes it to special ELF sections in the object
7188 file. When the object files are linked together, all the function
7189 bodies are read from these ELF sections and instantiated as if they
7190 had been part of the same translation unit.
7192 To use the link-timer optimizer, @option{-flto} needs to be specified at
7193 compile time and during the final link. For example,
7196 gcc -c -O2 -flto foo.c
7197 gcc -c -O2 -flto bar.c
7198 gcc -o myprog -flto -O2 foo.o bar.o
7201 The first two invocations to GCC will save a bytecode representation
7202 of GIMPLE into special ELF sections inside @file{foo.o} and
7203 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7204 @file{foo.o} and @file{bar.o}, merge the two files into a single
7205 internal image, and compile the result as usual. Since both
7206 @file{foo.o} and @file{bar.o} are merged into a single image, this
7207 causes all the inter-procedural analyses and optimizations in GCC to
7208 work across the two files as if they were a single one. This means,
7209 for example, that the inliner will be able to inline functions in
7210 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7212 Another (simpler) way to enable link-time optimization is,
7215 gcc -o myprog -flto -O2 foo.c bar.c
7218 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7219 merge them together into a single GIMPLE representation and optimize
7220 them as usual to produce @file{myprog}.
7222 The only important thing to keep in mind is that to enable link-time
7223 optimizations the @option{-flto} flag needs to be passed to both the
7224 compile and the link commands.
7226 Note that when a file is compiled with @option{-flto}, the generated
7227 object file will be larger than a regular object file because it will
7228 contain GIMPLE bytecodes and the usual final code. This means that
7229 object files with LTO information can be linked as a normal object
7230 file. So, in the previous example, if the final link is done with
7233 gcc -o myprog foo.o bar.o
7236 The only difference will be that no inter-procedural optimizations
7237 will be applied to produce @file{myprog}. The two object files
7238 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7241 Additionally, the optimization flags used to compile individual files
7242 are not necessarily related to those used at link-time. For instance,
7245 gcc -c -O0 -flto foo.c
7246 gcc -c -O0 -flto bar.c
7247 gcc -o myprog -flto -O3 foo.o bar.o
7250 This will produce individual object files with unoptimized assembler
7251 code, but the resulting binary @file{myprog} will be optimized at
7252 @option{-O3}. Now, if the final binary is generated without
7253 @option{-flto}, then @file{myprog} will not be optimized.
7255 When producing the final binary with @option{-flto}, GCC will only
7256 apply link-time optimizations to those files that contain bytecode.
7257 Therefore, you can mix and match object files and libraries with
7258 GIMPLE bytecodes and final object code. GCC will automatically select
7259 which files to optimize in LTO mode and which files to link without
7262 There are some code generation flags that GCC will preserve when
7263 generating bytecodes, as they need to be used during the final link
7264 stage. Currently, the following options are saved into the GIMPLE
7265 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7266 @option{-m} target flags.
7268 At link time, these options are read-in and reapplied. Note that the
7269 current implementation makes no attempt at recognizing conflicting
7270 values for these options. If two or more files have a conflicting
7271 value (e.g., one file is compiled with @option{-fPIC} and another
7272 isn't), the compiler will simply use the last value read from the
7273 bytecode files. It is recommended, then, that all the files
7274 participating in the same link be compiled with the same options.
7276 Another feature of LTO is that it is possible to apply interprocedural
7277 optimizations on files written in different languages. This requires
7278 some support in the language front end. Currently, the C, C++ and
7279 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7280 something like this should work
7285 gfortran -c -flto baz.f90
7286 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7289 Notice that the final link is done with @command{g++} to get the C++
7290 runtime libraries and @option{-lgfortran} is added to get the Fortran
7291 runtime libraries. In general, when mixing languages in LTO mode, you
7292 should use the same link command used when mixing languages in a
7293 regular (non-LTO) compilation. This means that if your build process
7294 was mixing languages before, all you need to add is @option{-flto} to
7295 all the compile and link commands.
7297 If LTO encounters objects with C linkage declared with incompatible
7298 types in separate translation units to be linked together (undefined
7299 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7300 issued. The behavior is still undefined at runtime.
7302 If object files containing GIMPLE bytecode are stored in a library
7303 archive, say @file{libfoo.a}, it is possible to extract and use them
7304 in an LTO link if you are using @command{gold} as the linker (which,
7305 in turn requires GCC to be configured with @option{--enable-gold}).
7306 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7310 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7313 With the linker plugin enabled, @command{gold} will extract the needed
7314 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7315 to make them part of the aggregated GIMPLE image to be optimized.
7317 If you are not using @command{gold} and/or do not specify
7318 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7319 will be extracted and linked as usual, but they will not participate
7320 in the LTO optimization process.
7322 Link time optimizations do not require the presence of the whole
7323 program to operate. If the program does not require any symbols to
7324 be exported, it is possible to combine @option{-flto} and
7325 @option{-fwhopr} with @option{-fwhole-program} to allow the
7326 interprocedural optimizers to use more aggressive assumptions which
7327 may lead to improved optimization opportunities.
7329 Regarding portability: the current implementation of LTO makes no
7330 attempt at generating bytecode that can be ported between different
7331 types of hosts. The bytecode files are versioned and there is a
7332 strict version check, so bytecode files generated in one version of
7333 GCC will not work with an older/newer version of GCC.
7335 Link time optimization does not play well with generating debugging
7336 information. Combining @option{-flto} or @option{-fwhopr} with
7337 @option{-g} is experimental.
7339 This option is disabled by default.
7343 This option is identical in functionality to @option{-flto} but it
7344 differs in how the final link stage is executed. Instead of loading
7345 all the function bodies in memory, the callgraph is analyzed and
7346 optimization decisions are made (whole program analysis or WPA). Once
7347 optimization decisions are made, the callgraph is partitioned and the
7348 different sections are compiled separately (local transformations or
7349 LTRANS)@. This process allows optimizations on very large programs
7350 that otherwise would not fit in memory. This option enables
7351 @option{-fwpa} and @option{-fltrans} automatically.
7353 Disabled by default.
7355 This option is experimental.
7359 This is an internal option used by GCC when compiling with
7360 @option{-fwhopr}. You should never need to use it.
7362 This option runs the link-time optimizer in the whole-program-analysis
7363 (WPA) mode, which reads in summary information from all inputs and
7364 performs a whole-program analysis based on summary information only.
7365 It generates object files for subsequent runs of the link-time
7366 optimizer where individual object files are optimized using both
7367 summary information from the WPA mode and the actual function bodies.
7368 It then drives the LTRANS phase.
7370 Disabled by default.
7374 This is an internal option used by GCC when compiling with
7375 @option{-fwhopr}. You should never need to use it.
7377 This option runs the link-time optimizer in the local-transformation (LTRANS)
7378 mode, which reads in output from a previous run of the LTO in WPA mode.
7379 In the LTRANS mode, LTO optimizes an object and produces the final assembly.
7381 Disabled by default.
7383 @item -fltrans-output-list=@var{file}
7384 @opindex fltrans-output-list
7385 This is an internal option used by GCC when compiling with
7386 @option{-fwhopr}. You should never need to use it.
7388 This option specifies a file to which the names of LTRANS output files are
7389 written. This option is only meaningful in conjunction with @option{-fwpa}.
7391 Disabled by default.
7393 @item -flto-compression-level=@var{n}
7394 This option specifies the level of compression used for intermediate
7395 language written to LTO object files, and is only meaningful in
7396 conjunction with LTO mode (@option{-fwhopr}, @option{-flto}). Valid
7397 values are 0 (no compression) to 9 (maximum compression). Values
7398 outside this range are clamped to either 0 or 9. If the option is not
7399 given, a default balanced compression setting is used.
7402 Prints a report with internal details on the workings of the link-time
7403 optimizer. The contents of this report vary from version to version,
7404 it is meant to be useful to GCC developers when processing object
7405 files in LTO mode (via @option{-fwhopr} or @option{-flto}).
7407 Disabled by default.
7409 @item -fuse-linker-plugin
7410 Enables the extraction of objects with GIMPLE bytecode information
7411 from library archives. This option relies on features available only
7412 in @command{gold}, so to use this you must configure GCC with
7413 @option{--enable-gold}. See @option{-flto} for a description on the
7414 effect of this flag and how to use it.
7416 Disabled by default.
7418 @item -fcprop-registers
7419 @opindex fcprop-registers
7420 After register allocation and post-register allocation instruction splitting,
7421 we perform a copy-propagation pass to try to reduce scheduling dependencies
7422 and occasionally eliminate the copy.
7424 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7426 @item -fprofile-correction
7427 @opindex fprofile-correction
7428 Profiles collected using an instrumented binary for multi-threaded programs may
7429 be inconsistent due to missed counter updates. When this option is specified,
7430 GCC will use heuristics to correct or smooth out such inconsistencies. By
7431 default, GCC will emit an error message when an inconsistent profile is detected.
7433 @item -fprofile-dir=@var{path}
7434 @opindex fprofile-dir
7436 Set the directory to search the profile data files in to @var{path}.
7437 This option affects only the profile data generated by
7438 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7439 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7440 and its related options.
7441 By default, GCC will use the current directory as @var{path}
7442 thus the profile data file will appear in the same directory as the object file.
7444 @item -fprofile-generate
7445 @itemx -fprofile-generate=@var{path}
7446 @opindex fprofile-generate
7448 Enable options usually used for instrumenting application to produce
7449 profile useful for later recompilation with profile feedback based
7450 optimization. You must use @option{-fprofile-generate} both when
7451 compiling and when linking your program.
7453 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7455 If @var{path} is specified, GCC will look at the @var{path} to find
7456 the profile feedback data files. See @option{-fprofile-dir}.
7459 @itemx -fprofile-use=@var{path}
7460 @opindex fprofile-use
7461 Enable profile feedback directed optimizations, and optimizations
7462 generally profitable only with profile feedback available.
7464 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7465 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7467 By default, GCC emits an error message if the feedback profiles do not
7468 match the source code. This error can be turned into a warning by using
7469 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7472 If @var{path} is specified, GCC will look at the @var{path} to find
7473 the profile feedback data files. See @option{-fprofile-dir}.
7476 The following options control compiler behavior regarding floating
7477 point arithmetic. These options trade off between speed and
7478 correctness. All must be specifically enabled.
7482 @opindex ffloat-store
7483 Do not store floating point variables in registers, and inhibit other
7484 options that might change whether a floating point value is taken from a
7487 @cindex floating point precision
7488 This option prevents undesirable excess precision on machines such as
7489 the 68000 where the floating registers (of the 68881) keep more
7490 precision than a @code{double} is supposed to have. Similarly for the
7491 x86 architecture. For most programs, the excess precision does only
7492 good, but a few programs rely on the precise definition of IEEE floating
7493 point. Use @option{-ffloat-store} for such programs, after modifying
7494 them to store all pertinent intermediate computations into variables.
7496 @item -fexcess-precision=@var{style}
7497 @opindex fexcess-precision
7498 This option allows further control over excess precision on machines
7499 where floating-point registers have more precision than the IEEE
7500 @code{float} and @code{double} types and the processor does not
7501 support operations rounding to those types. By default,
7502 @option{-fexcess-precision=fast} is in effect; this means that
7503 operations are carried out in the precision of the registers and that
7504 it is unpredictable when rounding to the types specified in the source
7505 code takes place. When compiling C, if
7506 @option{-fexcess-precision=standard} is specified then excess
7507 precision will follow the rules specified in ISO C99; in particular,
7508 both casts and assignments cause values to be rounded to their
7509 semantic types (whereas @option{-ffloat-store} only affects
7510 assignments). This option is enabled by default for C if a strict
7511 conformance option such as @option{-std=c99} is used.
7514 @option{-fexcess-precision=standard} is not implemented for languages
7515 other than C, and has no effect if
7516 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7517 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7518 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7519 semantics apply without excess precision, and in the latter, rounding
7524 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7525 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7526 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7528 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7530 This option is not turned on by any @option{-O} option since
7531 it can result in incorrect output for programs which depend on
7532 an exact implementation of IEEE or ISO rules/specifications for
7533 math functions. It may, however, yield faster code for programs
7534 that do not require the guarantees of these specifications.
7536 @item -fno-math-errno
7537 @opindex fno-math-errno
7538 Do not set ERRNO after calling math functions that are executed
7539 with a single instruction, e.g., sqrt. A program that relies on
7540 IEEE exceptions for math error handling may want to use this flag
7541 for speed while maintaining IEEE arithmetic compatibility.
7543 This option is not turned on by any @option{-O} option since
7544 it can result in incorrect output for programs which depend on
7545 an exact implementation of IEEE or ISO rules/specifications for
7546 math functions. It may, however, yield faster code for programs
7547 that do not require the guarantees of these specifications.
7549 The default is @option{-fmath-errno}.
7551 On Darwin systems, the math library never sets @code{errno}. There is
7552 therefore no reason for the compiler to consider the possibility that
7553 it might, and @option{-fno-math-errno} is the default.
7555 @item -funsafe-math-optimizations
7556 @opindex funsafe-math-optimizations
7558 Allow optimizations for floating-point arithmetic that (a) assume
7559 that arguments and results are valid and (b) may violate IEEE or
7560 ANSI standards. When used at link-time, it may include libraries
7561 or startup files that change the default FPU control word or other
7562 similar optimizations.
7564 This option is not turned on by any @option{-O} option since
7565 it can result in incorrect output for programs which depend on
7566 an exact implementation of IEEE or ISO rules/specifications for
7567 math functions. It may, however, yield faster code for programs
7568 that do not require the guarantees of these specifications.
7569 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7570 @option{-fassociative-math} and @option{-freciprocal-math}.
7572 The default is @option{-fno-unsafe-math-optimizations}.
7574 @item -fassociative-math
7575 @opindex fassociative-math
7577 Allow re-association of operands in series of floating-point operations.
7578 This violates the ISO C and C++ language standard by possibly changing
7579 computation result. NOTE: re-ordering may change the sign of zero as
7580 well as ignore NaNs and inhibit or create underflow or overflow (and
7581 thus cannot be used on a code which relies on rounding behavior like
7582 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7583 and thus may not be used when ordered comparisons are required.
7584 This option requires that both @option{-fno-signed-zeros} and
7585 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7586 much sense with @option{-frounding-math}. For Fortran the option
7587 is automatically enabled when both @option{-fno-signed-zeros} and
7588 @option{-fno-trapping-math} are in effect.
7590 The default is @option{-fno-associative-math}.
7592 @item -freciprocal-math
7593 @opindex freciprocal-math
7595 Allow the reciprocal of a value to be used instead of dividing by
7596 the value if this enables optimizations. For example @code{x / y}
7597 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7598 is subject to common subexpression elimination. Note that this loses
7599 precision and increases the number of flops operating on the value.
7601 The default is @option{-fno-reciprocal-math}.
7603 @item -ffinite-math-only
7604 @opindex ffinite-math-only
7605 Allow optimizations for floating-point arithmetic that assume
7606 that arguments and results are not NaNs or +-Infs.
7608 This option is not turned on by any @option{-O} option since
7609 it can result in incorrect output for programs which depend on
7610 an exact implementation of IEEE or ISO rules/specifications for
7611 math functions. It may, however, yield faster code for programs
7612 that do not require the guarantees of these specifications.
7614 The default is @option{-fno-finite-math-only}.
7616 @item -fno-signed-zeros
7617 @opindex fno-signed-zeros
7618 Allow optimizations for floating point arithmetic that ignore the
7619 signedness of zero. IEEE arithmetic specifies the behavior of
7620 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7621 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7622 This option implies that the sign of a zero result isn't significant.
7624 The default is @option{-fsigned-zeros}.
7626 @item -fno-trapping-math
7627 @opindex fno-trapping-math
7628 Compile code assuming that floating-point operations cannot generate
7629 user-visible traps. These traps include division by zero, overflow,
7630 underflow, inexact result and invalid operation. This option requires
7631 that @option{-fno-signaling-nans} be in effect. Setting this option may
7632 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7634 This option should never be turned on by any @option{-O} option since
7635 it can result in incorrect output for programs which depend on
7636 an exact implementation of IEEE or ISO rules/specifications for
7639 The default is @option{-ftrapping-math}.
7641 @item -frounding-math
7642 @opindex frounding-math
7643 Disable transformations and optimizations that assume default floating
7644 point rounding behavior. This is round-to-zero for all floating point
7645 to integer conversions, and round-to-nearest for all other arithmetic
7646 truncations. This option should be specified for programs that change
7647 the FP rounding mode dynamically, or that may be executed with a
7648 non-default rounding mode. This option disables constant folding of
7649 floating point expressions at compile-time (which may be affected by
7650 rounding mode) and arithmetic transformations that are unsafe in the
7651 presence of sign-dependent rounding modes.
7653 The default is @option{-fno-rounding-math}.
7655 This option is experimental and does not currently guarantee to
7656 disable all GCC optimizations that are affected by rounding mode.
7657 Future versions of GCC may provide finer control of this setting
7658 using C99's @code{FENV_ACCESS} pragma. This command line option
7659 will be used to specify the default state for @code{FENV_ACCESS}.
7661 @item -fsignaling-nans
7662 @opindex fsignaling-nans
7663 Compile code assuming that IEEE signaling NaNs may generate user-visible
7664 traps during floating-point operations. Setting this option disables
7665 optimizations that may change the number of exceptions visible with
7666 signaling NaNs. This option implies @option{-ftrapping-math}.
7668 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7671 The default is @option{-fno-signaling-nans}.
7673 This option is experimental and does not currently guarantee to
7674 disable all GCC optimizations that affect signaling NaN behavior.
7676 @item -fsingle-precision-constant
7677 @opindex fsingle-precision-constant
7678 Treat floating point constant as single precision constant instead of
7679 implicitly converting it to double precision constant.
7681 @item -fcx-limited-range
7682 @opindex fcx-limited-range
7683 When enabled, this option states that a range reduction step is not
7684 needed when performing complex division. Also, there is no checking
7685 whether the result of a complex multiplication or division is @code{NaN
7686 + I*NaN}, with an attempt to rescue the situation in that case. The
7687 default is @option{-fno-cx-limited-range}, but is enabled by
7688 @option{-ffast-math}.
7690 This option controls the default setting of the ISO C99
7691 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7694 @item -fcx-fortran-rules
7695 @opindex fcx-fortran-rules
7696 Complex multiplication and division follow Fortran rules. Range
7697 reduction is done as part of complex division, but there is no checking
7698 whether the result of a complex multiplication or division is @code{NaN
7699 + I*NaN}, with an attempt to rescue the situation in that case.
7701 The default is @option{-fno-cx-fortran-rules}.
7705 The following options control optimizations that may improve
7706 performance, but are not enabled by any @option{-O} options. This
7707 section includes experimental options that may produce broken code.
7710 @item -fbranch-probabilities
7711 @opindex fbranch-probabilities
7712 After running a program compiled with @option{-fprofile-arcs}
7713 (@pxref{Debugging Options,, Options for Debugging Your Program or
7714 @command{gcc}}), you can compile it a second time using
7715 @option{-fbranch-probabilities}, to improve optimizations based on
7716 the number of times each branch was taken. When the program
7717 compiled with @option{-fprofile-arcs} exits it saves arc execution
7718 counts to a file called @file{@var{sourcename}.gcda} for each source
7719 file. The information in this data file is very dependent on the
7720 structure of the generated code, so you must use the same source code
7721 and the same optimization options for both compilations.
7723 With @option{-fbranch-probabilities}, GCC puts a
7724 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7725 These can be used to improve optimization. Currently, they are only
7726 used in one place: in @file{reorg.c}, instead of guessing which path a
7727 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7728 exactly determine which path is taken more often.
7730 @item -fprofile-values
7731 @opindex fprofile-values
7732 If combined with @option{-fprofile-arcs}, it adds code so that some
7733 data about values of expressions in the program is gathered.
7735 With @option{-fbranch-probabilities}, it reads back the data gathered
7736 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7737 notes to instructions for their later usage in optimizations.
7739 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7743 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7744 a code to gather information about values of expressions.
7746 With @option{-fbranch-probabilities}, it reads back the data gathered
7747 and actually performs the optimizations based on them.
7748 Currently the optimizations include specialization of division operation
7749 using the knowledge about the value of the denominator.
7751 @item -frename-registers
7752 @opindex frename-registers
7753 Attempt to avoid false dependencies in scheduled code by making use
7754 of registers left over after register allocation. This optimization
7755 will most benefit processors with lots of registers. Depending on the
7756 debug information format adopted by the target, however, it can
7757 make debugging impossible, since variables will no longer stay in
7758 a ``home register''.
7760 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
7764 Perform tail duplication to enlarge superblock size. This transformation
7765 simplifies the control flow of the function allowing other optimizations to do
7768 Enabled with @option{-fprofile-use}.
7770 @item -funroll-loops
7771 @opindex funroll-loops
7772 Unroll loops whose number of iterations can be determined at compile time or
7773 upon entry to the loop. @option{-funroll-loops} implies
7774 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7775 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7776 small constant number of iterations). This option makes code larger, and may
7777 or may not make it run faster.
7779 Enabled with @option{-fprofile-use}.
7781 @item -funroll-all-loops
7782 @opindex funroll-all-loops
7783 Unroll all loops, even if their number of iterations is uncertain when
7784 the loop is entered. This usually makes programs run more slowly.
7785 @option{-funroll-all-loops} implies the same options as
7786 @option{-funroll-loops}.
7789 @opindex fpeel-loops
7790 Peels the loops for that there is enough information that they do not
7791 roll much (from profile feedback). It also turns on complete loop peeling
7792 (i.e.@: complete removal of loops with small constant number of iterations).
7794 Enabled with @option{-fprofile-use}.
7796 @item -fmove-loop-invariants
7797 @opindex fmove-loop-invariants
7798 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7799 at level @option{-O1}
7801 @item -funswitch-loops
7802 @opindex funswitch-loops
7803 Move branches with loop invariant conditions out of the loop, with duplicates
7804 of the loop on both branches (modified according to result of the condition).
7806 @item -ffunction-sections
7807 @itemx -fdata-sections
7808 @opindex ffunction-sections
7809 @opindex fdata-sections
7810 Place each function or data item into its own section in the output
7811 file if the target supports arbitrary sections. The name of the
7812 function or the name of the data item determines the section's name
7815 Use these options on systems where the linker can perform optimizations
7816 to improve locality of reference in the instruction space. Most systems
7817 using the ELF object format and SPARC processors running Solaris 2 have
7818 linkers with such optimizations. AIX may have these optimizations in
7821 Only use these options when there are significant benefits from doing
7822 so. When you specify these options, the assembler and linker will
7823 create larger object and executable files and will also be slower.
7824 You will not be able to use @code{gprof} on all systems if you
7825 specify this option and you may have problems with debugging if
7826 you specify both this option and @option{-g}.
7828 @item -fbranch-target-load-optimize
7829 @opindex fbranch-target-load-optimize
7830 Perform branch target register load optimization before prologue / epilogue
7832 The use of target registers can typically be exposed only during reload,
7833 thus hoisting loads out of loops and doing inter-block scheduling needs
7834 a separate optimization pass.
7836 @item -fbranch-target-load-optimize2
7837 @opindex fbranch-target-load-optimize2
7838 Perform branch target register load optimization after prologue / epilogue
7841 @item -fbtr-bb-exclusive
7842 @opindex fbtr-bb-exclusive
7843 When performing branch target register load optimization, don't reuse
7844 branch target registers in within any basic block.
7846 @item -fstack-protector
7847 @opindex fstack-protector
7848 Emit extra code to check for buffer overflows, such as stack smashing
7849 attacks. This is done by adding a guard variable to functions with
7850 vulnerable objects. This includes functions that call alloca, and
7851 functions with buffers larger than 8 bytes. The guards are initialized
7852 when a function is entered and then checked when the function exits.
7853 If a guard check fails, an error message is printed and the program exits.
7855 @item -fstack-protector-all
7856 @opindex fstack-protector-all
7857 Like @option{-fstack-protector} except that all functions are protected.
7859 @item -fsection-anchors
7860 @opindex fsection-anchors
7861 Try to reduce the number of symbolic address calculations by using
7862 shared ``anchor'' symbols to address nearby objects. This transformation
7863 can help to reduce the number of GOT entries and GOT accesses on some
7866 For example, the implementation of the following function @code{foo}:
7870 int foo (void) @{ return a + b + c; @}
7873 would usually calculate the addresses of all three variables, but if you
7874 compile it with @option{-fsection-anchors}, it will access the variables
7875 from a common anchor point instead. The effect is similar to the
7876 following pseudocode (which isn't valid C):
7881 register int *xr = &x;
7882 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7886 Not all targets support this option.
7888 @item --param @var{name}=@var{value}
7890 In some places, GCC uses various constants to control the amount of
7891 optimization that is done. For example, GCC will not inline functions
7892 that contain more that a certain number of instructions. You can
7893 control some of these constants on the command-line using the
7894 @option{--param} option.
7896 The names of specific parameters, and the meaning of the values, are
7897 tied to the internals of the compiler, and are subject to change
7898 without notice in future releases.
7900 In each case, the @var{value} is an integer. The allowable choices for
7901 @var{name} are given in the following table:
7904 @item struct-reorg-cold-struct-ratio
7905 The threshold ratio (as a percentage) between a structure frequency
7906 and the frequency of the hottest structure in the program. This parameter
7907 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7908 We say that if the ratio of a structure frequency, calculated by profiling,
7909 to the hottest structure frequency in the program is less than this
7910 parameter, then structure reorganization is not applied to this structure.
7913 @item predictable-branch-outcome
7914 When branch is predicted to be taken with probability lower than this threshold
7915 (in percent), then it is considered well predictable. The default is 10.
7917 @item max-crossjump-edges
7918 The maximum number of incoming edges to consider for crossjumping.
7919 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7920 the number of edges incoming to each block. Increasing values mean
7921 more aggressive optimization, making the compile time increase with
7922 probably small improvement in executable size.
7924 @item min-crossjump-insns
7925 The minimum number of instructions which must be matched at the end
7926 of two blocks before crossjumping will be performed on them. This
7927 value is ignored in the case where all instructions in the block being
7928 crossjumped from are matched. The default value is 5.
7930 @item max-grow-copy-bb-insns
7931 The maximum code size expansion factor when copying basic blocks
7932 instead of jumping. The expansion is relative to a jump instruction.
7933 The default value is 8.
7935 @item max-goto-duplication-insns
7936 The maximum number of instructions to duplicate to a block that jumps
7937 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7938 passes, GCC factors computed gotos early in the compilation process,
7939 and unfactors them as late as possible. Only computed jumps at the
7940 end of a basic blocks with no more than max-goto-duplication-insns are
7941 unfactored. The default value is 8.
7943 @item max-delay-slot-insn-search
7944 The maximum number of instructions to consider when looking for an
7945 instruction to fill a delay slot. If more than this arbitrary number of
7946 instructions is searched, the time savings from filling the delay slot
7947 will be minimal so stop searching. Increasing values mean more
7948 aggressive optimization, making the compile time increase with probably
7949 small improvement in executable run time.
7951 @item max-delay-slot-live-search
7952 When trying to fill delay slots, the maximum number of instructions to
7953 consider when searching for a block with valid live register
7954 information. Increasing this arbitrarily chosen value means more
7955 aggressive optimization, increasing the compile time. This parameter
7956 should be removed when the delay slot code is rewritten to maintain the
7959 @item max-gcse-memory
7960 The approximate maximum amount of memory that will be allocated in
7961 order to perform the global common subexpression elimination
7962 optimization. If more memory than specified is required, the
7963 optimization will not be done.
7965 @item max-pending-list-length
7966 The maximum number of pending dependencies scheduling will allow
7967 before flushing the current state and starting over. Large functions
7968 with few branches or calls can create excessively large lists which
7969 needlessly consume memory and resources.
7971 @item max-inline-insns-single
7972 Several parameters control the tree inliner used in gcc.
7973 This number sets the maximum number of instructions (counted in GCC's
7974 internal representation) in a single function that the tree inliner
7975 will consider for inlining. This only affects functions declared
7976 inline and methods implemented in a class declaration (C++).
7977 The default value is 300.
7979 @item max-inline-insns-auto
7980 When you use @option{-finline-functions} (included in @option{-O3}),
7981 a lot of functions that would otherwise not be considered for inlining
7982 by the compiler will be investigated. To those functions, a different
7983 (more restrictive) limit compared to functions declared inline can
7985 The default value is 50.
7987 @item large-function-insns
7988 The limit specifying really large functions. For functions larger than this
7989 limit after inlining, inlining is constrained by
7990 @option{--param large-function-growth}. This parameter is useful primarily
7991 to avoid extreme compilation time caused by non-linear algorithms used by the
7993 The default value is 2700.
7995 @item large-function-growth
7996 Specifies maximal growth of large function caused by inlining in percents.
7997 The default value is 100 which limits large function growth to 2.0 times
8000 @item large-unit-insns
8001 The limit specifying large translation unit. Growth caused by inlining of
8002 units larger than this limit is limited by @option{--param inline-unit-growth}.
8003 For small units this might be too tight (consider unit consisting of function A
8004 that is inline and B that just calls A three time. If B is small relative to
8005 A, the growth of unit is 300\% and yet such inlining is very sane. For very
8006 large units consisting of small inlineable functions however the overall unit
8007 growth limit is needed to avoid exponential explosion of code size. Thus for
8008 smaller units, the size is increased to @option{--param large-unit-insns}
8009 before applying @option{--param inline-unit-growth}. The default is 10000
8011 @item inline-unit-growth
8012 Specifies maximal overall growth of the compilation unit caused by inlining.
8013 The default value is 30 which limits unit growth to 1.3 times the original
8016 @item ipcp-unit-growth
8017 Specifies maximal overall growth of the compilation unit caused by
8018 interprocedural constant propagation. The default value is 10 which limits
8019 unit growth to 1.1 times the original size.
8021 @item large-stack-frame
8022 The limit specifying large stack frames. While inlining the algorithm is trying
8023 to not grow past this limit too much. Default value is 256 bytes.
8025 @item large-stack-frame-growth
8026 Specifies maximal growth of large stack frames caused by inlining in percents.
8027 The default value is 1000 which limits large stack frame growth to 11 times
8030 @item max-inline-insns-recursive
8031 @itemx max-inline-insns-recursive-auto
8032 Specifies maximum number of instructions out-of-line copy of self recursive inline
8033 function can grow into by performing recursive inlining.
8035 For functions declared inline @option{--param max-inline-insns-recursive} is
8036 taken into account. For function not declared inline, recursive inlining
8037 happens only when @option{-finline-functions} (included in @option{-O3}) is
8038 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8039 default value is 450.
8041 @item max-inline-recursive-depth
8042 @itemx max-inline-recursive-depth-auto
8043 Specifies maximum recursion depth used by the recursive inlining.
8045 For functions declared inline @option{--param max-inline-recursive-depth} is
8046 taken into account. For function not declared inline, recursive inlining
8047 happens only when @option{-finline-functions} (included in @option{-O3}) is
8048 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8051 @item min-inline-recursive-probability
8052 Recursive inlining is profitable only for function having deep recursion
8053 in average and can hurt for function having little recursion depth by
8054 increasing the prologue size or complexity of function body to other
8057 When profile feedback is available (see @option{-fprofile-generate}) the actual
8058 recursion depth can be guessed from probability that function will recurse via
8059 given call expression. This parameter limits inlining only to call expression
8060 whose probability exceeds given threshold (in percents). The default value is
8063 @item early-inlining-insns
8064 Specify growth that early inliner can make. In effect it increases amount of
8065 inlining for code having large abstraction penalty. The default value is 8.
8067 @item max-early-inliner-iterations
8068 @itemx max-early-inliner-iterations
8069 Limit of iterations of early inliner. This basically bounds number of nested
8070 indirect calls early inliner can resolve. Deeper chains are still handled by
8073 @item min-vect-loop-bound
8074 The minimum number of iterations under which a loop will not get vectorized
8075 when @option{-ftree-vectorize} is used. The number of iterations after
8076 vectorization needs to be greater than the value specified by this option
8077 to allow vectorization. The default value is 0.
8079 @item max-unrolled-insns
8080 The maximum number of instructions that a loop should have if that loop
8081 is unrolled, and if the loop is unrolled, it determines how many times
8082 the loop code is unrolled.
8084 @item max-average-unrolled-insns
8085 The maximum number of instructions biased by probabilities of their execution
8086 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8087 it determines how many times the loop code is unrolled.
8089 @item max-unroll-times
8090 The maximum number of unrollings of a single loop.
8092 @item max-peeled-insns
8093 The maximum number of instructions that a loop should have if that loop
8094 is peeled, and if the loop is peeled, it determines how many times
8095 the loop code is peeled.
8097 @item max-peel-times
8098 The maximum number of peelings of a single loop.
8100 @item max-completely-peeled-insns
8101 The maximum number of insns of a completely peeled loop.
8103 @item max-completely-peel-times
8104 The maximum number of iterations of a loop to be suitable for complete peeling.
8106 @item max-completely-peel-loop-nest-depth
8107 The maximum depth of a loop nest suitable for complete peeling.
8109 @item max-unswitch-insns
8110 The maximum number of insns of an unswitched loop.
8112 @item max-unswitch-level
8113 The maximum number of branches unswitched in a single loop.
8116 The minimum cost of an expensive expression in the loop invariant motion.
8118 @item iv-consider-all-candidates-bound
8119 Bound on number of candidates for induction variables below that
8120 all candidates are considered for each use in induction variable
8121 optimizations. Only the most relevant candidates are considered
8122 if there are more candidates, to avoid quadratic time complexity.
8124 @item iv-max-considered-uses
8125 The induction variable optimizations give up on loops that contain more
8126 induction variable uses.
8128 @item iv-always-prune-cand-set-bound
8129 If number of candidates in the set is smaller than this value,
8130 we always try to remove unnecessary ivs from the set during its
8131 optimization when a new iv is added to the set.
8133 @item scev-max-expr-size
8134 Bound on size of expressions used in the scalar evolutions analyzer.
8135 Large expressions slow the analyzer.
8137 @item omega-max-vars
8138 The maximum number of variables in an Omega constraint system.
8139 The default value is 128.
8141 @item omega-max-geqs
8142 The maximum number of inequalities in an Omega constraint system.
8143 The default value is 256.
8146 The maximum number of equalities in an Omega constraint system.
8147 The default value is 128.
8149 @item omega-max-wild-cards
8150 The maximum number of wildcard variables that the Omega solver will
8151 be able to insert. The default value is 18.
8153 @item omega-hash-table-size
8154 The size of the hash table in the Omega solver. The default value is
8157 @item omega-max-keys
8158 The maximal number of keys used by the Omega solver. The default
8161 @item omega-eliminate-redundant-constraints
8162 When set to 1, use expensive methods to eliminate all redundant
8163 constraints. The default value is 0.
8165 @item vect-max-version-for-alignment-checks
8166 The maximum number of runtime checks that can be performed when
8167 doing loop versioning for alignment in the vectorizer. See option
8168 ftree-vect-loop-version for more information.
8170 @item vect-max-version-for-alias-checks
8171 The maximum number of runtime checks that can be performed when
8172 doing loop versioning for alias in the vectorizer. See option
8173 ftree-vect-loop-version for more information.
8175 @item max-iterations-to-track
8177 The maximum number of iterations of a loop the brute force algorithm
8178 for analysis of # of iterations of the loop tries to evaluate.
8180 @item hot-bb-count-fraction
8181 Select fraction of the maximal count of repetitions of basic block in program
8182 given basic block needs to have to be considered hot.
8184 @item hot-bb-frequency-fraction
8185 Select fraction of the maximal frequency of executions of basic block in
8186 function given basic block needs to have to be considered hot
8188 @item max-predicted-iterations
8189 The maximum number of loop iterations we predict statically. This is useful
8190 in cases where function contain single loop with known bound and other loop
8191 with unknown. We predict the known number of iterations correctly, while
8192 the unknown number of iterations average to roughly 10. This means that the
8193 loop without bounds would appear artificially cold relative to the other one.
8195 @item align-threshold
8197 Select fraction of the maximal frequency of executions of basic block in
8198 function given basic block will get aligned.
8200 @item align-loop-iterations
8202 A loop expected to iterate at lest the selected number of iterations will get
8205 @item tracer-dynamic-coverage
8206 @itemx tracer-dynamic-coverage-feedback
8208 This value is used to limit superblock formation once the given percentage of
8209 executed instructions is covered. This limits unnecessary code size
8212 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8213 feedback is available. The real profiles (as opposed to statically estimated
8214 ones) are much less balanced allowing the threshold to be larger value.
8216 @item tracer-max-code-growth
8217 Stop tail duplication once code growth has reached given percentage. This is
8218 rather hokey argument, as most of the duplicates will be eliminated later in
8219 cross jumping, so it may be set to much higher values than is the desired code
8222 @item tracer-min-branch-ratio
8224 Stop reverse growth when the reverse probability of best edge is less than this
8225 threshold (in percent).
8227 @item tracer-min-branch-ratio
8228 @itemx tracer-min-branch-ratio-feedback
8230 Stop forward growth if the best edge do have probability lower than this
8233 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8234 compilation for profile feedback and one for compilation without. The value
8235 for compilation with profile feedback needs to be more conservative (higher) in
8236 order to make tracer effective.
8238 @item max-cse-path-length
8240 Maximum number of basic blocks on path that cse considers. The default is 10.
8243 The maximum instructions CSE process before flushing. The default is 1000.
8245 @item ggc-min-expand
8247 GCC uses a garbage collector to manage its own memory allocation. This
8248 parameter specifies the minimum percentage by which the garbage
8249 collector's heap should be allowed to expand between collections.
8250 Tuning this may improve compilation speed; it has no effect on code
8253 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8254 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8255 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8256 GCC is not able to calculate RAM on a particular platform, the lower
8257 bound of 30% is used. Setting this parameter and
8258 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8259 every opportunity. This is extremely slow, but can be useful for
8262 @item ggc-min-heapsize
8264 Minimum size of the garbage collector's heap before it begins bothering
8265 to collect garbage. The first collection occurs after the heap expands
8266 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8267 tuning this may improve compilation speed, and has no effect on code
8270 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8271 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8272 with a lower bound of 4096 (four megabytes) and an upper bound of
8273 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8274 particular platform, the lower bound is used. Setting this parameter
8275 very large effectively disables garbage collection. Setting this
8276 parameter and @option{ggc-min-expand} to zero causes a full collection
8277 to occur at every opportunity.
8279 @item max-reload-search-insns
8280 The maximum number of instruction reload should look backward for equivalent
8281 register. Increasing values mean more aggressive optimization, making the
8282 compile time increase with probably slightly better performance. The default
8285 @item max-cselib-memory-locations
8286 The maximum number of memory locations cselib should take into account.
8287 Increasing values mean more aggressive optimization, making the compile time
8288 increase with probably slightly better performance. The default value is 500.
8290 @item reorder-blocks-duplicate
8291 @itemx reorder-blocks-duplicate-feedback
8293 Used by basic block reordering pass to decide whether to use unconditional
8294 branch or duplicate the code on its destination. Code is duplicated when its
8295 estimated size is smaller than this value multiplied by the estimated size of
8296 unconditional jump in the hot spots of the program.
8298 The @option{reorder-block-duplicate-feedback} is used only when profile
8299 feedback is available and may be set to higher values than
8300 @option{reorder-block-duplicate} since information about the hot spots is more
8303 @item max-sched-ready-insns
8304 The maximum number of instructions ready to be issued the scheduler should
8305 consider at any given time during the first scheduling pass. Increasing
8306 values mean more thorough searches, making the compilation time increase
8307 with probably little benefit. The default value is 100.
8309 @item max-sched-region-blocks
8310 The maximum number of blocks in a region to be considered for
8311 interblock scheduling. The default value is 10.
8313 @item max-pipeline-region-blocks
8314 The maximum number of blocks in a region to be considered for
8315 pipelining in the selective scheduler. The default value is 15.
8317 @item max-sched-region-insns
8318 The maximum number of insns in a region to be considered for
8319 interblock scheduling. The default value is 100.
8321 @item max-pipeline-region-insns
8322 The maximum number of insns in a region to be considered for
8323 pipelining in the selective scheduler. The default value is 200.
8326 The minimum probability (in percents) of reaching a source block
8327 for interblock speculative scheduling. The default value is 40.
8329 @item max-sched-extend-regions-iters
8330 The maximum number of iterations through CFG to extend regions.
8331 0 - disable region extension,
8332 N - do at most N iterations.
8333 The default value is 0.
8335 @item max-sched-insn-conflict-delay
8336 The maximum conflict delay for an insn to be considered for speculative motion.
8337 The default value is 3.
8339 @item sched-spec-prob-cutoff
8340 The minimal probability of speculation success (in percents), so that
8341 speculative insn will be scheduled.
8342 The default value is 40.
8344 @item sched-mem-true-dep-cost
8345 Minimal distance (in CPU cycles) between store and load targeting same
8346 memory locations. The default value is 1.
8348 @item selsched-max-lookahead
8349 The maximum size of the lookahead window of selective scheduling. It is a
8350 depth of search for available instructions.
8351 The default value is 50.
8353 @item selsched-max-sched-times
8354 The maximum number of times that an instruction will be scheduled during
8355 selective scheduling. This is the limit on the number of iterations
8356 through which the instruction may be pipelined. The default value is 2.
8358 @item selsched-max-insns-to-rename
8359 The maximum number of best instructions in the ready list that are considered
8360 for renaming in the selective scheduler. The default value is 2.
8362 @item max-last-value-rtl
8363 The maximum size measured as number of RTLs that can be recorded in an expression
8364 in combiner for a pseudo register as last known value of that register. The default
8367 @item integer-share-limit
8368 Small integer constants can use a shared data structure, reducing the
8369 compiler's memory usage and increasing its speed. This sets the maximum
8370 value of a shared integer constant. The default value is 256.
8372 @item min-virtual-mappings
8373 Specifies the minimum number of virtual mappings in the incremental
8374 SSA updater that should be registered to trigger the virtual mappings
8375 heuristic defined by virtual-mappings-ratio. The default value is
8378 @item virtual-mappings-ratio
8379 If the number of virtual mappings is virtual-mappings-ratio bigger
8380 than the number of virtual symbols to be updated, then the incremental
8381 SSA updater switches to a full update for those symbols. The default
8384 @item ssp-buffer-size
8385 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8386 protection when @option{-fstack-protection} is used.
8388 @item max-jump-thread-duplication-stmts
8389 Maximum number of statements allowed in a block that needs to be
8390 duplicated when threading jumps.
8392 @item max-fields-for-field-sensitive
8393 Maximum number of fields in a structure we will treat in
8394 a field sensitive manner during pointer analysis. The default is zero
8395 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8397 @item prefetch-latency
8398 Estimate on average number of instructions that are executed before
8399 prefetch finishes. The distance we prefetch ahead is proportional
8400 to this constant. Increasing this number may also lead to less
8401 streams being prefetched (see @option{simultaneous-prefetches}).
8403 @item simultaneous-prefetches
8404 Maximum number of prefetches that can run at the same time.
8406 @item l1-cache-line-size
8407 The size of cache line in L1 cache, in bytes.
8410 The size of L1 cache, in kilobytes.
8413 The size of L2 cache, in kilobytes.
8415 @item min-insn-to-prefetch-ratio
8416 The minimum ratio between the number of instructions and the
8417 number of prefetches to enable prefetching in a loop with an
8420 @item prefetch-min-insn-to-mem-ratio
8421 The minimum ratio between the number of instructions and the
8422 number of memory references to enable prefetching in a loop.
8424 @item use-canonical-types
8425 Whether the compiler should use the ``canonical'' type system. By
8426 default, this should always be 1, which uses a more efficient internal
8427 mechanism for comparing types in C++ and Objective-C++. However, if
8428 bugs in the canonical type system are causing compilation failures,
8429 set this value to 0 to disable canonical types.
8431 @item switch-conversion-max-branch-ratio
8432 Switch initialization conversion will refuse to create arrays that are
8433 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8434 branches in the switch.
8436 @item max-partial-antic-length
8437 Maximum length of the partial antic set computed during the tree
8438 partial redundancy elimination optimization (@option{-ftree-pre}) when
8439 optimizing at @option{-O3} and above. For some sorts of source code
8440 the enhanced partial redundancy elimination optimization can run away,
8441 consuming all of the memory available on the host machine. This
8442 parameter sets a limit on the length of the sets that are computed,
8443 which prevents the runaway behavior. Setting a value of 0 for
8444 this parameter will allow an unlimited set length.
8446 @item sccvn-max-scc-size
8447 Maximum size of a strongly connected component (SCC) during SCCVN
8448 processing. If this limit is hit, SCCVN processing for the whole
8449 function will not be done and optimizations depending on it will
8450 be disabled. The default maximum SCC size is 10000.
8452 @item ira-max-loops-num
8453 IRA uses a regional register allocation by default. If a function
8454 contains loops more than number given by the parameter, only at most
8455 given number of the most frequently executed loops will form regions
8456 for the regional register allocation. The default value of the
8459 @item ira-max-conflict-table-size
8460 Although IRA uses a sophisticated algorithm of compression conflict
8461 table, the table can be still big for huge functions. If the conflict
8462 table for a function could be more than size in MB given by the
8463 parameter, the conflict table is not built and faster, simpler, and
8464 lower quality register allocation algorithm will be used. The
8465 algorithm do not use pseudo-register conflicts. The default value of
8466 the parameter is 2000.
8468 @item ira-loop-reserved-regs
8469 IRA can be used to evaluate more accurate register pressure in loops
8470 for decision to move loop invariants (see @option{-O3}). The number
8471 of available registers reserved for some other purposes is described
8472 by this parameter. The default value of the parameter is 2 which is
8473 minimal number of registers needed for execution of typical
8474 instruction. This value is the best found from numerous experiments.
8476 @item loop-invariant-max-bbs-in-loop
8477 Loop invariant motion can be very expensive, both in compile time and
8478 in amount of needed compile time memory, with very large loops. Loops
8479 with more basic blocks than this parameter won't have loop invariant
8480 motion optimization performed on them. The default value of the
8481 parameter is 1000 for -O1 and 10000 for -O2 and above.
8483 @item max-vartrack-size
8484 Sets a maximum number of hash table slots to use during variable
8485 tracking dataflow analysis of any function. If this limit is exceeded
8486 with variable tracking at assignments enabled, analysis for that
8487 function is retried without it, after removing all debug insns from
8488 the function. If the limit is exceeded even without debug insns, var
8489 tracking analysis is completely disabled for the function. Setting
8490 the parameter to zero makes it unlimited.
8492 @item min-nondebug-insn-uid
8493 Use uids starting at this parameter for nondebug insns. The range below
8494 the parameter is reserved exclusively for debug insns created by
8495 @option{-fvar-tracking-assignments}, but debug insns may get
8496 (non-overlapping) uids above it if the reserved range is exhausted.
8498 @item ipa-sra-ptr-growth-factor
8499 IPA-SRA will replace a pointer to an aggregate with one or more new
8500 parameters only when their cumulative size is less or equal to
8501 @option{ipa-sra-ptr-growth-factor} times the size of the original
8504 @item graphite-max-nb-scop-params
8505 To avoid exponential effects in the Graphite loop transforms, the
8506 number of parameters in a Static Control Part (SCoP) is bounded. The
8507 default value is 10 parameters. A variable whose value is unknown at
8508 compile time and defined outside a SCoP is a parameter of the SCoP.
8510 @item graphite-max-bbs-per-function
8511 To avoid exponential effects in the detection of SCoPs, the size of
8512 the functions analyzed by Graphite is bounded. The default value is
8515 @item loop-block-tile-size
8516 Loop blocking or strip mining transforms, enabled with
8517 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
8518 loop in the loop nest by a given number of iterations. The strip
8519 length can be changed using the @option{loop-block-tile-size}
8520 parameter. The default value is 51 iterations.
8525 @node Preprocessor Options
8526 @section Options Controlling the Preprocessor
8527 @cindex preprocessor options
8528 @cindex options, preprocessor
8530 These options control the C preprocessor, which is run on each C source
8531 file before actual compilation.
8533 If you use the @option{-E} option, nothing is done except preprocessing.
8534 Some of these options make sense only together with @option{-E} because
8535 they cause the preprocessor output to be unsuitable for actual
8539 @item -Wp,@var{option}
8541 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8542 and pass @var{option} directly through to the preprocessor. If
8543 @var{option} contains commas, it is split into multiple options at the
8544 commas. However, many options are modified, translated or interpreted
8545 by the compiler driver before being passed to the preprocessor, and
8546 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8547 interface is undocumented and subject to change, so whenever possible
8548 you should avoid using @option{-Wp} and let the driver handle the
8551 @item -Xpreprocessor @var{option}
8552 @opindex Xpreprocessor
8553 Pass @var{option} as an option to the preprocessor. You can use this to
8554 supply system-specific preprocessor options which GCC does not know how to
8557 If you want to pass an option that takes an argument, you must use
8558 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8561 @include cppopts.texi
8563 @node Assembler Options
8564 @section Passing Options to the Assembler
8566 @c prevent bad page break with this line
8567 You can pass options to the assembler.
8570 @item -Wa,@var{option}
8572 Pass @var{option} as an option to the assembler. If @var{option}
8573 contains commas, it is split into multiple options at the commas.
8575 @item -Xassembler @var{option}
8577 Pass @var{option} as an option to the assembler. You can use this to
8578 supply system-specific assembler options which GCC does not know how to
8581 If you want to pass an option that takes an argument, you must use
8582 @option{-Xassembler} twice, once for the option and once for the argument.
8587 @section Options for Linking
8588 @cindex link options
8589 @cindex options, linking
8591 These options come into play when the compiler links object files into
8592 an executable output file. They are meaningless if the compiler is
8593 not doing a link step.
8597 @item @var{object-file-name}
8598 A file name that does not end in a special recognized suffix is
8599 considered to name an object file or library. (Object files are
8600 distinguished from libraries by the linker according to the file
8601 contents.) If linking is done, these object files are used as input
8610 If any of these options is used, then the linker is not run, and
8611 object file names should not be used as arguments. @xref{Overall
8615 @item -l@var{library}
8616 @itemx -l @var{library}
8618 Search the library named @var{library} when linking. (The second
8619 alternative with the library as a separate argument is only for
8620 POSIX compliance and is not recommended.)
8622 It makes a difference where in the command you write this option; the
8623 linker searches and processes libraries and object files in the order they
8624 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8625 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8626 to functions in @samp{z}, those functions may not be loaded.
8628 The linker searches a standard list of directories for the library,
8629 which is actually a file named @file{lib@var{library}.a}. The linker
8630 then uses this file as if it had been specified precisely by name.
8632 The directories searched include several standard system directories
8633 plus any that you specify with @option{-L}.
8635 Normally the files found this way are library files---archive files
8636 whose members are object files. The linker handles an archive file by
8637 scanning through it for members which define symbols that have so far
8638 been referenced but not defined. But if the file that is found is an
8639 ordinary object file, it is linked in the usual fashion. The only
8640 difference between using an @option{-l} option and specifying a file name
8641 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8642 and searches several directories.
8646 You need this special case of the @option{-l} option in order to
8647 link an Objective-C or Objective-C++ program.
8650 @opindex nostartfiles
8651 Do not use the standard system startup files when linking.
8652 The standard system libraries are used normally, unless @option{-nostdlib}
8653 or @option{-nodefaultlibs} is used.
8655 @item -nodefaultlibs
8656 @opindex nodefaultlibs
8657 Do not use the standard system libraries when linking.
8658 Only the libraries you specify will be passed to the linker, options
8659 specifying linkage of the system libraries, such as @code{-static-libgcc}
8660 or @code{-shared-libgcc}, will be ignored.
8661 The standard startup files are used normally, unless @option{-nostartfiles}
8662 is used. The compiler may generate calls to @code{memcmp},
8663 @code{memset}, @code{memcpy} and @code{memmove}.
8664 These entries are usually resolved by entries in
8665 libc. These entry points should be supplied through some other
8666 mechanism when this option is specified.
8670 Do not use the standard system startup files or libraries when linking.
8671 No startup files and only the libraries you specify will be passed to
8672 the linker, options specifying linkage of the system libraries, such as
8673 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8674 The compiler may generate calls to @code{memcmp}, @code{memset},
8675 @code{memcpy} and @code{memmove}.
8676 These entries are usually resolved by entries in
8677 libc. These entry points should be supplied through some other
8678 mechanism when this option is specified.
8680 @cindex @option{-lgcc}, use with @option{-nostdlib}
8681 @cindex @option{-nostdlib} and unresolved references
8682 @cindex unresolved references and @option{-nostdlib}
8683 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8684 @cindex @option{-nodefaultlibs} and unresolved references
8685 @cindex unresolved references and @option{-nodefaultlibs}
8686 One of the standard libraries bypassed by @option{-nostdlib} and
8687 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8688 that GCC uses to overcome shortcomings of particular machines, or special
8689 needs for some languages.
8690 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8691 Collection (GCC) Internals},
8692 for more discussion of @file{libgcc.a}.)
8693 In most cases, you need @file{libgcc.a} even when you want to avoid
8694 other standard libraries. In other words, when you specify @option{-nostdlib}
8695 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8696 This ensures that you have no unresolved references to internal GCC
8697 library subroutines. (For example, @samp{__main}, used to ensure C++
8698 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8699 GNU Compiler Collection (GCC) Internals}.)
8703 Produce a position independent executable on targets which support it.
8704 For predictable results, you must also specify the same set of options
8705 that were used to generate code (@option{-fpie}, @option{-fPIE},
8706 or model suboptions) when you specify this option.
8710 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8711 that support it. This instructs the linker to add all symbols, not
8712 only used ones, to the dynamic symbol table. This option is needed
8713 for some uses of @code{dlopen} or to allow obtaining backtraces
8714 from within a program.
8718 Remove all symbol table and relocation information from the executable.
8722 On systems that support dynamic linking, this prevents linking with the shared
8723 libraries. On other systems, this option has no effect.
8727 Produce a shared object which can then be linked with other objects to
8728 form an executable. Not all systems support this option. For predictable
8729 results, you must also specify the same set of options that were used to
8730 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8731 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8732 needs to build supplementary stub code for constructors to work. On
8733 multi-libbed systems, @samp{gcc -shared} must select the correct support
8734 libraries to link against. Failing to supply the correct flags may lead
8735 to subtle defects. Supplying them in cases where they are not necessary
8738 @item -shared-libgcc
8739 @itemx -static-libgcc
8740 @opindex shared-libgcc
8741 @opindex static-libgcc
8742 On systems that provide @file{libgcc} as a shared library, these options
8743 force the use of either the shared or static version respectively.
8744 If no shared version of @file{libgcc} was built when the compiler was
8745 configured, these options have no effect.
8747 There are several situations in which an application should use the
8748 shared @file{libgcc} instead of the static version. The most common
8749 of these is when the application wishes to throw and catch exceptions
8750 across different shared libraries. In that case, each of the libraries
8751 as well as the application itself should use the shared @file{libgcc}.
8753 Therefore, the G++ and GCJ drivers automatically add
8754 @option{-shared-libgcc} whenever you build a shared library or a main
8755 executable, because C++ and Java programs typically use exceptions, so
8756 this is the right thing to do.
8758 If, instead, you use the GCC driver to create shared libraries, you may
8759 find that they will not always be linked with the shared @file{libgcc}.
8760 If GCC finds, at its configuration time, that you have a non-GNU linker
8761 or a GNU linker that does not support option @option{--eh-frame-hdr},
8762 it will link the shared version of @file{libgcc} into shared libraries
8763 by default. Otherwise, it will take advantage of the linker and optimize
8764 away the linking with the shared version of @file{libgcc}, linking with
8765 the static version of libgcc by default. This allows exceptions to
8766 propagate through such shared libraries, without incurring relocation
8767 costs at library load time.
8769 However, if a library or main executable is supposed to throw or catch
8770 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8771 for the languages used in the program, or using the option
8772 @option{-shared-libgcc}, such that it is linked with the shared
8775 @item -static-libstdc++
8776 When the @command{g++} program is used to link a C++ program, it will
8777 normally automatically link against @option{libstdc++}. If
8778 @file{libstdc++} is available as a shared library, and the
8779 @option{-static} option is not used, then this will link against the
8780 shared version of @file{libstdc++}. That is normally fine. However, it
8781 is sometimes useful to freeze the version of @file{libstdc++} used by
8782 the program without going all the way to a fully static link. The
8783 @option{-static-libstdc++} option directs the @command{g++} driver to
8784 link @file{libstdc++} statically, without necessarily linking other
8785 libraries statically.
8789 Bind references to global symbols when building a shared object. Warn
8790 about any unresolved references (unless overridden by the link editor
8791 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8794 @item -T @var{script}
8796 @cindex linker script
8797 Use @var{script} as the linker script. This option is supported by most
8798 systems using the GNU linker. On some targets, such as bare-board
8799 targets without an operating system, the @option{-T} option may be required
8800 when linking to avoid references to undefined symbols.
8802 @item -Xlinker @var{option}
8804 Pass @var{option} as an option to the linker. You can use this to
8805 supply system-specific linker options which GCC does not know how to
8808 If you want to pass an option that takes a separate argument, you must use
8809 @option{-Xlinker} twice, once for the option and once for the argument.
8810 For example, to pass @option{-assert definitions}, you must write
8811 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8812 @option{-Xlinker "-assert definitions"}, because this passes the entire
8813 string as a single argument, which is not what the linker expects.
8815 When using the GNU linker, it is usually more convenient to pass
8816 arguments to linker options using the @option{@var{option}=@var{value}}
8817 syntax than as separate arguments. For example, you can specify
8818 @samp{-Xlinker -Map=output.map} rather than
8819 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8820 this syntax for command-line options.
8822 @item -Wl,@var{option}
8824 Pass @var{option} as an option to the linker. If @var{option} contains
8825 commas, it is split into multiple options at the commas. You can use this
8826 syntax to pass an argument to the option.
8827 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8828 linker. When using the GNU linker, you can also get the same effect with
8829 @samp{-Wl,-Map=output.map}.
8831 @item -u @var{symbol}
8833 Pretend the symbol @var{symbol} is undefined, to force linking of
8834 library modules to define it. You can use @option{-u} multiple times with
8835 different symbols to force loading of additional library modules.
8838 @node Directory Options
8839 @section Options for Directory Search
8840 @cindex directory options
8841 @cindex options, directory search
8844 These options specify directories to search for header files, for
8845 libraries and for parts of the compiler:
8850 Add the directory @var{dir} to the head of the list of directories to be
8851 searched for header files. This can be used to override a system header
8852 file, substituting your own version, since these directories are
8853 searched before the system header file directories. However, you should
8854 not use this option to add directories that contain vendor-supplied
8855 system header files (use @option{-isystem} for that). If you use more than
8856 one @option{-I} option, the directories are scanned in left-to-right
8857 order; the standard system directories come after.
8859 If a standard system include directory, or a directory specified with
8860 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8861 option will be ignored. The directory will still be searched but as a
8862 system directory at its normal position in the system include chain.
8863 This is to ensure that GCC's procedure to fix buggy system headers and
8864 the ordering for the include_next directive are not inadvertently changed.
8865 If you really need to change the search order for system directories,
8866 use the @option{-nostdinc} and/or @option{-isystem} options.
8868 @item -iquote@var{dir}
8870 Add the directory @var{dir} to the head of the list of directories to
8871 be searched for header files only for the case of @samp{#include
8872 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8873 otherwise just like @option{-I}.
8877 Add directory @var{dir} to the list of directories to be searched
8880 @item -B@var{prefix}
8882 This option specifies where to find the executables, libraries,
8883 include files, and data files of the compiler itself.
8885 The compiler driver program runs one or more of the subprograms
8886 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8887 @var{prefix} as a prefix for each program it tries to run, both with and
8888 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8890 For each subprogram to be run, the compiler driver first tries the
8891 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8892 was not specified, the driver tries two standard prefixes, which are
8893 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8894 those results in a file name that is found, the unmodified program
8895 name is searched for using the directories specified in your
8896 @env{PATH} environment variable.
8898 The compiler will check to see if the path provided by the @option{-B}
8899 refers to a directory, and if necessary it will add a directory
8900 separator character at the end of the path.
8902 @option{-B} prefixes that effectively specify directory names also apply
8903 to libraries in the linker, because the compiler translates these
8904 options into @option{-L} options for the linker. They also apply to
8905 includes files in the preprocessor, because the compiler translates these
8906 options into @option{-isystem} options for the preprocessor. In this case,
8907 the compiler appends @samp{include} to the prefix.
8909 The run-time support file @file{libgcc.a} can also be searched for using
8910 the @option{-B} prefix, if needed. If it is not found there, the two
8911 standard prefixes above are tried, and that is all. The file is left
8912 out of the link if it is not found by those means.
8914 Another way to specify a prefix much like the @option{-B} prefix is to use
8915 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8918 As a special kludge, if the path provided by @option{-B} is
8919 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8920 9, then it will be replaced by @file{[dir/]include}. This is to help
8921 with boot-strapping the compiler.
8923 @item -specs=@var{file}
8925 Process @var{file} after the compiler reads in the standard @file{specs}
8926 file, in order to override the defaults that the @file{gcc} driver
8927 program uses when determining what switches to pass to @file{cc1},
8928 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8929 @option{-specs=@var{file}} can be specified on the command line, and they
8930 are processed in order, from left to right.
8932 @item --sysroot=@var{dir}
8934 Use @var{dir} as the logical root directory for headers and libraries.
8935 For example, if the compiler would normally search for headers in
8936 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8937 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8939 If you use both this option and the @option{-isysroot} option, then
8940 the @option{--sysroot} option will apply to libraries, but the
8941 @option{-isysroot} option will apply to header files.
8943 The GNU linker (beginning with version 2.16) has the necessary support
8944 for this option. If your linker does not support this option, the
8945 header file aspect of @option{--sysroot} will still work, but the
8946 library aspect will not.
8950 This option has been deprecated. Please use @option{-iquote} instead for
8951 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8952 Any directories you specify with @option{-I} options before the @option{-I-}
8953 option are searched only for the case of @samp{#include "@var{file}"};
8954 they are not searched for @samp{#include <@var{file}>}.
8956 If additional directories are specified with @option{-I} options after
8957 the @option{-I-}, these directories are searched for all @samp{#include}
8958 directives. (Ordinarily @emph{all} @option{-I} directories are used
8961 In addition, the @option{-I-} option inhibits the use of the current
8962 directory (where the current input file came from) as the first search
8963 directory for @samp{#include "@var{file}"}. There is no way to
8964 override this effect of @option{-I-}. With @option{-I.} you can specify
8965 searching the directory which was current when the compiler was
8966 invoked. That is not exactly the same as what the preprocessor does
8967 by default, but it is often satisfactory.
8969 @option{-I-} does not inhibit the use of the standard system directories
8970 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8977 @section Specifying subprocesses and the switches to pass to them
8980 @command{gcc} is a driver program. It performs its job by invoking a
8981 sequence of other programs to do the work of compiling, assembling and
8982 linking. GCC interprets its command-line parameters and uses these to
8983 deduce which programs it should invoke, and which command-line options
8984 it ought to place on their command lines. This behavior is controlled
8985 by @dfn{spec strings}. In most cases there is one spec string for each
8986 program that GCC can invoke, but a few programs have multiple spec
8987 strings to control their behavior. The spec strings built into GCC can
8988 be overridden by using the @option{-specs=} command-line switch to specify
8991 @dfn{Spec files} are plaintext files that are used to construct spec
8992 strings. They consist of a sequence of directives separated by blank
8993 lines. The type of directive is determined by the first non-whitespace
8994 character on the line and it can be one of the following:
8997 @item %@var{command}
8998 Issues a @var{command} to the spec file processor. The commands that can
9002 @item %include <@var{file}>
9004 Search for @var{file} and insert its text at the current point in the
9007 @item %include_noerr <@var{file}>
9008 @cindex %include_noerr
9009 Just like @samp{%include}, but do not generate an error message if the include
9010 file cannot be found.
9012 @item %rename @var{old_name} @var{new_name}
9014 Rename the spec string @var{old_name} to @var{new_name}.
9018 @item *[@var{spec_name}]:
9019 This tells the compiler to create, override or delete the named spec
9020 string. All lines after this directive up to the next directive or
9021 blank line are considered to be the text for the spec string. If this
9022 results in an empty string then the spec will be deleted. (Or, if the
9023 spec did not exist, then nothing will happened.) Otherwise, if the spec
9024 does not currently exist a new spec will be created. If the spec does
9025 exist then its contents will be overridden by the text of this
9026 directive, unless the first character of that text is the @samp{+}
9027 character, in which case the text will be appended to the spec.
9029 @item [@var{suffix}]:
9030 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
9031 and up to the next directive or blank line are considered to make up the
9032 spec string for the indicated suffix. When the compiler encounters an
9033 input file with the named suffix, it will processes the spec string in
9034 order to work out how to compile that file. For example:
9041 This says that any input file whose name ends in @samp{.ZZ} should be
9042 passed to the program @samp{z-compile}, which should be invoked with the
9043 command-line switch @option{-input} and with the result of performing the
9044 @samp{%i} substitution. (See below.)
9046 As an alternative to providing a spec string, the text that follows a
9047 suffix directive can be one of the following:
9050 @item @@@var{language}
9051 This says that the suffix is an alias for a known @var{language}. This is
9052 similar to using the @option{-x} command-line switch to GCC to specify a
9053 language explicitly. For example:
9060 Says that .ZZ files are, in fact, C++ source files.
9063 This causes an error messages saying:
9066 @var{name} compiler not installed on this system.
9070 GCC already has an extensive list of suffixes built into it.
9071 This directive will add an entry to the end of the list of suffixes, but
9072 since the list is searched from the end backwards, it is effectively
9073 possible to override earlier entries using this technique.
9077 GCC has the following spec strings built into it. Spec files can
9078 override these strings or create their own. Note that individual
9079 targets can also add their own spec strings to this list.
9082 asm Options to pass to the assembler
9083 asm_final Options to pass to the assembler post-processor
9084 cpp Options to pass to the C preprocessor
9085 cc1 Options to pass to the C compiler
9086 cc1plus Options to pass to the C++ compiler
9087 endfile Object files to include at the end of the link
9088 link Options to pass to the linker
9089 lib Libraries to include on the command line to the linker
9090 libgcc Decides which GCC support library to pass to the linker
9091 linker Sets the name of the linker
9092 predefines Defines to be passed to the C preprocessor
9093 signed_char Defines to pass to CPP to say whether @code{char} is signed
9095 startfile Object files to include at the start of the link
9098 Here is a small example of a spec file:
9104 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9107 This example renames the spec called @samp{lib} to @samp{old_lib} and
9108 then overrides the previous definition of @samp{lib} with a new one.
9109 The new definition adds in some extra command-line options before
9110 including the text of the old definition.
9112 @dfn{Spec strings} are a list of command-line options to be passed to their
9113 corresponding program. In addition, the spec strings can contain
9114 @samp{%}-prefixed sequences to substitute variable text or to
9115 conditionally insert text into the command line. Using these constructs
9116 it is possible to generate quite complex command lines.
9118 Here is a table of all defined @samp{%}-sequences for spec
9119 strings. Note that spaces are not generated automatically around the
9120 results of expanding these sequences. Therefore you can concatenate them
9121 together or combine them with constant text in a single argument.
9125 Substitute one @samp{%} into the program name or argument.
9128 Substitute the name of the input file being processed.
9131 Substitute the basename of the input file being processed.
9132 This is the substring up to (and not including) the last period
9133 and not including the directory.
9136 This is the same as @samp{%b}, but include the file suffix (text after
9140 Marks the argument containing or following the @samp{%d} as a
9141 temporary file name, so that that file will be deleted if GCC exits
9142 successfully. Unlike @samp{%g}, this contributes no text to the
9145 @item %g@var{suffix}
9146 Substitute a file name that has suffix @var{suffix} and is chosen
9147 once per compilation, and mark the argument in the same way as
9148 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9149 name is now chosen in a way that is hard to predict even when previously
9150 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9151 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9152 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9153 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9154 was simply substituted with a file name chosen once per compilation,
9155 without regard to any appended suffix (which was therefore treated
9156 just like ordinary text), making such attacks more likely to succeed.
9158 @item %u@var{suffix}
9159 Like @samp{%g}, but generates a new temporary file name even if
9160 @samp{%u@var{suffix}} was already seen.
9162 @item %U@var{suffix}
9163 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9164 new one if there is no such last file name. In the absence of any
9165 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9166 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9167 would involve the generation of two distinct file names, one
9168 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9169 simply substituted with a file name chosen for the previous @samp{%u},
9170 without regard to any appended suffix.
9172 @item %j@var{suffix}
9173 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9174 writable, and if save-temps is off; otherwise, substitute the name
9175 of a temporary file, just like @samp{%u}. This temporary file is not
9176 meant for communication between processes, but rather as a junk
9179 @item %|@var{suffix}
9180 @itemx %m@var{suffix}
9181 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9182 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9183 all. These are the two most common ways to instruct a program that it
9184 should read from standard input or write to standard output. If you
9185 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9186 construct: see for example @file{f/lang-specs.h}.
9188 @item %.@var{SUFFIX}
9189 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9190 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9191 terminated by the next space or %.
9194 Marks the argument containing or following the @samp{%w} as the
9195 designated output file of this compilation. This puts the argument
9196 into the sequence of arguments that @samp{%o} will substitute later.
9199 Substitutes the names of all the output files, with spaces
9200 automatically placed around them. You should write spaces
9201 around the @samp{%o} as well or the results are undefined.
9202 @samp{%o} is for use in the specs for running the linker.
9203 Input files whose names have no recognized suffix are not compiled
9204 at all, but they are included among the output files, so they will
9208 Substitutes the suffix for object files. Note that this is
9209 handled specially when it immediately follows @samp{%g, %u, or %U},
9210 because of the need for those to form complete file names. The
9211 handling is such that @samp{%O} is treated exactly as if it had already
9212 been substituted, except that @samp{%g, %u, and %U} do not currently
9213 support additional @var{suffix} characters following @samp{%O} as they would
9214 following, for example, @samp{.o}.
9217 Substitutes the standard macro predefinitions for the
9218 current target machine. Use this when running @code{cpp}.
9221 Like @samp{%p}, but puts @samp{__} before and after the name of each
9222 predefined macro, except for macros that start with @samp{__} or with
9223 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9227 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9228 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9229 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9230 and @option{-imultilib} as necessary.
9233 Current argument is the name of a library or startup file of some sort.
9234 Search for that file in a standard list of directories and substitute
9235 the full name found. The current working directory is included in the
9236 list of directories scanned.
9239 Current argument is the name of a linker script. Search for that file
9240 in the current list of directories to scan for libraries. If the file
9241 is located insert a @option{--script} option into the command line
9242 followed by the full path name found. If the file is not found then
9243 generate an error message. Note: the current working directory is not
9247 Print @var{str} as an error message. @var{str} is terminated by a newline.
9248 Use this when inconsistent options are detected.
9251 Substitute the contents of spec string @var{name} at this point.
9254 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9256 @item %x@{@var{option}@}
9257 Accumulate an option for @samp{%X}.
9260 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9264 Output the accumulated assembler options specified by @option{-Wa}.
9267 Output the accumulated preprocessor options specified by @option{-Wp}.
9270 Process the @code{asm} spec. This is used to compute the
9271 switches to be passed to the assembler.
9274 Process the @code{asm_final} spec. This is a spec string for
9275 passing switches to an assembler post-processor, if such a program is
9279 Process the @code{link} spec. This is the spec for computing the
9280 command line passed to the linker. Typically it will make use of the
9281 @samp{%L %G %S %D and %E} sequences.
9284 Dump out a @option{-L} option for each directory that GCC believes might
9285 contain startup files. If the target supports multilibs then the
9286 current multilib directory will be prepended to each of these paths.
9289 Process the @code{lib} spec. This is a spec string for deciding which
9290 libraries should be included on the command line to the linker.
9293 Process the @code{libgcc} spec. This is a spec string for deciding
9294 which GCC support library should be included on the command line to the linker.
9297 Process the @code{startfile} spec. This is a spec for deciding which
9298 object files should be the first ones passed to the linker. Typically
9299 this might be a file named @file{crt0.o}.
9302 Process the @code{endfile} spec. This is a spec string that specifies
9303 the last object files that will be passed to the linker.
9306 Process the @code{cpp} spec. This is used to construct the arguments
9307 to be passed to the C preprocessor.
9310 Process the @code{cc1} spec. This is used to construct the options to be
9311 passed to the actual C compiler (@samp{cc1}).
9314 Process the @code{cc1plus} spec. This is used to construct the options to be
9315 passed to the actual C++ compiler (@samp{cc1plus}).
9318 Substitute the variable part of a matched option. See below.
9319 Note that each comma in the substituted string is replaced by
9323 Remove all occurrences of @code{-S} from the command line. Note---this
9324 command is position dependent. @samp{%} commands in the spec string
9325 before this one will see @code{-S}, @samp{%} commands in the spec string
9326 after this one will not.
9328 @item %:@var{function}(@var{args})
9329 Call the named function @var{function}, passing it @var{args}.
9330 @var{args} is first processed as a nested spec string, then split
9331 into an argument vector in the usual fashion. The function returns
9332 a string which is processed as if it had appeared literally as part
9333 of the current spec.
9335 The following built-in spec functions are provided:
9339 The @code{getenv} spec function takes two arguments: an environment
9340 variable name and a string. If the environment variable is not
9341 defined, a fatal error is issued. Otherwise, the return value is the
9342 value of the environment variable concatenated with the string. For
9343 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9346 %:getenv(TOPDIR /include)
9349 expands to @file{/path/to/top/include}.
9351 @item @code{if-exists}
9352 The @code{if-exists} spec function takes one argument, an absolute
9353 pathname to a file. If the file exists, @code{if-exists} returns the
9354 pathname. Here is a small example of its usage:
9358 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9361 @item @code{if-exists-else}
9362 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9363 spec function, except that it takes two arguments. The first argument is
9364 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9365 returns the pathname. If it does not exist, it returns the second argument.
9366 This way, @code{if-exists-else} can be used to select one file or another,
9367 based on the existence of the first. Here is a small example of its usage:
9371 crt0%O%s %:if-exists(crti%O%s) \
9372 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9375 @item @code{replace-outfile}
9376 The @code{replace-outfile} spec function takes two arguments. It looks for the
9377 first argument in the outfiles array and replaces it with the second argument. Here
9378 is a small example of its usage:
9381 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9384 @item @code{print-asm-header}
9385 The @code{print-asm-header} function takes no arguments and simply
9386 prints a banner like:
9392 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9395 It is used to separate compiler options from assembler options
9396 in the @option{--target-help} output.
9400 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9401 If that switch was not specified, this substitutes nothing. Note that
9402 the leading dash is omitted when specifying this option, and it is
9403 automatically inserted if the substitution is performed. Thus the spec
9404 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9405 and would output the command line option @option{-foo}.
9407 @item %W@{@code{S}@}
9408 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9411 @item %@{@code{S}*@}
9412 Substitutes all the switches specified to GCC whose names start
9413 with @code{-S}, but which also take an argument. This is used for
9414 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9415 GCC considers @option{-o foo} as being
9416 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9417 text, including the space. Thus two arguments would be generated.
9419 @item %@{@code{S}*&@code{T}*@}
9420 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9421 (the order of @code{S} and @code{T} in the spec is not significant).
9422 There can be any number of ampersand-separated variables; for each the
9423 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9425 @item %@{@code{S}:@code{X}@}
9426 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9428 @item %@{!@code{S}:@code{X}@}
9429 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9431 @item %@{@code{S}*:@code{X}@}
9432 Substitutes @code{X} if one or more switches whose names start with
9433 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9434 once, no matter how many such switches appeared. However, if @code{%*}
9435 appears somewhere in @code{X}, then @code{X} will be substituted once
9436 for each matching switch, with the @code{%*} replaced by the part of
9437 that switch that matched the @code{*}.
9439 @item %@{.@code{S}:@code{X}@}
9440 Substitutes @code{X}, if processing a file with suffix @code{S}.
9442 @item %@{!.@code{S}:@code{X}@}
9443 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9445 @item %@{,@code{S}:@code{X}@}
9446 Substitutes @code{X}, if processing a file for language @code{S}.
9448 @item %@{!,@code{S}:@code{X}@}
9449 Substitutes @code{X}, if not processing a file for language @code{S}.
9451 @item %@{@code{S}|@code{P}:@code{X}@}
9452 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9453 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9454 @code{*} sequences as well, although they have a stronger binding than
9455 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9456 alternatives must be starred, and only the first matching alternative
9459 For example, a spec string like this:
9462 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9465 will output the following command-line options from the following input
9466 command-line options:
9471 -d fred.c -foo -baz -boggle
9472 -d jim.d -bar -baz -boggle
9475 @item %@{S:X; T:Y; :D@}
9477 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9478 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9479 be as many clauses as you need. This may be combined with @code{.},
9480 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9485 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9486 construct may contain other nested @samp{%} constructs or spaces, or
9487 even newlines. They are processed as usual, as described above.
9488 Trailing white space in @code{X} is ignored. White space may also
9489 appear anywhere on the left side of the colon in these constructs,
9490 except between @code{.} or @code{*} and the corresponding word.
9492 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9493 handled specifically in these constructs. If another value of
9494 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9495 @option{-W} switch is found later in the command line, the earlier
9496 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9497 just one letter, which passes all matching options.
9499 The character @samp{|} at the beginning of the predicate text is used to
9500 indicate that a command should be piped to the following command, but
9501 only if @option{-pipe} is specified.
9503 It is built into GCC which switches take arguments and which do not.
9504 (You might think it would be useful to generalize this to allow each
9505 compiler's spec to say which switches take arguments. But this cannot
9506 be done in a consistent fashion. GCC cannot even decide which input
9507 files have been specified without knowing which switches take arguments,
9508 and it must know which input files to compile in order to tell which
9511 GCC also knows implicitly that arguments starting in @option{-l} are to be
9512 treated as compiler output files, and passed to the linker in their
9513 proper position among the other output files.
9515 @c man begin OPTIONS
9517 @node Target Options
9518 @section Specifying Target Machine and Compiler Version
9519 @cindex target options
9520 @cindex cross compiling
9521 @cindex specifying machine version
9522 @cindex specifying compiler version and target machine
9523 @cindex compiler version, specifying
9524 @cindex target machine, specifying
9526 The usual way to run GCC is to run the executable called @file{gcc}, or
9527 @file{<machine>-gcc} when cross-compiling, or
9528 @file{<machine>-gcc-<version>} to run a version other than the one that
9529 was installed last. Sometimes this is inconvenient, so GCC provides
9530 options that will switch to another cross-compiler or version.
9533 @item -b @var{machine}
9535 The argument @var{machine} specifies the target machine for compilation.
9537 The value to use for @var{machine} is the same as was specified as the
9538 machine type when configuring GCC as a cross-compiler. For
9539 example, if a cross-compiler was configured with @samp{configure
9540 arm-elf}, meaning to compile for an arm processor with elf binaries,
9541 then you would specify @option{-b arm-elf} to run that cross compiler.
9542 Because there are other options beginning with @option{-b}, the
9543 configuration must contain a hyphen, or @option{-b} alone should be one
9544 argument followed by the configuration in the next argument.
9546 @item -V @var{version}
9548 The argument @var{version} specifies which version of GCC to run.
9549 This is useful when multiple versions are installed. For example,
9550 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
9553 The @option{-V} and @option{-b} options work by running the
9554 @file{<machine>-gcc-<version>} executable, so there's no real reason to
9555 use them if you can just run that directly.
9557 @node Submodel Options
9558 @section Hardware Models and Configurations
9559 @cindex submodel options
9560 @cindex specifying hardware config
9561 @cindex hardware models and configurations, specifying
9562 @cindex machine dependent options
9564 Earlier we discussed the standard option @option{-b} which chooses among
9565 different installed compilers for completely different target
9566 machines, such as VAX vs.@: 68000 vs.@: 80386.
9568 In addition, each of these target machine types can have its own
9569 special options, starting with @samp{-m}, to choose among various
9570 hardware models or configurations---for example, 68010 vs 68020,
9571 floating coprocessor or none. A single installed version of the
9572 compiler can compile for any model or configuration, according to the
9575 Some configurations of the compiler also support additional special
9576 options, usually for compatibility with other compilers on the same
9579 @c This list is ordered alphanumerically by subsection name.
9580 @c It should be the same order and spelling as these options are listed
9581 @c in Machine Dependent Options
9587 * Blackfin Options::
9591 * DEC Alpha Options::
9592 * DEC Alpha/VMS Options::
9595 * GNU/Linux Options::
9598 * i386 and x86-64 Options::
9599 * i386 and x86-64 Windows Options::
9601 * IA-64/VMS Options::
9613 * picoChip Options::
9615 * RS/6000 and PowerPC Options::
9617 * S/390 and zSeries Options::
9622 * System V Options::
9627 * Xstormy16 Options::
9633 @subsection ARC Options
9636 These options are defined for ARC implementations:
9641 Compile code for little endian mode. This is the default.
9645 Compile code for big endian mode.
9648 @opindex mmangle-cpu
9649 Prepend the name of the cpu to all public symbol names.
9650 In multiple-processor systems, there are many ARC variants with different
9651 instruction and register set characteristics. This flag prevents code
9652 compiled for one cpu to be linked with code compiled for another.
9653 No facility exists for handling variants that are ``almost identical''.
9654 This is an all or nothing option.
9656 @item -mcpu=@var{cpu}
9658 Compile code for ARC variant @var{cpu}.
9659 Which variants are supported depend on the configuration.
9660 All variants support @option{-mcpu=base}, this is the default.
9662 @item -mtext=@var{text-section}
9663 @itemx -mdata=@var{data-section}
9664 @itemx -mrodata=@var{readonly-data-section}
9668 Put functions, data, and readonly data in @var{text-section},
9669 @var{data-section}, and @var{readonly-data-section} respectively
9670 by default. This can be overridden with the @code{section} attribute.
9671 @xref{Variable Attributes}.
9673 @item -mfix-cortex-m3-ldrd
9674 @opindex mfix-cortex-m3-ldrd
9675 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
9676 with overlapping destination and base registers are used. This option avoids
9677 generating these instructions. This option is enabled by default when
9678 @option{-mcpu=cortex-m3} is specified.
9683 @subsection ARM Options
9686 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9690 @item -mabi=@var{name}
9692 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9693 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9696 @opindex mapcs-frame
9697 Generate a stack frame that is compliant with the ARM Procedure Call
9698 Standard for all functions, even if this is not strictly necessary for
9699 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9700 with this option will cause the stack frames not to be generated for
9701 leaf functions. The default is @option{-mno-apcs-frame}.
9705 This is a synonym for @option{-mapcs-frame}.
9708 @c not currently implemented
9709 @item -mapcs-stack-check
9710 @opindex mapcs-stack-check
9711 Generate code to check the amount of stack space available upon entry to
9712 every function (that actually uses some stack space). If there is
9713 insufficient space available then either the function
9714 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9715 called, depending upon the amount of stack space required. The run time
9716 system is required to provide these functions. The default is
9717 @option{-mno-apcs-stack-check}, since this produces smaller code.
9719 @c not currently implemented
9721 @opindex mapcs-float
9722 Pass floating point arguments using the float point registers. This is
9723 one of the variants of the APCS@. This option is recommended if the
9724 target hardware has a floating point unit or if a lot of floating point
9725 arithmetic is going to be performed by the code. The default is
9726 @option{-mno-apcs-float}, since integer only code is slightly increased in
9727 size if @option{-mapcs-float} is used.
9729 @c not currently implemented
9730 @item -mapcs-reentrant
9731 @opindex mapcs-reentrant
9732 Generate reentrant, position independent code. The default is
9733 @option{-mno-apcs-reentrant}.
9736 @item -mthumb-interwork
9737 @opindex mthumb-interwork
9738 Generate code which supports calling between the ARM and Thumb
9739 instruction sets. Without this option the two instruction sets cannot
9740 be reliably used inside one program. The default is
9741 @option{-mno-thumb-interwork}, since slightly larger code is generated
9742 when @option{-mthumb-interwork} is specified.
9744 @item -mno-sched-prolog
9745 @opindex mno-sched-prolog
9746 Prevent the reordering of instructions in the function prolog, or the
9747 merging of those instruction with the instructions in the function's
9748 body. This means that all functions will start with a recognizable set
9749 of instructions (or in fact one of a choice from a small set of
9750 different function prologues), and this information can be used to
9751 locate the start if functions inside an executable piece of code. The
9752 default is @option{-msched-prolog}.
9754 @item -mfloat-abi=@var{name}
9756 Specifies which floating-point ABI to use. Permissible values
9757 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9759 Specifying @samp{soft} causes GCC to generate output containing
9760 library calls for floating-point operations.
9761 @samp{softfp} allows the generation of code using hardware floating-point
9762 instructions, but still uses the soft-float calling conventions.
9763 @samp{hard} allows generation of floating-point instructions
9764 and uses FPU-specific calling conventions.
9766 The default depends on the specific target configuration. Note that
9767 the hard-float and soft-float ABIs are not link-compatible; you must
9768 compile your entire program with the same ABI, and link with a
9769 compatible set of libraries.
9772 @opindex mhard-float
9773 Equivalent to @option{-mfloat-abi=hard}.
9776 @opindex msoft-float
9777 Equivalent to @option{-mfloat-abi=soft}.
9779 @item -mlittle-endian
9780 @opindex mlittle-endian
9781 Generate code for a processor running in little-endian mode. This is
9782 the default for all standard configurations.
9785 @opindex mbig-endian
9786 Generate code for a processor running in big-endian mode; the default is
9787 to compile code for a little-endian processor.
9789 @item -mwords-little-endian
9790 @opindex mwords-little-endian
9791 This option only applies when generating code for big-endian processors.
9792 Generate code for a little-endian word order but a big-endian byte
9793 order. That is, a byte order of the form @samp{32107654}. Note: this
9794 option should only be used if you require compatibility with code for
9795 big-endian ARM processors generated by versions of the compiler prior to
9798 @item -mcpu=@var{name}
9800 This specifies the name of the target ARM processor. GCC uses this name
9801 to determine what kind of instructions it can emit when generating
9802 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9803 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9804 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9805 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9806 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9808 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9809 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9810 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9811 @samp{strongarm1110},
9812 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9813 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9814 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9815 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9816 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9817 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9818 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9819 @samp{cortex-a5}, @samp{cortex-a8}, @samp{cortex-a9},
9820 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9823 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9825 @item -mtune=@var{name}
9827 This option is very similar to the @option{-mcpu=} option, except that
9828 instead of specifying the actual target processor type, and hence
9829 restricting which instructions can be used, it specifies that GCC should
9830 tune the performance of the code as if the target were of the type
9831 specified in this option, but still choosing the instructions that it
9832 will generate based on the cpu specified by a @option{-mcpu=} option.
9833 For some ARM implementations better performance can be obtained by using
9836 @item -march=@var{name}
9838 This specifies the name of the target ARM architecture. GCC uses this
9839 name to determine what kind of instructions it can emit when generating
9840 assembly code. This option can be used in conjunction with or instead
9841 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9842 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9843 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9844 @samp{armv6}, @samp{armv6j},
9845 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9846 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9847 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9849 @item -mfpu=@var{name}
9850 @itemx -mfpe=@var{number}
9851 @itemx -mfp=@var{number}
9855 This specifies what floating point hardware (or hardware emulation) is
9856 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9857 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
9858 @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
9859 @samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
9860 @samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
9861 @option{-mfp} and @option{-mfpe} are synonyms for
9862 @option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
9865 If @option{-msoft-float} is specified this specifies the format of
9866 floating point values.
9868 @item -mfp16-format=@var{name}
9869 @opindex mfp16-format
9870 Specify the format of the @code{__fp16} half-precision floating-point type.
9871 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
9872 the default is @samp{none}, in which case the @code{__fp16} type is not
9873 defined. @xref{Half-Precision}, for more information.
9875 @item -mstructure-size-boundary=@var{n}
9876 @opindex mstructure-size-boundary
9877 The size of all structures and unions will be rounded up to a multiple
9878 of the number of bits set by this option. Permissible values are 8, 32
9879 and 64. The default value varies for different toolchains. For the COFF
9880 targeted toolchain the default value is 8. A value of 64 is only allowed
9881 if the underlying ABI supports it.
9883 Specifying the larger number can produce faster, more efficient code, but
9884 can also increase the size of the program. Different values are potentially
9885 incompatible. Code compiled with one value cannot necessarily expect to
9886 work with code or libraries compiled with another value, if they exchange
9887 information using structures or unions.
9889 @item -mabort-on-noreturn
9890 @opindex mabort-on-noreturn
9891 Generate a call to the function @code{abort} at the end of a
9892 @code{noreturn} function. It will be executed if the function tries to
9896 @itemx -mno-long-calls
9897 @opindex mlong-calls
9898 @opindex mno-long-calls
9899 Tells the compiler to perform function calls by first loading the
9900 address of the function into a register and then performing a subroutine
9901 call on this register. This switch is needed if the target function
9902 will lie outside of the 64 megabyte addressing range of the offset based
9903 version of subroutine call instruction.
9905 Even if this switch is enabled, not all function calls will be turned
9906 into long calls. The heuristic is that static functions, functions
9907 which have the @samp{short-call} attribute, functions that are inside
9908 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9909 definitions have already been compiled within the current compilation
9910 unit, will not be turned into long calls. The exception to this rule is
9911 that weak function definitions, functions with the @samp{long-call}
9912 attribute or the @samp{section} attribute, and functions that are within
9913 the scope of a @samp{#pragma long_calls} directive, will always be
9914 turned into long calls.
9916 This feature is not enabled by default. Specifying
9917 @option{-mno-long-calls} will restore the default behavior, as will
9918 placing the function calls within the scope of a @samp{#pragma
9919 long_calls_off} directive. Note these switches have no effect on how
9920 the compiler generates code to handle function calls via function
9923 @item -msingle-pic-base
9924 @opindex msingle-pic-base
9925 Treat the register used for PIC addressing as read-only, rather than
9926 loading it in the prologue for each function. The run-time system is
9927 responsible for initializing this register with an appropriate value
9928 before execution begins.
9930 @item -mpic-register=@var{reg}
9931 @opindex mpic-register
9932 Specify the register to be used for PIC addressing. The default is R10
9933 unless stack-checking is enabled, when R9 is used.
9935 @item -mcirrus-fix-invalid-insns
9936 @opindex mcirrus-fix-invalid-insns
9937 @opindex mno-cirrus-fix-invalid-insns
9938 Insert NOPs into the instruction stream to in order to work around
9939 problems with invalid Maverick instruction combinations. This option
9940 is only valid if the @option{-mcpu=ep9312} option has been used to
9941 enable generation of instructions for the Cirrus Maverick floating
9942 point co-processor. This option is not enabled by default, since the
9943 problem is only present in older Maverick implementations. The default
9944 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9947 @item -mpoke-function-name
9948 @opindex mpoke-function-name
9949 Write the name of each function into the text section, directly
9950 preceding the function prologue. The generated code is similar to this:
9954 .ascii "arm_poke_function_name", 0
9957 .word 0xff000000 + (t1 - t0)
9958 arm_poke_function_name
9960 stmfd sp!, @{fp, ip, lr, pc@}
9964 When performing a stack backtrace, code can inspect the value of
9965 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9966 location @code{pc - 12} and the top 8 bits are set, then we know that
9967 there is a function name embedded immediately preceding this location
9968 and has length @code{((pc[-3]) & 0xff000000)}.
9972 Generate code for the Thumb instruction set. The default is to
9973 use the 32-bit ARM instruction set.
9974 This option automatically enables either 16-bit Thumb-1 or
9975 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9976 and @option{-march=@var{name}} options. This option is not passed to the
9977 assembler. If you want to force assembler files to be interpreted as Thumb code,
9978 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
9979 option directly to the assembler by prefixing it with @option{-Wa}.
9982 @opindex mtpcs-frame
9983 Generate a stack frame that is compliant with the Thumb Procedure Call
9984 Standard for all non-leaf functions. (A leaf function is one that does
9985 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9987 @item -mtpcs-leaf-frame
9988 @opindex mtpcs-leaf-frame
9989 Generate a stack frame that is compliant with the Thumb Procedure Call
9990 Standard for all leaf functions. (A leaf function is one that does
9991 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9993 @item -mcallee-super-interworking
9994 @opindex mcallee-super-interworking
9995 Gives all externally visible functions in the file being compiled an ARM
9996 instruction set header which switches to Thumb mode before executing the
9997 rest of the function. This allows these functions to be called from
9998 non-interworking code. This option is not valid in AAPCS configurations
9999 because interworking is enabled by default.
10001 @item -mcaller-super-interworking
10002 @opindex mcaller-super-interworking
10003 Allows calls via function pointers (including virtual functions) to
10004 execute correctly regardless of whether the target code has been
10005 compiled for interworking or not. There is a small overhead in the cost
10006 of executing a function pointer if this option is enabled. This option
10007 is not valid in AAPCS configurations because interworking is enabled
10010 @item -mtp=@var{name}
10012 Specify the access model for the thread local storage pointer. The valid
10013 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10014 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
10015 (supported in the arm6k architecture), and @option{auto}, which uses the
10016 best available method for the selected processor. The default setting is
10019 @item -mword-relocations
10020 @opindex mword-relocations
10021 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
10022 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10023 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10029 @subsection AVR Options
10030 @cindex AVR Options
10032 These options are defined for AVR implementations:
10035 @item -mmcu=@var{mcu}
10037 Specify ATMEL AVR instruction set or MCU type.
10039 Instruction set avr1 is for the minimal AVR core, not supported by the C
10040 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10041 attiny11, attiny12, attiny15, attiny28).
10043 Instruction set avr2 (default) is for the classic AVR core with up to
10044 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10045 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10046 at90c8534, at90s8535).
10048 Instruction set avr3 is for the classic AVR core with up to 128K program
10049 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10051 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10052 memory space (MCU types: atmega8, atmega83, atmega85).
10054 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10055 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10056 atmega64, atmega128, at43usb355, at94k).
10058 @item -mno-interrupts
10059 @opindex mno-interrupts
10060 Generated code is not compatible with hardware interrupts.
10061 Code size will be smaller.
10063 @item -mcall-prologues
10064 @opindex mcall-prologues
10065 Functions prologues/epilogues expanded as call to appropriate
10066 subroutines. Code size will be smaller.
10069 @opindex mtiny-stack
10070 Change only the low 8 bits of the stack pointer.
10074 Assume int to be 8 bit integer. This affects the sizes of all types: A
10075 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10076 and long long will be 4 bytes. Please note that this option does not
10077 comply to the C standards, but it will provide you with smaller code
10081 @node Blackfin Options
10082 @subsection Blackfin Options
10083 @cindex Blackfin Options
10086 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10088 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10089 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10090 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10091 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10092 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10093 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10094 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10096 The optional @var{sirevision} specifies the silicon revision of the target
10097 Blackfin processor. Any workarounds available for the targeted silicon revision
10098 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10099 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10100 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10101 hexadecimal digits representing the major and minor numbers in the silicon
10102 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10103 is not defined. If @var{sirevision} is @samp{any}, the
10104 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10105 If this optional @var{sirevision} is not used, GCC assumes the latest known
10106 silicon revision of the targeted Blackfin processor.
10108 Support for @samp{bf561} is incomplete. For @samp{bf561},
10109 Only the processor macro is defined.
10110 Without this option, @samp{bf532} is used as the processor by default.
10111 The corresponding predefined processor macros for @var{cpu} is to
10112 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10113 provided by libgloss to be linked in if @option{-msim} is not given.
10117 Specifies that the program will be run on the simulator. This causes
10118 the simulator BSP provided by libgloss to be linked in. This option
10119 has effect only for @samp{bfin-elf} toolchain.
10120 Certain other options, such as @option{-mid-shared-library} and
10121 @option{-mfdpic}, imply @option{-msim}.
10123 @item -momit-leaf-frame-pointer
10124 @opindex momit-leaf-frame-pointer
10125 Don't keep the frame pointer in a register for leaf functions. This
10126 avoids the instructions to save, set up and restore frame pointers and
10127 makes an extra register available in leaf functions. The option
10128 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10129 which might make debugging harder.
10131 @item -mspecld-anomaly
10132 @opindex mspecld-anomaly
10133 When enabled, the compiler will ensure that the generated code does not
10134 contain speculative loads after jump instructions. If this option is used,
10135 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10137 @item -mno-specld-anomaly
10138 @opindex mno-specld-anomaly
10139 Don't generate extra code to prevent speculative loads from occurring.
10141 @item -mcsync-anomaly
10142 @opindex mcsync-anomaly
10143 When enabled, the compiler will ensure that the generated code does not
10144 contain CSYNC or SSYNC instructions too soon after conditional branches.
10145 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10147 @item -mno-csync-anomaly
10148 @opindex mno-csync-anomaly
10149 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10150 occurring too soon after a conditional branch.
10154 When enabled, the compiler is free to take advantage of the knowledge that
10155 the entire program fits into the low 64k of memory.
10158 @opindex mno-low-64k
10159 Assume that the program is arbitrarily large. This is the default.
10161 @item -mstack-check-l1
10162 @opindex mstack-check-l1
10163 Do stack checking using information placed into L1 scratchpad memory by the
10166 @item -mid-shared-library
10167 @opindex mid-shared-library
10168 Generate code that supports shared libraries via the library ID method.
10169 This allows for execute in place and shared libraries in an environment
10170 without virtual memory management. This option implies @option{-fPIC}.
10171 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10173 @item -mno-id-shared-library
10174 @opindex mno-id-shared-library
10175 Generate code that doesn't assume ID based shared libraries are being used.
10176 This is the default.
10178 @item -mleaf-id-shared-library
10179 @opindex mleaf-id-shared-library
10180 Generate code that supports shared libraries via the library ID method,
10181 but assumes that this library or executable won't link against any other
10182 ID shared libraries. That allows the compiler to use faster code for jumps
10185 @item -mno-leaf-id-shared-library
10186 @opindex mno-leaf-id-shared-library
10187 Do not assume that the code being compiled won't link against any ID shared
10188 libraries. Slower code will be generated for jump and call insns.
10190 @item -mshared-library-id=n
10191 @opindex mshared-library-id
10192 Specified the identification number of the ID based shared library being
10193 compiled. Specifying a value of 0 will generate more compact code, specifying
10194 other values will force the allocation of that number to the current
10195 library but is no more space or time efficient than omitting this option.
10199 Generate code that allows the data segment to be located in a different
10200 area of memory from the text segment. This allows for execute in place in
10201 an environment without virtual memory management by eliminating relocations
10202 against the text section.
10204 @item -mno-sep-data
10205 @opindex mno-sep-data
10206 Generate code that assumes that the data segment follows the text segment.
10207 This is the default.
10210 @itemx -mno-long-calls
10211 @opindex mlong-calls
10212 @opindex mno-long-calls
10213 Tells the compiler to perform function calls by first loading the
10214 address of the function into a register and then performing a subroutine
10215 call on this register. This switch is needed if the target function
10216 will lie outside of the 24 bit addressing range of the offset based
10217 version of subroutine call instruction.
10219 This feature is not enabled by default. Specifying
10220 @option{-mno-long-calls} will restore the default behavior. Note these
10221 switches have no effect on how the compiler generates code to handle
10222 function calls via function pointers.
10226 Link with the fast floating-point library. This library relaxes some of
10227 the IEEE floating-point standard's rules for checking inputs against
10228 Not-a-Number (NAN), in the interest of performance.
10231 @opindex minline-plt
10232 Enable inlining of PLT entries in function calls to functions that are
10233 not known to bind locally. It has no effect without @option{-mfdpic}.
10236 @opindex mmulticore
10237 Build standalone application for multicore Blackfin processor. Proper
10238 start files and link scripts will be used to support multicore.
10239 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10240 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10241 @option{-mcorea} or @option{-mcoreb}. If it's used without
10242 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10243 programming model is used. In this model, the main function of Core B
10244 should be named as coreb_main. If it's used with @option{-mcorea} or
10245 @option{-mcoreb}, one application per core programming model is used.
10246 If this option is not used, single core application programming
10251 Build standalone application for Core A of BF561 when using
10252 one application per core programming model. Proper start files
10253 and link scripts will be used to support Core A. This option
10254 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10258 Build standalone application for Core B of BF561 when using
10259 one application per core programming model. Proper start files
10260 and link scripts will be used to support Core B. This option
10261 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10262 should be used instead of main. It must be used with
10263 @option{-mmulticore}.
10267 Build standalone application for SDRAM. Proper start files and
10268 link scripts will be used to put the application into SDRAM.
10269 Loader should initialize SDRAM before loading the application
10270 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10274 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10275 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10276 are enabled; for standalone applications the default is off.
10280 @subsection CRIS Options
10281 @cindex CRIS Options
10283 These options are defined specifically for the CRIS ports.
10286 @item -march=@var{architecture-type}
10287 @itemx -mcpu=@var{architecture-type}
10290 Generate code for the specified architecture. The choices for
10291 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10292 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10293 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10296 @item -mtune=@var{architecture-type}
10298 Tune to @var{architecture-type} everything applicable about the generated
10299 code, except for the ABI and the set of available instructions. The
10300 choices for @var{architecture-type} are the same as for
10301 @option{-march=@var{architecture-type}}.
10303 @item -mmax-stack-frame=@var{n}
10304 @opindex mmax-stack-frame
10305 Warn when the stack frame of a function exceeds @var{n} bytes.
10311 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10312 @option{-march=v3} and @option{-march=v8} respectively.
10314 @item -mmul-bug-workaround
10315 @itemx -mno-mul-bug-workaround
10316 @opindex mmul-bug-workaround
10317 @opindex mno-mul-bug-workaround
10318 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10319 models where it applies. This option is active by default.
10323 Enable CRIS-specific verbose debug-related information in the assembly
10324 code. This option also has the effect to turn off the @samp{#NO_APP}
10325 formatted-code indicator to the assembler at the beginning of the
10330 Do not use condition-code results from previous instruction; always emit
10331 compare and test instructions before use of condition codes.
10333 @item -mno-side-effects
10334 @opindex mno-side-effects
10335 Do not emit instructions with side-effects in addressing modes other than
10338 @item -mstack-align
10339 @itemx -mno-stack-align
10340 @itemx -mdata-align
10341 @itemx -mno-data-align
10342 @itemx -mconst-align
10343 @itemx -mno-const-align
10344 @opindex mstack-align
10345 @opindex mno-stack-align
10346 @opindex mdata-align
10347 @opindex mno-data-align
10348 @opindex mconst-align
10349 @opindex mno-const-align
10350 These options (no-options) arranges (eliminate arrangements) for the
10351 stack-frame, individual data and constants to be aligned for the maximum
10352 single data access size for the chosen CPU model. The default is to
10353 arrange for 32-bit alignment. ABI details such as structure layout are
10354 not affected by these options.
10362 Similar to the stack- data- and const-align options above, these options
10363 arrange for stack-frame, writable data and constants to all be 32-bit,
10364 16-bit or 8-bit aligned. The default is 32-bit alignment.
10366 @item -mno-prologue-epilogue
10367 @itemx -mprologue-epilogue
10368 @opindex mno-prologue-epilogue
10369 @opindex mprologue-epilogue
10370 With @option{-mno-prologue-epilogue}, the normal function prologue and
10371 epilogue that sets up the stack-frame are omitted and no return
10372 instructions or return sequences are generated in the code. Use this
10373 option only together with visual inspection of the compiled code: no
10374 warnings or errors are generated when call-saved registers must be saved,
10375 or storage for local variable needs to be allocated.
10379 @opindex mno-gotplt
10381 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10382 instruction sequences that load addresses for functions from the PLT part
10383 of the GOT rather than (traditional on other architectures) calls to the
10384 PLT@. The default is @option{-mgotplt}.
10388 Legacy no-op option only recognized with the cris-axis-elf and
10389 cris-axis-linux-gnu targets.
10393 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10397 This option, recognized for the cris-axis-elf arranges
10398 to link with input-output functions from a simulator library. Code,
10399 initialized data and zero-initialized data are allocated consecutively.
10403 Like @option{-sim}, but pass linker options to locate initialized data at
10404 0x40000000 and zero-initialized data at 0x80000000.
10408 @subsection CRX Options
10409 @cindex CRX Options
10411 These options are defined specifically for the CRX ports.
10417 Enable the use of multiply-accumulate instructions. Disabled by default.
10420 @opindex mpush-args
10421 Push instructions will be used to pass outgoing arguments when functions
10422 are called. Enabled by default.
10425 @node Darwin Options
10426 @subsection Darwin Options
10427 @cindex Darwin options
10429 These options are defined for all architectures running the Darwin operating
10432 FSF GCC on Darwin does not create ``fat'' object files; it will create
10433 an object file for the single architecture that it was built to
10434 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10435 @option{-arch} options are used; it does so by running the compiler or
10436 linker multiple times and joining the results together with
10439 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10440 @samp{i686}) is determined by the flags that specify the ISA
10441 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10442 @option{-force_cpusubtype_ALL} option can be used to override this.
10444 The Darwin tools vary in their behavior when presented with an ISA
10445 mismatch. The assembler, @file{as}, will only permit instructions to
10446 be used that are valid for the subtype of the file it is generating,
10447 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10448 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10449 and print an error if asked to create a shared library with a less
10450 restrictive subtype than its input files (for instance, trying to put
10451 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10452 for executables, @file{ld}, will quietly give the executable the most
10453 restrictive subtype of any of its input files.
10458 Add the framework directory @var{dir} to the head of the list of
10459 directories to be searched for header files. These directories are
10460 interleaved with those specified by @option{-I} options and are
10461 scanned in a left-to-right order.
10463 A framework directory is a directory with frameworks in it. A
10464 framework is a directory with a @samp{"Headers"} and/or
10465 @samp{"PrivateHeaders"} directory contained directly in it that ends
10466 in @samp{".framework"}. The name of a framework is the name of this
10467 directory excluding the @samp{".framework"}. Headers associated with
10468 the framework are found in one of those two directories, with
10469 @samp{"Headers"} being searched first. A subframework is a framework
10470 directory that is in a framework's @samp{"Frameworks"} directory.
10471 Includes of subframework headers can only appear in a header of a
10472 framework that contains the subframework, or in a sibling subframework
10473 header. Two subframeworks are siblings if they occur in the same
10474 framework. A subframework should not have the same name as a
10475 framework, a warning will be issued if this is violated. Currently a
10476 subframework cannot have subframeworks, in the future, the mechanism
10477 may be extended to support this. The standard frameworks can be found
10478 in @samp{"/System/Library/Frameworks"} and
10479 @samp{"/Library/Frameworks"}. An example include looks like
10480 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10481 the name of the framework and header.h is found in the
10482 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10484 @item -iframework@var{dir}
10485 @opindex iframework
10486 Like @option{-F} except the directory is a treated as a system
10487 directory. The main difference between this @option{-iframework} and
10488 @option{-F} is that with @option{-iframework} the compiler does not
10489 warn about constructs contained within header files found via
10490 @var{dir}. This option is valid only for the C family of languages.
10494 Emit debugging information for symbols that are used. For STABS
10495 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10496 This is by default ON@.
10500 Emit debugging information for all symbols and types.
10502 @item -mmacosx-version-min=@var{version}
10503 The earliest version of MacOS X that this executable will run on
10504 is @var{version}. Typical values of @var{version} include @code{10.1},
10505 @code{10.2}, and @code{10.3.9}.
10507 If the compiler was built to use the system's headers by default,
10508 then the default for this option is the system version on which the
10509 compiler is running, otherwise the default is to make choices which
10510 are compatible with as many systems and code bases as possible.
10514 Enable kernel development mode. The @option{-mkernel} option sets
10515 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10516 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10517 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10518 applicable. This mode also sets @option{-mno-altivec},
10519 @option{-msoft-float}, @option{-fno-builtin} and
10520 @option{-mlong-branch} for PowerPC targets.
10522 @item -mone-byte-bool
10523 @opindex mone-byte-bool
10524 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10525 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10526 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10527 option has no effect on x86.
10529 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10530 to generate code that is not binary compatible with code generated
10531 without that switch. Using this switch may require recompiling all
10532 other modules in a program, including system libraries. Use this
10533 switch to conform to a non-default data model.
10535 @item -mfix-and-continue
10536 @itemx -ffix-and-continue
10537 @itemx -findirect-data
10538 @opindex mfix-and-continue
10539 @opindex ffix-and-continue
10540 @opindex findirect-data
10541 Generate code suitable for fast turn around development. Needed to
10542 enable gdb to dynamically load @code{.o} files into already running
10543 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10544 are provided for backwards compatibility.
10548 Loads all members of static archive libraries.
10549 See man ld(1) for more information.
10551 @item -arch_errors_fatal
10552 @opindex arch_errors_fatal
10553 Cause the errors having to do with files that have the wrong architecture
10556 @item -bind_at_load
10557 @opindex bind_at_load
10558 Causes the output file to be marked such that the dynamic linker will
10559 bind all undefined references when the file is loaded or launched.
10563 Produce a Mach-o bundle format file.
10564 See man ld(1) for more information.
10566 @item -bundle_loader @var{executable}
10567 @opindex bundle_loader
10568 This option specifies the @var{executable} that will be loading the build
10569 output file being linked. See man ld(1) for more information.
10572 @opindex dynamiclib
10573 When passed this option, GCC will produce a dynamic library instead of
10574 an executable when linking, using the Darwin @file{libtool} command.
10576 @item -force_cpusubtype_ALL
10577 @opindex force_cpusubtype_ALL
10578 This causes GCC's output file to have the @var{ALL} subtype, instead of
10579 one controlled by the @option{-mcpu} or @option{-march} option.
10581 @item -allowable_client @var{client_name}
10582 @itemx -client_name
10583 @itemx -compatibility_version
10584 @itemx -current_version
10586 @itemx -dependency-file
10588 @itemx -dylinker_install_name
10590 @itemx -exported_symbols_list
10592 @itemx -flat_namespace
10593 @itemx -force_flat_namespace
10594 @itemx -headerpad_max_install_names
10597 @itemx -install_name
10598 @itemx -keep_private_externs
10599 @itemx -multi_module
10600 @itemx -multiply_defined
10601 @itemx -multiply_defined_unused
10603 @itemx -no_dead_strip_inits_and_terms
10604 @itemx -nofixprebinding
10605 @itemx -nomultidefs
10607 @itemx -noseglinkedit
10608 @itemx -pagezero_size
10610 @itemx -prebind_all_twolevel_modules
10611 @itemx -private_bundle
10612 @itemx -read_only_relocs
10614 @itemx -sectobjectsymbols
10618 @itemx -sectobjectsymbols
10621 @itemx -segs_read_only_addr
10622 @itemx -segs_read_write_addr
10623 @itemx -seg_addr_table
10624 @itemx -seg_addr_table_filename
10625 @itemx -seglinkedit
10627 @itemx -segs_read_only_addr
10628 @itemx -segs_read_write_addr
10629 @itemx -single_module
10631 @itemx -sub_library
10632 @itemx -sub_umbrella
10633 @itemx -twolevel_namespace
10636 @itemx -unexported_symbols_list
10637 @itemx -weak_reference_mismatches
10638 @itemx -whatsloaded
10639 @opindex allowable_client
10640 @opindex client_name
10641 @opindex compatibility_version
10642 @opindex current_version
10643 @opindex dead_strip
10644 @opindex dependency-file
10645 @opindex dylib_file
10646 @opindex dylinker_install_name
10648 @opindex exported_symbols_list
10650 @opindex flat_namespace
10651 @opindex force_flat_namespace
10652 @opindex headerpad_max_install_names
10653 @opindex image_base
10655 @opindex install_name
10656 @opindex keep_private_externs
10657 @opindex multi_module
10658 @opindex multiply_defined
10659 @opindex multiply_defined_unused
10660 @opindex noall_load
10661 @opindex no_dead_strip_inits_and_terms
10662 @opindex nofixprebinding
10663 @opindex nomultidefs
10665 @opindex noseglinkedit
10666 @opindex pagezero_size
10668 @opindex prebind_all_twolevel_modules
10669 @opindex private_bundle
10670 @opindex read_only_relocs
10672 @opindex sectobjectsymbols
10675 @opindex sectcreate
10676 @opindex sectobjectsymbols
10679 @opindex segs_read_only_addr
10680 @opindex segs_read_write_addr
10681 @opindex seg_addr_table
10682 @opindex seg_addr_table_filename
10683 @opindex seglinkedit
10685 @opindex segs_read_only_addr
10686 @opindex segs_read_write_addr
10687 @opindex single_module
10689 @opindex sub_library
10690 @opindex sub_umbrella
10691 @opindex twolevel_namespace
10694 @opindex unexported_symbols_list
10695 @opindex weak_reference_mismatches
10696 @opindex whatsloaded
10697 These options are passed to the Darwin linker. The Darwin linker man page
10698 describes them in detail.
10701 @node DEC Alpha Options
10702 @subsection DEC Alpha Options
10704 These @samp{-m} options are defined for the DEC Alpha implementations:
10707 @item -mno-soft-float
10708 @itemx -msoft-float
10709 @opindex mno-soft-float
10710 @opindex msoft-float
10711 Use (do not use) the hardware floating-point instructions for
10712 floating-point operations. When @option{-msoft-float} is specified,
10713 functions in @file{libgcc.a} will be used to perform floating-point
10714 operations. Unless they are replaced by routines that emulate the
10715 floating-point operations, or compiled in such a way as to call such
10716 emulations routines, these routines will issue floating-point
10717 operations. If you are compiling for an Alpha without floating-point
10718 operations, you must ensure that the library is built so as not to call
10721 Note that Alpha implementations without floating-point operations are
10722 required to have floating-point registers.
10725 @itemx -mno-fp-regs
10727 @opindex mno-fp-regs
10728 Generate code that uses (does not use) the floating-point register set.
10729 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10730 register set is not used, floating point operands are passed in integer
10731 registers as if they were integers and floating-point results are passed
10732 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10733 so any function with a floating-point argument or return value called by code
10734 compiled with @option{-mno-fp-regs} must also be compiled with that
10737 A typical use of this option is building a kernel that does not use,
10738 and hence need not save and restore, any floating-point registers.
10742 The Alpha architecture implements floating-point hardware optimized for
10743 maximum performance. It is mostly compliant with the IEEE floating
10744 point standard. However, for full compliance, software assistance is
10745 required. This option generates code fully IEEE compliant code
10746 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10747 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10748 defined during compilation. The resulting code is less efficient but is
10749 able to correctly support denormalized numbers and exceptional IEEE
10750 values such as not-a-number and plus/minus infinity. Other Alpha
10751 compilers call this option @option{-ieee_with_no_inexact}.
10753 @item -mieee-with-inexact
10754 @opindex mieee-with-inexact
10755 This is like @option{-mieee} except the generated code also maintains
10756 the IEEE @var{inexact-flag}. Turning on this option causes the
10757 generated code to implement fully-compliant IEEE math. In addition to
10758 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10759 macro. On some Alpha implementations the resulting code may execute
10760 significantly slower than the code generated by default. Since there is
10761 very little code that depends on the @var{inexact-flag}, you should
10762 normally not specify this option. Other Alpha compilers call this
10763 option @option{-ieee_with_inexact}.
10765 @item -mfp-trap-mode=@var{trap-mode}
10766 @opindex mfp-trap-mode
10767 This option controls what floating-point related traps are enabled.
10768 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10769 The trap mode can be set to one of four values:
10773 This is the default (normal) setting. The only traps that are enabled
10774 are the ones that cannot be disabled in software (e.g., division by zero
10778 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10782 Like @samp{u}, but the instructions are marked to be safe for software
10783 completion (see Alpha architecture manual for details).
10786 Like @samp{su}, but inexact traps are enabled as well.
10789 @item -mfp-rounding-mode=@var{rounding-mode}
10790 @opindex mfp-rounding-mode
10791 Selects the IEEE rounding mode. Other Alpha compilers call this option
10792 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10797 Normal IEEE rounding mode. Floating point numbers are rounded towards
10798 the nearest machine number or towards the even machine number in case
10802 Round towards minus infinity.
10805 Chopped rounding mode. Floating point numbers are rounded towards zero.
10808 Dynamic rounding mode. A field in the floating point control register
10809 (@var{fpcr}, see Alpha architecture reference manual) controls the
10810 rounding mode in effect. The C library initializes this register for
10811 rounding towards plus infinity. Thus, unless your program modifies the
10812 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10815 @item -mtrap-precision=@var{trap-precision}
10816 @opindex mtrap-precision
10817 In the Alpha architecture, floating point traps are imprecise. This
10818 means without software assistance it is impossible to recover from a
10819 floating trap and program execution normally needs to be terminated.
10820 GCC can generate code that can assist operating system trap handlers
10821 in determining the exact location that caused a floating point trap.
10822 Depending on the requirements of an application, different levels of
10823 precisions can be selected:
10827 Program precision. This option is the default and means a trap handler
10828 can only identify which program caused a floating point exception.
10831 Function precision. The trap handler can determine the function that
10832 caused a floating point exception.
10835 Instruction precision. The trap handler can determine the exact
10836 instruction that caused a floating point exception.
10839 Other Alpha compilers provide the equivalent options called
10840 @option{-scope_safe} and @option{-resumption_safe}.
10842 @item -mieee-conformant
10843 @opindex mieee-conformant
10844 This option marks the generated code as IEEE conformant. You must not
10845 use this option unless you also specify @option{-mtrap-precision=i} and either
10846 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10847 is to emit the line @samp{.eflag 48} in the function prologue of the
10848 generated assembly file. Under DEC Unix, this has the effect that
10849 IEEE-conformant math library routines will be linked in.
10851 @item -mbuild-constants
10852 @opindex mbuild-constants
10853 Normally GCC examines a 32- or 64-bit integer constant to
10854 see if it can construct it from smaller constants in two or three
10855 instructions. If it cannot, it will output the constant as a literal and
10856 generate code to load it from the data segment at runtime.
10858 Use this option to require GCC to construct @emph{all} integer constants
10859 using code, even if it takes more instructions (the maximum is six).
10861 You would typically use this option to build a shared library dynamic
10862 loader. Itself a shared library, it must relocate itself in memory
10863 before it can find the variables and constants in its own data segment.
10869 Select whether to generate code to be assembled by the vendor-supplied
10870 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10888 Indicate whether GCC should generate code to use the optional BWX,
10889 CIX, FIX and MAX instruction sets. The default is to use the instruction
10890 sets supported by the CPU type specified via @option{-mcpu=} option or that
10891 of the CPU on which GCC was built if none was specified.
10894 @itemx -mfloat-ieee
10895 @opindex mfloat-vax
10896 @opindex mfloat-ieee
10897 Generate code that uses (does not use) VAX F and G floating point
10898 arithmetic instead of IEEE single and double precision.
10900 @item -mexplicit-relocs
10901 @itemx -mno-explicit-relocs
10902 @opindex mexplicit-relocs
10903 @opindex mno-explicit-relocs
10904 Older Alpha assemblers provided no way to generate symbol relocations
10905 except via assembler macros. Use of these macros does not allow
10906 optimal instruction scheduling. GNU binutils as of version 2.12
10907 supports a new syntax that allows the compiler to explicitly mark
10908 which relocations should apply to which instructions. This option
10909 is mostly useful for debugging, as GCC detects the capabilities of
10910 the assembler when it is built and sets the default accordingly.
10913 @itemx -mlarge-data
10914 @opindex msmall-data
10915 @opindex mlarge-data
10916 When @option{-mexplicit-relocs} is in effect, static data is
10917 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10918 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10919 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10920 16-bit relocations off of the @code{$gp} register. This limits the
10921 size of the small data area to 64KB, but allows the variables to be
10922 directly accessed via a single instruction.
10924 The default is @option{-mlarge-data}. With this option the data area
10925 is limited to just below 2GB@. Programs that require more than 2GB of
10926 data must use @code{malloc} or @code{mmap} to allocate the data in the
10927 heap instead of in the program's data segment.
10929 When generating code for shared libraries, @option{-fpic} implies
10930 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10933 @itemx -mlarge-text
10934 @opindex msmall-text
10935 @opindex mlarge-text
10936 When @option{-msmall-text} is used, the compiler assumes that the
10937 code of the entire program (or shared library) fits in 4MB, and is
10938 thus reachable with a branch instruction. When @option{-msmall-data}
10939 is used, the compiler can assume that all local symbols share the
10940 same @code{$gp} value, and thus reduce the number of instructions
10941 required for a function call from 4 to 1.
10943 The default is @option{-mlarge-text}.
10945 @item -mcpu=@var{cpu_type}
10947 Set the instruction set and instruction scheduling parameters for
10948 machine type @var{cpu_type}. You can specify either the @samp{EV}
10949 style name or the corresponding chip number. GCC supports scheduling
10950 parameters for the EV4, EV5 and EV6 family of processors and will
10951 choose the default values for the instruction set from the processor
10952 you specify. If you do not specify a processor type, GCC will default
10953 to the processor on which the compiler was built.
10955 Supported values for @var{cpu_type} are
10961 Schedules as an EV4 and has no instruction set extensions.
10965 Schedules as an EV5 and has no instruction set extensions.
10969 Schedules as an EV5 and supports the BWX extension.
10974 Schedules as an EV5 and supports the BWX and MAX extensions.
10978 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10982 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10985 Native Linux/GNU toolchains also support the value @samp{native},
10986 which selects the best architecture option for the host processor.
10987 @option{-mcpu=native} has no effect if GCC does not recognize
10990 @item -mtune=@var{cpu_type}
10992 Set only the instruction scheduling parameters for machine type
10993 @var{cpu_type}. The instruction set is not changed.
10995 Native Linux/GNU toolchains also support the value @samp{native},
10996 which selects the best architecture option for the host processor.
10997 @option{-mtune=native} has no effect if GCC does not recognize
11000 @item -mmemory-latency=@var{time}
11001 @opindex mmemory-latency
11002 Sets the latency the scheduler should assume for typical memory
11003 references as seen by the application. This number is highly
11004 dependent on the memory access patterns used by the application
11005 and the size of the external cache on the machine.
11007 Valid options for @var{time} are
11011 A decimal number representing clock cycles.
11017 The compiler contains estimates of the number of clock cycles for
11018 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
11019 (also called Dcache, Scache, and Bcache), as well as to main memory.
11020 Note that L3 is only valid for EV5.
11025 @node DEC Alpha/VMS Options
11026 @subsection DEC Alpha/VMS Options
11028 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
11031 @item -mvms-return-codes
11032 @opindex mvms-return-codes
11033 Return VMS condition codes from main. The default is to return POSIX
11034 style condition (e.g.@: error) codes.
11036 @item -mdebug-main=@var{prefix}
11037 @opindex mdebug-main=@var{prefix}
11038 Flag the first routine whose name starts with @var{prefix} as the main
11039 routine for the debugger.
11043 Default to 64bit memory allocation routines.
11047 @subsection FR30 Options
11048 @cindex FR30 Options
11050 These options are defined specifically for the FR30 port.
11054 @item -msmall-model
11055 @opindex msmall-model
11056 Use the small address space model. This can produce smaller code, but
11057 it does assume that all symbolic values and addresses will fit into a
11062 Assume that run-time support has been provided and so there is no need
11063 to include the simulator library (@file{libsim.a}) on the linker
11069 @subsection FRV Options
11070 @cindex FRV Options
11076 Only use the first 32 general purpose registers.
11081 Use all 64 general purpose registers.
11086 Use only the first 32 floating point registers.
11091 Use all 64 floating point registers
11094 @opindex mhard-float
11096 Use hardware instructions for floating point operations.
11099 @opindex msoft-float
11101 Use library routines for floating point operations.
11106 Dynamically allocate condition code registers.
11111 Do not try to dynamically allocate condition code registers, only
11112 use @code{icc0} and @code{fcc0}.
11117 Change ABI to use double word insns.
11122 Do not use double word instructions.
11127 Use floating point double instructions.
11130 @opindex mno-double
11132 Do not use floating point double instructions.
11137 Use media instructions.
11142 Do not use media instructions.
11147 Use multiply and add/subtract instructions.
11150 @opindex mno-muladd
11152 Do not use multiply and add/subtract instructions.
11157 Select the FDPIC ABI, that uses function descriptors to represent
11158 pointers to functions. Without any PIC/PIE-related options, it
11159 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11160 assumes GOT entries and small data are within a 12-bit range from the
11161 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11162 are computed with 32 bits.
11163 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11166 @opindex minline-plt
11168 Enable inlining of PLT entries in function calls to functions that are
11169 not known to bind locally. It has no effect without @option{-mfdpic}.
11170 It's enabled by default if optimizing for speed and compiling for
11171 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11172 optimization option such as @option{-O3} or above is present in the
11178 Assume a large TLS segment when generating thread-local code.
11183 Do not assume a large TLS segment when generating thread-local code.
11188 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11189 that is known to be in read-only sections. It's enabled by default,
11190 except for @option{-fpic} or @option{-fpie}: even though it may help
11191 make the global offset table smaller, it trades 1 instruction for 4.
11192 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11193 one of which may be shared by multiple symbols, and it avoids the need
11194 for a GOT entry for the referenced symbol, so it's more likely to be a
11195 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11197 @item -multilib-library-pic
11198 @opindex multilib-library-pic
11200 Link with the (library, not FD) pic libraries. It's implied by
11201 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11202 @option{-fpic} without @option{-mfdpic}. You should never have to use
11206 @opindex mlinked-fp
11208 Follow the EABI requirement of always creating a frame pointer whenever
11209 a stack frame is allocated. This option is enabled by default and can
11210 be disabled with @option{-mno-linked-fp}.
11213 @opindex mlong-calls
11215 Use indirect addressing to call functions outside the current
11216 compilation unit. This allows the functions to be placed anywhere
11217 within the 32-bit address space.
11219 @item -malign-labels
11220 @opindex malign-labels
11222 Try to align labels to an 8-byte boundary by inserting nops into the
11223 previous packet. This option only has an effect when VLIW packing
11224 is enabled. It doesn't create new packets; it merely adds nops to
11227 @item -mlibrary-pic
11228 @opindex mlibrary-pic
11230 Generate position-independent EABI code.
11235 Use only the first four media accumulator registers.
11240 Use all eight media accumulator registers.
11245 Pack VLIW instructions.
11250 Do not pack VLIW instructions.
11253 @opindex mno-eflags
11255 Do not mark ABI switches in e_flags.
11258 @opindex mcond-move
11260 Enable the use of conditional-move instructions (default).
11262 This switch is mainly for debugging the compiler and will likely be removed
11263 in a future version.
11265 @item -mno-cond-move
11266 @opindex mno-cond-move
11268 Disable the use of conditional-move instructions.
11270 This switch is mainly for debugging the compiler and will likely be removed
11271 in a future version.
11276 Enable the use of conditional set instructions (default).
11278 This switch is mainly for debugging the compiler and will likely be removed
11279 in a future version.
11284 Disable the use of conditional set instructions.
11286 This switch is mainly for debugging the compiler and will likely be removed
11287 in a future version.
11290 @opindex mcond-exec
11292 Enable the use of conditional execution (default).
11294 This switch is mainly for debugging the compiler and will likely be removed
11295 in a future version.
11297 @item -mno-cond-exec
11298 @opindex mno-cond-exec
11300 Disable the use of conditional execution.
11302 This switch is mainly for debugging the compiler and will likely be removed
11303 in a future version.
11305 @item -mvliw-branch
11306 @opindex mvliw-branch
11308 Run a pass to pack branches into VLIW instructions (default).
11310 This switch is mainly for debugging the compiler and will likely be removed
11311 in a future version.
11313 @item -mno-vliw-branch
11314 @opindex mno-vliw-branch
11316 Do not run a pass to pack branches into VLIW instructions.
11318 This switch is mainly for debugging the compiler and will likely be removed
11319 in a future version.
11321 @item -mmulti-cond-exec
11322 @opindex mmulti-cond-exec
11324 Enable optimization of @code{&&} and @code{||} in conditional execution
11327 This switch is mainly for debugging the compiler and will likely be removed
11328 in a future version.
11330 @item -mno-multi-cond-exec
11331 @opindex mno-multi-cond-exec
11333 Disable optimization of @code{&&} and @code{||} in conditional execution.
11335 This switch is mainly for debugging the compiler and will likely be removed
11336 in a future version.
11338 @item -mnested-cond-exec
11339 @opindex mnested-cond-exec
11341 Enable nested conditional execution optimizations (default).
11343 This switch is mainly for debugging the compiler and will likely be removed
11344 in a future version.
11346 @item -mno-nested-cond-exec
11347 @opindex mno-nested-cond-exec
11349 Disable nested conditional execution optimizations.
11351 This switch is mainly for debugging the compiler and will likely be removed
11352 in a future version.
11354 @item -moptimize-membar
11355 @opindex moptimize-membar
11357 This switch removes redundant @code{membar} instructions from the
11358 compiler generated code. It is enabled by default.
11360 @item -mno-optimize-membar
11361 @opindex mno-optimize-membar
11363 This switch disables the automatic removal of redundant @code{membar}
11364 instructions from the generated code.
11366 @item -mtomcat-stats
11367 @opindex mtomcat-stats
11369 Cause gas to print out tomcat statistics.
11371 @item -mcpu=@var{cpu}
11374 Select the processor type for which to generate code. Possible values are
11375 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11376 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11380 @node GNU/Linux Options
11381 @subsection GNU/Linux Options
11383 These @samp{-m} options are defined for GNU/Linux targets:
11388 Use the GNU C library instead of uClibc. This is the default except
11389 on @samp{*-*-linux-*uclibc*} targets.
11393 Use uClibc instead of the GNU C library. This is the default on
11394 @samp{*-*-linux-*uclibc*} targets.
11397 @node H8/300 Options
11398 @subsection H8/300 Options
11400 These @samp{-m} options are defined for the H8/300 implementations:
11405 Shorten some address references at link time, when possible; uses the
11406 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11407 ld, Using ld}, for a fuller description.
11411 Generate code for the H8/300H@.
11415 Generate code for the H8S@.
11419 Generate code for the H8S and H8/300H in the normal mode. This switch
11420 must be used either with @option{-mh} or @option{-ms}.
11424 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11428 Make @code{int} data 32 bits by default.
11431 @opindex malign-300
11432 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11433 The default for the H8/300H and H8S is to align longs and floats on 4
11435 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11436 This option has no effect on the H8/300.
11440 @subsection HPPA Options
11441 @cindex HPPA Options
11443 These @samp{-m} options are defined for the HPPA family of computers:
11446 @item -march=@var{architecture-type}
11448 Generate code for the specified architecture. The choices for
11449 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11450 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11451 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11452 architecture option for your machine. Code compiled for lower numbered
11453 architectures will run on higher numbered architectures, but not the
11456 @item -mpa-risc-1-0
11457 @itemx -mpa-risc-1-1
11458 @itemx -mpa-risc-2-0
11459 @opindex mpa-risc-1-0
11460 @opindex mpa-risc-1-1
11461 @opindex mpa-risc-2-0
11462 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11465 @opindex mbig-switch
11466 Generate code suitable for big switch tables. Use this option only if
11467 the assembler/linker complain about out of range branches within a switch
11470 @item -mjump-in-delay
11471 @opindex mjump-in-delay
11472 Fill delay slots of function calls with unconditional jump instructions
11473 by modifying the return pointer for the function call to be the target
11474 of the conditional jump.
11476 @item -mdisable-fpregs
11477 @opindex mdisable-fpregs
11478 Prevent floating point registers from being used in any manner. This is
11479 necessary for compiling kernels which perform lazy context switching of
11480 floating point registers. If you use this option and attempt to perform
11481 floating point operations, the compiler will abort.
11483 @item -mdisable-indexing
11484 @opindex mdisable-indexing
11485 Prevent the compiler from using indexing address modes. This avoids some
11486 rather obscure problems when compiling MIG generated code under MACH@.
11488 @item -mno-space-regs
11489 @opindex mno-space-regs
11490 Generate code that assumes the target has no space registers. This allows
11491 GCC to generate faster indirect calls and use unscaled index address modes.
11493 Such code is suitable for level 0 PA systems and kernels.
11495 @item -mfast-indirect-calls
11496 @opindex mfast-indirect-calls
11497 Generate code that assumes calls never cross space boundaries. This
11498 allows GCC to emit code which performs faster indirect calls.
11500 This option will not work in the presence of shared libraries or nested
11503 @item -mfixed-range=@var{register-range}
11504 @opindex mfixed-range
11505 Generate code treating the given register range as fixed registers.
11506 A fixed register is one that the register allocator can not use. This is
11507 useful when compiling kernel code. A register range is specified as
11508 two registers separated by a dash. Multiple register ranges can be
11509 specified separated by a comma.
11511 @item -mlong-load-store
11512 @opindex mlong-load-store
11513 Generate 3-instruction load and store sequences as sometimes required by
11514 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11517 @item -mportable-runtime
11518 @opindex mportable-runtime
11519 Use the portable calling conventions proposed by HP for ELF systems.
11523 Enable the use of assembler directives only GAS understands.
11525 @item -mschedule=@var{cpu-type}
11527 Schedule code according to the constraints for the machine type
11528 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11529 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11530 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11531 proper scheduling option for your machine. The default scheduling is
11535 @opindex mlinker-opt
11536 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11537 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11538 linkers in which they give bogus error messages when linking some programs.
11541 @opindex msoft-float
11542 Generate output containing library calls for floating point.
11543 @strong{Warning:} the requisite libraries are not available for all HPPA
11544 targets. Normally the facilities of the machine's usual C compiler are
11545 used, but this cannot be done directly in cross-compilation. You must make
11546 your own arrangements to provide suitable library functions for
11549 @option{-msoft-float} changes the calling convention in the output file;
11550 therefore, it is only useful if you compile @emph{all} of a program with
11551 this option. In particular, you need to compile @file{libgcc.a}, the
11552 library that comes with GCC, with @option{-msoft-float} in order for
11557 Generate the predefine, @code{_SIO}, for server IO@. The default is
11558 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11559 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11560 options are available under HP-UX and HI-UX@.
11564 Use GNU ld specific options. This passes @option{-shared} to ld when
11565 building a shared library. It is the default when GCC is configured,
11566 explicitly or implicitly, with the GNU linker. This option does not
11567 have any affect on which ld is called, it only changes what parameters
11568 are passed to that ld. The ld that is called is determined by the
11569 @option{--with-ld} configure option, GCC's program search path, and
11570 finally by the user's @env{PATH}. The linker used by GCC can be printed
11571 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11572 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11576 Use HP ld specific options. This passes @option{-b} to ld when building
11577 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11578 links. It is the default when GCC is configured, explicitly or
11579 implicitly, with the HP linker. This option does not have any affect on
11580 which ld is called, it only changes what parameters are passed to that
11581 ld. The ld that is called is determined by the @option{--with-ld}
11582 configure option, GCC's program search path, and finally by the user's
11583 @env{PATH}. The linker used by GCC can be printed using @samp{which
11584 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11585 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11588 @opindex mno-long-calls
11589 Generate code that uses long call sequences. This ensures that a call
11590 is always able to reach linker generated stubs. The default is to generate
11591 long calls only when the distance from the call site to the beginning
11592 of the function or translation unit, as the case may be, exceeds a
11593 predefined limit set by the branch type being used. The limits for
11594 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11595 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11598 Distances are measured from the beginning of functions when using the
11599 @option{-ffunction-sections} option, or when using the @option{-mgas}
11600 and @option{-mno-portable-runtime} options together under HP-UX with
11603 It is normally not desirable to use this option as it will degrade
11604 performance. However, it may be useful in large applications,
11605 particularly when partial linking is used to build the application.
11607 The types of long calls used depends on the capabilities of the
11608 assembler and linker, and the type of code being generated. The
11609 impact on systems that support long absolute calls, and long pic
11610 symbol-difference or pc-relative calls should be relatively small.
11611 However, an indirect call is used on 32-bit ELF systems in pic code
11612 and it is quite long.
11614 @item -munix=@var{unix-std}
11616 Generate compiler predefines and select a startfile for the specified
11617 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11618 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11619 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11620 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11621 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11624 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11625 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11626 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11627 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11628 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11629 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11631 It is @emph{important} to note that this option changes the interfaces
11632 for various library routines. It also affects the operational behavior
11633 of the C library. Thus, @emph{extreme} care is needed in using this
11636 Library code that is intended to operate with more than one UNIX
11637 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11638 as appropriate. Most GNU software doesn't provide this capability.
11642 Suppress the generation of link options to search libdld.sl when the
11643 @option{-static} option is specified on HP-UX 10 and later.
11647 The HP-UX implementation of setlocale in libc has a dependency on
11648 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11649 when the @option{-static} option is specified, special link options
11650 are needed to resolve this dependency.
11652 On HP-UX 10 and later, the GCC driver adds the necessary options to
11653 link with libdld.sl when the @option{-static} option is specified.
11654 This causes the resulting binary to be dynamic. On the 64-bit port,
11655 the linkers generate dynamic binaries by default in any case. The
11656 @option{-nolibdld} option can be used to prevent the GCC driver from
11657 adding these link options.
11661 Add support for multithreading with the @dfn{dce thread} library
11662 under HP-UX@. This option sets flags for both the preprocessor and
11666 @node i386 and x86-64 Options
11667 @subsection Intel 386 and AMD x86-64 Options
11668 @cindex i386 Options
11669 @cindex x86-64 Options
11670 @cindex Intel 386 Options
11671 @cindex AMD x86-64 Options
11673 These @samp{-m} options are defined for the i386 and x86-64 family of
11677 @item -mtune=@var{cpu-type}
11679 Tune to @var{cpu-type} everything applicable about the generated code, except
11680 for the ABI and the set of available instructions. The choices for
11681 @var{cpu-type} are:
11684 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11685 If you know the CPU on which your code will run, then you should use
11686 the corresponding @option{-mtune} option instead of
11687 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11688 of your application will have, then you should use this option.
11690 As new processors are deployed in the marketplace, the behavior of this
11691 option will change. Therefore, if you upgrade to a newer version of
11692 GCC, the code generated option will change to reflect the processors
11693 that were most common when that version of GCC was released.
11695 There is no @option{-march=generic} option because @option{-march}
11696 indicates the instruction set the compiler can use, and there is no
11697 generic instruction set applicable to all processors. In contrast,
11698 @option{-mtune} indicates the processor (or, in this case, collection of
11699 processors) for which the code is optimized.
11701 This selects the CPU to tune for at compilation time by determining
11702 the processor type of the compiling machine. Using @option{-mtune=native}
11703 will produce code optimized for the local machine under the constraints
11704 of the selected instruction set. Using @option{-march=native} will
11705 enable all instruction subsets supported by the local machine (hence
11706 the result might not run on different machines).
11708 Original Intel's i386 CPU@.
11710 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11711 @item i586, pentium
11712 Intel Pentium CPU with no MMX support.
11714 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11716 Intel PentiumPro CPU@.
11718 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11719 instruction set will be used, so the code will run on all i686 family chips.
11721 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11722 @item pentium3, pentium3m
11723 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11726 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11727 support. Used by Centrino notebooks.
11728 @item pentium4, pentium4m
11729 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11731 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11734 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11735 SSE2 and SSE3 instruction set support.
11737 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11738 instruction set support.
11740 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11741 instruction set support.
11743 AMD K6 CPU with MMX instruction set support.
11745 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
11746 @item athlon, athlon-tbird
11747 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
11749 @item athlon-4, athlon-xp, athlon-mp
11750 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
11751 instruction set support.
11752 @item k8, opteron, athlon64, athlon-fx
11753 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11754 MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
11755 @item k8-sse3, opteron-sse3, athlon64-sse3
11756 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11757 @item amdfam10, barcelona
11758 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11759 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
11760 instruction set extensions.)
11762 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11765 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
11766 instruction set support.
11768 Via C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
11769 implemented for this chip.)
11771 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11772 implemented for this chip.)
11774 Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
11777 While picking a specific @var{cpu-type} will schedule things appropriately
11778 for that particular chip, the compiler will not generate any code that
11779 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11782 @item -march=@var{cpu-type}
11784 Generate instructions for the machine type @var{cpu-type}. The choices
11785 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11786 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11788 @item -mcpu=@var{cpu-type}
11790 A deprecated synonym for @option{-mtune}.
11792 @item -mfpmath=@var{unit}
11794 Generate floating point arithmetics for selected unit @var{unit}. The choices
11795 for @var{unit} are:
11799 Use the standard 387 floating point coprocessor present majority of chips and
11800 emulated otherwise. Code compiled with this option will run almost everywhere.
11801 The temporary results are computed in 80bit precision instead of precision
11802 specified by the type resulting in slightly different results compared to most
11803 of other chips. See @option{-ffloat-store} for more detailed description.
11805 This is the default choice for i386 compiler.
11808 Use scalar floating point instructions present in the SSE instruction set.
11809 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11810 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11811 instruction set supports only single precision arithmetics, thus the double and
11812 extended precision arithmetics is still done using 387. Later version, present
11813 only in Pentium4 and the future AMD x86-64 chips supports double precision
11816 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11817 or @option{-msse2} switches to enable SSE extensions and make this option
11818 effective. For the x86-64 compiler, these extensions are enabled by default.
11820 The resulting code should be considerably faster in the majority of cases and avoid
11821 the numerical instability problems of 387 code, but may break some existing
11822 code that expects temporaries to be 80bit.
11824 This is the default choice for the x86-64 compiler.
11829 Attempt to utilize both instruction sets at once. This effectively double the
11830 amount of available registers and on chips with separate execution units for
11831 387 and SSE the execution resources too. Use this option with care, as it is
11832 still experimental, because the GCC register allocator does not model separate
11833 functional units well resulting in instable performance.
11836 @item -masm=@var{dialect}
11837 @opindex masm=@var{dialect}
11838 Output asm instructions using selected @var{dialect}. Supported
11839 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11840 not support @samp{intel}.
11843 @itemx -mno-ieee-fp
11845 @opindex mno-ieee-fp
11846 Control whether or not the compiler uses IEEE floating point
11847 comparisons. These handle correctly the case where the result of a
11848 comparison is unordered.
11851 @opindex msoft-float
11852 Generate output containing library calls for floating point.
11853 @strong{Warning:} the requisite libraries are not part of GCC@.
11854 Normally the facilities of the machine's usual C compiler are used, but
11855 this can't be done directly in cross-compilation. You must make your
11856 own arrangements to provide suitable library functions for
11859 On machines where a function returns floating point results in the 80387
11860 register stack, some floating point opcodes may be emitted even if
11861 @option{-msoft-float} is used.
11863 @item -mno-fp-ret-in-387
11864 @opindex mno-fp-ret-in-387
11865 Do not use the FPU registers for return values of functions.
11867 The usual calling convention has functions return values of types
11868 @code{float} and @code{double} in an FPU register, even if there
11869 is no FPU@. The idea is that the operating system should emulate
11872 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11873 in ordinary CPU registers instead.
11875 @item -mno-fancy-math-387
11876 @opindex mno-fancy-math-387
11877 Some 387 emulators do not support the @code{sin}, @code{cos} and
11878 @code{sqrt} instructions for the 387. Specify this option to avoid
11879 generating those instructions. This option is the default on FreeBSD,
11880 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11881 indicates that the target cpu will always have an FPU and so the
11882 instruction will not need emulation. As of revision 2.6.1, these
11883 instructions are not generated unless you also use the
11884 @option{-funsafe-math-optimizations} switch.
11886 @item -malign-double
11887 @itemx -mno-align-double
11888 @opindex malign-double
11889 @opindex mno-align-double
11890 Control whether GCC aligns @code{double}, @code{long double}, and
11891 @code{long long} variables on a two word boundary or a one word
11892 boundary. Aligning @code{double} variables on a two word boundary will
11893 produce code that runs somewhat faster on a @samp{Pentium} at the
11894 expense of more memory.
11896 On x86-64, @option{-malign-double} is enabled by default.
11898 @strong{Warning:} if you use the @option{-malign-double} switch,
11899 structures containing the above types will be aligned differently than
11900 the published application binary interface specifications for the 386
11901 and will not be binary compatible with structures in code compiled
11902 without that switch.
11904 @item -m96bit-long-double
11905 @itemx -m128bit-long-double
11906 @opindex m96bit-long-double
11907 @opindex m128bit-long-double
11908 These switches control the size of @code{long double} type. The i386
11909 application binary interface specifies the size to be 96 bits,
11910 so @option{-m96bit-long-double} is the default in 32 bit mode.
11912 Modern architectures (Pentium and newer) would prefer @code{long double}
11913 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11914 conforming to the ABI, this would not be possible. So specifying a
11915 @option{-m128bit-long-double} will align @code{long double}
11916 to a 16 byte boundary by padding the @code{long double} with an additional
11919 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11920 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11922 Notice that neither of these options enable any extra precision over the x87
11923 standard of 80 bits for a @code{long double}.
11925 @strong{Warning:} if you override the default value for your target ABI, the
11926 structures and arrays containing @code{long double} variables will change
11927 their size as well as function calling convention for function taking
11928 @code{long double} will be modified. Hence they will not be binary
11929 compatible with arrays or structures in code compiled without that switch.
11931 @item -mlarge-data-threshold=@var{number}
11932 @opindex mlarge-data-threshold=@var{number}
11933 When @option{-mcmodel=medium} is specified, the data greater than
11934 @var{threshold} are placed in large data section. This value must be the
11935 same across all object linked into the binary and defaults to 65535.
11939 Use a different function-calling convention, in which functions that
11940 take a fixed number of arguments return with the @code{ret} @var{num}
11941 instruction, which pops their arguments while returning. This saves one
11942 instruction in the caller since there is no need to pop the arguments
11945 You can specify that an individual function is called with this calling
11946 sequence with the function attribute @samp{stdcall}. You can also
11947 override the @option{-mrtd} option by using the function attribute
11948 @samp{cdecl}. @xref{Function Attributes}.
11950 @strong{Warning:} this calling convention is incompatible with the one
11951 normally used on Unix, so you cannot use it if you need to call
11952 libraries compiled with the Unix compiler.
11954 Also, you must provide function prototypes for all functions that
11955 take variable numbers of arguments (including @code{printf});
11956 otherwise incorrect code will be generated for calls to those
11959 In addition, seriously incorrect code will result if you call a
11960 function with too many arguments. (Normally, extra arguments are
11961 harmlessly ignored.)
11963 @item -mregparm=@var{num}
11965 Control how many registers are used to pass integer arguments. By
11966 default, no registers are used to pass arguments, and at most 3
11967 registers can be used. You can control this behavior for a specific
11968 function by using the function attribute @samp{regparm}.
11969 @xref{Function Attributes}.
11971 @strong{Warning:} if you use this switch, and
11972 @var{num} is nonzero, then you must build all modules with the same
11973 value, including any libraries. This includes the system libraries and
11977 @opindex msseregparm
11978 Use SSE register passing conventions for float and double arguments
11979 and return values. You can control this behavior for a specific
11980 function by using the function attribute @samp{sseregparm}.
11981 @xref{Function Attributes}.
11983 @strong{Warning:} if you use this switch then you must build all
11984 modules with the same value, including any libraries. This includes
11985 the system libraries and startup modules.
11994 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11995 is specified, the significands of results of floating-point operations are
11996 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11997 significands of results of floating-point operations to 53 bits (double
11998 precision) and @option{-mpc80} rounds the significands of results of
11999 floating-point operations to 64 bits (extended double precision), which is
12000 the default. When this option is used, floating-point operations in higher
12001 precisions are not available to the programmer without setting the FPU
12002 control word explicitly.
12004 Setting the rounding of floating-point operations to less than the default
12005 80 bits can speed some programs by 2% or more. Note that some mathematical
12006 libraries assume that extended precision (80 bit) floating-point operations
12007 are enabled by default; routines in such libraries could suffer significant
12008 loss of accuracy, typically through so-called "catastrophic cancellation",
12009 when this option is used to set the precision to less than extended precision.
12011 @item -mstackrealign
12012 @opindex mstackrealign
12013 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
12014 option will generate an alternate prologue and epilogue that realigns the
12015 runtime stack if necessary. This supports mixing legacy codes that keep
12016 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
12017 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
12018 applicable to individual functions.
12020 @item -mpreferred-stack-boundary=@var{num}
12021 @opindex mpreferred-stack-boundary
12022 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
12023 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
12024 the default is 4 (16 bytes or 128 bits).
12026 @item -mincoming-stack-boundary=@var{num}
12027 @opindex mincoming-stack-boundary
12028 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
12029 boundary. If @option{-mincoming-stack-boundary} is not specified,
12030 the one specified by @option{-mpreferred-stack-boundary} will be used.
12032 On Pentium and PentiumPro, @code{double} and @code{long double} values
12033 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
12034 suffer significant run time performance penalties. On Pentium III, the
12035 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
12036 properly if it is not 16 byte aligned.
12038 To ensure proper alignment of this values on the stack, the stack boundary
12039 must be as aligned as that required by any value stored on the stack.
12040 Further, every function must be generated such that it keeps the stack
12041 aligned. Thus calling a function compiled with a higher preferred
12042 stack boundary from a function compiled with a lower preferred stack
12043 boundary will most likely misalign the stack. It is recommended that
12044 libraries that use callbacks always use the default setting.
12046 This extra alignment does consume extra stack space, and generally
12047 increases code size. Code that is sensitive to stack space usage, such
12048 as embedded systems and operating system kernels, may want to reduce the
12049 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12093 These switches enable or disable the use of instructions in the MMX,
12094 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, FMA4, XOP,
12095 LWP, ABM or 3DNow!@: extended instruction sets.
12096 These extensions are also available as built-in functions: see
12097 @ref{X86 Built-in Functions}, for details of the functions enabled and
12098 disabled by these switches.
12100 To have SSE/SSE2 instructions generated automatically from floating-point
12101 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12103 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12104 generates new AVX instructions or AVX equivalence for all SSEx instructions
12107 These options will enable GCC to use these extended instructions in
12108 generated code, even without @option{-mfpmath=sse}. Applications which
12109 perform runtime CPU detection must compile separate files for each
12110 supported architecture, using the appropriate flags. In particular,
12111 the file containing the CPU detection code should be compiled without
12115 @itemx -mno-fused-madd
12116 @opindex mfused-madd
12117 @opindex mno-fused-madd
12118 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12119 instructions. The default is to use these instructions.
12123 This option instructs GCC to emit a @code{cld} instruction in the prologue
12124 of functions that use string instructions. String instructions depend on
12125 the DF flag to select between autoincrement or autodecrement mode. While the
12126 ABI specifies the DF flag to be cleared on function entry, some operating
12127 systems violate this specification by not clearing the DF flag in their
12128 exception dispatchers. The exception handler can be invoked with the DF flag
12129 set which leads to wrong direction mode, when string instructions are used.
12130 This option can be enabled by default on 32-bit x86 targets by configuring
12131 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12132 instructions can be suppressed with the @option{-mno-cld} compiler option
12137 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12138 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12139 data types. This is useful for high resolution counters that could be updated
12140 by multiple processors (or cores). This instruction is generated as part of
12141 atomic built-in functions: see @ref{Atomic Builtins} for details.
12145 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12146 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12147 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12148 SAHF are load and store instructions, respectively, for certain status flags.
12149 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12150 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12154 This option will enable GCC to use movbe instruction to implement
12155 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12159 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12160 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12161 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12165 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12166 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12167 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12168 variants) for single precision floating point arguments. These instructions
12169 are generated only when @option{-funsafe-math-optimizations} is enabled
12170 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12171 Note that while the throughput of the sequence is higher than the throughput
12172 of the non-reciprocal instruction, the precision of the sequence can be
12173 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12175 Note that GCC implements 1.0f/sqrtf(x) in terms of RSQRTSS (or RSQRTPS)
12176 already with @option{-ffast-math} (or the above option combination), and
12177 doesn't need @option{-mrecip}.
12179 @item -mveclibabi=@var{type}
12180 @opindex mveclibabi
12181 Specifies the ABI type to use for vectorizing intrinsics using an
12182 external library. Supported types are @code{svml} for the Intel short
12183 vector math library and @code{acml} for the AMD math core library style
12184 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12185 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12186 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12187 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12188 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12189 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12190 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12191 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12192 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12193 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12194 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12195 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12196 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12197 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12198 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12199 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12200 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12201 compatible library will have to be specified at link time.
12203 @item -mabi=@var{name}
12205 Generate code for the specified calling convention. Permissible values
12206 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12207 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12208 ABI when targeting Windows. On all other systems, the default is the
12209 SYSV ABI. You can control this behavior for a specific function by
12210 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12211 @xref{Function Attributes}.
12214 @itemx -mno-push-args
12215 @opindex mpush-args
12216 @opindex mno-push-args
12217 Use PUSH operations to store outgoing parameters. This method is shorter
12218 and usually equally fast as method using SUB/MOV operations and is enabled
12219 by default. In some cases disabling it may improve performance because of
12220 improved scheduling and reduced dependencies.
12222 @item -maccumulate-outgoing-args
12223 @opindex maccumulate-outgoing-args
12224 If enabled, the maximum amount of space required for outgoing arguments will be
12225 computed in the function prologue. This is faster on most modern CPUs
12226 because of reduced dependencies, improved scheduling and reduced stack usage
12227 when preferred stack boundary is not equal to 2. The drawback is a notable
12228 increase in code size. This switch implies @option{-mno-push-args}.
12232 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12233 on thread-safe exception handling must compile and link all code with the
12234 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12235 @option{-D_MT}; when linking, it links in a special thread helper library
12236 @option{-lmingwthrd} which cleans up per thread exception handling data.
12238 @item -mno-align-stringops
12239 @opindex mno-align-stringops
12240 Do not align destination of inlined string operations. This switch reduces
12241 code size and improves performance in case the destination is already aligned,
12242 but GCC doesn't know about it.
12244 @item -minline-all-stringops
12245 @opindex minline-all-stringops
12246 By default GCC inlines string operations only when destination is known to be
12247 aligned at least to 4 byte boundary. This enables more inlining, increase code
12248 size, but may improve performance of code that depends on fast memcpy, strlen
12249 and memset for short lengths.
12251 @item -minline-stringops-dynamically
12252 @opindex minline-stringops-dynamically
12253 For string operation of unknown size, inline runtime checks so for small
12254 blocks inline code is used, while for large blocks library call is used.
12256 @item -mstringop-strategy=@var{alg}
12257 @opindex mstringop-strategy=@var{alg}
12258 Overwrite internal decision heuristic about particular algorithm to inline
12259 string operation with. The allowed values are @code{rep_byte},
12260 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12261 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12262 expanding inline loop, @code{libcall} for always expanding library call.
12264 @item -momit-leaf-frame-pointer
12265 @opindex momit-leaf-frame-pointer
12266 Don't keep the frame pointer in a register for leaf functions. This
12267 avoids the instructions to save, set up and restore frame pointers and
12268 makes an extra register available in leaf functions. The option
12269 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12270 which might make debugging harder.
12272 @item -mtls-direct-seg-refs
12273 @itemx -mno-tls-direct-seg-refs
12274 @opindex mtls-direct-seg-refs
12275 Controls whether TLS variables may be accessed with offsets from the
12276 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12277 or whether the thread base pointer must be added. Whether or not this
12278 is legal depends on the operating system, and whether it maps the
12279 segment to cover the entire TLS area.
12281 For systems that use GNU libc, the default is on.
12284 @itemx -mno-sse2avx
12286 Specify that the assembler should encode SSE instructions with VEX
12287 prefix. The option @option{-mavx} turns this on by default.
12290 These @samp{-m} switches are supported in addition to the above
12291 on AMD x86-64 processors in 64-bit environments.
12298 Generate code for a 32-bit or 64-bit environment.
12299 The 32-bit environment sets int, long and pointer to 32 bits and
12300 generates code that runs on any i386 system.
12301 The 64-bit environment sets int to 32 bits and long and pointer
12302 to 64 bits and generates code for AMD's x86-64 architecture. For
12303 darwin only the -m64 option turns off the @option{-fno-pic} and
12304 @option{-mdynamic-no-pic} options.
12306 @item -mno-red-zone
12307 @opindex mno-red-zone
12308 Do not use a so called red zone for x86-64 code. The red zone is mandated
12309 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12310 stack pointer that will not be modified by signal or interrupt handlers
12311 and therefore can be used for temporary data without adjusting the stack
12312 pointer. The flag @option{-mno-red-zone} disables this red zone.
12314 @item -mcmodel=small
12315 @opindex mcmodel=small
12316 Generate code for the small code model: the program and its symbols must
12317 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12318 Programs can be statically or dynamically linked. This is the default
12321 @item -mcmodel=kernel
12322 @opindex mcmodel=kernel
12323 Generate code for the kernel code model. The kernel runs in the
12324 negative 2 GB of the address space.
12325 This model has to be used for Linux kernel code.
12327 @item -mcmodel=medium
12328 @opindex mcmodel=medium
12329 Generate code for the medium model: The program is linked in the lower 2
12330 GB of the address space. Small symbols are also placed there. Symbols
12331 with sizes larger than @option{-mlarge-data-threshold} are put into
12332 large data or bss sections and can be located above 2GB. Programs can
12333 be statically or dynamically linked.
12335 @item -mcmodel=large
12336 @opindex mcmodel=large
12337 Generate code for the large model: This model makes no assumptions
12338 about addresses and sizes of sections.
12341 @node IA-64 Options
12342 @subsection IA-64 Options
12343 @cindex IA-64 Options
12345 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12349 @opindex mbig-endian
12350 Generate code for a big endian target. This is the default for HP-UX@.
12352 @item -mlittle-endian
12353 @opindex mlittle-endian
12354 Generate code for a little endian target. This is the default for AIX5
12360 @opindex mno-gnu-as
12361 Generate (or don't) code for the GNU assembler. This is the default.
12362 @c Also, this is the default if the configure option @option{--with-gnu-as}
12368 @opindex mno-gnu-ld
12369 Generate (or don't) code for the GNU linker. This is the default.
12370 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12375 Generate code that does not use a global pointer register. The result
12376 is not position independent code, and violates the IA-64 ABI@.
12378 @item -mvolatile-asm-stop
12379 @itemx -mno-volatile-asm-stop
12380 @opindex mvolatile-asm-stop
12381 @opindex mno-volatile-asm-stop
12382 Generate (or don't) a stop bit immediately before and after volatile asm
12385 @item -mregister-names
12386 @itemx -mno-register-names
12387 @opindex mregister-names
12388 @opindex mno-register-names
12389 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12390 the stacked registers. This may make assembler output more readable.
12396 Disable (or enable) optimizations that use the small data section. This may
12397 be useful for working around optimizer bugs.
12399 @item -mconstant-gp
12400 @opindex mconstant-gp
12401 Generate code that uses a single constant global pointer value. This is
12402 useful when compiling kernel code.
12406 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12407 This is useful when compiling firmware code.
12409 @item -minline-float-divide-min-latency
12410 @opindex minline-float-divide-min-latency
12411 Generate code for inline divides of floating point values
12412 using the minimum latency algorithm.
12414 @item -minline-float-divide-max-throughput
12415 @opindex minline-float-divide-max-throughput
12416 Generate code for inline divides of floating point values
12417 using the maximum throughput algorithm.
12419 @item -mno-inline-float-divide
12420 @opindex mno-inline-float-divide
12421 Do not generate inline code for divides of floating point values.
12423 @item -minline-int-divide-min-latency
12424 @opindex minline-int-divide-min-latency
12425 Generate code for inline divides of integer values
12426 using the minimum latency algorithm.
12428 @item -minline-int-divide-max-throughput
12429 @opindex minline-int-divide-max-throughput
12430 Generate code for inline divides of integer values
12431 using the maximum throughput algorithm.
12433 @item -mno-inline-int-divide
12434 @opindex mno-inline-int-divide
12435 Do not generate inline code for divides of integer values.
12437 @item -minline-sqrt-min-latency
12438 @opindex minline-sqrt-min-latency
12439 Generate code for inline square roots
12440 using the minimum latency algorithm.
12442 @item -minline-sqrt-max-throughput
12443 @opindex minline-sqrt-max-throughput
12444 Generate code for inline square roots
12445 using the maximum throughput algorithm.
12447 @item -mno-inline-sqrt
12448 @opindex mno-inline-sqrt
12449 Do not generate inline code for sqrt.
12452 @itemx -mno-fused-madd
12453 @opindex mfused-madd
12454 @opindex mno-fused-madd
12455 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12456 instructions. The default is to use these instructions.
12458 @item -mno-dwarf2-asm
12459 @itemx -mdwarf2-asm
12460 @opindex mno-dwarf2-asm
12461 @opindex mdwarf2-asm
12462 Don't (or do) generate assembler code for the DWARF2 line number debugging
12463 info. This may be useful when not using the GNU assembler.
12465 @item -mearly-stop-bits
12466 @itemx -mno-early-stop-bits
12467 @opindex mearly-stop-bits
12468 @opindex mno-early-stop-bits
12469 Allow stop bits to be placed earlier than immediately preceding the
12470 instruction that triggered the stop bit. This can improve instruction
12471 scheduling, but does not always do so.
12473 @item -mfixed-range=@var{register-range}
12474 @opindex mfixed-range
12475 Generate code treating the given register range as fixed registers.
12476 A fixed register is one that the register allocator can not use. This is
12477 useful when compiling kernel code. A register range is specified as
12478 two registers separated by a dash. Multiple register ranges can be
12479 specified separated by a comma.
12481 @item -mtls-size=@var{tls-size}
12483 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12486 @item -mtune=@var{cpu-type}
12488 Tune the instruction scheduling for a particular CPU, Valid values are
12489 itanium, itanium1, merced, itanium2, and mckinley.
12495 Generate code for a 32-bit or 64-bit environment.
12496 The 32-bit environment sets int, long and pointer to 32 bits.
12497 The 64-bit environment sets int to 32 bits and long and pointer
12498 to 64 bits. These are HP-UX specific flags.
12500 @item -mno-sched-br-data-spec
12501 @itemx -msched-br-data-spec
12502 @opindex mno-sched-br-data-spec
12503 @opindex msched-br-data-spec
12504 (Dis/En)able data speculative scheduling before reload.
12505 This will result in generation of the ld.a instructions and
12506 the corresponding check instructions (ld.c / chk.a).
12507 The default is 'disable'.
12509 @item -msched-ar-data-spec
12510 @itemx -mno-sched-ar-data-spec
12511 @opindex msched-ar-data-spec
12512 @opindex mno-sched-ar-data-spec
12513 (En/Dis)able data speculative scheduling after reload.
12514 This will result in generation of the ld.a instructions and
12515 the corresponding check instructions (ld.c / chk.a).
12516 The default is 'enable'.
12518 @item -mno-sched-control-spec
12519 @itemx -msched-control-spec
12520 @opindex mno-sched-control-spec
12521 @opindex msched-control-spec
12522 (Dis/En)able control speculative scheduling. This feature is
12523 available only during region scheduling (i.e.@: before reload).
12524 This will result in generation of the ld.s instructions and
12525 the corresponding check instructions chk.s .
12526 The default is 'disable'.
12528 @item -msched-br-in-data-spec
12529 @itemx -mno-sched-br-in-data-spec
12530 @opindex msched-br-in-data-spec
12531 @opindex mno-sched-br-in-data-spec
12532 (En/Dis)able speculative scheduling of the instructions that
12533 are dependent on the data speculative loads before reload.
12534 This is effective only with @option{-msched-br-data-spec} enabled.
12535 The default is 'enable'.
12537 @item -msched-ar-in-data-spec
12538 @itemx -mno-sched-ar-in-data-spec
12539 @opindex msched-ar-in-data-spec
12540 @opindex mno-sched-ar-in-data-spec
12541 (En/Dis)able speculative scheduling of the instructions that
12542 are dependent on the data speculative loads after reload.
12543 This is effective only with @option{-msched-ar-data-spec} enabled.
12544 The default is 'enable'.
12546 @item -msched-in-control-spec
12547 @itemx -mno-sched-in-control-spec
12548 @opindex msched-in-control-spec
12549 @opindex mno-sched-in-control-spec
12550 (En/Dis)able speculative scheduling of the instructions that
12551 are dependent on the control speculative loads.
12552 This is effective only with @option{-msched-control-spec} enabled.
12553 The default is 'enable'.
12555 @item -mno-sched-prefer-non-data-spec-insns
12556 @itemx -msched-prefer-non-data-spec-insns
12557 @opindex mno-sched-prefer-non-data-spec-insns
12558 @opindex msched-prefer-non-data-spec-insns
12559 If enabled, data speculative instructions will be chosen for schedule
12560 only if there are no other choices at the moment. This will make
12561 the use of the data speculation much more conservative.
12562 The default is 'disable'.
12564 @item -mno-sched-prefer-non-control-spec-insns
12565 @itemx -msched-prefer-non-control-spec-insns
12566 @opindex mno-sched-prefer-non-control-spec-insns
12567 @opindex msched-prefer-non-control-spec-insns
12568 If enabled, control speculative instructions will be chosen for schedule
12569 only if there are no other choices at the moment. This will make
12570 the use of the control speculation much more conservative.
12571 The default is 'disable'.
12573 @item -mno-sched-count-spec-in-critical-path
12574 @itemx -msched-count-spec-in-critical-path
12575 @opindex mno-sched-count-spec-in-critical-path
12576 @opindex msched-count-spec-in-critical-path
12577 If enabled, speculative dependencies will be considered during
12578 computation of the instructions priorities. This will make the use of the
12579 speculation a bit more conservative.
12580 The default is 'disable'.
12582 @item -msched-spec-ldc
12583 @opindex msched-spec-ldc
12584 Use a simple data speculation check. This option is on by default.
12586 @item -msched-control-spec-ldc
12587 @opindex msched-spec-ldc
12588 Use a simple check for control speculation. This option is on by default.
12590 @item -msched-stop-bits-after-every-cycle
12591 @opindex msched-stop-bits-after-every-cycle
12592 Place a stop bit after every cycle when scheduling. This option is on
12595 @item -msched-fp-mem-deps-zero-cost
12596 @opindex msched-fp-mem-deps-zero-cost
12597 Assume that floating-point stores and loads are not likely to cause a conflict
12598 when placed into the same instruction group. This option is disabled by
12601 @item -msel-sched-dont-check-control-spec
12602 @opindex msel-sched-dont-check-control-spec
12603 Generate checks for control speculation in selective scheduling.
12604 This flag is disabled by default.
12606 @item -msched-max-memory-insns=@var{max-insns}
12607 @opindex msched-max-memory-insns
12608 Limit on the number of memory insns per instruction group, giving lower
12609 priority to subsequent memory insns attempting to schedule in the same
12610 instruction group. Frequently useful to prevent cache bank conflicts.
12611 The default value is 1.
12613 @item -msched-max-memory-insns-hard-limit
12614 @opindex msched-max-memory-insns-hard-limit
12615 Disallow more than `msched-max-memory-insns' in instruction group.
12616 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12617 when limit is reached but may still schedule memory operations.
12621 @node IA-64/VMS Options
12622 @subsection IA-64/VMS Options
12624 These @samp{-m} options are defined for the IA-64/VMS implementations:
12627 @item -mvms-return-codes
12628 @opindex mvms-return-codes
12629 Return VMS condition codes from main. The default is to return POSIX
12630 style condition (e.g.@ error) codes.
12632 @item -mdebug-main=@var{prefix}
12633 @opindex mdebug-main=@var{prefix}
12634 Flag the first routine whose name starts with @var{prefix} as the main
12635 routine for the debugger.
12639 Default to 64bit memory allocation routines.
12643 @subsection LM32 Options
12644 @cindex LM32 options
12646 These @option{-m} options are defined for the Lattice Mico32 architecture:
12649 @item -mbarrel-shift-enabled
12650 @opindex mbarrel-shift-enabled
12651 Enable barrel-shift instructions.
12653 @item -mdivide-enabled
12654 @opindex mdivide-enabled
12655 Enable divide and modulus instructions.
12657 @item -mmultiply-enabled
12658 @opindex multiply-enabled
12659 Enable multiply instructions.
12661 @item -msign-extend-enabled
12662 @opindex msign-extend-enabled
12663 Enable sign extend instructions.
12665 @item -muser-enabled
12666 @opindex muser-enabled
12667 Enable user-defined instructions.
12672 @subsection M32C Options
12673 @cindex M32C options
12676 @item -mcpu=@var{name}
12678 Select the CPU for which code is generated. @var{name} may be one of
12679 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12680 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12681 the M32C/80 series.
12685 Specifies that the program will be run on the simulator. This causes
12686 an alternate runtime library to be linked in which supports, for
12687 example, file I/O@. You must not use this option when generating
12688 programs that will run on real hardware; you must provide your own
12689 runtime library for whatever I/O functions are needed.
12691 @item -memregs=@var{number}
12693 Specifies the number of memory-based pseudo-registers GCC will use
12694 during code generation. These pseudo-registers will be used like real
12695 registers, so there is a tradeoff between GCC's ability to fit the
12696 code into available registers, and the performance penalty of using
12697 memory instead of registers. Note that all modules in a program must
12698 be compiled with the same value for this option. Because of that, you
12699 must not use this option with the default runtime libraries gcc
12704 @node M32R/D Options
12705 @subsection M32R/D Options
12706 @cindex M32R/D options
12708 These @option{-m} options are defined for Renesas M32R/D architectures:
12713 Generate code for the M32R/2@.
12717 Generate code for the M32R/X@.
12721 Generate code for the M32R@. This is the default.
12723 @item -mmodel=small
12724 @opindex mmodel=small
12725 Assume all objects live in the lower 16MB of memory (so that their addresses
12726 can be loaded with the @code{ld24} instruction), and assume all subroutines
12727 are reachable with the @code{bl} instruction.
12728 This is the default.
12730 The addressability of a particular object can be set with the
12731 @code{model} attribute.
12733 @item -mmodel=medium
12734 @opindex mmodel=medium
12735 Assume objects may be anywhere in the 32-bit address space (the compiler
12736 will generate @code{seth/add3} instructions to load their addresses), and
12737 assume all subroutines are reachable with the @code{bl} instruction.
12739 @item -mmodel=large
12740 @opindex mmodel=large
12741 Assume objects may be anywhere in the 32-bit address space (the compiler
12742 will generate @code{seth/add3} instructions to load their addresses), and
12743 assume subroutines may not be reachable with the @code{bl} instruction
12744 (the compiler will generate the much slower @code{seth/add3/jl}
12745 instruction sequence).
12748 @opindex msdata=none
12749 Disable use of the small data area. Variables will be put into
12750 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12751 @code{section} attribute has been specified).
12752 This is the default.
12754 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12755 Objects may be explicitly put in the small data area with the
12756 @code{section} attribute using one of these sections.
12758 @item -msdata=sdata
12759 @opindex msdata=sdata
12760 Put small global and static data in the small data area, but do not
12761 generate special code to reference them.
12764 @opindex msdata=use
12765 Put small global and static data in the small data area, and generate
12766 special instructions to reference them.
12770 @cindex smaller data references
12771 Put global and static objects less than or equal to @var{num} bytes
12772 into the small data or bss sections instead of the normal data or bss
12773 sections. The default value of @var{num} is 8.
12774 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12775 for this option to have any effect.
12777 All modules should be compiled with the same @option{-G @var{num}} value.
12778 Compiling with different values of @var{num} may or may not work; if it
12779 doesn't the linker will give an error message---incorrect code will not be
12784 Makes the M32R specific code in the compiler display some statistics
12785 that might help in debugging programs.
12787 @item -malign-loops
12788 @opindex malign-loops
12789 Align all loops to a 32-byte boundary.
12791 @item -mno-align-loops
12792 @opindex mno-align-loops
12793 Do not enforce a 32-byte alignment for loops. This is the default.
12795 @item -missue-rate=@var{number}
12796 @opindex missue-rate=@var{number}
12797 Issue @var{number} instructions per cycle. @var{number} can only be 1
12800 @item -mbranch-cost=@var{number}
12801 @opindex mbranch-cost=@var{number}
12802 @var{number} can only be 1 or 2. If it is 1 then branches will be
12803 preferred over conditional code, if it is 2, then the opposite will
12806 @item -mflush-trap=@var{number}
12807 @opindex mflush-trap=@var{number}
12808 Specifies the trap number to use to flush the cache. The default is
12809 12. Valid numbers are between 0 and 15 inclusive.
12811 @item -mno-flush-trap
12812 @opindex mno-flush-trap
12813 Specifies that the cache cannot be flushed by using a trap.
12815 @item -mflush-func=@var{name}
12816 @opindex mflush-func=@var{name}
12817 Specifies the name of the operating system function to call to flush
12818 the cache. The default is @emph{_flush_cache}, but a function call
12819 will only be used if a trap is not available.
12821 @item -mno-flush-func
12822 @opindex mno-flush-func
12823 Indicates that there is no OS function for flushing the cache.
12827 @node M680x0 Options
12828 @subsection M680x0 Options
12829 @cindex M680x0 options
12831 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
12832 The default settings depend on which architecture was selected when
12833 the compiler was configured; the defaults for the most common choices
12837 @item -march=@var{arch}
12839 Generate code for a specific M680x0 or ColdFire instruction set
12840 architecture. Permissible values of @var{arch} for M680x0
12841 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
12842 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
12843 architectures are selected according to Freescale's ISA classification
12844 and the permissible values are: @samp{isaa}, @samp{isaaplus},
12845 @samp{isab} and @samp{isac}.
12847 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
12848 code for a ColdFire target. The @var{arch} in this macro is one of the
12849 @option{-march} arguments given above.
12851 When used together, @option{-march} and @option{-mtune} select code
12852 that runs on a family of similar processors but that is optimized
12853 for a particular microarchitecture.
12855 @item -mcpu=@var{cpu}
12857 Generate code for a specific M680x0 or ColdFire processor.
12858 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12859 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12860 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12861 below, which also classifies the CPUs into families:
12863 @multitable @columnfractions 0.20 0.80
12864 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12865 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
12866 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12867 @item @samp{5206e} @tab @samp{5206e}
12868 @item @samp{5208} @tab @samp{5207} @samp{5208}
12869 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12870 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12871 @item @samp{5216} @tab @samp{5214} @samp{5216}
12872 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12873 @item @samp{5225} @tab @samp{5224} @samp{5225}
12874 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
12875 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12876 @item @samp{5249} @tab @samp{5249}
12877 @item @samp{5250} @tab @samp{5250}
12878 @item @samp{5271} @tab @samp{5270} @samp{5271}
12879 @item @samp{5272} @tab @samp{5272}
12880 @item @samp{5275} @tab @samp{5274} @samp{5275}
12881 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12882 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
12883 @item @samp{5307} @tab @samp{5307}
12884 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12885 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12886 @item @samp{5407} @tab @samp{5407}
12887 @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}
12890 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12891 @var{arch} is compatible with @var{cpu}. Other combinations of
12892 @option{-mcpu} and @option{-march} are rejected.
12894 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12895 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12896 where the value of @var{family} is given by the table above.
12898 @item -mtune=@var{tune}
12900 Tune the code for a particular microarchitecture, within the
12901 constraints set by @option{-march} and @option{-mcpu}.
12902 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12903 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12904 and @samp{cpu32}. The ColdFire microarchitectures
12905 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12907 You can also use @option{-mtune=68020-40} for code that needs
12908 to run relatively well on 68020, 68030 and 68040 targets.
12909 @option{-mtune=68020-60} is similar but includes 68060 targets
12910 as well. These two options select the same tuning decisions as
12911 @option{-m68020-40} and @option{-m68020-60} respectively.
12913 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12914 when tuning for 680x0 architecture @var{arch}. It also defines
12915 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12916 option is used. If gcc is tuning for a range of architectures,
12917 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12918 it defines the macros for every architecture in the range.
12920 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12921 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12922 of the arguments given above.
12928 Generate output for a 68000. This is the default
12929 when the compiler is configured for 68000-based systems.
12930 It is equivalent to @option{-march=68000}.
12932 Use this option for microcontrollers with a 68000 or EC000 core,
12933 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12937 Generate output for a 68010. This is the default
12938 when the compiler is configured for 68010-based systems.
12939 It is equivalent to @option{-march=68010}.
12945 Generate output for a 68020. This is the default
12946 when the compiler is configured for 68020-based systems.
12947 It is equivalent to @option{-march=68020}.
12951 Generate output for a 68030. This is the default when the compiler is
12952 configured for 68030-based systems. It is equivalent to
12953 @option{-march=68030}.
12957 Generate output for a 68040. This is the default when the compiler is
12958 configured for 68040-based systems. It is equivalent to
12959 @option{-march=68040}.
12961 This option inhibits the use of 68881/68882 instructions that have to be
12962 emulated by software on the 68040. Use this option if your 68040 does not
12963 have code to emulate those instructions.
12967 Generate output for a 68060. This is the default when the compiler is
12968 configured for 68060-based systems. It is equivalent to
12969 @option{-march=68060}.
12971 This option inhibits the use of 68020 and 68881/68882 instructions that
12972 have to be emulated by software on the 68060. Use this option if your 68060
12973 does not have code to emulate those instructions.
12977 Generate output for a CPU32. This is the default
12978 when the compiler is configured for CPU32-based systems.
12979 It is equivalent to @option{-march=cpu32}.
12981 Use this option for microcontrollers with a
12982 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12983 68336, 68340, 68341, 68349 and 68360.
12987 Generate output for a 520X ColdFire CPU@. This is the default
12988 when the compiler is configured for 520X-based systems.
12989 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12990 in favor of that option.
12992 Use this option for microcontroller with a 5200 core, including
12993 the MCF5202, MCF5203, MCF5204 and MCF5206.
12997 Generate output for a 5206e ColdFire CPU@. The option is now
12998 deprecated in favor of the equivalent @option{-mcpu=5206e}.
13002 Generate output for a member of the ColdFire 528X family.
13003 The option is now deprecated in favor of the equivalent
13004 @option{-mcpu=528x}.
13008 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
13009 in favor of the equivalent @option{-mcpu=5307}.
13013 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
13014 in favor of the equivalent @option{-mcpu=5407}.
13018 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
13019 This includes use of hardware floating point instructions.
13020 The option is equivalent to @option{-mcpu=547x}, and is now
13021 deprecated in favor of that option.
13025 Generate output for a 68040, without using any of the new instructions.
13026 This results in code which can run relatively efficiently on either a
13027 68020/68881 or a 68030 or a 68040. The generated code does use the
13028 68881 instructions that are emulated on the 68040.
13030 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
13034 Generate output for a 68060, without using any of the new instructions.
13035 This results in code which can run relatively efficiently on either a
13036 68020/68881 or a 68030 or a 68040. The generated code does use the
13037 68881 instructions that are emulated on the 68060.
13039 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
13043 @opindex mhard-float
13045 Generate floating-point instructions. This is the default for 68020
13046 and above, and for ColdFire devices that have an FPU@. It defines the
13047 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
13048 on ColdFire targets.
13051 @opindex msoft-float
13052 Do not generate floating-point instructions; use library calls instead.
13053 This is the default for 68000, 68010, and 68832 targets. It is also
13054 the default for ColdFire devices that have no FPU.
13060 Generate (do not generate) ColdFire hardware divide and remainder
13061 instructions. If @option{-march} is used without @option{-mcpu},
13062 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
13063 architectures. Otherwise, the default is taken from the target CPU
13064 (either the default CPU, or the one specified by @option{-mcpu}). For
13065 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
13066 @option{-mcpu=5206e}.
13068 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
13072 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13073 Additionally, parameters passed on the stack are also aligned to a
13074 16-bit boundary even on targets whose API mandates promotion to 32-bit.
13078 Do not consider type @code{int} to be 16 bits wide. This is the default.
13081 @itemx -mno-bitfield
13082 @opindex mnobitfield
13083 @opindex mno-bitfield
13084 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
13085 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
13089 Do use the bit-field instructions. The @option{-m68020} option implies
13090 @option{-mbitfield}. This is the default if you use a configuration
13091 designed for a 68020.
13095 Use a different function-calling convention, in which functions
13096 that take a fixed number of arguments return with the @code{rtd}
13097 instruction, which pops their arguments while returning. This
13098 saves one instruction in the caller since there is no need to pop
13099 the arguments there.
13101 This calling convention is incompatible with the one normally
13102 used on Unix, so you cannot use it if you need to call libraries
13103 compiled with the Unix compiler.
13105 Also, you must provide function prototypes for all functions that
13106 take variable numbers of arguments (including @code{printf});
13107 otherwise incorrect code will be generated for calls to those
13110 In addition, seriously incorrect code will result if you call a
13111 function with too many arguments. (Normally, extra arguments are
13112 harmlessly ignored.)
13114 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13115 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13119 Do not use the calling conventions selected by @option{-mrtd}.
13120 This is the default.
13123 @itemx -mno-align-int
13124 @opindex malign-int
13125 @opindex mno-align-int
13126 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13127 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13128 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13129 Aligning variables on 32-bit boundaries produces code that runs somewhat
13130 faster on processors with 32-bit busses at the expense of more memory.
13132 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13133 align structures containing the above types differently than
13134 most published application binary interface specifications for the m68k.
13138 Use the pc-relative addressing mode of the 68000 directly, instead of
13139 using a global offset table. At present, this option implies @option{-fpic},
13140 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13141 not presently supported with @option{-mpcrel}, though this could be supported for
13142 68020 and higher processors.
13144 @item -mno-strict-align
13145 @itemx -mstrict-align
13146 @opindex mno-strict-align
13147 @opindex mstrict-align
13148 Do not (do) assume that unaligned memory references will be handled by
13152 Generate code that allows the data segment to be located in a different
13153 area of memory from the text segment. This allows for execute in place in
13154 an environment without virtual memory management. This option implies
13157 @item -mno-sep-data
13158 Generate code that assumes that the data segment follows the text segment.
13159 This is the default.
13161 @item -mid-shared-library
13162 Generate code that supports shared libraries via the library ID method.
13163 This allows for execute in place and shared libraries in an environment
13164 without virtual memory management. This option implies @option{-fPIC}.
13166 @item -mno-id-shared-library
13167 Generate code that doesn't assume ID based shared libraries are being used.
13168 This is the default.
13170 @item -mshared-library-id=n
13171 Specified the identification number of the ID based shared library being
13172 compiled. Specifying a value of 0 will generate more compact code, specifying
13173 other values will force the allocation of that number to the current
13174 library but is no more space or time efficient than omitting this option.
13180 When generating position-independent code for ColdFire, generate code
13181 that works if the GOT has more than 8192 entries. This code is
13182 larger and slower than code generated without this option. On M680x0
13183 processors, this option is not needed; @option{-fPIC} suffices.
13185 GCC normally uses a single instruction to load values from the GOT@.
13186 While this is relatively efficient, it only works if the GOT
13187 is smaller than about 64k. Anything larger causes the linker
13188 to report an error such as:
13190 @cindex relocation truncated to fit (ColdFire)
13192 relocation truncated to fit: R_68K_GOT16O foobar
13195 If this happens, you should recompile your code with @option{-mxgot}.
13196 It should then work with very large GOTs. However, code generated with
13197 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13198 the value of a global symbol.
13200 Note that some linkers, including newer versions of the GNU linker,
13201 can create multiple GOTs and sort GOT entries. If you have such a linker,
13202 you should only need to use @option{-mxgot} when compiling a single
13203 object file that accesses more than 8192 GOT entries. Very few do.
13205 These options have no effect unless GCC is generating
13206 position-independent code.
13210 @node M68hc1x Options
13211 @subsection M68hc1x Options
13212 @cindex M68hc1x options
13214 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13215 microcontrollers. The default values for these options depends on
13216 which style of microcontroller was selected when the compiler was configured;
13217 the defaults for the most common choices are given below.
13224 Generate output for a 68HC11. This is the default
13225 when the compiler is configured for 68HC11-based systems.
13231 Generate output for a 68HC12. This is the default
13232 when the compiler is configured for 68HC12-based systems.
13238 Generate output for a 68HCS12.
13240 @item -mauto-incdec
13241 @opindex mauto-incdec
13242 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13249 Enable the use of 68HC12 min and max instructions.
13252 @itemx -mno-long-calls
13253 @opindex mlong-calls
13254 @opindex mno-long-calls
13255 Treat all calls as being far away (near). If calls are assumed to be
13256 far away, the compiler will use the @code{call} instruction to
13257 call a function and the @code{rtc} instruction for returning.
13261 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13263 @item -msoft-reg-count=@var{count}
13264 @opindex msoft-reg-count
13265 Specify the number of pseudo-soft registers which are used for the
13266 code generation. The maximum number is 32. Using more pseudo-soft
13267 register may or may not result in better code depending on the program.
13268 The default is 4 for 68HC11 and 2 for 68HC12.
13272 @node MCore Options
13273 @subsection MCore Options
13274 @cindex MCore options
13276 These are the @samp{-m} options defined for the Motorola M*Core
13282 @itemx -mno-hardlit
13284 @opindex mno-hardlit
13285 Inline constants into the code stream if it can be done in two
13286 instructions or less.
13292 Use the divide instruction. (Enabled by default).
13294 @item -mrelax-immediate
13295 @itemx -mno-relax-immediate
13296 @opindex mrelax-immediate
13297 @opindex mno-relax-immediate
13298 Allow arbitrary sized immediates in bit operations.
13300 @item -mwide-bitfields
13301 @itemx -mno-wide-bitfields
13302 @opindex mwide-bitfields
13303 @opindex mno-wide-bitfields
13304 Always treat bit-fields as int-sized.
13306 @item -m4byte-functions
13307 @itemx -mno-4byte-functions
13308 @opindex m4byte-functions
13309 @opindex mno-4byte-functions
13310 Force all functions to be aligned to a four byte boundary.
13312 @item -mcallgraph-data
13313 @itemx -mno-callgraph-data
13314 @opindex mcallgraph-data
13315 @opindex mno-callgraph-data
13316 Emit callgraph information.
13319 @itemx -mno-slow-bytes
13320 @opindex mslow-bytes
13321 @opindex mno-slow-bytes
13322 Prefer word access when reading byte quantities.
13324 @item -mlittle-endian
13325 @itemx -mbig-endian
13326 @opindex mlittle-endian
13327 @opindex mbig-endian
13328 Generate code for a little endian target.
13334 Generate code for the 210 processor.
13338 Assume that run-time support has been provided and so omit the
13339 simulator library (@file{libsim.a)} from the linker command line.
13341 @item -mstack-increment=@var{size}
13342 @opindex mstack-increment
13343 Set the maximum amount for a single stack increment operation. Large
13344 values can increase the speed of programs which contain functions
13345 that need a large amount of stack space, but they can also trigger a
13346 segmentation fault if the stack is extended too much. The default
13352 @subsection MeP Options
13353 @cindex MeP options
13359 Enables the @code{abs} instruction, which is the absolute difference
13360 between two registers.
13364 Enables all the optional instructions - average, multiply, divide, bit
13365 operations, leading zero, absolute difference, min/max, clip, and
13371 Enables the @code{ave} instruction, which computes the average of two
13374 @item -mbased=@var{n}
13376 Variables of size @var{n} bytes or smaller will be placed in the
13377 @code{.based} section by default. Based variables use the @code{$tp}
13378 register as a base register, and there is a 128 byte limit to the
13379 @code{.based} section.
13383 Enables the bit operation instructions - bit test (@code{btstm}), set
13384 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13385 test-and-set (@code{tas}).
13387 @item -mc=@var{name}
13389 Selects which section constant data will be placed in. @var{name} may
13390 be @code{tiny}, @code{near}, or @code{far}.
13394 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13395 useful unless you also provide @code{-mminmax}.
13397 @item -mconfig=@var{name}
13399 Selects one of the build-in core configurations. Each MeP chip has
13400 one or more modules in it; each module has a core CPU and a variety of
13401 coprocessors, optional instructions, and peripherals. The
13402 @code{MeP-Integrator} tool, not part of GCC, provides these
13403 configurations through this option; using this option is the same as
13404 using all the corresponding command line options. The default
13405 configuration is @code{default}.
13409 Enables the coprocessor instructions. By default, this is a 32-bit
13410 coprocessor. Note that the coprocessor is normally enabled via the
13411 @code{-mconfig=} option.
13415 Enables the 32-bit coprocessor's instructions.
13419 Enables the 64-bit coprocessor's instructions.
13423 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13427 Causes constant variables to be placed in the @code{.near} section.
13431 Enables the @code{div} and @code{divu} instructions.
13435 Generate big-endian code.
13439 Generate little-endian code.
13441 @item -mio-volatile
13442 @opindex mio-volatile
13443 Tells the compiler that any variable marked with the @code{io}
13444 attribute is to be considered volatile.
13448 Causes variables to be assigned to the @code{.far} section by default.
13452 Enables the @code{leadz} (leading zero) instruction.
13456 Causes variables to be assigned to the @code{.near} section by default.
13460 Enables the @code{min} and @code{max} instructions.
13464 Enables the multiplication and multiply-accumulate instructions.
13468 Disables all the optional instructions enabled by @code{-mall-opts}.
13472 Enables the @code{repeat} and @code{erepeat} instructions, used for
13473 low-overhead looping.
13477 Causes all variables to default to the @code{.tiny} section. Note
13478 that there is a 65536 byte limit to this section. Accesses to these
13479 variables use the @code{%gp} base register.
13483 Enables the saturation instructions. Note that the compiler does not
13484 currently generate these itself, but this option is included for
13485 compatibility with other tools, like @code{as}.
13489 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13493 Link the simulator runtime libraries.
13497 Link the simulator runtime libraries, excluding built-in support
13498 for reset and exception vectors and tables.
13502 Causes all functions to default to the @code{.far} section. Without
13503 this option, functions default to the @code{.near} section.
13505 @item -mtiny=@var{n}
13507 Variables that are @var{n} bytes or smaller will be allocated to the
13508 @code{.tiny} section. These variables use the @code{$gp} base
13509 register. The default for this option is 4, but note that there's a
13510 65536 byte limit to the @code{.tiny} section.
13515 @subsection MIPS Options
13516 @cindex MIPS options
13522 Generate big-endian code.
13526 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13529 @item -march=@var{arch}
13531 Generate code that will run on @var{arch}, which can be the name of a
13532 generic MIPS ISA, or the name of a particular processor.
13534 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13535 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13536 The processor names are:
13537 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13538 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13539 @samp{5kc}, @samp{5kf},
13541 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13542 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13543 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13544 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13545 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13546 @samp{loongson2e}, @samp{loongson2f},
13550 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13551 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13552 @samp{rm7000}, @samp{rm9000},
13553 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13556 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13557 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13559 The special value @samp{from-abi} selects the
13560 most compatible architecture for the selected ABI (that is,
13561 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13563 Native Linux/GNU toolchains also support the value @samp{native},
13564 which selects the best architecture option for the host processor.
13565 @option{-march=native} has no effect if GCC does not recognize
13568 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13569 (for example, @samp{-march=r2k}). Prefixes are optional, and
13570 @samp{vr} may be written @samp{r}.
13572 Names of the form @samp{@var{n}f2_1} refer to processors with
13573 FPUs clocked at half the rate of the core, names of the form
13574 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13575 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13576 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13577 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13578 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13579 accepted as synonyms for @samp{@var{n}f1_1}.
13581 GCC defines two macros based on the value of this option. The first
13582 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13583 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13584 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13585 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13586 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13588 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13589 above. In other words, it will have the full prefix and will not
13590 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13591 the macro names the resolved architecture (either @samp{"mips1"} or
13592 @samp{"mips3"}). It names the default architecture when no
13593 @option{-march} option is given.
13595 @item -mtune=@var{arch}
13597 Optimize for @var{arch}. Among other things, this option controls
13598 the way instructions are scheduled, and the perceived cost of arithmetic
13599 operations. The list of @var{arch} values is the same as for
13602 When this option is not used, GCC will optimize for the processor
13603 specified by @option{-march}. By using @option{-march} and
13604 @option{-mtune} together, it is possible to generate code that will
13605 run on a family of processors, but optimize the code for one
13606 particular member of that family.
13608 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13609 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13610 @samp{-march} ones described above.
13614 Equivalent to @samp{-march=mips1}.
13618 Equivalent to @samp{-march=mips2}.
13622 Equivalent to @samp{-march=mips3}.
13626 Equivalent to @samp{-march=mips4}.
13630 Equivalent to @samp{-march=mips32}.
13634 Equivalent to @samp{-march=mips32r2}.
13638 Equivalent to @samp{-march=mips64}.
13642 Equivalent to @samp{-march=mips64r2}.
13647 @opindex mno-mips16
13648 Generate (do not generate) MIPS16 code. If GCC is targetting a
13649 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13651 MIPS16 code generation can also be controlled on a per-function basis
13652 by means of @code{mips16} and @code{nomips16} attributes.
13653 @xref{Function Attributes}, for more information.
13655 @item -mflip-mips16
13656 @opindex mflip-mips16
13657 Generate MIPS16 code on alternating functions. This option is provided
13658 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13659 not intended for ordinary use in compiling user code.
13661 @item -minterlink-mips16
13662 @itemx -mno-interlink-mips16
13663 @opindex minterlink-mips16
13664 @opindex mno-interlink-mips16
13665 Require (do not require) that non-MIPS16 code be link-compatible with
13668 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13669 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13670 therefore disables direct jumps unless GCC knows that the target of the
13671 jump is not MIPS16.
13683 Generate code for the given ABI@.
13685 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13686 generates 64-bit code when you select a 64-bit architecture, but you
13687 can use @option{-mgp32} to get 32-bit code instead.
13689 For information about the O64 ABI, see
13690 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13692 GCC supports a variant of the o32 ABI in which floating-point registers
13693 are 64 rather than 32 bits wide. You can select this combination with
13694 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13695 and @samp{mfhc1} instructions and is therefore only supported for
13696 MIPS32R2 processors.
13698 The register assignments for arguments and return values remain the
13699 same, but each scalar value is passed in a single 64-bit register
13700 rather than a pair of 32-bit registers. For example, scalar
13701 floating-point values are returned in @samp{$f0} only, not a
13702 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13703 remains the same, but all 64 bits are saved.
13706 @itemx -mno-abicalls
13708 @opindex mno-abicalls
13709 Generate (do not generate) code that is suitable for SVR4-style
13710 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13715 Generate (do not generate) code that is fully position-independent,
13716 and that can therefore be linked into shared libraries. This option
13717 only affects @option{-mabicalls}.
13719 All @option{-mabicalls} code has traditionally been position-independent,
13720 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13721 as an extension, the GNU toolchain allows executables to use absolute
13722 accesses for locally-binding symbols. It can also use shorter GP
13723 initialization sequences and generate direct calls to locally-defined
13724 functions. This mode is selected by @option{-mno-shared}.
13726 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13727 objects that can only be linked by the GNU linker. However, the option
13728 does not affect the ABI of the final executable; it only affects the ABI
13729 of relocatable objects. Using @option{-mno-shared} will generally make
13730 executables both smaller and quicker.
13732 @option{-mshared} is the default.
13738 Assume (do not assume) that the static and dynamic linkers
13739 support PLTs and copy relocations. This option only affects
13740 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13741 has no effect without @samp{-msym32}.
13743 You can make @option{-mplt} the default by configuring
13744 GCC with @option{--with-mips-plt}. The default is
13745 @option{-mno-plt} otherwise.
13751 Lift (do not lift) the usual restrictions on the size of the global
13754 GCC normally uses a single instruction to load values from the GOT@.
13755 While this is relatively efficient, it will only work if the GOT
13756 is smaller than about 64k. Anything larger will cause the linker
13757 to report an error such as:
13759 @cindex relocation truncated to fit (MIPS)
13761 relocation truncated to fit: R_MIPS_GOT16 foobar
13764 If this happens, you should recompile your code with @option{-mxgot}.
13765 It should then work with very large GOTs, although it will also be
13766 less efficient, since it will take three instructions to fetch the
13767 value of a global symbol.
13769 Note that some linkers can create multiple GOTs. If you have such a
13770 linker, you should only need to use @option{-mxgot} when a single object
13771 file accesses more than 64k's worth of GOT entries. Very few do.
13773 These options have no effect unless GCC is generating position
13778 Assume that general-purpose registers are 32 bits wide.
13782 Assume that general-purpose registers are 64 bits wide.
13786 Assume that floating-point registers are 32 bits wide.
13790 Assume that floating-point registers are 64 bits wide.
13793 @opindex mhard-float
13794 Use floating-point coprocessor instructions.
13797 @opindex msoft-float
13798 Do not use floating-point coprocessor instructions. Implement
13799 floating-point calculations using library calls instead.
13801 @item -msingle-float
13802 @opindex msingle-float
13803 Assume that the floating-point coprocessor only supports single-precision
13806 @item -mdouble-float
13807 @opindex mdouble-float
13808 Assume that the floating-point coprocessor supports double-precision
13809 operations. This is the default.
13815 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13816 implement atomic memory built-in functions. When neither option is
13817 specified, GCC will use the instructions if the target architecture
13820 @option{-mllsc} is useful if the runtime environment can emulate the
13821 instructions and @option{-mno-llsc} can be useful when compiling for
13822 nonstandard ISAs. You can make either option the default by
13823 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
13824 respectively. @option{--with-llsc} is the default for some
13825 configurations; see the installation documentation for details.
13831 Use (do not use) revision 1 of the MIPS DSP ASE@.
13832 @xref{MIPS DSP Built-in Functions}. This option defines the
13833 preprocessor macro @samp{__mips_dsp}. It also defines
13834 @samp{__mips_dsp_rev} to 1.
13840 Use (do not use) revision 2 of the MIPS DSP ASE@.
13841 @xref{MIPS DSP Built-in Functions}. This option defines the
13842 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
13843 It also defines @samp{__mips_dsp_rev} to 2.
13846 @itemx -mno-smartmips
13847 @opindex msmartmips
13848 @opindex mno-smartmips
13849 Use (do not use) the MIPS SmartMIPS ASE.
13851 @item -mpaired-single
13852 @itemx -mno-paired-single
13853 @opindex mpaired-single
13854 @opindex mno-paired-single
13855 Use (do not use) paired-single floating-point instructions.
13856 @xref{MIPS Paired-Single Support}. This option requires
13857 hardware floating-point support to be enabled.
13863 Use (do not use) MIPS Digital Media Extension instructions.
13864 This option can only be used when generating 64-bit code and requires
13865 hardware floating-point support to be enabled.
13870 @opindex mno-mips3d
13871 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
13872 The option @option{-mips3d} implies @option{-mpaired-single}.
13878 Use (do not use) MT Multithreading instructions.
13882 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
13883 an explanation of the default and the way that the pointer size is
13888 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
13890 The default size of @code{int}s, @code{long}s and pointers depends on
13891 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
13892 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
13893 32-bit @code{long}s. Pointers are the same size as @code{long}s,
13894 or the same size as integer registers, whichever is smaller.
13900 Assume (do not assume) that all symbols have 32-bit values, regardless
13901 of the selected ABI@. This option is useful in combination with
13902 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
13903 to generate shorter and faster references to symbolic addresses.
13907 Put definitions of externally-visible data in a small data section
13908 if that data is no bigger than @var{num} bytes. GCC can then access
13909 the data more efficiently; see @option{-mgpopt} for details.
13911 The default @option{-G} option depends on the configuration.
13913 @item -mlocal-sdata
13914 @itemx -mno-local-sdata
13915 @opindex mlocal-sdata
13916 @opindex mno-local-sdata
13917 Extend (do not extend) the @option{-G} behavior to local data too,
13918 such as to static variables in C@. @option{-mlocal-sdata} is the
13919 default for all configurations.
13921 If the linker complains that an application is using too much small data,
13922 you might want to try rebuilding the less performance-critical parts with
13923 @option{-mno-local-sdata}. You might also want to build large
13924 libraries with @option{-mno-local-sdata}, so that the libraries leave
13925 more room for the main program.
13927 @item -mextern-sdata
13928 @itemx -mno-extern-sdata
13929 @opindex mextern-sdata
13930 @opindex mno-extern-sdata
13931 Assume (do not assume) that externally-defined data will be in
13932 a small data section if that data is within the @option{-G} limit.
13933 @option{-mextern-sdata} is the default for all configurations.
13935 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
13936 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
13937 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
13938 is placed in a small data section. If @var{Var} is defined by another
13939 module, you must either compile that module with a high-enough
13940 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
13941 definition. If @var{Var} is common, you must link the application
13942 with a high-enough @option{-G} setting.
13944 The easiest way of satisfying these restrictions is to compile
13945 and link every module with the same @option{-G} option. However,
13946 you may wish to build a library that supports several different
13947 small data limits. You can do this by compiling the library with
13948 the highest supported @option{-G} setting and additionally using
13949 @option{-mno-extern-sdata} to stop the library from making assumptions
13950 about externally-defined data.
13956 Use (do not use) GP-relative accesses for symbols that are known to be
13957 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
13958 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
13961 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
13962 might not hold the value of @code{_gp}. For example, if the code is
13963 part of a library that might be used in a boot monitor, programs that
13964 call boot monitor routines will pass an unknown value in @code{$gp}.
13965 (In such situations, the boot monitor itself would usually be compiled
13966 with @option{-G0}.)
13968 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
13969 @option{-mno-extern-sdata}.
13971 @item -membedded-data
13972 @itemx -mno-embedded-data
13973 @opindex membedded-data
13974 @opindex mno-embedded-data
13975 Allocate variables to the read-only data section first if possible, then
13976 next in the small data section if possible, otherwise in data. This gives
13977 slightly slower code than the default, but reduces the amount of RAM required
13978 when executing, and thus may be preferred for some embedded systems.
13980 @item -muninit-const-in-rodata
13981 @itemx -mno-uninit-const-in-rodata
13982 @opindex muninit-const-in-rodata
13983 @opindex mno-uninit-const-in-rodata
13984 Put uninitialized @code{const} variables in the read-only data section.
13985 This option is only meaningful in conjunction with @option{-membedded-data}.
13987 @item -mcode-readable=@var{setting}
13988 @opindex mcode-readable
13989 Specify whether GCC may generate code that reads from executable sections.
13990 There are three possible settings:
13993 @item -mcode-readable=yes
13994 Instructions may freely access executable sections. This is the
13997 @item -mcode-readable=pcrel
13998 MIPS16 PC-relative load instructions can access executable sections,
13999 but other instructions must not do so. This option is useful on 4KSc
14000 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
14001 It is also useful on processors that can be configured to have a dual
14002 instruction/data SRAM interface and that, like the M4K, automatically
14003 redirect PC-relative loads to the instruction RAM.
14005 @item -mcode-readable=no
14006 Instructions must not access executable sections. This option can be
14007 useful on targets that are configured to have a dual instruction/data
14008 SRAM interface but that (unlike the M4K) do not automatically redirect
14009 PC-relative loads to the instruction RAM.
14012 @item -msplit-addresses
14013 @itemx -mno-split-addresses
14014 @opindex msplit-addresses
14015 @opindex mno-split-addresses
14016 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
14017 relocation operators. This option has been superseded by
14018 @option{-mexplicit-relocs} but is retained for backwards compatibility.
14020 @item -mexplicit-relocs
14021 @itemx -mno-explicit-relocs
14022 @opindex mexplicit-relocs
14023 @opindex mno-explicit-relocs
14024 Use (do not use) assembler relocation operators when dealing with symbolic
14025 addresses. The alternative, selected by @option{-mno-explicit-relocs},
14026 is to use assembler macros instead.
14028 @option{-mexplicit-relocs} is the default if GCC was configured
14029 to use an assembler that supports relocation operators.
14031 @item -mcheck-zero-division
14032 @itemx -mno-check-zero-division
14033 @opindex mcheck-zero-division
14034 @opindex mno-check-zero-division
14035 Trap (do not trap) on integer division by zero.
14037 The default is @option{-mcheck-zero-division}.
14039 @item -mdivide-traps
14040 @itemx -mdivide-breaks
14041 @opindex mdivide-traps
14042 @opindex mdivide-breaks
14043 MIPS systems check for division by zero by generating either a
14044 conditional trap or a break instruction. Using traps results in
14045 smaller code, but is only supported on MIPS II and later. Also, some
14046 versions of the Linux kernel have a bug that prevents trap from
14047 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
14048 allow conditional traps on architectures that support them and
14049 @option{-mdivide-breaks} to force the use of breaks.
14051 The default is usually @option{-mdivide-traps}, but this can be
14052 overridden at configure time using @option{--with-divide=breaks}.
14053 Divide-by-zero checks can be completely disabled using
14054 @option{-mno-check-zero-division}.
14059 @opindex mno-memcpy
14060 Force (do not force) the use of @code{memcpy()} for non-trivial block
14061 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
14062 most constant-sized copies.
14065 @itemx -mno-long-calls
14066 @opindex mlong-calls
14067 @opindex mno-long-calls
14068 Disable (do not disable) use of the @code{jal} instruction. Calling
14069 functions using @code{jal} is more efficient but requires the caller
14070 and callee to be in the same 256 megabyte segment.
14072 This option has no effect on abicalls code. The default is
14073 @option{-mno-long-calls}.
14079 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
14080 instructions, as provided by the R4650 ISA@.
14083 @itemx -mno-fused-madd
14084 @opindex mfused-madd
14085 @opindex mno-fused-madd
14086 Enable (disable) use of the floating point multiply-accumulate
14087 instructions, when they are available. The default is
14088 @option{-mfused-madd}.
14090 When multiply-accumulate instructions are used, the intermediate
14091 product is calculated to infinite precision and is not subject to
14092 the FCSR Flush to Zero bit. This may be undesirable in some
14097 Tell the MIPS assembler to not run its preprocessor over user
14098 assembler files (with a @samp{.s} suffix) when assembling them.
14101 @itemx -mno-fix-r4000
14102 @opindex mfix-r4000
14103 @opindex mno-fix-r4000
14104 Work around certain R4000 CPU errata:
14107 A double-word or a variable shift may give an incorrect result if executed
14108 immediately after starting an integer division.
14110 A double-word or a variable shift may give an incorrect result if executed
14111 while an integer multiplication is in progress.
14113 An integer division may give an incorrect result if started in a delay slot
14114 of a taken branch or a jump.
14118 @itemx -mno-fix-r4400
14119 @opindex mfix-r4400
14120 @opindex mno-fix-r4400
14121 Work around certain R4400 CPU errata:
14124 A double-word or a variable shift may give an incorrect result if executed
14125 immediately after starting an integer division.
14129 @itemx -mno-fix-r10000
14130 @opindex mfix-r10000
14131 @opindex mno-fix-r10000
14132 Work around certain R10000 errata:
14135 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14136 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14139 This option can only be used if the target architecture supports
14140 branch-likely instructions. @option{-mfix-r10000} is the default when
14141 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14145 @itemx -mno-fix-vr4120
14146 @opindex mfix-vr4120
14147 Work around certain VR4120 errata:
14150 @code{dmultu} does not always produce the correct result.
14152 @code{div} and @code{ddiv} do not always produce the correct result if one
14153 of the operands is negative.
14155 The workarounds for the division errata rely on special functions in
14156 @file{libgcc.a}. At present, these functions are only provided by
14157 the @code{mips64vr*-elf} configurations.
14159 Other VR4120 errata require a nop to be inserted between certain pairs of
14160 instructions. These errata are handled by the assembler, not by GCC itself.
14163 @opindex mfix-vr4130
14164 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14165 workarounds are implemented by the assembler rather than by GCC,
14166 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14167 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14168 instructions are available instead.
14171 @itemx -mno-fix-sb1
14173 Work around certain SB-1 CPU core errata.
14174 (This flag currently works around the SB-1 revision 2
14175 ``F1'' and ``F2'' floating point errata.)
14177 @item -mr10k-cache-barrier=@var{setting}
14178 @opindex mr10k-cache-barrier
14179 Specify whether GCC should insert cache barriers to avoid the
14180 side-effects of speculation on R10K processors.
14182 In common with many processors, the R10K tries to predict the outcome
14183 of a conditional branch and speculatively executes instructions from
14184 the ``taken'' branch. It later aborts these instructions if the
14185 predicted outcome was wrong. However, on the R10K, even aborted
14186 instructions can have side effects.
14188 This problem only affects kernel stores and, depending on the system,
14189 kernel loads. As an example, a speculatively-executed store may load
14190 the target memory into cache and mark the cache line as dirty, even if
14191 the store itself is later aborted. If a DMA operation writes to the
14192 same area of memory before the ``dirty'' line is flushed, the cached
14193 data will overwrite the DMA-ed data. See the R10K processor manual
14194 for a full description, including other potential problems.
14196 One workaround is to insert cache barrier instructions before every memory
14197 access that might be speculatively executed and that might have side
14198 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14199 controls GCC's implementation of this workaround. It assumes that
14200 aborted accesses to any byte in the following regions will not have
14205 the memory occupied by the current function's stack frame;
14208 the memory occupied by an incoming stack argument;
14211 the memory occupied by an object with a link-time-constant address.
14214 It is the kernel's responsibility to ensure that speculative
14215 accesses to these regions are indeed safe.
14217 If the input program contains a function declaration such as:
14223 then the implementation of @code{foo} must allow @code{j foo} and
14224 @code{jal foo} to be executed speculatively. GCC honors this
14225 restriction for functions it compiles itself. It expects non-GCC
14226 functions (such as hand-written assembly code) to do the same.
14228 The option has three forms:
14231 @item -mr10k-cache-barrier=load-store
14232 Insert a cache barrier before a load or store that might be
14233 speculatively executed and that might have side effects even
14236 @item -mr10k-cache-barrier=store
14237 Insert a cache barrier before a store that might be speculatively
14238 executed and that might have side effects even if aborted.
14240 @item -mr10k-cache-barrier=none
14241 Disable the insertion of cache barriers. This is the default setting.
14244 @item -mflush-func=@var{func}
14245 @itemx -mno-flush-func
14246 @opindex mflush-func
14247 Specifies the function to call to flush the I and D caches, or to not
14248 call any such function. If called, the function must take the same
14249 arguments as the common @code{_flush_func()}, that is, the address of the
14250 memory range for which the cache is being flushed, the size of the
14251 memory range, and the number 3 (to flush both caches). The default
14252 depends on the target GCC was configured for, but commonly is either
14253 @samp{_flush_func} or @samp{__cpu_flush}.
14255 @item mbranch-cost=@var{num}
14256 @opindex mbranch-cost
14257 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14258 This cost is only a heuristic and is not guaranteed to produce
14259 consistent results across releases. A zero cost redundantly selects
14260 the default, which is based on the @option{-mtune} setting.
14262 @item -mbranch-likely
14263 @itemx -mno-branch-likely
14264 @opindex mbranch-likely
14265 @opindex mno-branch-likely
14266 Enable or disable use of Branch Likely instructions, regardless of the
14267 default for the selected architecture. By default, Branch Likely
14268 instructions may be generated if they are supported by the selected
14269 architecture. An exception is for the MIPS32 and MIPS64 architectures
14270 and processors which implement those architectures; for those, Branch
14271 Likely instructions will not be generated by default because the MIPS32
14272 and MIPS64 architectures specifically deprecate their use.
14274 @item -mfp-exceptions
14275 @itemx -mno-fp-exceptions
14276 @opindex mfp-exceptions
14277 Specifies whether FP exceptions are enabled. This affects how we schedule
14278 FP instructions for some processors. The default is that FP exceptions are
14281 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14282 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14285 @item -mvr4130-align
14286 @itemx -mno-vr4130-align
14287 @opindex mvr4130-align
14288 The VR4130 pipeline is two-way superscalar, but can only issue two
14289 instructions together if the first one is 8-byte aligned. When this
14290 option is enabled, GCC will align pairs of instructions that it
14291 thinks should execute in parallel.
14293 This option only has an effect when optimizing for the VR4130.
14294 It normally makes code faster, but at the expense of making it bigger.
14295 It is enabled by default at optimization level @option{-O3}.
14300 Enable (disable) generation of @code{synci} instructions on
14301 architectures that support it. The @code{synci} instructions (if
14302 enabled) will be generated when @code{__builtin___clear_cache()} is
14305 This option defaults to @code{-mno-synci}, but the default can be
14306 overridden by configuring with @code{--with-synci}.
14308 When compiling code for single processor systems, it is generally safe
14309 to use @code{synci}. However, on many multi-core (SMP) systems, it
14310 will not invalidate the instruction caches on all cores and may lead
14311 to undefined behavior.
14313 @item -mrelax-pic-calls
14314 @itemx -mno-relax-pic-calls
14315 @opindex mrelax-pic-calls
14316 Try to turn PIC calls that are normally dispatched via register
14317 @code{$25} into direct calls. This is only possible if the linker can
14318 resolve the destination at link-time and if the destination is within
14319 range for a direct call.
14321 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14322 an assembler and a linker that supports the @code{.reloc} assembly
14323 directive and @code{-mexplicit-relocs} is in effect. With
14324 @code{-mno-explicit-relocs}, this optimization can be performed by the
14325 assembler and the linker alone without help from the compiler.
14327 @item -mmcount-ra-address
14328 @itemx -mno-mcount-ra-address
14329 @opindex mmcount-ra-address
14330 @opindex mno-mcount-ra-address
14331 Emit (do not emit) code that allows @code{_mcount} to modify the
14332 calling function's return address. When enabled, this option extends
14333 the usual @code{_mcount} interface with a new @var{ra-address}
14334 parameter, which has type @code{intptr_t *} and is passed in register
14335 @code{$12}. @code{_mcount} can then modify the return address by
14336 doing both of the following:
14339 Returning the new address in register @code{$31}.
14341 Storing the new address in @code{*@var{ra-address}},
14342 if @var{ra-address} is nonnull.
14345 The default is @option{-mno-mcount-ra-address}.
14350 @subsection MMIX Options
14351 @cindex MMIX Options
14353 These options are defined for the MMIX:
14357 @itemx -mno-libfuncs
14359 @opindex mno-libfuncs
14360 Specify that intrinsic library functions are being compiled, passing all
14361 values in registers, no matter the size.
14364 @itemx -mno-epsilon
14366 @opindex mno-epsilon
14367 Generate floating-point comparison instructions that compare with respect
14368 to the @code{rE} epsilon register.
14370 @item -mabi=mmixware
14372 @opindex mabi=mmixware
14374 Generate code that passes function parameters and return values that (in
14375 the called function) are seen as registers @code{$0} and up, as opposed to
14376 the GNU ABI which uses global registers @code{$231} and up.
14378 @item -mzero-extend
14379 @itemx -mno-zero-extend
14380 @opindex mzero-extend
14381 @opindex mno-zero-extend
14382 When reading data from memory in sizes shorter than 64 bits, use (do not
14383 use) zero-extending load instructions by default, rather than
14384 sign-extending ones.
14387 @itemx -mno-knuthdiv
14389 @opindex mno-knuthdiv
14390 Make the result of a division yielding a remainder have the same sign as
14391 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14392 remainder follows the sign of the dividend. Both methods are
14393 arithmetically valid, the latter being almost exclusively used.
14395 @item -mtoplevel-symbols
14396 @itemx -mno-toplevel-symbols
14397 @opindex mtoplevel-symbols
14398 @opindex mno-toplevel-symbols
14399 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14400 code can be used with the @code{PREFIX} assembly directive.
14404 Generate an executable in the ELF format, rather than the default
14405 @samp{mmo} format used by the @command{mmix} simulator.
14407 @item -mbranch-predict
14408 @itemx -mno-branch-predict
14409 @opindex mbranch-predict
14410 @opindex mno-branch-predict
14411 Use (do not use) the probable-branch instructions, when static branch
14412 prediction indicates a probable branch.
14414 @item -mbase-addresses
14415 @itemx -mno-base-addresses
14416 @opindex mbase-addresses
14417 @opindex mno-base-addresses
14418 Generate (do not generate) code that uses @emph{base addresses}. Using a
14419 base address automatically generates a request (handled by the assembler
14420 and the linker) for a constant to be set up in a global register. The
14421 register is used for one or more base address requests within the range 0
14422 to 255 from the value held in the register. The generally leads to short
14423 and fast code, but the number of different data items that can be
14424 addressed is limited. This means that a program that uses lots of static
14425 data may require @option{-mno-base-addresses}.
14427 @item -msingle-exit
14428 @itemx -mno-single-exit
14429 @opindex msingle-exit
14430 @opindex mno-single-exit
14431 Force (do not force) generated code to have a single exit point in each
14435 @node MN10300 Options
14436 @subsection MN10300 Options
14437 @cindex MN10300 options
14439 These @option{-m} options are defined for Matsushita MN10300 architectures:
14444 Generate code to avoid bugs in the multiply instructions for the MN10300
14445 processors. This is the default.
14447 @item -mno-mult-bug
14448 @opindex mno-mult-bug
14449 Do not generate code to avoid bugs in the multiply instructions for the
14450 MN10300 processors.
14454 Generate code which uses features specific to the AM33 processor.
14458 Do not generate code which uses features specific to the AM33 processor. This
14461 @item -mreturn-pointer-on-d0
14462 @opindex mreturn-pointer-on-d0
14463 When generating a function which returns a pointer, return the pointer
14464 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14465 only in a0, and attempts to call such functions without a prototype
14466 would result in errors. Note that this option is on by default; use
14467 @option{-mno-return-pointer-on-d0} to disable it.
14471 Do not link in the C run-time initialization object file.
14475 Indicate to the linker that it should perform a relaxation optimization pass
14476 to shorten branches, calls and absolute memory addresses. This option only
14477 has an effect when used on the command line for the final link step.
14479 This option makes symbolic debugging impossible.
14482 @node PDP-11 Options
14483 @subsection PDP-11 Options
14484 @cindex PDP-11 Options
14486 These options are defined for the PDP-11:
14491 Use hardware FPP floating point. This is the default. (FIS floating
14492 point on the PDP-11/40 is not supported.)
14495 @opindex msoft-float
14496 Do not use hardware floating point.
14500 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14504 Return floating-point results in memory. This is the default.
14508 Generate code for a PDP-11/40.
14512 Generate code for a PDP-11/45. This is the default.
14516 Generate code for a PDP-11/10.
14518 @item -mbcopy-builtin
14519 @opindex mbcopy-builtin
14520 Use inline @code{movmemhi} patterns for copying memory. This is the
14525 Do not use inline @code{movmemhi} patterns for copying memory.
14531 Use 16-bit @code{int}. This is the default.
14537 Use 32-bit @code{int}.
14540 @itemx -mno-float32
14542 @opindex mno-float32
14543 Use 64-bit @code{float}. This is the default.
14546 @itemx -mno-float64
14548 @opindex mno-float64
14549 Use 32-bit @code{float}.
14553 Use @code{abshi2} pattern. This is the default.
14557 Do not use @code{abshi2} pattern.
14559 @item -mbranch-expensive
14560 @opindex mbranch-expensive
14561 Pretend that branches are expensive. This is for experimenting with
14562 code generation only.
14564 @item -mbranch-cheap
14565 @opindex mbranch-cheap
14566 Do not pretend that branches are expensive. This is the default.
14570 Generate code for a system with split I&D@.
14574 Generate code for a system without split I&D@. This is the default.
14578 Use Unix assembler syntax. This is the default when configured for
14579 @samp{pdp11-*-bsd}.
14583 Use DEC assembler syntax. This is the default when configured for any
14584 PDP-11 target other than @samp{pdp11-*-bsd}.
14587 @node picoChip Options
14588 @subsection picoChip Options
14589 @cindex picoChip options
14591 These @samp{-m} options are defined for picoChip implementations:
14595 @item -mae=@var{ae_type}
14597 Set the instruction set, register set, and instruction scheduling
14598 parameters for array element type @var{ae_type}. Supported values
14599 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14601 @option{-mae=ANY} selects a completely generic AE type. Code
14602 generated with this option will run on any of the other AE types. The
14603 code will not be as efficient as it would be if compiled for a specific
14604 AE type, and some types of operation (e.g., multiplication) will not
14605 work properly on all types of AE.
14607 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14608 for compiled code, and is the default.
14610 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14611 option may suffer from poor performance of byte (char) manipulation,
14612 since the DSP AE does not provide hardware support for byte load/stores.
14614 @item -msymbol-as-address
14615 Enable the compiler to directly use a symbol name as an address in a
14616 load/store instruction, without first loading it into a
14617 register. Typically, the use of this option will generate larger
14618 programs, which run faster than when the option isn't used. However, the
14619 results vary from program to program, so it is left as a user option,
14620 rather than being permanently enabled.
14622 @item -mno-inefficient-warnings
14623 Disables warnings about the generation of inefficient code. These
14624 warnings can be generated, for example, when compiling code which
14625 performs byte-level memory operations on the MAC AE type. The MAC AE has
14626 no hardware support for byte-level memory operations, so all byte
14627 load/stores must be synthesized from word load/store operations. This is
14628 inefficient and a warning will be generated indicating to the programmer
14629 that they should rewrite the code to avoid byte operations, or to target
14630 an AE type which has the necessary hardware support. This option enables
14631 the warning to be turned off.
14635 @node PowerPC Options
14636 @subsection PowerPC Options
14637 @cindex PowerPC options
14639 These are listed under @xref{RS/6000 and PowerPC Options}.
14641 @node RS/6000 and PowerPC Options
14642 @subsection IBM RS/6000 and PowerPC Options
14643 @cindex RS/6000 and PowerPC Options
14644 @cindex IBM RS/6000 and PowerPC Options
14646 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14653 @itemx -mno-powerpc
14654 @itemx -mpowerpc-gpopt
14655 @itemx -mno-powerpc-gpopt
14656 @itemx -mpowerpc-gfxopt
14657 @itemx -mno-powerpc-gfxopt
14659 @itemx -mno-powerpc64
14663 @itemx -mno-popcntb
14665 @itemx -mno-popcntd
14673 @itemx -mno-hard-dfp
14677 @opindex mno-power2
14679 @opindex mno-powerpc
14680 @opindex mpowerpc-gpopt
14681 @opindex mno-powerpc-gpopt
14682 @opindex mpowerpc-gfxopt
14683 @opindex mno-powerpc-gfxopt
14684 @opindex mpowerpc64
14685 @opindex mno-powerpc64
14689 @opindex mno-popcntb
14691 @opindex mno-popcntd
14697 @opindex mno-mfpgpr
14699 @opindex mno-hard-dfp
14700 GCC supports two related instruction set architectures for the
14701 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14702 instructions supported by the @samp{rios} chip set used in the original
14703 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14704 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14705 the IBM 4xx, 6xx, and follow-on microprocessors.
14707 Neither architecture is a subset of the other. However there is a
14708 large common subset of instructions supported by both. An MQ
14709 register is included in processors supporting the POWER architecture.
14711 You use these options to specify which instructions are available on the
14712 processor you are using. The default value of these options is
14713 determined when configuring GCC@. Specifying the
14714 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14715 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14716 rather than the options listed above.
14718 The @option{-mpower} option allows GCC to generate instructions that
14719 are found only in the POWER architecture and to use the MQ register.
14720 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14721 to generate instructions that are present in the POWER2 architecture but
14722 not the original POWER architecture.
14724 The @option{-mpowerpc} option allows GCC to generate instructions that
14725 are found only in the 32-bit subset of the PowerPC architecture.
14726 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14727 GCC to use the optional PowerPC architecture instructions in the
14728 General Purpose group, including floating-point square root. Specifying
14729 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14730 use the optional PowerPC architecture instructions in the Graphics
14731 group, including floating-point select.
14733 The @option{-mmfcrf} option allows GCC to generate the move from
14734 condition register field instruction implemented on the POWER4
14735 processor and other processors that support the PowerPC V2.01
14737 The @option{-mpopcntb} option allows GCC to generate the popcount and
14738 double precision FP reciprocal estimate instruction implemented on the
14739 POWER5 processor and other processors that support the PowerPC V2.02
14741 The @option{-mpopcntd} option allows GCC to generate the popcount
14742 instruction implemented on the POWER7 processor and other processors
14743 that support the PowerPC V2.06 architecture.
14744 The @option{-mfprnd} option allows GCC to generate the FP round to
14745 integer instructions implemented on the POWER5+ processor and other
14746 processors that support the PowerPC V2.03 architecture.
14747 The @option{-mcmpb} option allows GCC to generate the compare bytes
14748 instruction implemented on the POWER6 processor and other processors
14749 that support the PowerPC V2.05 architecture.
14750 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14751 general purpose register instructions implemented on the POWER6X
14752 processor and other processors that support the extended PowerPC V2.05
14754 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14755 point instructions implemented on some POWER processors.
14757 The @option{-mpowerpc64} option allows GCC to generate the additional
14758 64-bit instructions that are found in the full PowerPC64 architecture
14759 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14760 @option{-mno-powerpc64}.
14762 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14763 will use only the instructions in the common subset of both
14764 architectures plus some special AIX common-mode calls, and will not use
14765 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14766 permits GCC to use any instruction from either architecture and to
14767 allow use of the MQ register; specify this for the Motorola MPC601.
14769 @item -mnew-mnemonics
14770 @itemx -mold-mnemonics
14771 @opindex mnew-mnemonics
14772 @opindex mold-mnemonics
14773 Select which mnemonics to use in the generated assembler code. With
14774 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14775 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14776 assembler mnemonics defined for the POWER architecture. Instructions
14777 defined in only one architecture have only one mnemonic; GCC uses that
14778 mnemonic irrespective of which of these options is specified.
14780 GCC defaults to the mnemonics appropriate for the architecture in
14781 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14782 value of these option. Unless you are building a cross-compiler, you
14783 should normally not specify either @option{-mnew-mnemonics} or
14784 @option{-mold-mnemonics}, but should instead accept the default.
14786 @item -mcpu=@var{cpu_type}
14788 Set architecture type, register usage, choice of mnemonics, and
14789 instruction scheduling parameters for machine type @var{cpu_type}.
14790 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14791 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14792 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
14793 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
14794 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14795 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
14796 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
14797 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
14798 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
14799 @samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14800 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14802 @option{-mcpu=common} selects a completely generic processor. Code
14803 generated under this option will run on any POWER or PowerPC processor.
14804 GCC will use only the instructions in the common subset of both
14805 architectures, and will not use the MQ register. GCC assumes a generic
14806 processor model for scheduling purposes.
14808 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14809 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14810 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14811 types, with an appropriate, generic processor model assumed for
14812 scheduling purposes.
14814 The other options specify a specific processor. Code generated under
14815 those options will run best on that processor, and may not run at all on
14818 The @option{-mcpu} options automatically enable or disable the
14821 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
14822 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
14823 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
14824 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
14826 The particular options set for any particular CPU will vary between
14827 compiler versions, depending on what setting seems to produce optimal
14828 code for that CPU; it doesn't necessarily reflect the actual hardware's
14829 capabilities. If you wish to set an individual option to a particular
14830 value, you may specify it after the @option{-mcpu} option, like
14831 @samp{-mcpu=970 -mno-altivec}.
14833 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
14834 not enabled or disabled by the @option{-mcpu} option at present because
14835 AIX does not have full support for these options. You may still
14836 enable or disable them individually if you're sure it'll work in your
14839 @item -mtune=@var{cpu_type}
14841 Set the instruction scheduling parameters for machine type
14842 @var{cpu_type}, but do not set the architecture type, register usage, or
14843 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
14844 values for @var{cpu_type} are used for @option{-mtune} as for
14845 @option{-mcpu}. If both are specified, the code generated will use the
14846 architecture, registers, and mnemonics set by @option{-mcpu}, but the
14847 scheduling parameters set by @option{-mtune}.
14853 Generate code to compute division as reciprocal estimate and iterative
14854 refinement, creating opportunities for increased throughput. This
14855 feature requires: optional PowerPC Graphics instruction set for single
14856 precision and FRE instruction for double precision, assuming divides
14857 cannot generate user-visible traps, and the domain values not include
14858 Infinities, denormals or zero denominator.
14861 @itemx -mno-altivec
14863 @opindex mno-altivec
14864 Generate code that uses (does not use) AltiVec instructions, and also
14865 enable the use of built-in functions that allow more direct access to
14866 the AltiVec instruction set. You may also need to set
14867 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
14873 @opindex mno-vrsave
14874 Generate VRSAVE instructions when generating AltiVec code.
14876 @item -mgen-cell-microcode
14877 @opindex mgen-cell-microcode
14878 Generate Cell microcode instructions
14880 @item -mwarn-cell-microcode
14881 @opindex mwarn-cell-microcode
14882 Warning when a Cell microcode instruction is going to emitted. An example
14883 of a Cell microcode instruction is a variable shift.
14886 @opindex msecure-plt
14887 Generate code that allows ld and ld.so to build executables and shared
14888 libraries with non-exec .plt and .got sections. This is a PowerPC
14889 32-bit SYSV ABI option.
14893 Generate code that uses a BSS .plt section that ld.so fills in, and
14894 requires .plt and .got sections that are both writable and executable.
14895 This is a PowerPC 32-bit SYSV ABI option.
14901 This switch enables or disables the generation of ISEL instructions.
14903 @item -misel=@var{yes/no}
14904 This switch has been deprecated. Use @option{-misel} and
14905 @option{-mno-isel} instead.
14911 This switch enables or disables the generation of SPE simd
14917 @opindex mno-paired
14918 This switch enables or disables the generation of PAIRED simd
14921 @item -mspe=@var{yes/no}
14922 This option has been deprecated. Use @option{-mspe} and
14923 @option{-mno-spe} instead.
14929 Generate code that uses (does not use) vector/scalar (VSX)
14930 instructions, and also enable the use of built-in functions that allow
14931 more direct access to the VSX instruction set.
14933 @item -mfloat-gprs=@var{yes/single/double/no}
14934 @itemx -mfloat-gprs
14935 @opindex mfloat-gprs
14936 This switch enables or disables the generation of floating point
14937 operations on the general purpose registers for architectures that
14940 The argument @var{yes} or @var{single} enables the use of
14941 single-precision floating point operations.
14943 The argument @var{double} enables the use of single and
14944 double-precision floating point operations.
14946 The argument @var{no} disables floating point operations on the
14947 general purpose registers.
14949 This option is currently only available on the MPC854x.
14955 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
14956 targets (including GNU/Linux). The 32-bit environment sets int, long
14957 and pointer to 32 bits and generates code that runs on any PowerPC
14958 variant. The 64-bit environment sets int to 32 bits and long and
14959 pointer to 64 bits, and generates code for PowerPC64, as for
14960 @option{-mpowerpc64}.
14963 @itemx -mno-fp-in-toc
14964 @itemx -mno-sum-in-toc
14965 @itemx -mminimal-toc
14967 @opindex mno-fp-in-toc
14968 @opindex mno-sum-in-toc
14969 @opindex mminimal-toc
14970 Modify generation of the TOC (Table Of Contents), which is created for
14971 every executable file. The @option{-mfull-toc} option is selected by
14972 default. In that case, GCC will allocate at least one TOC entry for
14973 each unique non-automatic variable reference in your program. GCC
14974 will also place floating-point constants in the TOC@. However, only
14975 16,384 entries are available in the TOC@.
14977 If you receive a linker error message that saying you have overflowed
14978 the available TOC space, you can reduce the amount of TOC space used
14979 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
14980 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
14981 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
14982 generate code to calculate the sum of an address and a constant at
14983 run-time instead of putting that sum into the TOC@. You may specify one
14984 or both of these options. Each causes GCC to produce very slightly
14985 slower and larger code at the expense of conserving TOC space.
14987 If you still run out of space in the TOC even when you specify both of
14988 these options, specify @option{-mminimal-toc} instead. This option causes
14989 GCC to make only one TOC entry for every file. When you specify this
14990 option, GCC will produce code that is slower and larger but which
14991 uses extremely little TOC space. You may wish to use this option
14992 only on files that contain less frequently executed code.
14998 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
14999 @code{long} type, and the infrastructure needed to support them.
15000 Specifying @option{-maix64} implies @option{-mpowerpc64} and
15001 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
15002 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
15005 @itemx -mno-xl-compat
15006 @opindex mxl-compat
15007 @opindex mno-xl-compat
15008 Produce code that conforms more closely to IBM XL compiler semantics
15009 when using AIX-compatible ABI@. Pass floating-point arguments to
15010 prototyped functions beyond the register save area (RSA) on the stack
15011 in addition to argument FPRs. Do not assume that most significant
15012 double in 128-bit long double value is properly rounded when comparing
15013 values and converting to double. Use XL symbol names for long double
15016 The AIX calling convention was extended but not initially documented to
15017 handle an obscure K&R C case of calling a function that takes the
15018 address of its arguments with fewer arguments than declared. IBM XL
15019 compilers access floating point arguments which do not fit in the
15020 RSA from the stack when a subroutine is compiled without
15021 optimization. Because always storing floating-point arguments on the
15022 stack is inefficient and rarely needed, this option is not enabled by
15023 default and only is necessary when calling subroutines compiled by IBM
15024 XL compilers without optimization.
15028 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
15029 application written to use message passing with special startup code to
15030 enable the application to run. The system must have PE installed in the
15031 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
15032 must be overridden with the @option{-specs=} option to specify the
15033 appropriate directory location. The Parallel Environment does not
15034 support threads, so the @option{-mpe} option and the @option{-pthread}
15035 option are incompatible.
15037 @item -malign-natural
15038 @itemx -malign-power
15039 @opindex malign-natural
15040 @opindex malign-power
15041 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
15042 @option{-malign-natural} overrides the ABI-defined alignment of larger
15043 types, such as floating-point doubles, on their natural size-based boundary.
15044 The option @option{-malign-power} instructs GCC to follow the ABI-specified
15045 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
15047 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
15051 @itemx -mhard-float
15052 @opindex msoft-float
15053 @opindex mhard-float
15054 Generate code that does not use (uses) the floating-point register set.
15055 Software floating point emulation is provided if you use the
15056 @option{-msoft-float} option, and pass the option to GCC when linking.
15058 @item -msingle-float
15059 @itemx -mdouble-float
15060 @opindex msingle-float
15061 @opindex mdouble-float
15062 Generate code for single or double-precision floating point operations.
15063 @option{-mdouble-float} implies @option{-msingle-float}.
15066 @opindex msimple-fpu
15067 Do not generate sqrt and div instructions for hardware floating point unit.
15071 Specify type of floating point unit. Valid values are @var{sp_lite}
15072 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
15073 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
15074 and @var{dp_full} (equivalent to -mdouble-float).
15077 @opindex mxilinx-fpu
15078 Perform optimizations for floating point unit on Xilinx PPC 405/440.
15081 @itemx -mno-multiple
15083 @opindex mno-multiple
15084 Generate code that uses (does not use) the load multiple word
15085 instructions and the store multiple word instructions. These
15086 instructions are generated by default on POWER systems, and not
15087 generated on PowerPC systems. Do not use @option{-mmultiple} on little
15088 endian PowerPC systems, since those instructions do not work when the
15089 processor is in little endian mode. The exceptions are PPC740 and
15090 PPC750 which permit the instructions usage in little endian mode.
15095 @opindex mno-string
15096 Generate code that uses (does not use) the load string instructions
15097 and the store string word instructions to save multiple registers and
15098 do small block moves. These instructions are generated by default on
15099 POWER systems, and not generated on PowerPC systems. Do not use
15100 @option{-mstring} on little endian PowerPC systems, since those
15101 instructions do not work when the processor is in little endian mode.
15102 The exceptions are PPC740 and PPC750 which permit the instructions
15103 usage in little endian mode.
15108 @opindex mno-update
15109 Generate code that uses (does not use) the load or store instructions
15110 that update the base register to the address of the calculated memory
15111 location. These instructions are generated by default. If you use
15112 @option{-mno-update}, there is a small window between the time that the
15113 stack pointer is updated and the address of the previous frame is
15114 stored, which means code that walks the stack frame across interrupts or
15115 signals may get corrupted data.
15117 @item -mavoid-indexed-addresses
15118 @itemx -mno-avoid-indexed-addresses
15119 @opindex mavoid-indexed-addresses
15120 @opindex mno-avoid-indexed-addresses
15121 Generate code that tries to avoid (not avoid) the use of indexed load
15122 or store instructions. These instructions can incur a performance
15123 penalty on Power6 processors in certain situations, such as when
15124 stepping through large arrays that cross a 16M boundary. This option
15125 is enabled by default when targetting Power6 and disabled otherwise.
15128 @itemx -mno-fused-madd
15129 @opindex mfused-madd
15130 @opindex mno-fused-madd
15131 Generate code that uses (does not use) the floating point multiply and
15132 accumulate instructions. These instructions are generated by default if
15133 hardware floating is used.
15139 Generate code that uses (does not use) the half-word multiply and
15140 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15141 These instructions are generated by default when targetting those
15148 Generate code that uses (does not use) the string-search @samp{dlmzb}
15149 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15150 generated by default when targetting those processors.
15152 @item -mno-bit-align
15154 @opindex mno-bit-align
15155 @opindex mbit-align
15156 On System V.4 and embedded PowerPC systems do not (do) force structures
15157 and unions that contain bit-fields to be aligned to the base type of the
15160 For example, by default a structure containing nothing but 8
15161 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15162 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15163 the structure would be aligned to a 1 byte boundary and be one byte in
15166 @item -mno-strict-align
15167 @itemx -mstrict-align
15168 @opindex mno-strict-align
15169 @opindex mstrict-align
15170 On System V.4 and embedded PowerPC systems do not (do) assume that
15171 unaligned memory references will be handled by the system.
15173 @item -mrelocatable
15174 @itemx -mno-relocatable
15175 @opindex mrelocatable
15176 @opindex mno-relocatable
15177 On embedded PowerPC systems generate code that allows (does not allow)
15178 the program to be relocated to a different address at runtime. If you
15179 use @option{-mrelocatable} on any module, all objects linked together must
15180 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
15182 @item -mrelocatable-lib
15183 @itemx -mno-relocatable-lib
15184 @opindex mrelocatable-lib
15185 @opindex mno-relocatable-lib
15186 On embedded PowerPC systems generate code that allows (does not allow)
15187 the program to be relocated to a different address at runtime. Modules
15188 compiled with @option{-mrelocatable-lib} can be linked with either modules
15189 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
15190 with modules compiled with the @option{-mrelocatable} options.
15196 On System V.4 and embedded PowerPC systems do not (do) assume that
15197 register 2 contains a pointer to a global area pointing to the addresses
15198 used in the program.
15201 @itemx -mlittle-endian
15203 @opindex mlittle-endian
15204 On System V.4 and embedded PowerPC systems compile code for the
15205 processor in little endian mode. The @option{-mlittle-endian} option is
15206 the same as @option{-mlittle}.
15209 @itemx -mbig-endian
15211 @opindex mbig-endian
15212 On System V.4 and embedded PowerPC systems compile code for the
15213 processor in big endian mode. The @option{-mbig-endian} option is
15214 the same as @option{-mbig}.
15216 @item -mdynamic-no-pic
15217 @opindex mdynamic-no-pic
15218 On Darwin and Mac OS X systems, compile code so that it is not
15219 relocatable, but that its external references are relocatable. The
15220 resulting code is suitable for applications, but not shared
15223 @item -mprioritize-restricted-insns=@var{priority}
15224 @opindex mprioritize-restricted-insns
15225 This option controls the priority that is assigned to
15226 dispatch-slot restricted instructions during the second scheduling
15227 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15228 @var{no/highest/second-highest} priority to dispatch slot restricted
15231 @item -msched-costly-dep=@var{dependence_type}
15232 @opindex msched-costly-dep
15233 This option controls which dependences are considered costly
15234 by the target during instruction scheduling. The argument
15235 @var{dependence_type} takes one of the following values:
15236 @var{no}: no dependence is costly,
15237 @var{all}: all dependences are costly,
15238 @var{true_store_to_load}: a true dependence from store to load is costly,
15239 @var{store_to_load}: any dependence from store to load is costly,
15240 @var{number}: any dependence which latency >= @var{number} is costly.
15242 @item -minsert-sched-nops=@var{scheme}
15243 @opindex minsert-sched-nops
15244 This option controls which nop insertion scheme will be used during
15245 the second scheduling pass. The argument @var{scheme} takes one of the
15247 @var{no}: Don't insert nops.
15248 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15249 according to the scheduler's grouping.
15250 @var{regroup_exact}: Insert nops to force costly dependent insns into
15251 separate groups. Insert exactly as many nops as needed to force an insn
15252 to a new group, according to the estimated processor grouping.
15253 @var{number}: Insert nops to force costly dependent insns into
15254 separate groups. Insert @var{number} nops to force an insn to a new group.
15257 @opindex mcall-sysv
15258 On System V.4 and embedded PowerPC systems compile code using calling
15259 conventions that adheres to the March 1995 draft of the System V
15260 Application Binary Interface, PowerPC processor supplement. This is the
15261 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15263 @item -mcall-sysv-eabi
15265 @opindex mcall-sysv-eabi
15266 @opindex mcall-eabi
15267 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15269 @item -mcall-sysv-noeabi
15270 @opindex mcall-sysv-noeabi
15271 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15273 @item -mcall-aixdesc
15275 On System V.4 and embedded PowerPC systems compile code for the AIX
15279 @opindex mcall-linux
15280 On System V.4 and embedded PowerPC systems compile code for the
15281 Linux-based GNU system.
15285 On System V.4 and embedded PowerPC systems compile code for the
15286 Hurd-based GNU system.
15288 @item -mcall-freebsd
15289 @opindex mcall-freebsd
15290 On System V.4 and embedded PowerPC systems compile code for the
15291 FreeBSD operating system.
15293 @item -mcall-netbsd
15294 @opindex mcall-netbsd
15295 On System V.4 and embedded PowerPC systems compile code for the
15296 NetBSD operating system.
15298 @item -mcall-openbsd
15299 @opindex mcall-netbsd
15300 On System V.4 and embedded PowerPC systems compile code for the
15301 OpenBSD operating system.
15303 @item -maix-struct-return
15304 @opindex maix-struct-return
15305 Return all structures in memory (as specified by the AIX ABI)@.
15307 @item -msvr4-struct-return
15308 @opindex msvr4-struct-return
15309 Return structures smaller than 8 bytes in registers (as specified by the
15312 @item -mabi=@var{abi-type}
15314 Extend the current ABI with a particular extension, or remove such extension.
15315 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15316 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15320 Extend the current ABI with SPE ABI extensions. This does not change
15321 the default ABI, instead it adds the SPE ABI extensions to the current
15325 @opindex mabi=no-spe
15326 Disable Booke SPE ABI extensions for the current ABI@.
15328 @item -mabi=ibmlongdouble
15329 @opindex mabi=ibmlongdouble
15330 Change the current ABI to use IBM extended precision long double.
15331 This is a PowerPC 32-bit SYSV ABI option.
15333 @item -mabi=ieeelongdouble
15334 @opindex mabi=ieeelongdouble
15335 Change the current ABI to use IEEE extended precision long double.
15336 This is a PowerPC 32-bit Linux ABI option.
15339 @itemx -mno-prototype
15340 @opindex mprototype
15341 @opindex mno-prototype
15342 On System V.4 and embedded PowerPC systems assume that all calls to
15343 variable argument functions are properly prototyped. Otherwise, the
15344 compiler must insert an instruction before every non prototyped call to
15345 set or clear bit 6 of the condition code register (@var{CR}) to
15346 indicate whether floating point values were passed in the floating point
15347 registers in case the function takes a variable arguments. With
15348 @option{-mprototype}, only calls to prototyped variable argument functions
15349 will set or clear the bit.
15353 On embedded PowerPC systems, assume that the startup module is called
15354 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15355 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15360 On embedded PowerPC systems, assume that the startup module is called
15361 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15366 On embedded PowerPC systems, assume that the startup module is called
15367 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15370 @item -myellowknife
15371 @opindex myellowknife
15372 On embedded PowerPC systems, assume that the startup module is called
15373 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15378 On System V.4 and embedded PowerPC systems, specify that you are
15379 compiling for a VxWorks system.
15383 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15384 header to indicate that @samp{eabi} extended relocations are used.
15390 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15391 Embedded Applications Binary Interface (eabi) which is a set of
15392 modifications to the System V.4 specifications. Selecting @option{-meabi}
15393 means that the stack is aligned to an 8 byte boundary, a function
15394 @code{__eabi} is called to from @code{main} to set up the eabi
15395 environment, and the @option{-msdata} option can use both @code{r2} and
15396 @code{r13} to point to two separate small data areas. Selecting
15397 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15398 do not call an initialization function from @code{main}, and the
15399 @option{-msdata} option will only use @code{r13} to point to a single
15400 small data area. The @option{-meabi} option is on by default if you
15401 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15404 @opindex msdata=eabi
15405 On System V.4 and embedded PowerPC systems, put small initialized
15406 @code{const} global and static data in the @samp{.sdata2} section, which
15407 is pointed to by register @code{r2}. Put small initialized
15408 non-@code{const} global and static data in the @samp{.sdata} section,
15409 which is pointed to by register @code{r13}. Put small uninitialized
15410 global and static data in the @samp{.sbss} section, which is adjacent to
15411 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15412 incompatible with the @option{-mrelocatable} option. The
15413 @option{-msdata=eabi} option also sets the @option{-memb} option.
15416 @opindex msdata=sysv
15417 On System V.4 and embedded PowerPC systems, put small global and static
15418 data in the @samp{.sdata} section, which is pointed to by register
15419 @code{r13}. Put small uninitialized global and static data in the
15420 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15421 The @option{-msdata=sysv} option is incompatible with the
15422 @option{-mrelocatable} option.
15424 @item -msdata=default
15426 @opindex msdata=default
15428 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15429 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15430 same as @option{-msdata=sysv}.
15433 @opindex msdata=data
15434 On System V.4 and embedded PowerPC systems, put small global
15435 data in the @samp{.sdata} section. Put small uninitialized global
15436 data in the @samp{.sbss} section. Do not use register @code{r13}
15437 to address small data however. This is the default behavior unless
15438 other @option{-msdata} options are used.
15442 @opindex msdata=none
15444 On embedded PowerPC systems, put all initialized global and static data
15445 in the @samp{.data} section, and all uninitialized data in the
15446 @samp{.bss} section.
15450 @cindex smaller data references (PowerPC)
15451 @cindex .sdata/.sdata2 references (PowerPC)
15452 On embedded PowerPC systems, put global and static items less than or
15453 equal to @var{num} bytes into the small data or bss sections instead of
15454 the normal data or bss section. By default, @var{num} is 8. The
15455 @option{-G @var{num}} switch is also passed to the linker.
15456 All modules should be compiled with the same @option{-G @var{num}} value.
15459 @itemx -mno-regnames
15461 @opindex mno-regnames
15462 On System V.4 and embedded PowerPC systems do (do not) emit register
15463 names in the assembly language output using symbolic forms.
15466 @itemx -mno-longcall
15468 @opindex mno-longcall
15469 By default assume that all calls are far away so that a longer more
15470 expensive calling sequence is required. This is required for calls
15471 further than 32 megabytes (33,554,432 bytes) from the current location.
15472 A short call will be generated if the compiler knows
15473 the call cannot be that far away. This setting can be overridden by
15474 the @code{shortcall} function attribute, or by @code{#pragma
15477 Some linkers are capable of detecting out-of-range calls and generating
15478 glue code on the fly. On these systems, long calls are unnecessary and
15479 generate slower code. As of this writing, the AIX linker can do this,
15480 as can the GNU linker for PowerPC/64. It is planned to add this feature
15481 to the GNU linker for 32-bit PowerPC systems as well.
15483 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15484 callee, L42'', plus a ``branch island'' (glue code). The two target
15485 addresses represent the callee and the ``branch island''. The
15486 Darwin/PPC linker will prefer the first address and generate a ``bl
15487 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15488 otherwise, the linker will generate ``bl L42'' to call the ``branch
15489 island''. The ``branch island'' is appended to the body of the
15490 calling function; it computes the full 32-bit address of the callee
15493 On Mach-O (Darwin) systems, this option directs the compiler emit to
15494 the glue for every direct call, and the Darwin linker decides whether
15495 to use or discard it.
15497 In the future, we may cause GCC to ignore all longcall specifications
15498 when the linker is known to generate glue.
15500 @item -mtls-markers
15501 @itemx -mno-tls-markers
15502 @opindex mtls-markers
15503 @opindex mno-tls-markers
15504 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15505 specifying the function argument. The relocation allows ld to
15506 reliably associate function call with argument setup instructions for
15507 TLS optimization, which in turn allows gcc to better schedule the
15512 Adds support for multithreading with the @dfn{pthreads} library.
15513 This option sets flags for both the preprocessor and linker.
15518 @subsection RX Options
15521 These command line options are defined for RX targets:
15524 @item -m64bit-doubles
15525 @itemx -m32bit-doubles
15526 @opindex m64bit-doubles
15527 @opindex m32bit-doubles
15528 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
15529 or 32-bits (@option{-m32bit-doubles}) in size. The default is
15530 @option{-m32bit-doubles}. @emph{Note} RX floating point hardware only
15531 works on 32-bit values, which is why the default is
15532 @option{-m32bit-doubles}.
15538 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
15539 floating point hardware. The default is enabled for the @var{RX600}
15540 series and disabled for the @var{RX200} series.
15542 Floating point instructions will only be generated for 32-bit floating
15543 point values however, so if the @option{-m64bit-doubles} option is in
15544 use then the FPU hardware will not be used for doubles.
15546 @emph{Note} If the @option{-fpu} option is enabled then
15547 @option{-funsafe-math-optimizations} is also enabled automatically.
15548 This is because the RX FPU instructions are themselves unsafe.
15550 @item -mcpu=@var{name}
15551 @itemx -patch=@var{name}
15554 Selects the type of RX CPU to be targeted. Currently three types are
15555 supported, the generic @var{RX600} and @var{RX200} series hardware and
15556 the specific @var{RX610} cpu. The default is @var{RX600}.
15558 The only difference between @var{RX600} and @var{RX610} is that the
15559 @var{RX610} does not support the @code{MVTIPL} instruction.
15561 The @var{RX200} series does not have a hardware floating point unit
15562 and so @option{-nofpu} is enabled by default when this type is
15565 @item -mbig-endian-data
15566 @itemx -mlittle-endian-data
15567 @opindex mbig-endian-data
15568 @opindex mlittle-endian-data
15569 Store data (but not code) in the big-endian format. The default is
15570 @option{-mlittle-endian-data}, ie to store data in the little endian
15573 @item -msmall-data-limit=@var{N}
15574 @opindex msmall-data-limit
15575 Specifies the maximum size in bytes of global and static variables
15576 which can be placed into the small data area. Using the small data
15577 area can lead to smaller and faster code, but the size of area is
15578 limited and it is up to the programmer to ensure that the area does
15579 not overflow. Also when the small data area is used one of the RX's
15580 registers (@code{r13}) is reserved for use pointing to this area, so
15581 it is no longer available for use by the compiler. This could result
15582 in slower and/or larger code if variables which once could have been
15583 held in @code{r13} are now pushed onto the stack.
15585 Note, common variables (variables which have not been initialised) and
15586 constants are not placed into the small data area as they are assigned
15587 to other sections in the output executable.
15589 The default value is zero, which disables this feature. Note, this
15590 feature is not enabled by default with higher optimization levels
15591 (@option{-O2} etc) because of the potentially detrimental effects of
15592 reserving register @code{r13}. It is up to the programmer to
15593 experiment and discover whether this feature is of benefit to their
15600 Use the simulator runtime. The default is to use the libgloss board
15603 @item -mas100-syntax
15604 @itemx -mno-as100-syntax
15605 @opindex mas100-syntax
15606 @opindex mno-as100-syntax
15607 When generating assembler output use a syntax that is compatible with
15608 Renesas's AS100 assembler. This syntax can also be handled by the GAS
15609 assembler but it has some restrictions so generating it is not the
15612 @item -mmax-constant-size=@var{N}
15613 @opindex mmax-constant-size
15614 Specifies the maximum size, in bytes, of a constant that can be used as
15615 an operand in a RX instruction. Although the RX instruction set does
15616 allow constants of up to 4 bytes in length to be used in instructions,
15617 a longer value equates to a longer instruction. Thus in some
15618 circumstances it can be beneficial to restrict the size of constants
15619 that are used in instructions. Constants that are too big are instead
15620 placed into a constant pool and referenced via register indirection.
15622 The value @var{N} can be between 0 and 4. A value of 0 (the default)
15623 or 4 means that constants of any size are allowed.
15627 Enable linker relaxation. Linker relaxation is a process whereby the
15628 linker will attempt to reduce the size of a program by finding shorter
15629 versions of various instructions. Disabled by default.
15631 @item -mint-register=@var{N}
15632 @opindex mint-register
15633 Specify the number of registers to reserve for fast interrupt handler
15634 functions. The value @var{N} can be between 0 and 4. A value of 1
15635 means that register @code{r13} will be reserved for the exclusive use
15636 of fast interrupt handlers. A value of 2 reserves @code{r13} and
15637 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
15638 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
15639 A value of 0, the default, does not reserve any registers.
15641 @item -msave-acc-in-interrupts
15642 @opindex msave-acc-in-interrupts
15643 Specifies that interrupt handler functions should preserve the
15644 accumulator register. This is only necessary if normal code might use
15645 the accumulator register, for example because it performs 64-bit
15646 multiplications. The default is to ignore the accumulator as this
15647 makes the interrupt handlers faster.
15651 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
15652 has special significance to the RX port when used with the
15653 @code{interrupt} function attribute. This attribute indicates a
15654 function intended to process fast interrupts. GCC will will ensure
15655 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
15656 and/or @code{r13} and only provided that the normal use of the
15657 corresponding registers have been restricted via the
15658 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
15661 @node S/390 and zSeries Options
15662 @subsection S/390 and zSeries Options
15663 @cindex S/390 and zSeries Options
15665 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15669 @itemx -msoft-float
15670 @opindex mhard-float
15671 @opindex msoft-float
15672 Use (do not use) the hardware floating-point instructions and registers
15673 for floating-point operations. When @option{-msoft-float} is specified,
15674 functions in @file{libgcc.a} will be used to perform floating-point
15675 operations. When @option{-mhard-float} is specified, the compiler
15676 generates IEEE floating-point instructions. This is the default.
15679 @itemx -mno-hard-dfp
15681 @opindex mno-hard-dfp
15682 Use (do not use) the hardware decimal-floating-point instructions for
15683 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15684 specified, functions in @file{libgcc.a} will be used to perform
15685 decimal-floating-point operations. When @option{-mhard-dfp} is
15686 specified, the compiler generates decimal-floating-point hardware
15687 instructions. This is the default for @option{-march=z9-ec} or higher.
15689 @item -mlong-double-64
15690 @itemx -mlong-double-128
15691 @opindex mlong-double-64
15692 @opindex mlong-double-128
15693 These switches control the size of @code{long double} type. A size
15694 of 64bit makes the @code{long double} type equivalent to the @code{double}
15695 type. This is the default.
15698 @itemx -mno-backchain
15699 @opindex mbackchain
15700 @opindex mno-backchain
15701 Store (do not store) the address of the caller's frame as backchain pointer
15702 into the callee's stack frame.
15703 A backchain may be needed to allow debugging using tools that do not understand
15704 DWARF-2 call frame information.
15705 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15706 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15707 the backchain is placed into the topmost word of the 96/160 byte register
15710 In general, code compiled with @option{-mbackchain} is call-compatible with
15711 code compiled with @option{-mmo-backchain}; however, use of the backchain
15712 for debugging purposes usually requires that the whole binary is built with
15713 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15714 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15715 to build a linux kernel use @option{-msoft-float}.
15717 The default is to not maintain the backchain.
15719 @item -mpacked-stack
15720 @itemx -mno-packed-stack
15721 @opindex mpacked-stack
15722 @opindex mno-packed-stack
15723 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15724 specified, the compiler uses the all fields of the 96/160 byte register save
15725 area only for their default purpose; unused fields still take up stack space.
15726 When @option{-mpacked-stack} is specified, register save slots are densely
15727 packed at the top of the register save area; unused space is reused for other
15728 purposes, allowing for more efficient use of the available stack space.
15729 However, when @option{-mbackchain} is also in effect, the topmost word of
15730 the save area is always used to store the backchain, and the return address
15731 register is always saved two words below the backchain.
15733 As long as the stack frame backchain is not used, code generated with
15734 @option{-mpacked-stack} is call-compatible with code generated with
15735 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15736 S/390 or zSeries generated code that uses the stack frame backchain at run
15737 time, not just for debugging purposes. Such code is not call-compatible
15738 with code compiled with @option{-mpacked-stack}. Also, note that the
15739 combination of @option{-mbackchain},
15740 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15741 to build a linux kernel use @option{-msoft-float}.
15743 The default is to not use the packed stack layout.
15746 @itemx -mno-small-exec
15747 @opindex msmall-exec
15748 @opindex mno-small-exec
15749 Generate (or do not generate) code using the @code{bras} instruction
15750 to do subroutine calls.
15751 This only works reliably if the total executable size does not
15752 exceed 64k. The default is to use the @code{basr} instruction instead,
15753 which does not have this limitation.
15759 When @option{-m31} is specified, generate code compliant to the
15760 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15761 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15762 particular to generate 64-bit instructions. For the @samp{s390}
15763 targets, the default is @option{-m31}, while the @samp{s390x}
15764 targets default to @option{-m64}.
15770 When @option{-mzarch} is specified, generate code using the
15771 instructions available on z/Architecture.
15772 When @option{-mesa} is specified, generate code using the
15773 instructions available on ESA/390. Note that @option{-mesa} is
15774 not possible with @option{-m64}.
15775 When generating code compliant to the GNU/Linux for S/390 ABI,
15776 the default is @option{-mesa}. When generating code compliant
15777 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
15783 Generate (or do not generate) code using the @code{mvcle} instruction
15784 to perform block moves. When @option{-mno-mvcle} is specified,
15785 use a @code{mvc} loop instead. This is the default unless optimizing for
15792 Print (or do not print) additional debug information when compiling.
15793 The default is to not print debug information.
15795 @item -march=@var{cpu-type}
15797 Generate code that will run on @var{cpu-type}, which is the name of a system
15798 representing a certain processor type. Possible values for
15799 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
15800 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
15801 When generating code using the instructions available on z/Architecture,
15802 the default is @option{-march=z900}. Otherwise, the default is
15803 @option{-march=g5}.
15805 @item -mtune=@var{cpu-type}
15807 Tune to @var{cpu-type} everything applicable about the generated code,
15808 except for the ABI and the set of available instructions.
15809 The list of @var{cpu-type} values is the same as for @option{-march}.
15810 The default is the value used for @option{-march}.
15813 @itemx -mno-tpf-trace
15814 @opindex mtpf-trace
15815 @opindex mno-tpf-trace
15816 Generate code that adds (does not add) in TPF OS specific branches to trace
15817 routines in the operating system. This option is off by default, even
15818 when compiling for the TPF OS@.
15821 @itemx -mno-fused-madd
15822 @opindex mfused-madd
15823 @opindex mno-fused-madd
15824 Generate code that uses (does not use) the floating point multiply and
15825 accumulate instructions. These instructions are generated by default if
15826 hardware floating point is used.
15828 @item -mwarn-framesize=@var{framesize}
15829 @opindex mwarn-framesize
15830 Emit a warning if the current function exceeds the given frame size. Because
15831 this is a compile time check it doesn't need to be a real problem when the program
15832 runs. It is intended to identify functions which most probably cause
15833 a stack overflow. It is useful to be used in an environment with limited stack
15834 size e.g.@: the linux kernel.
15836 @item -mwarn-dynamicstack
15837 @opindex mwarn-dynamicstack
15838 Emit a warning if the function calls alloca or uses dynamically
15839 sized arrays. This is generally a bad idea with a limited stack size.
15841 @item -mstack-guard=@var{stack-guard}
15842 @itemx -mstack-size=@var{stack-size}
15843 @opindex mstack-guard
15844 @opindex mstack-size
15845 If these options are provided the s390 back end emits additional instructions in
15846 the function prologue which trigger a trap if the stack size is @var{stack-guard}
15847 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
15848 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
15849 the frame size of the compiled function is chosen.
15850 These options are intended to be used to help debugging stack overflow problems.
15851 The additionally emitted code causes only little overhead and hence can also be
15852 used in production like systems without greater performance degradation. The given
15853 values have to be exact powers of 2 and @var{stack-size} has to be greater than
15854 @var{stack-guard} without exceeding 64k.
15855 In order to be efficient the extra code makes the assumption that the stack starts
15856 at an address aligned to the value given by @var{stack-size}.
15857 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
15860 @node Score Options
15861 @subsection Score Options
15862 @cindex Score Options
15864 These options are defined for Score implementations:
15869 Compile code for big endian mode. This is the default.
15873 Compile code for little endian mode.
15877 Disable generate bcnz instruction.
15881 Enable generate unaligned load and store instruction.
15885 Enable the use of multiply-accumulate instructions. Disabled by default.
15889 Specify the SCORE5 as the target architecture.
15893 Specify the SCORE5U of the target architecture.
15897 Specify the SCORE7 as the target architecture. This is the default.
15901 Specify the SCORE7D as the target architecture.
15905 @subsection SH Options
15907 These @samp{-m} options are defined for the SH implementations:
15912 Generate code for the SH1.
15916 Generate code for the SH2.
15919 Generate code for the SH2e.
15923 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
15924 that the floating-point unit is not used.
15926 @item -m2a-single-only
15927 @opindex m2a-single-only
15928 Generate code for the SH2a-FPU, in such a way that no double-precision
15929 floating point operations are used.
15932 @opindex m2a-single
15933 Generate code for the SH2a-FPU assuming the floating-point unit is in
15934 single-precision mode by default.
15938 Generate code for the SH2a-FPU assuming the floating-point unit is in
15939 double-precision mode by default.
15943 Generate code for the SH3.
15947 Generate code for the SH3e.
15951 Generate code for the SH4 without a floating-point unit.
15953 @item -m4-single-only
15954 @opindex m4-single-only
15955 Generate code for the SH4 with a floating-point unit that only
15956 supports single-precision arithmetic.
15960 Generate code for the SH4 assuming the floating-point unit is in
15961 single-precision mode by default.
15965 Generate code for the SH4.
15969 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
15970 floating-point unit is not used.
15972 @item -m4a-single-only
15973 @opindex m4a-single-only
15974 Generate code for the SH4a, in such a way that no double-precision
15975 floating point operations are used.
15978 @opindex m4a-single
15979 Generate code for the SH4a assuming the floating-point unit is in
15980 single-precision mode by default.
15984 Generate code for the SH4a.
15988 Same as @option{-m4a-nofpu}, except that it implicitly passes
15989 @option{-dsp} to the assembler. GCC doesn't generate any DSP
15990 instructions at the moment.
15994 Compile code for the processor in big endian mode.
15998 Compile code for the processor in little endian mode.
16002 Align doubles at 64-bit boundaries. Note that this changes the calling
16003 conventions, and thus some functions from the standard C library will
16004 not work unless you recompile it first with @option{-mdalign}.
16008 Shorten some address references at link time, when possible; uses the
16009 linker option @option{-relax}.
16013 Use 32-bit offsets in @code{switch} tables. The default is to use
16018 Enable the use of bit manipulation instructions on SH2A.
16022 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
16023 alignment constraints.
16027 Comply with the calling conventions defined by Renesas.
16031 Comply with the calling conventions defined by Renesas.
16035 Comply with the calling conventions defined for GCC before the Renesas
16036 conventions were available. This option is the default for all
16037 targets of the SH toolchain except for @samp{sh-symbianelf}.
16040 @opindex mnomacsave
16041 Mark the @code{MAC} register as call-clobbered, even if
16042 @option{-mhitachi} is given.
16046 Increase IEEE-compliance of floating-point code.
16047 At the moment, this is equivalent to @option{-fno-finite-math-only}.
16048 When generating 16 bit SH opcodes, getting IEEE-conforming results for
16049 comparisons of NANs / infinities incurs extra overhead in every
16050 floating point comparison, therefore the default is set to
16051 @option{-ffinite-math-only}.
16053 @item -minline-ic_invalidate
16054 @opindex minline-ic_invalidate
16055 Inline code to invalidate instruction cache entries after setting up
16056 nested function trampolines.
16057 This option has no effect if -musermode is in effect and the selected
16058 code generation option (e.g. -m4) does not allow the use of the icbi
16060 If the selected code generation option does not allow the use of the icbi
16061 instruction, and -musermode is not in effect, the inlined code will
16062 manipulate the instruction cache address array directly with an associative
16063 write. This not only requires privileged mode, but it will also
16064 fail if the cache line had been mapped via the TLB and has become unmapped.
16068 Dump instruction size and location in the assembly code.
16071 @opindex mpadstruct
16072 This option is deprecated. It pads structures to multiple of 4 bytes,
16073 which is incompatible with the SH ABI@.
16077 Optimize for space instead of speed. Implied by @option{-Os}.
16080 @opindex mprefergot
16081 When generating position-independent code, emit function calls using
16082 the Global Offset Table instead of the Procedure Linkage Table.
16086 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
16087 if the inlined code would not work in user mode.
16088 This is the default when the target is @code{sh-*-linux*}.
16090 @item -multcost=@var{number}
16091 @opindex multcost=@var{number}
16092 Set the cost to assume for a multiply insn.
16094 @item -mdiv=@var{strategy}
16095 @opindex mdiv=@var{strategy}
16096 Set the division strategy to use for SHmedia code. @var{strategy} must be
16097 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
16098 inv:call2, inv:fp .
16099 "fp" performs the operation in floating point. This has a very high latency,
16100 but needs only a few instructions, so it might be a good choice if
16101 your code has enough easily exploitable ILP to allow the compiler to
16102 schedule the floating point instructions together with other instructions.
16103 Division by zero causes a floating point exception.
16104 "inv" uses integer operations to calculate the inverse of the divisor,
16105 and then multiplies the dividend with the inverse. This strategy allows
16106 cse and hoisting of the inverse calculation. Division by zero calculates
16107 an unspecified result, but does not trap.
16108 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
16109 have been found, or if the entire operation has been hoisted to the same
16110 place, the last stages of the inverse calculation are intertwined with the
16111 final multiply to reduce the overall latency, at the expense of using a few
16112 more instructions, and thus offering fewer scheduling opportunities with
16114 "call" calls a library function that usually implements the inv:minlat
16116 This gives high code density for m5-*media-nofpu compilations.
16117 "call2" uses a different entry point of the same library function, where it
16118 assumes that a pointer to a lookup table has already been set up, which
16119 exposes the pointer load to cse / code hoisting optimizations.
16120 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
16121 code generation, but if the code stays unoptimized, revert to the "call",
16122 "call2", or "fp" strategies, respectively. Note that the
16123 potentially-trapping side effect of division by zero is carried by a
16124 separate instruction, so it is possible that all the integer instructions
16125 are hoisted out, but the marker for the side effect stays where it is.
16126 A recombination to fp operations or a call is not possible in that case.
16127 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
16128 that the inverse calculation was nor separated from the multiply, they speed
16129 up division where the dividend fits into 20 bits (plus sign where applicable),
16130 by inserting a test to skip a number of operations in this case; this test
16131 slows down the case of larger dividends. inv20u assumes the case of a such
16132 a small dividend to be unlikely, and inv20l assumes it to be likely.
16134 @item -mdivsi3_libfunc=@var{name}
16135 @opindex mdivsi3_libfunc=@var{name}
16136 Set the name of the library function used for 32 bit signed division to
16137 @var{name}. This only affect the name used in the call and inv:call
16138 division strategies, and the compiler will still expect the same
16139 sets of input/output/clobbered registers as if this option was not present.
16141 @item -mfixed-range=@var{register-range}
16142 @opindex mfixed-range
16143 Generate code treating the given register range as fixed registers.
16144 A fixed register is one that the register allocator can not use. This is
16145 useful when compiling kernel code. A register range is specified as
16146 two registers separated by a dash. Multiple register ranges can be
16147 specified separated by a comma.
16149 @item -madjust-unroll
16150 @opindex madjust-unroll
16151 Throttle unrolling to avoid thrashing target registers.
16152 This option only has an effect if the gcc code base supports the
16153 TARGET_ADJUST_UNROLL_MAX target hook.
16155 @item -mindexed-addressing
16156 @opindex mindexed-addressing
16157 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16158 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16159 semantics for the indexed addressing mode. The architecture allows the
16160 implementation of processors with 64 bit MMU, which the OS could use to
16161 get 32 bit addressing, but since no current hardware implementation supports
16162 this or any other way to make the indexed addressing mode safe to use in
16163 the 32 bit ABI, the default is -mno-indexed-addressing.
16165 @item -mgettrcost=@var{number}
16166 @opindex mgettrcost=@var{number}
16167 Set the cost assumed for the gettr instruction to @var{number}.
16168 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16172 Assume pt* instructions won't trap. This will generally generate better
16173 scheduled code, but is unsafe on current hardware. The current architecture
16174 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16175 This has the unintentional effect of making it unsafe to schedule ptabs /
16176 ptrel before a branch, or hoist it out of a loop. For example,
16177 __do_global_ctors, a part of libgcc that runs constructors at program
16178 startup, calls functions in a list which is delimited by @minus{}1. With the
16179 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16180 That means that all the constructors will be run a bit quicker, but when
16181 the loop comes to the end of the list, the program crashes because ptabs
16182 loads @minus{}1 into a target register. Since this option is unsafe for any
16183 hardware implementing the current architecture specification, the default
16184 is -mno-pt-fixed. Unless the user specifies a specific cost with
16185 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16186 this deters register allocation using target registers for storing
16189 @item -minvalid-symbols
16190 @opindex minvalid-symbols
16191 Assume symbols might be invalid. Ordinary function symbols generated by
16192 the compiler will always be valid to load with movi/shori/ptabs or
16193 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16194 to generate symbols that will cause ptabs / ptrel to trap.
16195 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16196 It will then prevent cross-basic-block cse, hoisting and most scheduling
16197 of symbol loads. The default is @option{-mno-invalid-symbols}.
16200 @node SPARC Options
16201 @subsection SPARC Options
16202 @cindex SPARC options
16204 These @samp{-m} options are supported on the SPARC:
16207 @item -mno-app-regs
16209 @opindex mno-app-regs
16211 Specify @option{-mapp-regs} to generate output using the global registers
16212 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16215 To be fully SVR4 ABI compliant at the cost of some performance loss,
16216 specify @option{-mno-app-regs}. You should compile libraries and system
16217 software with this option.
16220 @itemx -mhard-float
16222 @opindex mhard-float
16223 Generate output containing floating point instructions. This is the
16227 @itemx -msoft-float
16229 @opindex msoft-float
16230 Generate output containing library calls for floating point.
16231 @strong{Warning:} the requisite libraries are not available for all SPARC
16232 targets. Normally the facilities of the machine's usual C compiler are
16233 used, but this cannot be done directly in cross-compilation. You must make
16234 your own arrangements to provide suitable library functions for
16235 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16236 @samp{sparclite-*-*} do provide software floating point support.
16238 @option{-msoft-float} changes the calling convention in the output file;
16239 therefore, it is only useful if you compile @emph{all} of a program with
16240 this option. In particular, you need to compile @file{libgcc.a}, the
16241 library that comes with GCC, with @option{-msoft-float} in order for
16244 @item -mhard-quad-float
16245 @opindex mhard-quad-float
16246 Generate output containing quad-word (long double) floating point
16249 @item -msoft-quad-float
16250 @opindex msoft-quad-float
16251 Generate output containing library calls for quad-word (long double)
16252 floating point instructions. The functions called are those specified
16253 in the SPARC ABI@. This is the default.
16255 As of this writing, there are no SPARC implementations that have hardware
16256 support for the quad-word floating point instructions. They all invoke
16257 a trap handler for one of these instructions, and then the trap handler
16258 emulates the effect of the instruction. Because of the trap handler overhead,
16259 this is much slower than calling the ABI library routines. Thus the
16260 @option{-msoft-quad-float} option is the default.
16262 @item -mno-unaligned-doubles
16263 @itemx -munaligned-doubles
16264 @opindex mno-unaligned-doubles
16265 @opindex munaligned-doubles
16266 Assume that doubles have 8 byte alignment. This is the default.
16268 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16269 alignment only if they are contained in another type, or if they have an
16270 absolute address. Otherwise, it assumes they have 4 byte alignment.
16271 Specifying this option avoids some rare compatibility problems with code
16272 generated by other compilers. It is not the default because it results
16273 in a performance loss, especially for floating point code.
16275 @item -mno-faster-structs
16276 @itemx -mfaster-structs
16277 @opindex mno-faster-structs
16278 @opindex mfaster-structs
16279 With @option{-mfaster-structs}, the compiler assumes that structures
16280 should have 8 byte alignment. This enables the use of pairs of
16281 @code{ldd} and @code{std} instructions for copies in structure
16282 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16283 However, the use of this changed alignment directly violates the SPARC
16284 ABI@. Thus, it's intended only for use on targets where the developer
16285 acknowledges that their resulting code will not be directly in line with
16286 the rules of the ABI@.
16288 @item -mimpure-text
16289 @opindex mimpure-text
16290 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16291 the compiler to not pass @option{-z text} to the linker when linking a
16292 shared object. Using this option, you can link position-dependent
16293 code into a shared object.
16295 @option{-mimpure-text} suppresses the ``relocations remain against
16296 allocatable but non-writable sections'' linker error message.
16297 However, the necessary relocations will trigger copy-on-write, and the
16298 shared object is not actually shared across processes. Instead of
16299 using @option{-mimpure-text}, you should compile all source code with
16300 @option{-fpic} or @option{-fPIC}.
16302 This option is only available on SunOS and Solaris.
16304 @item -mcpu=@var{cpu_type}
16306 Set the instruction set, register set, and instruction scheduling parameters
16307 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16308 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16309 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16310 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16311 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16313 Default instruction scheduling parameters are used for values that select
16314 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16315 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16317 Here is a list of each supported architecture and their supported
16322 v8: supersparc, hypersparc
16323 sparclite: f930, f934, sparclite86x
16325 v9: ultrasparc, ultrasparc3, niagara, niagara2
16328 By default (unless configured otherwise), GCC generates code for the V7
16329 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16330 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16331 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16332 SPARCStation 1, 2, IPX etc.
16334 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16335 architecture. The only difference from V7 code is that the compiler emits
16336 the integer multiply and integer divide instructions which exist in SPARC-V8
16337 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16338 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16341 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16342 the SPARC architecture. This adds the integer multiply, integer divide step
16343 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16344 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16345 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16346 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16347 MB86934 chip, which is the more recent SPARClite with FPU@.
16349 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16350 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16351 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16352 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16353 optimizes it for the TEMIC SPARClet chip.
16355 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16356 architecture. This adds 64-bit integer and floating-point move instructions,
16357 3 additional floating-point condition code registers and conditional move
16358 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16359 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16360 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
16361 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
16362 @option{-mcpu=niagara}, the compiler additionally optimizes it for
16363 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
16364 additionally optimizes it for Sun UltraSPARC T2 chips.
16366 @item -mtune=@var{cpu_type}
16368 Set the instruction scheduling parameters for machine type
16369 @var{cpu_type}, but do not set the instruction set or register set that the
16370 option @option{-mcpu=@var{cpu_type}} would.
16372 The same values for @option{-mcpu=@var{cpu_type}} can be used for
16373 @option{-mtune=@var{cpu_type}}, but the only useful values are those
16374 that select a particular cpu implementation. Those are @samp{cypress},
16375 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
16376 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
16377 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
16382 @opindex mno-v8plus
16383 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
16384 difference from the V8 ABI is that the global and out registers are
16385 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
16386 mode for all SPARC-V9 processors.
16392 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
16393 Visual Instruction Set extensions. The default is @option{-mno-vis}.
16396 These @samp{-m} options are supported in addition to the above
16397 on SPARC-V9 processors in 64-bit environments:
16400 @item -mlittle-endian
16401 @opindex mlittle-endian
16402 Generate code for a processor running in little-endian mode. It is only
16403 available for a few configurations and most notably not on Solaris and Linux.
16409 Generate code for a 32-bit or 64-bit environment.
16410 The 32-bit environment sets int, long and pointer to 32 bits.
16411 The 64-bit environment sets int to 32 bits and long and pointer
16414 @item -mcmodel=medlow
16415 @opindex mcmodel=medlow
16416 Generate code for the Medium/Low code model: 64-bit addresses, programs
16417 must be linked in the low 32 bits of memory. Programs can be statically
16418 or dynamically linked.
16420 @item -mcmodel=medmid
16421 @opindex mcmodel=medmid
16422 Generate code for the Medium/Middle code model: 64-bit addresses, programs
16423 must be linked in the low 44 bits of memory, the text and data segments must
16424 be less than 2GB in size and the data segment must be located within 2GB of
16427 @item -mcmodel=medany
16428 @opindex mcmodel=medany
16429 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
16430 may be linked anywhere in memory, the text and data segments must be less
16431 than 2GB in size and the data segment must be located within 2GB of the
16434 @item -mcmodel=embmedany
16435 @opindex mcmodel=embmedany
16436 Generate code for the Medium/Anywhere code model for embedded systems:
16437 64-bit addresses, the text and data segments must be less than 2GB in
16438 size, both starting anywhere in memory (determined at link time). The
16439 global register %g4 points to the base of the data segment. Programs
16440 are statically linked and PIC is not supported.
16443 @itemx -mno-stack-bias
16444 @opindex mstack-bias
16445 @opindex mno-stack-bias
16446 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
16447 frame pointer if present, are offset by @minus{}2047 which must be added back
16448 when making stack frame references. This is the default in 64-bit mode.
16449 Otherwise, assume no such offset is present.
16452 These switches are supported in addition to the above on Solaris:
16457 Add support for multithreading using the Solaris threads library. This
16458 option sets flags for both the preprocessor and linker. This option does
16459 not affect the thread safety of object code produced by the compiler or
16460 that of libraries supplied with it.
16464 Add support for multithreading using the POSIX threads library. This
16465 option sets flags for both the preprocessor and linker. This option does
16466 not affect the thread safety of object code produced by the compiler or
16467 that of libraries supplied with it.
16471 This is a synonym for @option{-pthreads}.
16475 @subsection SPU Options
16476 @cindex SPU options
16478 These @samp{-m} options are supported on the SPU:
16482 @itemx -merror-reloc
16483 @opindex mwarn-reloc
16484 @opindex merror-reloc
16486 The loader for SPU does not handle dynamic relocations. By default, GCC
16487 will give an error when it generates code that requires a dynamic
16488 relocation. @option{-mno-error-reloc} disables the error,
16489 @option{-mwarn-reloc} will generate a warning instead.
16492 @itemx -munsafe-dma
16494 @opindex munsafe-dma
16496 Instructions which initiate or test completion of DMA must not be
16497 reordered with respect to loads and stores of the memory which is being
16498 accessed. Users typically address this problem using the volatile
16499 keyword, but that can lead to inefficient code in places where the
16500 memory is known to not change. Rather than mark the memory as volatile
16501 we treat the DMA instructions as potentially effecting all memory. With
16502 @option{-munsafe-dma} users must use the volatile keyword to protect
16505 @item -mbranch-hints
16506 @opindex mbranch-hints
16508 By default, GCC will generate a branch hint instruction to avoid
16509 pipeline stalls for always taken or probably taken branches. A hint
16510 will not be generated closer than 8 instructions away from its branch.
16511 There is little reason to disable them, except for debugging purposes,
16512 or to make an object a little bit smaller.
16516 @opindex msmall-mem
16517 @opindex mlarge-mem
16519 By default, GCC generates code assuming that addresses are never larger
16520 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
16521 a full 32 bit address.
16526 By default, GCC links against startup code that assumes the SPU-style
16527 main function interface (which has an unconventional parameter list).
16528 With @option{-mstdmain}, GCC will link your program against startup
16529 code that assumes a C99-style interface to @code{main}, including a
16530 local copy of @code{argv} strings.
16532 @item -mfixed-range=@var{register-range}
16533 @opindex mfixed-range
16534 Generate code treating the given register range as fixed registers.
16535 A fixed register is one that the register allocator can not use. This is
16536 useful when compiling kernel code. A register range is specified as
16537 two registers separated by a dash. Multiple register ranges can be
16538 specified separated by a comma.
16544 Compile code assuming that pointers to the PPU address space accessed
16545 via the @code{__ea} named address space qualifier are either 32 or 64
16546 bits wide. The default is 32 bits. As this is an ABI changing option,
16547 all object code in an executable must be compiled with the same setting.
16549 @item -maddress-space-conversion
16550 @itemx -mno-address-space-conversion
16551 @opindex maddress-space-conversion
16552 @opindex mno-address-space-conversion
16553 Allow/disallow treating the @code{__ea} address space as superset
16554 of the generic address space. This enables explicit type casts
16555 between @code{__ea} and generic pointer as well as implicit
16556 conversions of generic pointers to @code{__ea} pointers. The
16557 default is to allow address space pointer conversions.
16559 @item -mcache-size=@var{cache-size}
16560 @opindex mcache-size
16561 This option controls the version of libgcc that the compiler links to an
16562 executable and selects a software-managed cache for accessing variables
16563 in the @code{__ea} address space with a particular cache size. Possible
16564 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
16565 and @samp{128}. The default cache size is 64KB.
16567 @item -matomic-updates
16568 @itemx -mno-atomic-updates
16569 @opindex matomic-updates
16570 @opindex mno-atomic-updates
16571 This option controls the version of libgcc that the compiler links to an
16572 executable and selects whether atomic updates to the software-managed
16573 cache of PPU-side variables are used. If you use atomic updates, changes
16574 to a PPU variable from SPU code using the @code{__ea} named address space
16575 qualifier will not interfere with changes to other PPU variables residing
16576 in the same cache line from PPU code. If you do not use atomic updates,
16577 such interference may occur; however, writing back cache lines will be
16578 more efficient. The default behavior is to use atomic updates.
16581 @itemx -mdual-nops=@var{n}
16582 @opindex mdual-nops
16583 By default, GCC will insert nops to increase dual issue when it expects
16584 it to increase performance. @var{n} can be a value from 0 to 10. A
16585 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
16586 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
16588 @item -mhint-max-nops=@var{n}
16589 @opindex mhint-max-nops
16590 Maximum number of nops to insert for a branch hint. A branch hint must
16591 be at least 8 instructions away from the branch it is effecting. GCC
16592 will insert up to @var{n} nops to enforce this, otherwise it will not
16593 generate the branch hint.
16595 @item -mhint-max-distance=@var{n}
16596 @opindex mhint-max-distance
16597 The encoding of the branch hint instruction limits the hint to be within
16598 256 instructions of the branch it is effecting. By default, GCC makes
16599 sure it is within 125.
16602 @opindex msafe-hints
16603 Work around a hardware bug which causes the SPU to stall indefinitely.
16604 By default, GCC will insert the @code{hbrp} instruction to make sure
16605 this stall won't happen.
16609 @node System V Options
16610 @subsection Options for System V
16612 These additional options are available on System V Release 4 for
16613 compatibility with other compilers on those systems:
16618 Create a shared object.
16619 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16623 Identify the versions of each tool used by the compiler, in a
16624 @code{.ident} assembler directive in the output.
16628 Refrain from adding @code{.ident} directives to the output file (this is
16631 @item -YP,@var{dirs}
16633 Search the directories @var{dirs}, and no others, for libraries
16634 specified with @option{-l}.
16636 @item -Ym,@var{dir}
16638 Look in the directory @var{dir} to find the M4 preprocessor.
16639 The assembler uses this option.
16640 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16641 @c the generic assembler that comes with Solaris takes just -Ym.
16645 @subsection V850 Options
16646 @cindex V850 Options
16648 These @samp{-m} options are defined for V850 implementations:
16652 @itemx -mno-long-calls
16653 @opindex mlong-calls
16654 @opindex mno-long-calls
16655 Treat all calls as being far away (near). If calls are assumed to be
16656 far away, the compiler will always load the functions address up into a
16657 register, and call indirect through the pointer.
16663 Do not optimize (do optimize) basic blocks that use the same index
16664 pointer 4 or more times to copy pointer into the @code{ep} register, and
16665 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16666 option is on by default if you optimize.
16668 @item -mno-prolog-function
16669 @itemx -mprolog-function
16670 @opindex mno-prolog-function
16671 @opindex mprolog-function
16672 Do not use (do use) external functions to save and restore registers
16673 at the prologue and epilogue of a function. The external functions
16674 are slower, but use less code space if more than one function saves
16675 the same number of registers. The @option{-mprolog-function} option
16676 is on by default if you optimize.
16680 Try to make the code as small as possible. At present, this just turns
16681 on the @option{-mep} and @option{-mprolog-function} options.
16683 @item -mtda=@var{n}
16685 Put static or global variables whose size is @var{n} bytes or less into
16686 the tiny data area that register @code{ep} points to. The tiny data
16687 area can hold up to 256 bytes in total (128 bytes for byte references).
16689 @item -msda=@var{n}
16691 Put static or global variables whose size is @var{n} bytes or less into
16692 the small data area that register @code{gp} points to. The small data
16693 area can hold up to 64 kilobytes.
16695 @item -mzda=@var{n}
16697 Put static or global variables whose size is @var{n} bytes or less into
16698 the first 32 kilobytes of memory.
16702 Specify that the target processor is the V850.
16705 @opindex mbig-switch
16706 Generate code suitable for big switch tables. Use this option only if
16707 the assembler/linker complain about out of range branches within a switch
16712 This option will cause r2 and r5 to be used in the code generated by
16713 the compiler. This setting is the default.
16715 @item -mno-app-regs
16716 @opindex mno-app-regs
16717 This option will cause r2 and r5 to be treated as fixed registers.
16721 Specify that the target processor is the V850E1. The preprocessor
16722 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16723 this option is used.
16727 Specify that the target processor is the V850E@. The preprocessor
16728 constant @samp{__v850e__} will be defined if this option is used.
16730 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16731 are defined then a default target processor will be chosen and the
16732 relevant @samp{__v850*__} preprocessor constant will be defined.
16734 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16735 defined, regardless of which processor variant is the target.
16737 @item -mdisable-callt
16738 @opindex mdisable-callt
16739 This option will suppress generation of the CALLT instruction for the
16740 v850e and v850e1 flavors of the v850 architecture. The default is
16741 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16746 @subsection VAX Options
16747 @cindex VAX options
16749 These @samp{-m} options are defined for the VAX:
16754 Do not output certain jump instructions (@code{aobleq} and so on)
16755 that the Unix assembler for the VAX cannot handle across long
16760 Do output those jump instructions, on the assumption that you
16761 will assemble with the GNU assembler.
16765 Output code for g-format floating point numbers instead of d-format.
16768 @node VxWorks Options
16769 @subsection VxWorks Options
16770 @cindex VxWorks Options
16772 The options in this section are defined for all VxWorks targets.
16773 Options specific to the target hardware are listed with the other
16774 options for that target.
16779 GCC can generate code for both VxWorks kernels and real time processes
16780 (RTPs). This option switches from the former to the latter. It also
16781 defines the preprocessor macro @code{__RTP__}.
16784 @opindex non-static
16785 Link an RTP executable against shared libraries rather than static
16786 libraries. The options @option{-static} and @option{-shared} can
16787 also be used for RTPs (@pxref{Link Options}); @option{-static}
16794 These options are passed down to the linker. They are defined for
16795 compatibility with Diab.
16798 @opindex Xbind-lazy
16799 Enable lazy binding of function calls. This option is equivalent to
16800 @option{-Wl,-z,now} and is defined for compatibility with Diab.
16804 Disable lazy binding of function calls. This option is the default and
16805 is defined for compatibility with Diab.
16808 @node x86-64 Options
16809 @subsection x86-64 Options
16810 @cindex x86-64 options
16812 These are listed under @xref{i386 and x86-64 Options}.
16814 @node i386 and x86-64 Windows Options
16815 @subsection i386 and x86-64 Windows Options
16816 @cindex i386 and x86-64 Windows Options
16818 These additional options are available for Windows targets:
16823 This option is available for Cygwin and MinGW targets. It
16824 specifies that a console application is to be generated, by
16825 instructing the linker to set the PE header subsystem type
16826 required for console applications.
16827 This is the default behavior for Cygwin and MinGW targets.
16831 This option is available for Cygwin targets. It specifies that
16832 the Cygwin internal interface is to be used for predefined
16833 preprocessor macros, C runtime libraries and related linker
16834 paths and options. For Cygwin targets this is the default behavior.
16835 This option is deprecated and will be removed in a future release.
16838 @opindex mno-cygwin
16839 This option is available for Cygwin targets. It specifies that
16840 the MinGW internal interface is to be used instead of Cygwin's, by
16841 setting MinGW-related predefined macros and linker paths and default
16843 This option is deprecated and will be removed in a future release.
16847 This option is available for Cygwin and MinGW targets. It
16848 specifies that a DLL - a dynamic link library - is to be
16849 generated, enabling the selection of the required runtime
16850 startup object and entry point.
16852 @item -mnop-fun-dllimport
16853 @opindex mnop-fun-dllimport
16854 This option is available for Cygwin and MinGW targets. It
16855 specifies that the dllimport attribute should be ignored.
16859 This option is available for MinGW targets. It specifies
16860 that MinGW-specific thread support is to be used.
16864 This option is available for mingw-w64 targets. It specifies
16865 that the UNICODE macro is getting pre-defined and that the
16866 unicode capable runtime startup code is chosen.
16870 This option is available for Cygwin and MinGW targets. It
16871 specifies that the typical Windows pre-defined macros are to
16872 be set in the pre-processor, but does not influence the choice
16873 of runtime library/startup code.
16877 This option is available for Cygwin and MinGW targets. It
16878 specifies that a GUI application is to be generated by
16879 instructing the linker to set the PE header subsystem type
16882 @item -fno-set-stack-executable
16883 @opindex fno-set-stack-executable
16884 This option is available for MinGW targets. It specifies that
16885 the executable flag for stack used by nested functions isn't
16886 set. This is necessary for binaries running in kernel mode of
16887 Windows, as there the user32 API, which is used to set executable
16888 privileges, isn't available.
16890 @item -mpe-aligned-commons
16891 @opindex mpe-aligned-commons
16892 This option is available for Cygwin and MinGW targets. It
16893 specifies that the GNU extension to the PE file format that
16894 permits the correct alignment of COMMON variables should be
16895 used when generating code. It will be enabled by default if
16896 GCC detects that the target assembler found during configuration
16897 supports the feature.
16900 See also under @ref{i386 and x86-64 Options} for standard options.
16902 @node Xstormy16 Options
16903 @subsection Xstormy16 Options
16904 @cindex Xstormy16 Options
16906 These options are defined for Xstormy16:
16911 Choose startup files and linker script suitable for the simulator.
16914 @node Xtensa Options
16915 @subsection Xtensa Options
16916 @cindex Xtensa Options
16918 These options are supported for Xtensa targets:
16922 @itemx -mno-const16
16924 @opindex mno-const16
16925 Enable or disable use of @code{CONST16} instructions for loading
16926 constant values. The @code{CONST16} instruction is currently not a
16927 standard option from Tensilica. When enabled, @code{CONST16}
16928 instructions are always used in place of the standard @code{L32R}
16929 instructions. The use of @code{CONST16} is enabled by default only if
16930 the @code{L32R} instruction is not available.
16933 @itemx -mno-fused-madd
16934 @opindex mfused-madd
16935 @opindex mno-fused-madd
16936 Enable or disable use of fused multiply/add and multiply/subtract
16937 instructions in the floating-point option. This has no effect if the
16938 floating-point option is not also enabled. Disabling fused multiply/add
16939 and multiply/subtract instructions forces the compiler to use separate
16940 instructions for the multiply and add/subtract operations. This may be
16941 desirable in some cases where strict IEEE 754-compliant results are
16942 required: the fused multiply add/subtract instructions do not round the
16943 intermediate result, thereby producing results with @emph{more} bits of
16944 precision than specified by the IEEE standard. Disabling fused multiply
16945 add/subtract instructions also ensures that the program output is not
16946 sensitive to the compiler's ability to combine multiply and add/subtract
16949 @item -mserialize-volatile
16950 @itemx -mno-serialize-volatile
16951 @opindex mserialize-volatile
16952 @opindex mno-serialize-volatile
16953 When this option is enabled, GCC inserts @code{MEMW} instructions before
16954 @code{volatile} memory references to guarantee sequential consistency.
16955 The default is @option{-mserialize-volatile}. Use
16956 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
16958 @item -mtext-section-literals
16959 @itemx -mno-text-section-literals
16960 @opindex mtext-section-literals
16961 @opindex mno-text-section-literals
16962 Control the treatment of literal pools. The default is
16963 @option{-mno-text-section-literals}, which places literals in a separate
16964 section in the output file. This allows the literal pool to be placed
16965 in a data RAM/ROM, and it also allows the linker to combine literal
16966 pools from separate object files to remove redundant literals and
16967 improve code size. With @option{-mtext-section-literals}, the literals
16968 are interspersed in the text section in order to keep them as close as
16969 possible to their references. This may be necessary for large assembly
16972 @item -mtarget-align
16973 @itemx -mno-target-align
16974 @opindex mtarget-align
16975 @opindex mno-target-align
16976 When this option is enabled, GCC instructs the assembler to
16977 automatically align instructions to reduce branch penalties at the
16978 expense of some code density. The assembler attempts to widen density
16979 instructions to align branch targets and the instructions following call
16980 instructions. If there are not enough preceding safe density
16981 instructions to align a target, no widening will be performed. The
16982 default is @option{-mtarget-align}. These options do not affect the
16983 treatment of auto-aligned instructions like @code{LOOP}, which the
16984 assembler will always align, either by widening density instructions or
16985 by inserting no-op instructions.
16988 @itemx -mno-longcalls
16989 @opindex mlongcalls
16990 @opindex mno-longcalls
16991 When this option is enabled, GCC instructs the assembler to translate
16992 direct calls to indirect calls unless it can determine that the target
16993 of a direct call is in the range allowed by the call instruction. This
16994 translation typically occurs for calls to functions in other source
16995 files. Specifically, the assembler translates a direct @code{CALL}
16996 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
16997 The default is @option{-mno-longcalls}. This option should be used in
16998 programs where the call target can potentially be out of range. This
16999 option is implemented in the assembler, not the compiler, so the
17000 assembly code generated by GCC will still show direct call
17001 instructions---look at the disassembled object code to see the actual
17002 instructions. Note that the assembler will use an indirect call for
17003 every cross-file call, not just those that really will be out of range.
17006 @node zSeries Options
17007 @subsection zSeries Options
17008 @cindex zSeries options
17010 These are listed under @xref{S/390 and zSeries Options}.
17012 @node Code Gen Options
17013 @section Options for Code Generation Conventions
17014 @cindex code generation conventions
17015 @cindex options, code generation
17016 @cindex run-time options
17018 These machine-independent options control the interface conventions
17019 used in code generation.
17021 Most of them have both positive and negative forms; the negative form
17022 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
17023 one of the forms is listed---the one which is not the default. You
17024 can figure out the other form by either removing @samp{no-} or adding
17028 @item -fbounds-check
17029 @opindex fbounds-check
17030 For front-ends that support it, generate additional code to check that
17031 indices used to access arrays are within the declared range. This is
17032 currently only supported by the Java and Fortran front-ends, where
17033 this option defaults to true and false respectively.
17037 This option generates traps for signed overflow on addition, subtraction,
17038 multiplication operations.
17042 This option instructs the compiler to assume that signed arithmetic
17043 overflow of addition, subtraction and multiplication wraps around
17044 using twos-complement representation. This flag enables some optimizations
17045 and disables others. This option is enabled by default for the Java
17046 front-end, as required by the Java language specification.
17049 @opindex fexceptions
17050 Enable exception handling. Generates extra code needed to propagate
17051 exceptions. For some targets, this implies GCC will generate frame
17052 unwind information for all functions, which can produce significant data
17053 size overhead, although it does not affect execution. If you do not
17054 specify this option, GCC will enable it by default for languages like
17055 C++ which normally require exception handling, and disable it for
17056 languages like C that do not normally require it. However, you may need
17057 to enable this option when compiling C code that needs to interoperate
17058 properly with exception handlers written in C++. You may also wish to
17059 disable this option if you are compiling older C++ programs that don't
17060 use exception handling.
17062 @item -fnon-call-exceptions
17063 @opindex fnon-call-exceptions
17064 Generate code that allows trapping instructions to throw exceptions.
17065 Note that this requires platform-specific runtime support that does
17066 not exist everywhere. Moreover, it only allows @emph{trapping}
17067 instructions to throw exceptions, i.e.@: memory references or floating
17068 point instructions. It does not allow exceptions to be thrown from
17069 arbitrary signal handlers such as @code{SIGALRM}.
17071 @item -funwind-tables
17072 @opindex funwind-tables
17073 Similar to @option{-fexceptions}, except that it will just generate any needed
17074 static data, but will not affect the generated code in any other way.
17075 You will normally not enable this option; instead, a language processor
17076 that needs this handling would enable it on your behalf.
17078 @item -fasynchronous-unwind-tables
17079 @opindex fasynchronous-unwind-tables
17080 Generate unwind table in dwarf2 format, if supported by target machine. The
17081 table is exact at each instruction boundary, so it can be used for stack
17082 unwinding from asynchronous events (such as debugger or garbage collector).
17084 @item -fpcc-struct-return
17085 @opindex fpcc-struct-return
17086 Return ``short'' @code{struct} and @code{union} values in memory like
17087 longer ones, rather than in registers. This convention is less
17088 efficient, but it has the advantage of allowing intercallability between
17089 GCC-compiled files and files compiled with other compilers, particularly
17090 the Portable C Compiler (pcc).
17092 The precise convention for returning structures in memory depends
17093 on the target configuration macros.
17095 Short structures and unions are those whose size and alignment match
17096 that of some integer type.
17098 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
17099 switch is not binary compatible with code compiled with the
17100 @option{-freg-struct-return} switch.
17101 Use it to conform to a non-default application binary interface.
17103 @item -freg-struct-return
17104 @opindex freg-struct-return
17105 Return @code{struct} and @code{union} values in registers when possible.
17106 This is more efficient for small structures than
17107 @option{-fpcc-struct-return}.
17109 If you specify neither @option{-fpcc-struct-return} nor
17110 @option{-freg-struct-return}, GCC defaults to whichever convention is
17111 standard for the target. If there is no standard convention, GCC
17112 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
17113 the principal compiler. In those cases, we can choose the standard, and
17114 we chose the more efficient register return alternative.
17116 @strong{Warning:} code compiled with the @option{-freg-struct-return}
17117 switch is not binary compatible with code compiled with the
17118 @option{-fpcc-struct-return} switch.
17119 Use it to conform to a non-default application binary interface.
17121 @item -fshort-enums
17122 @opindex fshort-enums
17123 Allocate to an @code{enum} type only as many bytes as it needs for the
17124 declared range of possible values. Specifically, the @code{enum} type
17125 will be equivalent to the smallest integer type which has enough room.
17127 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
17128 code that is not binary compatible with code generated without that switch.
17129 Use it to conform to a non-default application binary interface.
17131 @item -fshort-double
17132 @opindex fshort-double
17133 Use the same size for @code{double} as for @code{float}.
17135 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
17136 code that is not binary compatible with code generated without that switch.
17137 Use it to conform to a non-default application binary interface.
17139 @item -fshort-wchar
17140 @opindex fshort-wchar
17141 Override the underlying type for @samp{wchar_t} to be @samp{short
17142 unsigned int} instead of the default for the target. This option is
17143 useful for building programs to run under WINE@.
17145 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
17146 code that is not binary compatible with code generated without that switch.
17147 Use it to conform to a non-default application binary interface.
17150 @opindex fno-common
17151 In C code, controls the placement of uninitialized global variables.
17152 Unix C compilers have traditionally permitted multiple definitions of
17153 such variables in different compilation units by placing the variables
17155 This is the behavior specified by @option{-fcommon}, and is the default
17156 for GCC on most targets.
17157 On the other hand, this behavior is not required by ISO C, and on some
17158 targets may carry a speed or code size penalty on variable references.
17159 The @option{-fno-common} option specifies that the compiler should place
17160 uninitialized global variables in the data section of the object file,
17161 rather than generating them as common blocks.
17162 This has the effect that if the same variable is declared
17163 (without @code{extern}) in two different compilations,
17164 you will get a multiple-definition error when you link them.
17165 In this case, you must compile with @option{-fcommon} instead.
17166 Compiling with @option{-fno-common} is useful on targets for which
17167 it provides better performance, or if you wish to verify that the
17168 program will work on other systems which always treat uninitialized
17169 variable declarations this way.
17173 Ignore the @samp{#ident} directive.
17175 @item -finhibit-size-directive
17176 @opindex finhibit-size-directive
17177 Don't output a @code{.size} assembler directive, or anything else that
17178 would cause trouble if the function is split in the middle, and the
17179 two halves are placed at locations far apart in memory. This option is
17180 used when compiling @file{crtstuff.c}; you should not need to use it
17183 @item -fverbose-asm
17184 @opindex fverbose-asm
17185 Put extra commentary information in the generated assembly code to
17186 make it more readable. This option is generally only of use to those
17187 who actually need to read the generated assembly code (perhaps while
17188 debugging the compiler itself).
17190 @option{-fno-verbose-asm}, the default, causes the
17191 extra information to be omitted and is useful when comparing two assembler
17194 @item -frecord-gcc-switches
17195 @opindex frecord-gcc-switches
17196 This switch causes the command line that was used to invoke the
17197 compiler to be recorded into the object file that is being created.
17198 This switch is only implemented on some targets and the exact format
17199 of the recording is target and binary file format dependent, but it
17200 usually takes the form of a section containing ASCII text. This
17201 switch is related to the @option{-fverbose-asm} switch, but that
17202 switch only records information in the assembler output file as
17203 comments, so it never reaches the object file.
17207 @cindex global offset table
17209 Generate position-independent code (PIC) suitable for use in a shared
17210 library, if supported for the target machine. Such code accesses all
17211 constant addresses through a global offset table (GOT)@. The dynamic
17212 loader resolves the GOT entries when the program starts (the dynamic
17213 loader is not part of GCC; it is part of the operating system). If
17214 the GOT size for the linked executable exceeds a machine-specific
17215 maximum size, you get an error message from the linker indicating that
17216 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17217 instead. (These maximums are 8k on the SPARC and 32k
17218 on the m68k and RS/6000. The 386 has no such limit.)
17220 Position-independent code requires special support, and therefore works
17221 only on certain machines. For the 386, GCC supports PIC for System V
17222 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17223 position-independent.
17225 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17230 If supported for the target machine, emit position-independent code,
17231 suitable for dynamic linking and avoiding any limit on the size of the
17232 global offset table. This option makes a difference on the m68k,
17233 PowerPC and SPARC@.
17235 Position-independent code requires special support, and therefore works
17236 only on certain machines.
17238 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17245 These options are similar to @option{-fpic} and @option{-fPIC}, but
17246 generated position independent code can be only linked into executables.
17247 Usually these options are used when @option{-pie} GCC option will be
17248 used during linking.
17250 @option{-fpie} and @option{-fPIE} both define the macros
17251 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17252 for @option{-fpie} and 2 for @option{-fPIE}.
17254 @item -fno-jump-tables
17255 @opindex fno-jump-tables
17256 Do not use jump tables for switch statements even where it would be
17257 more efficient than other code generation strategies. This option is
17258 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17259 building code which forms part of a dynamic linker and cannot
17260 reference the address of a jump table. On some targets, jump tables
17261 do not require a GOT and this option is not needed.
17263 @item -ffixed-@var{reg}
17265 Treat the register named @var{reg} as a fixed register; generated code
17266 should never refer to it (except perhaps as a stack pointer, frame
17267 pointer or in some other fixed role).
17269 @var{reg} must be the name of a register. The register names accepted
17270 are machine-specific and are defined in the @code{REGISTER_NAMES}
17271 macro in the machine description macro file.
17273 This flag does not have a negative form, because it specifies a
17276 @item -fcall-used-@var{reg}
17277 @opindex fcall-used
17278 Treat the register named @var{reg} as an allocable register that is
17279 clobbered by function calls. It may be allocated for temporaries or
17280 variables that do not live across a call. Functions compiled this way
17281 will not save and restore the register @var{reg}.
17283 It is an error to used this flag with the frame pointer or stack pointer.
17284 Use of this flag for other registers that have fixed pervasive roles in
17285 the machine's execution model will produce disastrous results.
17287 This flag does not have a negative form, because it specifies a
17290 @item -fcall-saved-@var{reg}
17291 @opindex fcall-saved
17292 Treat the register named @var{reg} as an allocable register saved by
17293 functions. It may be allocated even for temporaries or variables that
17294 live across a call. Functions compiled this way will save and restore
17295 the register @var{reg} if they use it.
17297 It is an error to used this flag with the frame pointer or stack pointer.
17298 Use of this flag for other registers that have fixed pervasive roles in
17299 the machine's execution model will produce disastrous results.
17301 A different sort of disaster will result from the use of this flag for
17302 a register in which function values may be returned.
17304 This flag does not have a negative form, because it specifies a
17307 @item -fpack-struct[=@var{n}]
17308 @opindex fpack-struct
17309 Without a value specified, pack all structure members together without
17310 holes. When a value is specified (which must be a small power of two), pack
17311 structure members according to this value, representing the maximum
17312 alignment (that is, objects with default alignment requirements larger than
17313 this will be output potentially unaligned at the next fitting location.
17315 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17316 code that is not binary compatible with code generated without that switch.
17317 Additionally, it makes the code suboptimal.
17318 Use it to conform to a non-default application binary interface.
17320 @item -finstrument-functions
17321 @opindex finstrument-functions
17322 Generate instrumentation calls for entry and exit to functions. Just
17323 after function entry and just before function exit, the following
17324 profiling functions will be called with the address of the current
17325 function and its call site. (On some platforms,
17326 @code{__builtin_return_address} does not work beyond the current
17327 function, so the call site information may not be available to the
17328 profiling functions otherwise.)
17331 void __cyg_profile_func_enter (void *this_fn,
17333 void __cyg_profile_func_exit (void *this_fn,
17337 The first argument is the address of the start of the current function,
17338 which may be looked up exactly in the symbol table.
17340 This instrumentation is also done for functions expanded inline in other
17341 functions. The profiling calls will indicate where, conceptually, the
17342 inline function is entered and exited. This means that addressable
17343 versions of such functions must be available. If all your uses of a
17344 function are expanded inline, this may mean an additional expansion of
17345 code size. If you use @samp{extern inline} in your C code, an
17346 addressable version of such functions must be provided. (This is
17347 normally the case anyways, but if you get lucky and the optimizer always
17348 expands the functions inline, you might have gotten away without
17349 providing static copies.)
17351 A function may be given the attribute @code{no_instrument_function}, in
17352 which case this instrumentation will not be done. This can be used, for
17353 example, for the profiling functions listed above, high-priority
17354 interrupt routines, and any functions from which the profiling functions
17355 cannot safely be called (perhaps signal handlers, if the profiling
17356 routines generate output or allocate memory).
17358 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17359 @opindex finstrument-functions-exclude-file-list
17361 Set the list of functions that are excluded from instrumentation (see
17362 the description of @code{-finstrument-functions}). If the file that
17363 contains a function definition matches with one of @var{file}, then
17364 that function is not instrumented. The match is done on substrings:
17365 if the @var{file} parameter is a substring of the file name, it is
17366 considered to be a match.
17369 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
17370 will exclude any inline function defined in files whose pathnames
17371 contain @code{/bits/stl} or @code{include/sys}.
17373 If, for some reason, you want to include letter @code{','} in one of
17374 @var{sym}, write @code{'\,'}. For example,
17375 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17376 (note the single quote surrounding the option).
17378 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
17379 @opindex finstrument-functions-exclude-function-list
17381 This is similar to @code{-finstrument-functions-exclude-file-list},
17382 but this option sets the list of function names to be excluded from
17383 instrumentation. The function name to be matched is its user-visible
17384 name, such as @code{vector<int> blah(const vector<int> &)}, not the
17385 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
17386 match is done on substrings: if the @var{sym} parameter is a substring
17387 of the function name, it is considered to be a match. For C99 and C++
17388 extended identifiers, the function name must be given in UTF-8, not
17389 using universal character names.
17391 @item -fstack-check
17392 @opindex fstack-check
17393 Generate code to verify that you do not go beyond the boundary of the
17394 stack. You should specify this flag if you are running in an
17395 environment with multiple threads, but only rarely need to specify it in
17396 a single-threaded environment since stack overflow is automatically
17397 detected on nearly all systems if there is only one stack.
17399 Note that this switch does not actually cause checking to be done; the
17400 operating system or the language runtime must do that. The switch causes
17401 generation of code to ensure that they see the stack being extended.
17403 You can additionally specify a string parameter: @code{no} means no
17404 checking, @code{generic} means force the use of old-style checking,
17405 @code{specific} means use the best checking method and is equivalent
17406 to bare @option{-fstack-check}.
17408 Old-style checking is a generic mechanism that requires no specific
17409 target support in the compiler but comes with the following drawbacks:
17413 Modified allocation strategy for large objects: they will always be
17414 allocated dynamically if their size exceeds a fixed threshold.
17417 Fixed limit on the size of the static frame of functions: when it is
17418 topped by a particular function, stack checking is not reliable and
17419 a warning is issued by the compiler.
17422 Inefficiency: because of both the modified allocation strategy and the
17423 generic implementation, the performances of the code are hampered.
17426 Note that old-style stack checking is also the fallback method for
17427 @code{specific} if no target support has been added in the compiler.
17429 @item -fstack-limit-register=@var{reg}
17430 @itemx -fstack-limit-symbol=@var{sym}
17431 @itemx -fno-stack-limit
17432 @opindex fstack-limit-register
17433 @opindex fstack-limit-symbol
17434 @opindex fno-stack-limit
17435 Generate code to ensure that the stack does not grow beyond a certain value,
17436 either the value of a register or the address of a symbol. If the stack
17437 would grow beyond the value, a signal is raised. For most targets,
17438 the signal is raised before the stack overruns the boundary, so
17439 it is possible to catch the signal without taking special precautions.
17441 For instance, if the stack starts at absolute address @samp{0x80000000}
17442 and grows downwards, you can use the flags
17443 @option{-fstack-limit-symbol=__stack_limit} and
17444 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
17445 of 128KB@. Note that this may only work with the GNU linker.
17447 @cindex aliasing of parameters
17448 @cindex parameters, aliased
17449 @item -fargument-alias
17450 @itemx -fargument-noalias
17451 @itemx -fargument-noalias-global
17452 @itemx -fargument-noalias-anything
17453 @opindex fargument-alias
17454 @opindex fargument-noalias
17455 @opindex fargument-noalias-global
17456 @opindex fargument-noalias-anything
17457 Specify the possible relationships among parameters and between
17458 parameters and global data.
17460 @option{-fargument-alias} specifies that arguments (parameters) may
17461 alias each other and may alias global storage.@*
17462 @option{-fargument-noalias} specifies that arguments do not alias
17463 each other, but may alias global storage.@*
17464 @option{-fargument-noalias-global} specifies that arguments do not
17465 alias each other and do not alias global storage.
17466 @option{-fargument-noalias-anything} specifies that arguments do not
17467 alias any other storage.
17469 Each language will automatically use whatever option is required by
17470 the language standard. You should not need to use these options yourself.
17472 @item -fleading-underscore
17473 @opindex fleading-underscore
17474 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
17475 change the way C symbols are represented in the object file. One use
17476 is to help link with legacy assembly code.
17478 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
17479 generate code that is not binary compatible with code generated without that
17480 switch. Use it to conform to a non-default application binary interface.
17481 Not all targets provide complete support for this switch.
17483 @item -ftls-model=@var{model}
17484 @opindex ftls-model
17485 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
17486 The @var{model} argument should be one of @code{global-dynamic},
17487 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
17489 The default without @option{-fpic} is @code{initial-exec}; with
17490 @option{-fpic} the default is @code{global-dynamic}.
17492 @item -fvisibility=@var{default|internal|hidden|protected}
17493 @opindex fvisibility
17494 Set the default ELF image symbol visibility to the specified option---all
17495 symbols will be marked with this unless overridden within the code.
17496 Using this feature can very substantially improve linking and
17497 load times of shared object libraries, produce more optimized
17498 code, provide near-perfect API export and prevent symbol clashes.
17499 It is @strong{strongly} recommended that you use this in any shared objects
17502 Despite the nomenclature, @code{default} always means public ie;
17503 available to be linked against from outside the shared object.
17504 @code{protected} and @code{internal} are pretty useless in real-world
17505 usage so the only other commonly used option will be @code{hidden}.
17506 The default if @option{-fvisibility} isn't specified is
17507 @code{default}, i.e., make every
17508 symbol public---this causes the same behavior as previous versions of
17511 A good explanation of the benefits offered by ensuring ELF
17512 symbols have the correct visibility is given by ``How To Write
17513 Shared Libraries'' by Ulrich Drepper (which can be found at
17514 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
17515 solution made possible by this option to marking things hidden when
17516 the default is public is to make the default hidden and mark things
17517 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
17518 and @code{__attribute__ ((visibility("default")))} instead of
17519 @code{__declspec(dllexport)} you get almost identical semantics with
17520 identical syntax. This is a great boon to those working with
17521 cross-platform projects.
17523 For those adding visibility support to existing code, you may find
17524 @samp{#pragma GCC visibility} of use. This works by you enclosing
17525 the declarations you wish to set visibility for with (for example)
17526 @samp{#pragma GCC visibility push(hidden)} and
17527 @samp{#pragma GCC visibility pop}.
17528 Bear in mind that symbol visibility should be viewed @strong{as
17529 part of the API interface contract} and thus all new code should
17530 always specify visibility when it is not the default ie; declarations
17531 only for use within the local DSO should @strong{always} be marked explicitly
17532 as hidden as so to avoid PLT indirection overheads---making this
17533 abundantly clear also aids readability and self-documentation of the code.
17534 Note that due to ISO C++ specification requirements, operator new and
17535 operator delete must always be of default visibility.
17537 Be aware that headers from outside your project, in particular system
17538 headers and headers from any other library you use, may not be
17539 expecting to be compiled with visibility other than the default. You
17540 may need to explicitly say @samp{#pragma GCC visibility push(default)}
17541 before including any such headers.
17543 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
17544 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
17545 no modifications. However, this means that calls to @samp{extern}
17546 functions with no explicit visibility will use the PLT, so it is more
17547 effective to use @samp{__attribute ((visibility))} and/or
17548 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
17549 declarations should be treated as hidden.
17551 Note that @samp{-fvisibility} does affect C++ vague linkage
17552 entities. This means that, for instance, an exception class that will
17553 be thrown between DSOs must be explicitly marked with default
17554 visibility so that the @samp{type_info} nodes will be unified between
17557 An overview of these techniques, their benefits and how to use them
17558 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
17564 @node Environment Variables
17565 @section Environment Variables Affecting GCC
17566 @cindex environment variables
17568 @c man begin ENVIRONMENT
17569 This section describes several environment variables that affect how GCC
17570 operates. Some of them work by specifying directories or prefixes to use
17571 when searching for various kinds of files. Some are used to specify other
17572 aspects of the compilation environment.
17574 Note that you can also specify places to search using options such as
17575 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
17576 take precedence over places specified using environment variables, which
17577 in turn take precedence over those specified by the configuration of GCC@.
17578 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
17579 GNU Compiler Collection (GCC) Internals}.
17584 @c @itemx LC_COLLATE
17586 @c @itemx LC_MONETARY
17587 @c @itemx LC_NUMERIC
17592 @c @findex LC_COLLATE
17593 @findex LC_MESSAGES
17594 @c @findex LC_MONETARY
17595 @c @findex LC_NUMERIC
17599 These environment variables control the way that GCC uses
17600 localization information that allow GCC to work with different
17601 national conventions. GCC inspects the locale categories
17602 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
17603 so. These locale categories can be set to any value supported by your
17604 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
17605 Kingdom encoded in UTF-8.
17607 The @env{LC_CTYPE} environment variable specifies character
17608 classification. GCC uses it to determine the character boundaries in
17609 a string; this is needed for some multibyte encodings that contain quote
17610 and escape characters that would otherwise be interpreted as a string
17613 The @env{LC_MESSAGES} environment variable specifies the language to
17614 use in diagnostic messages.
17616 If the @env{LC_ALL} environment variable is set, it overrides the value
17617 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17618 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17619 environment variable. If none of these variables are set, GCC
17620 defaults to traditional C English behavior.
17624 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17625 files. GCC uses temporary files to hold the output of one stage of
17626 compilation which is to be used as input to the next stage: for example,
17627 the output of the preprocessor, which is the input to the compiler
17630 @item GCC_EXEC_PREFIX
17631 @findex GCC_EXEC_PREFIX
17632 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17633 names of the subprograms executed by the compiler. No slash is added
17634 when this prefix is combined with the name of a subprogram, but you can
17635 specify a prefix that ends with a slash if you wish.
17637 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17638 an appropriate prefix to use based on the pathname it was invoked with.
17640 If GCC cannot find the subprogram using the specified prefix, it
17641 tries looking in the usual places for the subprogram.
17643 The default value of @env{GCC_EXEC_PREFIX} is
17644 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17645 the installed compiler. In many cases @var{prefix} is the value
17646 of @code{prefix} when you ran the @file{configure} script.
17648 Other prefixes specified with @option{-B} take precedence over this prefix.
17650 This prefix is also used for finding files such as @file{crt0.o} that are
17653 In addition, the prefix is used in an unusual way in finding the
17654 directories to search for header files. For each of the standard
17655 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17656 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17657 replacing that beginning with the specified prefix to produce an
17658 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17659 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17660 These alternate directories are searched first; the standard directories
17661 come next. If a standard directory begins with the configured
17662 @var{prefix} then the value of @var{prefix} is replaced by
17663 @env{GCC_EXEC_PREFIX} when looking for header files.
17665 @item COMPILER_PATH
17666 @findex COMPILER_PATH
17667 The value of @env{COMPILER_PATH} is a colon-separated list of
17668 directories, much like @env{PATH}. GCC tries the directories thus
17669 specified when searching for subprograms, if it can't find the
17670 subprograms using @env{GCC_EXEC_PREFIX}.
17673 @findex LIBRARY_PATH
17674 The value of @env{LIBRARY_PATH} is a colon-separated list of
17675 directories, much like @env{PATH}. When configured as a native compiler,
17676 GCC tries the directories thus specified when searching for special
17677 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17678 using GCC also uses these directories when searching for ordinary
17679 libraries for the @option{-l} option (but directories specified with
17680 @option{-L} come first).
17684 @cindex locale definition
17685 This variable is used to pass locale information to the compiler. One way in
17686 which this information is used is to determine the character set to be used
17687 when character literals, string literals and comments are parsed in C and C++.
17688 When the compiler is configured to allow multibyte characters,
17689 the following values for @env{LANG} are recognized:
17693 Recognize JIS characters.
17695 Recognize SJIS characters.
17697 Recognize EUCJP characters.
17700 If @env{LANG} is not defined, or if it has some other value, then the
17701 compiler will use mblen and mbtowc as defined by the default locale to
17702 recognize and translate multibyte characters.
17706 Some additional environments variables affect the behavior of the
17709 @include cppenv.texi
17713 @node Precompiled Headers
17714 @section Using Precompiled Headers
17715 @cindex precompiled headers
17716 @cindex speed of compilation
17718 Often large projects have many header files that are included in every
17719 source file. The time the compiler takes to process these header files
17720 over and over again can account for nearly all of the time required to
17721 build the project. To make builds faster, GCC allows users to
17722 `precompile' a header file; then, if builds can use the precompiled
17723 header file they will be much faster.
17725 To create a precompiled header file, simply compile it as you would any
17726 other file, if necessary using the @option{-x} option to make the driver
17727 treat it as a C or C++ header file. You will probably want to use a
17728 tool like @command{make} to keep the precompiled header up-to-date when
17729 the headers it contains change.
17731 A precompiled header file will be searched for when @code{#include} is
17732 seen in the compilation. As it searches for the included file
17733 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17734 compiler looks for a precompiled header in each directory just before it
17735 looks for the include file in that directory. The name searched for is
17736 the name specified in the @code{#include} with @samp{.gch} appended. If
17737 the precompiled header file can't be used, it is ignored.
17739 For instance, if you have @code{#include "all.h"}, and you have
17740 @file{all.h.gch} in the same directory as @file{all.h}, then the
17741 precompiled header file will be used if possible, and the original
17742 header will be used otherwise.
17744 Alternatively, you might decide to put the precompiled header file in a
17745 directory and use @option{-I} to ensure that directory is searched
17746 before (or instead of) the directory containing the original header.
17747 Then, if you want to check that the precompiled header file is always
17748 used, you can put a file of the same name as the original header in this
17749 directory containing an @code{#error} command.
17751 This also works with @option{-include}. So yet another way to use
17752 precompiled headers, good for projects not designed with precompiled
17753 header files in mind, is to simply take most of the header files used by
17754 a project, include them from another header file, precompile that header
17755 file, and @option{-include} the precompiled header. If the header files
17756 have guards against multiple inclusion, they will be skipped because
17757 they've already been included (in the precompiled header).
17759 If you need to precompile the same header file for different
17760 languages, targets, or compiler options, you can instead make a
17761 @emph{directory} named like @file{all.h.gch}, and put each precompiled
17762 header in the directory, perhaps using @option{-o}. It doesn't matter
17763 what you call the files in the directory, every precompiled header in
17764 the directory will be considered. The first precompiled header
17765 encountered in the directory that is valid for this compilation will
17766 be used; they're searched in no particular order.
17768 There are many other possibilities, limited only by your imagination,
17769 good sense, and the constraints of your build system.
17771 A precompiled header file can be used only when these conditions apply:
17775 Only one precompiled header can be used in a particular compilation.
17778 A precompiled header can't be used once the first C token is seen. You
17779 can have preprocessor directives before a precompiled header; you can
17780 even include a precompiled header from inside another header, so long as
17781 there are no C tokens before the @code{#include}.
17784 The precompiled header file must be produced for the same language as
17785 the current compilation. You can't use a C precompiled header for a C++
17789 The precompiled header file must have been produced by the same compiler
17790 binary as the current compilation is using.
17793 Any macros defined before the precompiled header is included must
17794 either be defined in the same way as when the precompiled header was
17795 generated, or must not affect the precompiled header, which usually
17796 means that they don't appear in the precompiled header at all.
17798 The @option{-D} option is one way to define a macro before a
17799 precompiled header is included; using a @code{#define} can also do it.
17800 There are also some options that define macros implicitly, like
17801 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
17804 @item If debugging information is output when using the precompiled
17805 header, using @option{-g} or similar, the same kind of debugging information
17806 must have been output when building the precompiled header. However,
17807 a precompiled header built using @option{-g} can be used in a compilation
17808 when no debugging information is being output.
17810 @item The same @option{-m} options must generally be used when building
17811 and using the precompiled header. @xref{Submodel Options},
17812 for any cases where this rule is relaxed.
17814 @item Each of the following options must be the same when building and using
17815 the precompiled header:
17817 @gccoptlist{-fexceptions}
17820 Some other command-line options starting with @option{-f},
17821 @option{-p}, or @option{-O} must be defined in the same way as when
17822 the precompiled header was generated. At present, it's not clear
17823 which options are safe to change and which are not; the safest choice
17824 is to use exactly the same options when generating and using the
17825 precompiled header. The following are known to be safe:
17827 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
17828 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
17829 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
17834 For all of these except the last, the compiler will automatically
17835 ignore the precompiled header if the conditions aren't met. If you
17836 find an option combination that doesn't work and doesn't cause the
17837 precompiled header to be ignored, please consider filing a bug report,
17840 If you do use differing options when generating and using the
17841 precompiled header, the actual behavior will be a mixture of the
17842 behavior for the options. For instance, if you use @option{-g} to
17843 generate the precompiled header but not when using it, you may or may
17844 not get debugging information for routines in the precompiled header.