1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
156 @section Option Summary
158 Here is a summary of all the options, grouped by type. Explanations are
159 in the following sections.
162 @item Overall Options
163 @xref{Overall Options,,Options Controlling the Kind of Output}.
164 @gccoptlist{-c -S -E -o @var{file} -combine -no-canonical-prefixes @gol
165 -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
167 --version -wrapper@@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -fno-pretty-templates @gol
193 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
194 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
195 -fno-default-inline -fvisibility-inlines-hidden @gol
196 -fvisibility-ms-compat @gol
197 -Wabi -Wctor-dtor-privacy @gol
198 -Wnon-virtual-dtor -Wreorder @gol
199 -Weffc++ -Wstrict-null-sentinel @gol
200 -Wno-non-template-friend -Wold-style-cast @gol
201 -Woverloaded-virtual -Wno-pmf-conversions @gol
204 @item Objective-C and Objective-C++ Language Options
205 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
206 Objective-C and Objective-C++ Dialects}.
207 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
208 -fgnu-runtime -fnext-runtime @gol
209 -fno-nil-receivers @gol
210 -fobjc-call-cxx-cdtors @gol
211 -fobjc-direct-dispatch @gol
212 -fobjc-exceptions @gol
214 -freplace-objc-classes @gol
217 -Wassign-intercept @gol
218 -Wno-protocol -Wselector @gol
219 -Wstrict-selector-match @gol
220 -Wundeclared-selector}
222 @item Language Independent Options
223 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
224 @gccoptlist{-fmessage-length=@var{n} @gol
225 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
226 -fdiagnostics-show-option}
228 @item Warning Options
229 @xref{Warning Options,,Options to Request or Suppress Warnings}.
230 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
231 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
232 -Wno-attributes -Wno-builtin-macro-redefined @gol
233 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
234 -Wchar-subscripts -Wclobbered -Wcomment @gol
235 -Wconversion -Wcoverage-mismatch -Wno-deprecated @gol
236 -Wno-deprecated-declarations -Wdisabled-optimization @gol
237 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
238 -Werror -Werror=* @gol
239 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
240 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
241 -Wformat-security -Wformat-y2k @gol
242 -Wframe-larger-than=@var{len} -Wjump-misses-init -Wignored-qualifiers @gol
243 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
244 -Winit-self -Winline @gol
245 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
246 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
247 -Wlogical-op -Wlong-long @gol
248 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
249 -Wmissing-format-attribute -Wmissing-include-dirs @gol
250 -Wmissing-noreturn -Wno-mudflap @gol
251 -Wno-multichar -Wnonnull -Wno-overflow @gol
252 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
253 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
254 -Wpointer-arith -Wno-pointer-to-int-cast @gol
255 -Wredundant-decls @gol
256 -Wreturn-type -Wsequence-point -Wshadow @gol
257 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
258 -Wstrict-aliasing -Wstrict-aliasing=n @gol
259 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
260 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
261 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
262 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
263 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
264 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value -Wunused-variable @gol
265 -Wvariadic-macros -Wvla @gol
266 -Wvolatile-register-var -Wwrite-strings}
268 @item C and Objective-C-only Warning Options
269 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
270 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
271 -Wold-style-declaration -Wold-style-definition @gol
272 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
273 -Wdeclaration-after-statement -Wpointer-sign}
275 @item Debugging Options
276 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
277 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
278 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
279 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
280 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
281 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
282 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
283 -fdump-statistics @gol
285 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
289 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-nrv -fdump-tree-vect @gol
299 -fdump-tree-sink @gol
300 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
301 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
302 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
303 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
304 -ftree-vectorizer-verbose=@var{n} @gol
305 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
306 -fdump-final-insns=@var{file} @gol
307 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
308 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
309 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
310 -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-assigments -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 @gol
371 -fsched2-use-traces -fsched-pressure @gol
372 -fsched-spec-load -fsched-spec-load-dangerous @gol
373 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
374 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
375 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
376 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
377 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
378 -fselective-scheduling -fselective-scheduling2 @gol
379 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
380 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
381 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
382 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
383 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
384 -ftree-copyrename -ftree-dce @gol
385 -ftree-dominator-opts -ftree-dse -ftree-forwprop -ftree-fre -ftree-loop-im @gol
386 -ftree-phiprop -ftree-loop-distribution @gol
387 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
388 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
389 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
390 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
391 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
392 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
393 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
394 -fwhole-program -fwhopr -fwpa -fuse-linker-plugin @gol
395 --param @var{name}=@var{value}
396 -O -O0 -O1 -O2 -O3 -Os}
398 @item Preprocessor Options
399 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
400 @gccoptlist{-A@var{question}=@var{answer} @gol
401 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
402 -C -dD -dI -dM -dN @gol
403 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
404 -idirafter @var{dir} @gol
405 -include @var{file} -imacros @var{file} @gol
406 -iprefix @var{file} -iwithprefix @var{dir} @gol
407 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
408 -imultilib @var{dir} -isysroot @var{dir} @gol
409 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
410 -P -fworking-directory -remap @gol
411 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
412 -Xpreprocessor @var{option}}
414 @item Assembler Option
415 @xref{Assembler Options,,Passing Options to the Assembler}.
416 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
419 @xref{Link Options,,Options for Linking}.
420 @gccoptlist{@var{object-file-name} -l@var{library} @gol
421 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
422 -s -static -static-libgcc -static-libstdc++ -shared @gol
423 -shared-libgcc -symbolic @gol
424 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
427 @item Directory Options
428 @xref{Directory Options,,Options for Directory Search}.
429 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
430 -specs=@var{file} -I- --sysroot=@var{dir}}
433 @c I wrote this xref this way to avoid overfull hbox. -- rms
434 @xref{Target Options}.
435 @gccoptlist{-V @var{version} -b @var{machine}}
437 @item Machine Dependent Options
438 @xref{Submodel Options,,Hardware Models and Configurations}.
439 @c This list is ordered alphanumerically by subsection name.
440 @c Try and put the significant identifier (CPU or system) first,
441 @c so users have a clue at guessing where the ones they want will be.
444 @gccoptlist{-EB -EL @gol
445 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
446 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
449 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
450 -mabi=@var{name} @gol
451 -mapcs-stack-check -mno-apcs-stack-check @gol
452 -mapcs-float -mno-apcs-float @gol
453 -mapcs-reentrant -mno-apcs-reentrant @gol
454 -msched-prolog -mno-sched-prolog @gol
455 -mlittle-endian -mbig-endian -mwords-little-endian @gol
456 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
457 -mfp16-format=@var{name}
458 -mthumb-interwork -mno-thumb-interwork @gol
459 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
460 -mstructure-size-boundary=@var{n} @gol
461 -mabort-on-noreturn @gol
462 -mlong-calls -mno-long-calls @gol
463 -msingle-pic-base -mno-single-pic-base @gol
464 -mpic-register=@var{reg} @gol
465 -mnop-fun-dllimport @gol
466 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
467 -mpoke-function-name @gol
469 -mtpcs-frame -mtpcs-leaf-frame @gol
470 -mcaller-super-interworking -mcallee-super-interworking @gol
472 -mword-relocations @gol
473 -mfix-cortex-m3-ldrd}
476 @gccoptlist{-mmcu=@var{mcu} -mno-interrupts @gol
477 -mcall-prologues -mtiny-stack -mint8}
479 @emph{Blackfin Options}
480 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
481 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
482 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
483 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
484 -mno-id-shared-library -mshared-library-id=@var{n} @gol
485 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
486 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
487 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
491 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
492 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
493 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
494 -mstack-align -mdata-align -mconst-align @gol
495 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
496 -melf -maout -melinux -mlinux -sim -sim2 @gol
497 -mmul-bug-workaround -mno-mul-bug-workaround}
500 @gccoptlist{-mmac -mpush-args}
502 @emph{Darwin Options}
503 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
504 -arch_only -bind_at_load -bundle -bundle_loader @gol
505 -client_name -compatibility_version -current_version @gol
507 -dependency-file -dylib_file -dylinker_install_name @gol
508 -dynamic -dynamiclib -exported_symbols_list @gol
509 -filelist -flat_namespace -force_cpusubtype_ALL @gol
510 -force_flat_namespace -headerpad_max_install_names @gol
512 -image_base -init -install_name -keep_private_externs @gol
513 -multi_module -multiply_defined -multiply_defined_unused @gol
514 -noall_load -no_dead_strip_inits_and_terms @gol
515 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
516 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
517 -private_bundle -read_only_relocs -sectalign @gol
518 -sectobjectsymbols -whyload -seg1addr @gol
519 -sectcreate -sectobjectsymbols -sectorder @gol
520 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
521 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
522 -segprot -segs_read_only_addr -segs_read_write_addr @gol
523 -single_module -static -sub_library -sub_umbrella @gol
524 -twolevel_namespace -umbrella -undefined @gol
525 -unexported_symbols_list -weak_reference_mismatches @gol
526 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
527 -mkernel -mone-byte-bool}
529 @emph{DEC Alpha Options}
530 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
531 -mieee -mieee-with-inexact -mieee-conformant @gol
532 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
533 -mtrap-precision=@var{mode} -mbuild-constants @gol
534 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
535 -mbwx -mmax -mfix -mcix @gol
536 -mfloat-vax -mfloat-ieee @gol
537 -mexplicit-relocs -msmall-data -mlarge-data @gol
538 -msmall-text -mlarge-text @gol
539 -mmemory-latency=@var{time}}
541 @emph{DEC Alpha/VMS Options}
542 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
545 @gccoptlist{-msmall-model -mno-lsim}
548 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
549 -mhard-float -msoft-float @gol
550 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
551 -mdouble -mno-double @gol
552 -mmedia -mno-media -mmuladd -mno-muladd @gol
553 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
554 -mlinked-fp -mlong-calls -malign-labels @gol
555 -mlibrary-pic -macc-4 -macc-8 @gol
556 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
557 -moptimize-membar -mno-optimize-membar @gol
558 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
559 -mvliw-branch -mno-vliw-branch @gol
560 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
561 -mno-nested-cond-exec -mtomcat-stats @gol
565 @emph{GNU/Linux Options}
566 @gccoptlist{-muclibc}
568 @emph{H8/300 Options}
569 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
572 @gccoptlist{-march=@var{architecture-type} @gol
573 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
574 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
575 -mfixed-range=@var{register-range} @gol
576 -mjump-in-delay -mlinker-opt -mlong-calls @gol
577 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
578 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
579 -mno-jump-in-delay -mno-long-load-store @gol
580 -mno-portable-runtime -mno-soft-float @gol
581 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
582 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
583 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
584 -munix=@var{unix-std} -nolibdld -static -threads}
586 @emph{i386 and x86-64 Options}
587 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
588 -mfpmath=@var{unit} @gol
589 -masm=@var{dialect} -mno-fancy-math-387 @gol
590 -mno-fp-ret-in-387 -msoft-float @gol
591 -mno-wide-multiply -mrtd -malign-double @gol
592 -mpreferred-stack-boundary=@var{num}
593 -mincoming-stack-boundary=@var{num}
594 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip @gol
595 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
597 -msse4a -m3dnow -mpopcnt -mabm -mfma4 @gol
598 -mthreads -mno-align-stringops -minline-all-stringops @gol
599 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
600 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
601 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
602 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
603 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
604 -mcmodel=@var{code-model} -mabi=@var{name} @gol
605 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
609 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
610 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
611 -mconstant-gp -mauto-pic -mfused-madd @gol
612 -minline-float-divide-min-latency @gol
613 -minline-float-divide-max-throughput @gol
614 -mno-inline-float-divide @gol
615 -minline-int-divide-min-latency @gol
616 -minline-int-divide-max-throughput @gol
617 -mno-inline-int-divide @gol
618 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
619 -mno-inline-sqrt @gol
620 -mdwarf2-asm -mearly-stop-bits @gol
621 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
622 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
623 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
624 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
625 -msched-spec-ldc -msched-spec-control-ldc @gol
626 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
627 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
628 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
629 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
631 @emph{IA-64/VMS Options}
632 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
634 @emph{M32R/D Options}
635 @gccoptlist{-m32r2 -m32rx -m32r @gol
637 -malign-loops -mno-align-loops @gol
638 -missue-rate=@var{number} @gol
639 -mbranch-cost=@var{number} @gol
640 -mmodel=@var{code-size-model-type} @gol
641 -msdata=@var{sdata-type} @gol
642 -mno-flush-func -mflush-func=@var{name} @gol
643 -mno-flush-trap -mflush-trap=@var{number} @gol
647 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
649 @emph{M680x0 Options}
650 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
651 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
652 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
653 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
654 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
655 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
656 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
657 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
660 @emph{M68hc1x Options}
661 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
662 -mauto-incdec -minmax -mlong-calls -mshort @gol
663 -msoft-reg-count=@var{count}}
666 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
667 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
668 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
669 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
670 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
673 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
674 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
675 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
676 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
680 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
681 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
682 -mips64 -mips64r2 @gol
683 -mips16 -mno-mips16 -mflip-mips16 @gol
684 -minterlink-mips16 -mno-interlink-mips16 @gol
685 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
686 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
687 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
688 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
689 -mfpu=@var{fpu-type} @gol
690 -msmartmips -mno-smartmips @gol
691 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
692 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
693 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
694 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
695 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
696 -membedded-data -mno-embedded-data @gol
697 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
698 -mcode-readable=@var{setting} @gol
699 -msplit-addresses -mno-split-addresses @gol
700 -mexplicit-relocs -mno-explicit-relocs @gol
701 -mcheck-zero-division -mno-check-zero-division @gol
702 -mdivide-traps -mdivide-breaks @gol
703 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
704 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
705 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
706 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
707 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
708 -mflush-func=@var{func} -mno-flush-func @gol
709 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
710 -mfp-exceptions -mno-fp-exceptions @gol
711 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
712 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
715 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
716 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
717 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
718 -mno-base-addresses -msingle-exit -mno-single-exit}
720 @emph{MN10300 Options}
721 @gccoptlist{-mmult-bug -mno-mult-bug @gol
722 -mam33 -mno-am33 @gol
723 -mam33-2 -mno-am33-2 @gol
724 -mreturn-pointer-on-d0 @gol
727 @emph{PDP-11 Options}
728 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
729 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
730 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
731 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
732 -mbranch-expensive -mbranch-cheap @gol
733 -msplit -mno-split -munix-asm -mdec-asm}
735 @emph{picoChip Options}
736 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
737 -msymbol-as-address -mno-inefficient-warnings}
739 @emph{PowerPC Options}
740 See RS/6000 and PowerPC Options.
742 @emph{RS/6000 and PowerPC Options}
743 @gccoptlist{-mcpu=@var{cpu-type} @gol
744 -mtune=@var{cpu-type} @gol
745 -mpower -mno-power -mpower2 -mno-power2 @gol
746 -mpowerpc -mpowerpc64 -mno-powerpc @gol
747 -maltivec -mno-altivec @gol
748 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
749 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
750 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
751 -mfprnd -mno-fprnd @gol
752 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
753 -mnew-mnemonics -mold-mnemonics @gol
754 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
755 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
756 -malign-power -malign-natural @gol
757 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
758 -msingle-float -mdouble-float -msimple-fpu @gol
759 -mstring -mno-string -mupdate -mno-update @gol
760 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
761 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
762 -mstrict-align -mno-strict-align -mrelocatable @gol
763 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
764 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
765 -mdynamic-no-pic -maltivec -mswdiv @gol
766 -mprioritize-restricted-insns=@var{priority} @gol
767 -msched-costly-dep=@var{dependence_type} @gol
768 -minsert-sched-nops=@var{scheme} @gol
769 -mcall-sysv -mcall-netbsd @gol
770 -maix-struct-return -msvr4-struct-return @gol
771 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
772 -misel -mno-isel @gol
773 -misel=yes -misel=no @gol
775 -mspe=yes -mspe=no @gol
777 -mgen-cell-microcode -mwarn-cell-microcode @gol
778 -mvrsave -mno-vrsave @gol
779 -mmulhw -mno-mulhw @gol
780 -mdlmzb -mno-dlmzb @gol
781 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
782 -mprototype -mno-prototype @gol
783 -msim -mmvme -mads -myellowknife -memb -msdata @gol
784 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
787 @gccoptlist{-m64bit-doubles -m32bit-doubles -mieee -mno-ieee@gol
788 -mbig-endian-data -mlittle-endian-data @gol
791 -mas100-syntax -mno-as100-syntax@gol
793 -mmax-constant-size=@gol
796 @emph{S/390 and zSeries Options}
797 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
798 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
799 -mlong-double-64 -mlong-double-128 @gol
800 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
801 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
802 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
803 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
804 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
807 @gccoptlist{-meb -mel @gol
811 -mscore5 -mscore5u -mscore7 -mscore7d}
814 @gccoptlist{-m1 -m2 -m2e @gol
815 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
817 -m4-nofpu -m4-single-only -m4-single -m4 @gol
818 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
819 -m5-64media -m5-64media-nofpu @gol
820 -m5-32media -m5-32media-nofpu @gol
821 -m5-compact -m5-compact-nofpu @gol
822 -mb -ml -mdalign -mrelax @gol
823 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
824 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
825 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
826 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
827 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
831 @gccoptlist{-mcpu=@var{cpu-type} @gol
832 -mtune=@var{cpu-type} @gol
833 -mcmodel=@var{code-model} @gol
834 -m32 -m64 -mapp-regs -mno-app-regs @gol
835 -mfaster-structs -mno-faster-structs @gol
836 -mfpu -mno-fpu -mhard-float -msoft-float @gol
837 -mhard-quad-float -msoft-quad-float @gol
838 -mimpure-text -mno-impure-text -mlittle-endian @gol
839 -mstack-bias -mno-stack-bias @gol
840 -munaligned-doubles -mno-unaligned-doubles @gol
841 -mv8plus -mno-v8plus -mvis -mno-vis
842 -threads -pthreads -pthread}
845 @gccoptlist{-mwarn-reloc -merror-reloc @gol
846 -msafe-dma -munsafe-dma @gol
848 -msmall-mem -mlarge-mem -mstdmain @gol
849 -mfixed-range=@var{register-range} @gol
851 -maddress-space-conversion -mno-address-space-conversion @gol
852 -mcache-size=@var{cache-size} @gol
853 -matomic-updates -mno-atomic-updates}
855 @emph{System V Options}
856 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
859 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
860 -mprolog-function -mno-prolog-function -mspace @gol
861 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
862 -mapp-regs -mno-app-regs @gol
863 -mdisable-callt -mno-disable-callt @gol
869 @gccoptlist{-mg -mgnu -munix}
871 @emph{VxWorks Options}
872 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
873 -Xbind-lazy -Xbind-now}
875 @emph{x86-64 Options}
876 See i386 and x86-64 Options.
878 @emph{i386 and x86-64 Windows Options}
879 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
880 -mnop-fun-dllimport -mthread -municode -mwin32 -mwindows
881 -fno-set-stack-executable}
883 @emph{Xstormy16 Options}
886 @emph{Xtensa Options}
887 @gccoptlist{-mconst16 -mno-const16 @gol
888 -mfused-madd -mno-fused-madd @gol
889 -mserialize-volatile -mno-serialize-volatile @gol
890 -mtext-section-literals -mno-text-section-literals @gol
891 -mtarget-align -mno-target-align @gol
892 -mlongcalls -mno-longcalls}
894 @emph{zSeries Options}
895 See S/390 and zSeries Options.
897 @item Code Generation Options
898 @xref{Code Gen Options,,Options for Code Generation Conventions}.
899 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
900 -ffixed-@var{reg} -fexceptions @gol
901 -fnon-call-exceptions -funwind-tables @gol
902 -fasynchronous-unwind-tables @gol
903 -finhibit-size-directive -finstrument-functions @gol
904 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
905 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
906 -fno-common -fno-ident @gol
907 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
908 -fno-jump-tables @gol
909 -frecord-gcc-switches @gol
910 -freg-struct-return -fshort-enums @gol
911 -fshort-double -fshort-wchar @gol
912 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
913 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
914 -fno-stack-limit -fargument-alias -fargument-noalias @gol
915 -fargument-noalias-global -fargument-noalias-anything @gol
916 -fleading-underscore -ftls-model=@var{model} @gol
917 -ftrapv -fwrapv -fbounds-check @gol
922 * Overall Options:: Controlling the kind of output:
923 an executable, object files, assembler files,
924 or preprocessed source.
925 * C Dialect Options:: Controlling the variant of C language compiled.
926 * C++ Dialect Options:: Variations on C++.
927 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
929 * Language Independent Options:: Controlling how diagnostics should be
931 * Warning Options:: How picky should the compiler be?
932 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
933 * Optimize Options:: How much optimization?
934 * Preprocessor Options:: Controlling header files and macro definitions.
935 Also, getting dependency information for Make.
936 * Assembler Options:: Passing options to the assembler.
937 * Link Options:: Specifying libraries and so on.
938 * Directory Options:: Where to find header files and libraries.
939 Where to find the compiler executable files.
940 * Spec Files:: How to pass switches to sub-processes.
941 * Target Options:: Running a cross-compiler, or an old version of GCC.
944 @node Overall Options
945 @section Options Controlling the Kind of Output
947 Compilation can involve up to four stages: preprocessing, compilation
948 proper, assembly and linking, always in that order. GCC is capable of
949 preprocessing and compiling several files either into several
950 assembler input files, or into one assembler input file; then each
951 assembler input file produces an object file, and linking combines all
952 the object files (those newly compiled, and those specified as input)
953 into an executable file.
955 @cindex file name suffix
956 For any given input file, the file name suffix determines what kind of
961 C source code which must be preprocessed.
964 C source code which should not be preprocessed.
967 C++ source code which should not be preprocessed.
970 Objective-C source code. Note that you must link with the @file{libobjc}
971 library to make an Objective-C program work.
974 Objective-C source code which should not be preprocessed.
978 Objective-C++ source code. Note that you must link with the @file{libobjc}
979 library to make an Objective-C++ program work. Note that @samp{.M} refers
980 to a literal capital M@.
983 Objective-C++ source code which should not be preprocessed.
986 C, C++, Objective-C or Objective-C++ header file to be turned into a
991 @itemx @var{file}.cxx
992 @itemx @var{file}.cpp
993 @itemx @var{file}.CPP
994 @itemx @var{file}.c++
996 C++ source code which must be preprocessed. Note that in @samp{.cxx},
997 the last two letters must both be literally @samp{x}. Likewise,
998 @samp{.C} refers to a literal capital C@.
1002 Objective-C++ source code which must be preprocessed.
1004 @item @var{file}.mii
1005 Objective-C++ source code which should not be preprocessed.
1009 @itemx @var{file}.hp
1010 @itemx @var{file}.hxx
1011 @itemx @var{file}.hpp
1012 @itemx @var{file}.HPP
1013 @itemx @var{file}.h++
1014 @itemx @var{file}.tcc
1015 C++ header file to be turned into a precompiled header.
1018 @itemx @var{file}.for
1019 @itemx @var{file}.ftn
1020 Fixed form Fortran source code which should not be preprocessed.
1023 @itemx @var{file}.FOR
1024 @itemx @var{file}.fpp
1025 @itemx @var{file}.FPP
1026 @itemx @var{file}.FTN
1027 Fixed form Fortran source code which must be preprocessed (with the traditional
1030 @item @var{file}.f90
1031 @itemx @var{file}.f95
1032 @itemx @var{file}.f03
1033 @itemx @var{file}.f08
1034 Free form Fortran source code which should not be preprocessed.
1036 @item @var{file}.F90
1037 @itemx @var{file}.F95
1038 @itemx @var{file}.F03
1039 @itemx @var{file}.F08
1040 Free form Fortran source code which must be preprocessed (with the
1041 traditional preprocessor).
1043 @c FIXME: Descriptions of Java file types.
1049 @item @var{file}.ads
1050 Ada source code file which contains a library unit declaration (a
1051 declaration of a package, subprogram, or generic, or a generic
1052 instantiation), or a library unit renaming declaration (a package,
1053 generic, or subprogram renaming declaration). Such files are also
1056 @item @var{file}.adb
1057 Ada source code file containing a library unit body (a subprogram or
1058 package body). Such files are also called @dfn{bodies}.
1060 @c GCC also knows about some suffixes for languages not yet included:
1071 @itemx @var{file}.sx
1072 Assembler code which must be preprocessed.
1075 An object file to be fed straight into linking.
1076 Any file name with no recognized suffix is treated this way.
1080 You can specify the input language explicitly with the @option{-x} option:
1083 @item -x @var{language}
1084 Specify explicitly the @var{language} for the following input files
1085 (rather than letting the compiler choose a default based on the file
1086 name suffix). This option applies to all following input files until
1087 the next @option{-x} option. Possible values for @var{language} are:
1089 c c-header c-cpp-output
1090 c++ c++-header c++-cpp-output
1091 objective-c objective-c-header objective-c-cpp-output
1092 objective-c++ objective-c++-header objective-c++-cpp-output
1093 assembler assembler-with-cpp
1095 f77 f77-cpp-input f95 f95-cpp-input
1100 Turn off any specification of a language, so that subsequent files are
1101 handled according to their file name suffixes (as they are if @option{-x}
1102 has not been used at all).
1104 @item -pass-exit-codes
1105 @opindex pass-exit-codes
1106 Normally the @command{gcc} program will exit with the code of 1 if any
1107 phase of the compiler returns a non-success return code. If you specify
1108 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1109 numerically highest error produced by any phase that returned an error
1110 indication. The C, C++, and Fortran frontends return 4, if an internal
1111 compiler error is encountered.
1114 If you only want some of the stages of compilation, you can use
1115 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1116 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1117 @command{gcc} is to stop. Note that some combinations (for example,
1118 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1123 Compile or assemble the source files, but do not link. The linking
1124 stage simply is not done. The ultimate output is in the form of an
1125 object file for each source file.
1127 By default, the object file name for a source file is made by replacing
1128 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1130 Unrecognized input files, not requiring compilation or assembly, are
1135 Stop after the stage of compilation proper; do not assemble. The output
1136 is in the form of an assembler code file for each non-assembler input
1139 By default, the assembler file name for a source file is made by
1140 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1142 Input files that don't require compilation are ignored.
1146 Stop after the preprocessing stage; do not run the compiler proper. The
1147 output is in the form of preprocessed source code, which is sent to the
1150 Input files which don't require preprocessing are ignored.
1152 @cindex output file option
1155 Place output in file @var{file}. This applies regardless to whatever
1156 sort of output is being produced, whether it be an executable file,
1157 an object file, an assembler file or preprocessed C code.
1159 If @option{-o} is not specified, the default is to put an executable
1160 file in @file{a.out}, the object file for
1161 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1162 assembler file in @file{@var{source}.s}, a precompiled header file in
1163 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1168 Print (on standard error output) the commands executed to run the stages
1169 of compilation. Also print the version number of the compiler driver
1170 program and of the preprocessor and the compiler proper.
1174 Like @option{-v} except the commands are not executed and all command
1175 arguments are quoted. This is useful for shell scripts to capture the
1176 driver-generated command lines.
1180 Use pipes rather than temporary files for communication between the
1181 various stages of compilation. This fails to work on some systems where
1182 the assembler is unable to read from a pipe; but the GNU assembler has
1187 If you are compiling multiple source files, this option tells the driver
1188 to pass all the source files to the compiler at once (for those
1189 languages for which the compiler can handle this). This will allow
1190 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1191 language for which this is supported is C@. If you pass source files for
1192 multiple languages to the driver, using this option, the driver will invoke
1193 the compiler(s) that support IMA once each, passing each compiler all the
1194 source files appropriate for it. For those languages that do not support
1195 IMA this option will be ignored, and the compiler will be invoked once for
1196 each source file in that language. If you use this option in conjunction
1197 with @option{-save-temps}, the compiler will generate multiple
1199 (one for each source file), but only one (combined) @file{.o} or
1204 Print (on the standard output) a description of the command line options
1205 understood by @command{gcc}. If the @option{-v} option is also specified
1206 then @option{--help} will also be passed on to the various processes
1207 invoked by @command{gcc}, so that they can display the command line options
1208 they accept. If the @option{-Wextra} option has also been specified
1209 (prior to the @option{--help} option), then command line options which
1210 have no documentation associated with them will also be displayed.
1213 @opindex target-help
1214 Print (on the standard output) a description of target-specific command
1215 line options for each tool. For some targets extra target-specific
1216 information may also be printed.
1218 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1219 Print (on the standard output) a description of the command line
1220 options understood by the compiler that fit into all specified classes
1221 and qualifiers. These are the supported classes:
1224 @item @samp{optimizers}
1225 This will display all of the optimization options supported by the
1228 @item @samp{warnings}
1229 This will display all of the options controlling warning messages
1230 produced by the compiler.
1233 This will display target-specific options. Unlike the
1234 @option{--target-help} option however, target-specific options of the
1235 linker and assembler will not be displayed. This is because those
1236 tools do not currently support the extended @option{--help=} syntax.
1239 This will display the values recognized by the @option{--param}
1242 @item @var{language}
1243 This will display the options supported for @var{language}, where
1244 @var{language} is the name of one of the languages supported in this
1248 This will display the options that are common to all languages.
1251 These are the supported qualifiers:
1254 @item @samp{undocumented}
1255 Display only those options which are undocumented.
1258 Display options which take an argument that appears after an equal
1259 sign in the same continuous piece of text, such as:
1260 @samp{--help=target}.
1262 @item @samp{separate}
1263 Display options which take an argument that appears as a separate word
1264 following the original option, such as: @samp{-o output-file}.
1267 Thus for example to display all the undocumented target-specific
1268 switches supported by the compiler the following can be used:
1271 --help=target,undocumented
1274 The sense of a qualifier can be inverted by prefixing it with the
1275 @samp{^} character, so for example to display all binary warning
1276 options (i.e., ones that are either on or off and that do not take an
1277 argument), which have a description the following can be used:
1280 --help=warnings,^joined,^undocumented
1283 The argument to @option{--help=} should not consist solely of inverted
1286 Combining several classes is possible, although this usually
1287 restricts the output by so much that there is nothing to display. One
1288 case where it does work however is when one of the classes is
1289 @var{target}. So for example to display all the target-specific
1290 optimization options the following can be used:
1293 --help=target,optimizers
1296 The @option{--help=} option can be repeated on the command line. Each
1297 successive use will display its requested class of options, skipping
1298 those that have already been displayed.
1300 If the @option{-Q} option appears on the command line before the
1301 @option{--help=} option, then the descriptive text displayed by
1302 @option{--help=} is changed. Instead of describing the displayed
1303 options, an indication is given as to whether the option is enabled,
1304 disabled or set to a specific value (assuming that the compiler
1305 knows this at the point where the @option{--help=} option is used).
1307 Here is a truncated example from the ARM port of @command{gcc}:
1310 % gcc -Q -mabi=2 --help=target -c
1311 The following options are target specific:
1313 -mabort-on-noreturn [disabled]
1317 The output is sensitive to the effects of previous command line
1318 options, so for example it is possible to find out which optimizations
1319 are enabled at @option{-O2} by using:
1322 -Q -O2 --help=optimizers
1325 Alternatively you can discover which binary optimizations are enabled
1326 by @option{-O3} by using:
1329 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1330 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1331 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1334 @item -no-canonical-prefixes
1335 @opindex no-canonical-prefixes
1336 Do not expand any symbolic links, resolve references to @samp{/../}
1337 or @samp{/./}, or make the path absolute when generating a relative
1342 Display the version number and copyrights of the invoked GCC@.
1346 Invoke all subcommands under a wrapper program. It takes a single
1347 comma separated list as an argument, which will be used to invoke
1351 gcc -c t.c -wrapper gdb,--args
1354 This will invoke all subprograms of gcc under "gdb --args",
1355 thus cc1 invocation will be "gdb --args cc1 ...".
1357 @item -fplugin=@var{name}.so
1358 Load the plugin code in file @var{name}.so, assumed to be a
1359 shared object to be dlopen'd by the compiler. The base name of
1360 the shared object file is used to identify the plugin for the
1361 purposes of argument parsing (See
1362 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1363 Each plugin should define the callback functions specified in the
1366 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1367 Define an argument called @var{key} with a value of @var{value}
1368 for the plugin called @var{name}.
1370 @include @value{srcdir}/../libiberty/at-file.texi
1374 @section Compiling C++ Programs
1376 @cindex suffixes for C++ source
1377 @cindex C++ source file suffixes
1378 C++ source files conventionally use one of the suffixes @samp{.C},
1379 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1380 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1381 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1382 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1383 files with these names and compiles them as C++ programs even if you
1384 call the compiler the same way as for compiling C programs (usually
1385 with the name @command{gcc}).
1389 However, the use of @command{gcc} does not add the C++ library.
1390 @command{g++} is a program that calls GCC and treats @samp{.c},
1391 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1392 files unless @option{-x} is used, and automatically specifies linking
1393 against the C++ library. This program is also useful when
1394 precompiling a C header file with a @samp{.h} extension for use in C++
1395 compilations. On many systems, @command{g++} is also installed with
1396 the name @command{c++}.
1398 @cindex invoking @command{g++}
1399 When you compile C++ programs, you may specify many of the same
1400 command-line options that you use for compiling programs in any
1401 language; or command-line options meaningful for C and related
1402 languages; or options that are meaningful only for C++ programs.
1403 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1404 explanations of options for languages related to C@.
1405 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1406 explanations of options that are meaningful only for C++ programs.
1408 @node C Dialect Options
1409 @section Options Controlling C Dialect
1410 @cindex dialect options
1411 @cindex language dialect options
1412 @cindex options, dialect
1414 The following options control the dialect of C (or languages derived
1415 from C, such as C++, Objective-C and Objective-C++) that the compiler
1419 @cindex ANSI support
1423 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1424 equivalent to @samp{-std=c++98}.
1426 This turns off certain features of GCC that are incompatible with ISO
1427 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1428 such as the @code{asm} and @code{typeof} keywords, and
1429 predefined macros such as @code{unix} and @code{vax} that identify the
1430 type of system you are using. It also enables the undesirable and
1431 rarely used ISO trigraph feature. For the C compiler,
1432 it disables recognition of C++ style @samp{//} comments as well as
1433 the @code{inline} keyword.
1435 The alternate keywords @code{__asm__}, @code{__extension__},
1436 @code{__inline__} and @code{__typeof__} continue to work despite
1437 @option{-ansi}. You would not want to use them in an ISO C program, of
1438 course, but it is useful to put them in header files that might be included
1439 in compilations done with @option{-ansi}. Alternate predefined macros
1440 such as @code{__unix__} and @code{__vax__} are also available, with or
1441 without @option{-ansi}.
1443 The @option{-ansi} option does not cause non-ISO programs to be
1444 rejected gratuitously. For that, @option{-pedantic} is required in
1445 addition to @option{-ansi}. @xref{Warning Options}.
1447 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1448 option is used. Some header files may notice this macro and refrain
1449 from declaring certain functions or defining certain macros that the
1450 ISO standard doesn't call for; this is to avoid interfering with any
1451 programs that might use these names for other things.
1453 Functions that would normally be built in but do not have semantics
1454 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1455 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1456 built-in functions provided by GCC}, for details of the functions
1461 Determine the language standard. @xref{Standards,,Language Standards
1462 Supported by GCC}, for details of these standard versions. This option
1463 is currently only supported when compiling C or C++.
1465 The compiler can accept several base standards, such as @samp{c89} or
1466 @samp{c++98}, and GNU dialects of those standards, such as
1467 @samp{gnu89} or @samp{gnu++98}. By specifying a base standard, the
1468 compiler will accept all programs following that standard and those
1469 using GNU extensions that do not contradict it. For example,
1470 @samp{-std=c89} turns off certain features of GCC that are
1471 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1472 keywords, but not other GNU extensions that do not have a meaning in
1473 ISO C90, such as omitting the middle term of a @code{?:}
1474 expression. On the other hand, by specifying a GNU dialect of a
1475 standard, all features the compiler support are enabled, even when
1476 those features change the meaning of the base standard and some
1477 strict-conforming programs may be rejected. The particular standard
1478 is used by @option{-pedantic} to identify which features are GNU
1479 extensions given that version of the standard. For example
1480 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1481 comments, while @samp{-std=gnu99 -pedantic} would not.
1483 A value for this option must be provided; possible values are
1488 Support all ISO C90 programs (certain GNU extensions that conflict
1489 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1491 @item iso9899:199409
1492 ISO C90 as modified in amendment 1.
1498 ISO C99. Note that this standard is not yet fully supported; see
1499 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1500 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1503 GNU dialect of ISO C90 (including some C99 features). This
1504 is the default for C code.
1508 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1509 this will become the default. The name @samp{gnu9x} is deprecated.
1512 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1516 GNU dialect of @option{-std=c++98}. This is the default for
1520 The working draft of the upcoming ISO C++0x standard. This option
1521 enables experimental features that are likely to be included in
1522 C++0x. The working draft is constantly changing, and any feature that is
1523 enabled by this flag may be removed from future versions of GCC if it is
1524 not part of the C++0x standard.
1527 GNU dialect of @option{-std=c++0x}. This option enables
1528 experimental features that may be removed in future versions of GCC.
1531 @item -fgnu89-inline
1532 @opindex fgnu89-inline
1533 The option @option{-fgnu89-inline} tells GCC to use the traditional
1534 GNU semantics for @code{inline} functions when in C99 mode.
1535 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1536 is accepted and ignored by GCC versions 4.1.3 up to but not including
1537 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1538 C99 mode. Using this option is roughly equivalent to adding the
1539 @code{gnu_inline} function attribute to all inline functions
1540 (@pxref{Function Attributes}).
1542 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1543 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1544 specifies the default behavior). This option was first supported in
1545 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1547 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1548 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1549 in effect for @code{inline} functions. @xref{Common Predefined
1550 Macros,,,cpp,The C Preprocessor}.
1552 @item -aux-info @var{filename}
1554 Output to the given filename prototyped declarations for all functions
1555 declared and/or defined in a translation unit, including those in header
1556 files. This option is silently ignored in any language other than C@.
1558 Besides declarations, the file indicates, in comments, the origin of
1559 each declaration (source file and line), whether the declaration was
1560 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1561 @samp{O} for old, respectively, in the first character after the line
1562 number and the colon), and whether it came from a declaration or a
1563 definition (@samp{C} or @samp{F}, respectively, in the following
1564 character). In the case of function definitions, a K&R-style list of
1565 arguments followed by their declarations is also provided, inside
1566 comments, after the declaration.
1570 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1571 keyword, so that code can use these words as identifiers. You can use
1572 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1573 instead. @option{-ansi} implies @option{-fno-asm}.
1575 In C++, this switch only affects the @code{typeof} keyword, since
1576 @code{asm} and @code{inline} are standard keywords. You may want to
1577 use the @option{-fno-gnu-keywords} flag instead, which has the same
1578 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1579 switch only affects the @code{asm} and @code{typeof} keywords, since
1580 @code{inline} is a standard keyword in ISO C99.
1583 @itemx -fno-builtin-@var{function}
1584 @opindex fno-builtin
1585 @cindex built-in functions
1586 Don't recognize built-in functions that do not begin with
1587 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1588 functions provided by GCC}, for details of the functions affected,
1589 including those which are not built-in functions when @option{-ansi} or
1590 @option{-std} options for strict ISO C conformance are used because they
1591 do not have an ISO standard meaning.
1593 GCC normally generates special code to handle certain built-in functions
1594 more efficiently; for instance, calls to @code{alloca} may become single
1595 instructions that adjust the stack directly, and calls to @code{memcpy}
1596 may become inline copy loops. The resulting code is often both smaller
1597 and faster, but since the function calls no longer appear as such, you
1598 cannot set a breakpoint on those calls, nor can you change the behavior
1599 of the functions by linking with a different library. In addition,
1600 when a function is recognized as a built-in function, GCC may use
1601 information about that function to warn about problems with calls to
1602 that function, or to generate more efficient code, even if the
1603 resulting code still contains calls to that function. For example,
1604 warnings are given with @option{-Wformat} for bad calls to
1605 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1606 known not to modify global memory.
1608 With the @option{-fno-builtin-@var{function}} option
1609 only the built-in function @var{function} is
1610 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1611 function is named that is not built-in in this version of GCC, this
1612 option is ignored. There is no corresponding
1613 @option{-fbuiltin-@var{function}} option; if you wish to enable
1614 built-in functions selectively when using @option{-fno-builtin} or
1615 @option{-ffreestanding}, you may define macros such as:
1618 #define abs(n) __builtin_abs ((n))
1619 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1624 @cindex hosted environment
1626 Assert that compilation takes place in a hosted environment. This implies
1627 @option{-fbuiltin}. A hosted environment is one in which the
1628 entire standard library is available, and in which @code{main} has a return
1629 type of @code{int}. Examples are nearly everything except a kernel.
1630 This is equivalent to @option{-fno-freestanding}.
1632 @item -ffreestanding
1633 @opindex ffreestanding
1634 @cindex hosted environment
1636 Assert that compilation takes place in a freestanding environment. This
1637 implies @option{-fno-builtin}. A freestanding environment
1638 is one in which the standard library may not exist, and program startup may
1639 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1640 This is equivalent to @option{-fno-hosted}.
1642 @xref{Standards,,Language Standards Supported by GCC}, for details of
1643 freestanding and hosted environments.
1647 @cindex openmp parallel
1648 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1649 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1650 compiler generates parallel code according to the OpenMP Application
1651 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1652 implies @option{-pthread}, and thus is only supported on targets that
1653 have support for @option{-pthread}.
1655 @item -fms-extensions
1656 @opindex fms-extensions
1657 Accept some non-standard constructs used in Microsoft header files.
1659 Some cases of unnamed fields in structures and unions are only
1660 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1661 fields within structs/unions}, for details.
1665 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1666 options for strict ISO C conformance) implies @option{-trigraphs}.
1668 @item -no-integrated-cpp
1669 @opindex no-integrated-cpp
1670 Performs a compilation in two passes: preprocessing and compiling. This
1671 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1672 @option{-B} option. The user supplied compilation step can then add in
1673 an additional preprocessing step after normal preprocessing but before
1674 compiling. The default is to use the integrated cpp (internal cpp)
1676 The semantics of this option will change if "cc1", "cc1plus", and
1677 "cc1obj" are merged.
1679 @cindex traditional C language
1680 @cindex C language, traditional
1682 @itemx -traditional-cpp
1683 @opindex traditional-cpp
1684 @opindex traditional
1685 Formerly, these options caused GCC to attempt to emulate a pre-standard
1686 C compiler. They are now only supported with the @option{-E} switch.
1687 The preprocessor continues to support a pre-standard mode. See the GNU
1688 CPP manual for details.
1690 @item -fcond-mismatch
1691 @opindex fcond-mismatch
1692 Allow conditional expressions with mismatched types in the second and
1693 third arguments. The value of such an expression is void. This option
1694 is not supported for C++.
1696 @item -flax-vector-conversions
1697 @opindex flax-vector-conversions
1698 Allow implicit conversions between vectors with differing numbers of
1699 elements and/or incompatible element types. This option should not be
1702 @item -funsigned-char
1703 @opindex funsigned-char
1704 Let the type @code{char} be unsigned, like @code{unsigned char}.
1706 Each kind of machine has a default for what @code{char} should
1707 be. It is either like @code{unsigned char} by default or like
1708 @code{signed char} by default.
1710 Ideally, a portable program should always use @code{signed char} or
1711 @code{unsigned char} when it depends on the signedness of an object.
1712 But many programs have been written to use plain @code{char} and
1713 expect it to be signed, or expect it to be unsigned, depending on the
1714 machines they were written for. This option, and its inverse, let you
1715 make such a program work with the opposite default.
1717 The type @code{char} is always a distinct type from each of
1718 @code{signed char} or @code{unsigned char}, even though its behavior
1719 is always just like one of those two.
1722 @opindex fsigned-char
1723 Let the type @code{char} be signed, like @code{signed char}.
1725 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1726 the negative form of @option{-funsigned-char}. Likewise, the option
1727 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1729 @item -fsigned-bitfields
1730 @itemx -funsigned-bitfields
1731 @itemx -fno-signed-bitfields
1732 @itemx -fno-unsigned-bitfields
1733 @opindex fsigned-bitfields
1734 @opindex funsigned-bitfields
1735 @opindex fno-signed-bitfields
1736 @opindex fno-unsigned-bitfields
1737 These options control whether a bit-field is signed or unsigned, when the
1738 declaration does not use either @code{signed} or @code{unsigned}. By
1739 default, such a bit-field is signed, because this is consistent: the
1740 basic integer types such as @code{int} are signed types.
1743 @node C++ Dialect Options
1744 @section Options Controlling C++ Dialect
1746 @cindex compiler options, C++
1747 @cindex C++ options, command line
1748 @cindex options, C++
1749 This section describes the command-line options that are only meaningful
1750 for C++ programs; but you can also use most of the GNU compiler options
1751 regardless of what language your program is in. For example, you
1752 might compile a file @code{firstClass.C} like this:
1755 g++ -g -frepo -O -c firstClass.C
1759 In this example, only @option{-frepo} is an option meant
1760 only for C++ programs; you can use the other options with any
1761 language supported by GCC@.
1763 Here is a list of options that are @emph{only} for compiling C++ programs:
1767 @item -fabi-version=@var{n}
1768 @opindex fabi-version
1769 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1770 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1771 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1772 the version that conforms most closely to the C++ ABI specification.
1773 Therefore, the ABI obtained using version 0 will change as ABI bugs
1776 The default is version 2.
1778 @item -fno-access-control
1779 @opindex fno-access-control
1780 Turn off all access checking. This switch is mainly useful for working
1781 around bugs in the access control code.
1785 Check that the pointer returned by @code{operator new} is non-null
1786 before attempting to modify the storage allocated. This check is
1787 normally unnecessary because the C++ standard specifies that
1788 @code{operator new} will only return @code{0} if it is declared
1789 @samp{throw()}, in which case the compiler will always check the
1790 return value even without this option. In all other cases, when
1791 @code{operator new} has a non-empty exception specification, memory
1792 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1793 @samp{new (nothrow)}.
1795 @item -fconserve-space
1796 @opindex fconserve-space
1797 Put uninitialized or runtime-initialized global variables into the
1798 common segment, as C does. This saves space in the executable at the
1799 cost of not diagnosing duplicate definitions. If you compile with this
1800 flag and your program mysteriously crashes after @code{main()} has
1801 completed, you may have an object that is being destroyed twice because
1802 two definitions were merged.
1804 This option is no longer useful on most targets, now that support has
1805 been added for putting variables into BSS without making them common.
1807 @item -fno-deduce-init-list
1808 @opindex fno-deduce-init-list
1809 Disable deduction of a template type parameter as
1810 std::initializer_list from a brace-enclosed initializer list, i.e.
1813 template <class T> auto forward(T t) -> decltype (realfn (t))
1820 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1824 This option is present because this deduction is an extension to the
1825 current specification in the C++0x working draft, and there was
1826 some concern about potential overload resolution problems.
1828 @item -ffriend-injection
1829 @opindex ffriend-injection
1830 Inject friend functions into the enclosing namespace, so that they are
1831 visible outside the scope of the class in which they are declared.
1832 Friend functions were documented to work this way in the old Annotated
1833 C++ Reference Manual, and versions of G++ before 4.1 always worked
1834 that way. However, in ISO C++ a friend function which is not declared
1835 in an enclosing scope can only be found using argument dependent
1836 lookup. This option causes friends to be injected as they were in
1839 This option is for compatibility, and may be removed in a future
1842 @item -fno-elide-constructors
1843 @opindex fno-elide-constructors
1844 The C++ standard allows an implementation to omit creating a temporary
1845 which is only used to initialize another object of the same type.
1846 Specifying this option disables that optimization, and forces G++ to
1847 call the copy constructor in all cases.
1849 @item -fno-enforce-eh-specs
1850 @opindex fno-enforce-eh-specs
1851 Don't generate code to check for violation of exception specifications
1852 at runtime. This option violates the C++ standard, but may be useful
1853 for reducing code size in production builds, much like defining
1854 @samp{NDEBUG}. This does not give user code permission to throw
1855 exceptions in violation of the exception specifications; the compiler
1856 will still optimize based on the specifications, so throwing an
1857 unexpected exception will result in undefined behavior.
1860 @itemx -fno-for-scope
1862 @opindex fno-for-scope
1863 If @option{-ffor-scope} is specified, the scope of variables declared in
1864 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1865 as specified by the C++ standard.
1866 If @option{-fno-for-scope} is specified, the scope of variables declared in
1867 a @i{for-init-statement} extends to the end of the enclosing scope,
1868 as was the case in old versions of G++, and other (traditional)
1869 implementations of C++.
1871 The default if neither flag is given to follow the standard,
1872 but to allow and give a warning for old-style code that would
1873 otherwise be invalid, or have different behavior.
1875 @item -fno-gnu-keywords
1876 @opindex fno-gnu-keywords
1877 Do not recognize @code{typeof} as a keyword, so that code can use this
1878 word as an identifier. You can use the keyword @code{__typeof__} instead.
1879 @option{-ansi} implies @option{-fno-gnu-keywords}.
1881 @item -fno-implicit-templates
1882 @opindex fno-implicit-templates
1883 Never emit code for non-inline templates which are instantiated
1884 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1885 @xref{Template Instantiation}, for more information.
1887 @item -fno-implicit-inline-templates
1888 @opindex fno-implicit-inline-templates
1889 Don't emit code for implicit instantiations of inline templates, either.
1890 The default is to handle inlines differently so that compiles with and
1891 without optimization will need the same set of explicit instantiations.
1893 @item -fno-implement-inlines
1894 @opindex fno-implement-inlines
1895 To save space, do not emit out-of-line copies of inline functions
1896 controlled by @samp{#pragma implementation}. This will cause linker
1897 errors if these functions are not inlined everywhere they are called.
1899 @item -fms-extensions
1900 @opindex fms-extensions
1901 Disable pedantic warnings about constructs used in MFC, such as implicit
1902 int and getting a pointer to member function via non-standard syntax.
1904 @item -fno-nonansi-builtins
1905 @opindex fno-nonansi-builtins
1906 Disable built-in declarations of functions that are not mandated by
1907 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1908 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1910 @item -fno-operator-names
1911 @opindex fno-operator-names
1912 Do not treat the operator name keywords @code{and}, @code{bitand},
1913 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1914 synonyms as keywords.
1916 @item -fno-optional-diags
1917 @opindex fno-optional-diags
1918 Disable diagnostics that the standard says a compiler does not need to
1919 issue. Currently, the only such diagnostic issued by G++ is the one for
1920 a name having multiple meanings within a class.
1923 @opindex fpermissive
1924 Downgrade some diagnostics about nonconformant code from errors to
1925 warnings. Thus, using @option{-fpermissive} will allow some
1926 nonconforming code to compile.
1928 @item -fno-pretty-templates
1929 @opindex fno-pretty-templates
1930 When an error message refers to a specialization of a function
1931 template, the compiler will normally print the signature of the
1932 template followed by the template arguments and any typedefs or
1933 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1934 rather than @code{void f(int)}) so that it's clear which template is
1935 involved. When an error message refers to a specialization of a class
1936 template, the compiler will omit any template arguments which match
1937 the default template arguments for that template. If either of these
1938 behaviors make it harder to understand the error message rather than
1939 easier, using @option{-fno-pretty-templates} will disable them.
1943 Enable automatic template instantiation at link time. This option also
1944 implies @option{-fno-implicit-templates}. @xref{Template
1945 Instantiation}, for more information.
1949 Disable generation of information about every class with virtual
1950 functions for use by the C++ runtime type identification features
1951 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1952 of the language, you can save some space by using this flag. Note that
1953 exception handling uses the same information, but it will generate it as
1954 needed. The @samp{dynamic_cast} operator can still be used for casts that
1955 do not require runtime type information, i.e.@: casts to @code{void *} or to
1956 unambiguous base classes.
1960 Emit statistics about front-end processing at the end of the compilation.
1961 This information is generally only useful to the G++ development team.
1963 @item -ftemplate-depth-@var{n}
1964 @opindex ftemplate-depth
1965 Set the maximum instantiation depth for template classes to @var{n}.
1966 A limit on the template instantiation depth is needed to detect
1967 endless recursions during template class instantiation. ANSI/ISO C++
1968 conforming programs must not rely on a maximum depth greater than 17
1969 (changed to 1024 in C++0x).
1971 @item -fno-threadsafe-statics
1972 @opindex fno-threadsafe-statics
1973 Do not emit the extra code to use the routines specified in the C++
1974 ABI for thread-safe initialization of local statics. You can use this
1975 option to reduce code size slightly in code that doesn't need to be
1978 @item -fuse-cxa-atexit
1979 @opindex fuse-cxa-atexit
1980 Register destructors for objects with static storage duration with the
1981 @code{__cxa_atexit} function rather than the @code{atexit} function.
1982 This option is required for fully standards-compliant handling of static
1983 destructors, but will only work if your C library supports
1984 @code{__cxa_atexit}.
1986 @item -fno-use-cxa-get-exception-ptr
1987 @opindex fno-use-cxa-get-exception-ptr
1988 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1989 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1990 if the runtime routine is not available.
1992 @item -fvisibility-inlines-hidden
1993 @opindex fvisibility-inlines-hidden
1994 This switch declares that the user does not attempt to compare
1995 pointers to inline methods where the addresses of the two functions
1996 were taken in different shared objects.
1998 The effect of this is that GCC may, effectively, mark inline methods with
1999 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2000 appear in the export table of a DSO and do not require a PLT indirection
2001 when used within the DSO@. Enabling this option can have a dramatic effect
2002 on load and link times of a DSO as it massively reduces the size of the
2003 dynamic export table when the library makes heavy use of templates.
2005 The behavior of this switch is not quite the same as marking the
2006 methods as hidden directly, because it does not affect static variables
2007 local to the function or cause the compiler to deduce that
2008 the function is defined in only one shared object.
2010 You may mark a method as having a visibility explicitly to negate the
2011 effect of the switch for that method. For example, if you do want to
2012 compare pointers to a particular inline method, you might mark it as
2013 having default visibility. Marking the enclosing class with explicit
2014 visibility will have no effect.
2016 Explicitly instantiated inline methods are unaffected by this option
2017 as their linkage might otherwise cross a shared library boundary.
2018 @xref{Template Instantiation}.
2020 @item -fvisibility-ms-compat
2021 @opindex fvisibility-ms-compat
2022 This flag attempts to use visibility settings to make GCC's C++
2023 linkage model compatible with that of Microsoft Visual Studio.
2025 The flag makes these changes to GCC's linkage model:
2029 It sets the default visibility to @code{hidden}, like
2030 @option{-fvisibility=hidden}.
2033 Types, but not their members, are not hidden by default.
2036 The One Definition Rule is relaxed for types without explicit
2037 visibility specifications which are defined in more than one different
2038 shared object: those declarations are permitted if they would have
2039 been permitted when this option was not used.
2042 In new code it is better to use @option{-fvisibility=hidden} and
2043 export those classes which are intended to be externally visible.
2044 Unfortunately it is possible for code to rely, perhaps accidentally,
2045 on the Visual Studio behavior.
2047 Among the consequences of these changes are that static data members
2048 of the same type with the same name but defined in different shared
2049 objects will be different, so changing one will not change the other;
2050 and that pointers to function members defined in different shared
2051 objects may not compare equal. When this flag is given, it is a
2052 violation of the ODR to define types with the same name differently.
2056 Do not use weak symbol support, even if it is provided by the linker.
2057 By default, G++ will use weak symbols if they are available. This
2058 option exists only for testing, and should not be used by end-users;
2059 it will result in inferior code and has no benefits. This option may
2060 be removed in a future release of G++.
2064 Do not search for header files in the standard directories specific to
2065 C++, but do still search the other standard directories. (This option
2066 is used when building the C++ library.)
2069 In addition, these optimization, warning, and code generation options
2070 have meanings only for C++ programs:
2073 @item -fno-default-inline
2074 @opindex fno-default-inline
2075 Do not assume @samp{inline} for functions defined inside a class scope.
2076 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2077 functions will have linkage like inline functions; they just won't be
2080 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2083 Warn when G++ generates code that is probably not compatible with the
2084 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2085 all such cases, there are probably some cases that are not warned about,
2086 even though G++ is generating incompatible code. There may also be
2087 cases where warnings are emitted even though the code that is generated
2090 You should rewrite your code to avoid these warnings if you are
2091 concerned about the fact that code generated by G++ may not be binary
2092 compatible with code generated by other compilers.
2094 The known incompatibilities at this point include:
2099 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2100 pack data into the same byte as a base class. For example:
2103 struct A @{ virtual void f(); int f1 : 1; @};
2104 struct B : public A @{ int f2 : 1; @};
2108 In this case, G++ will place @code{B::f2} into the same byte
2109 as@code{A::f1}; other compilers will not. You can avoid this problem
2110 by explicitly padding @code{A} so that its size is a multiple of the
2111 byte size on your platform; that will cause G++ and other compilers to
2112 layout @code{B} identically.
2115 Incorrect handling of tail-padding for virtual bases. G++ does not use
2116 tail padding when laying out virtual bases. For example:
2119 struct A @{ virtual void f(); char c1; @};
2120 struct B @{ B(); char c2; @};
2121 struct C : public A, public virtual B @{@};
2125 In this case, G++ will not place @code{B} into the tail-padding for
2126 @code{A}; other compilers will. You can avoid this problem by
2127 explicitly padding @code{A} so that its size is a multiple of its
2128 alignment (ignoring virtual base classes); that will cause G++ and other
2129 compilers to layout @code{C} identically.
2132 Incorrect handling of bit-fields with declared widths greater than that
2133 of their underlying types, when the bit-fields appear in a union. For
2137 union U @{ int i : 4096; @};
2141 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2142 union too small by the number of bits in an @code{int}.
2145 Empty classes can be placed at incorrect offsets. For example:
2155 struct C : public B, public A @{@};
2159 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2160 it should be placed at offset zero. G++ mistakenly believes that the
2161 @code{A} data member of @code{B} is already at offset zero.
2164 Names of template functions whose types involve @code{typename} or
2165 template template parameters can be mangled incorrectly.
2168 template <typename Q>
2169 void f(typename Q::X) @{@}
2171 template <template <typename> class Q>
2172 void f(typename Q<int>::X) @{@}
2176 Instantiations of these templates may be mangled incorrectly.
2180 It also warns psABI related changes. The known psABI changes at this
2186 For SYSV/x86-64, when passing union with long double, it is changed to
2187 pass in memory as specified in psABI. For example:
2197 @code{union U} will always be passed in memory.
2201 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2202 @opindex Wctor-dtor-privacy
2203 @opindex Wno-ctor-dtor-privacy
2204 Warn when a class seems unusable because all the constructors or
2205 destructors in that class are private, and it has neither friends nor
2206 public static member functions.
2208 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2209 @opindex Wnon-virtual-dtor
2210 @opindex Wno-non-virtual-dtor
2211 Warn when a class has virtual functions and accessible non-virtual
2212 destructor, in which case it would be possible but unsafe to delete
2213 an instance of a derived class through a pointer to the base class.
2214 This warning is also enabled if -Weffc++ is specified.
2216 @item -Wreorder @r{(C++ and Objective-C++ only)}
2218 @opindex Wno-reorder
2219 @cindex reordering, warning
2220 @cindex warning for reordering of member initializers
2221 Warn when the order of member initializers given in the code does not
2222 match the order in which they must be executed. For instance:
2228 A(): j (0), i (1) @{ @}
2232 The compiler will rearrange the member initializers for @samp{i}
2233 and @samp{j} to match the declaration order of the members, emitting
2234 a warning to that effect. This warning is enabled by @option{-Wall}.
2237 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2240 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2243 Warn about violations of the following style guidelines from Scott Meyers'
2244 @cite{Effective C++} book:
2248 Item 11: Define a copy constructor and an assignment operator for classes
2249 with dynamically allocated memory.
2252 Item 12: Prefer initialization to assignment in constructors.
2255 Item 14: Make destructors virtual in base classes.
2258 Item 15: Have @code{operator=} return a reference to @code{*this}.
2261 Item 23: Don't try to return a reference when you must return an object.
2265 Also warn about violations of the following style guidelines from
2266 Scott Meyers' @cite{More Effective C++} book:
2270 Item 6: Distinguish between prefix and postfix forms of increment and
2271 decrement operators.
2274 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2278 When selecting this option, be aware that the standard library
2279 headers do not obey all of these guidelines; use @samp{grep -v}
2280 to filter out those warnings.
2282 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2283 @opindex Wstrict-null-sentinel
2284 @opindex Wno-strict-null-sentinel
2285 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2286 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2287 to @code{__null}. Although it is a null pointer constant not a null pointer,
2288 it is guaranteed to be of the same size as a pointer. But this use is
2289 not portable across different compilers.
2291 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2292 @opindex Wno-non-template-friend
2293 @opindex Wnon-template-friend
2294 Disable warnings when non-templatized friend functions are declared
2295 within a template. Since the advent of explicit template specification
2296 support in G++, if the name of the friend is an unqualified-id (i.e.,
2297 @samp{friend foo(int)}), the C++ language specification demands that the
2298 friend declare or define an ordinary, nontemplate function. (Section
2299 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2300 could be interpreted as a particular specialization of a templatized
2301 function. Because this non-conforming behavior is no longer the default
2302 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2303 check existing code for potential trouble spots and is on by default.
2304 This new compiler behavior can be turned off with
2305 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2306 but disables the helpful warning.
2308 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2309 @opindex Wold-style-cast
2310 @opindex Wno-old-style-cast
2311 Warn if an old-style (C-style) cast to a non-void type is used within
2312 a C++ program. The new-style casts (@samp{dynamic_cast},
2313 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2314 less vulnerable to unintended effects and much easier to search for.
2316 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2317 @opindex Woverloaded-virtual
2318 @opindex Wno-overloaded-virtual
2319 @cindex overloaded virtual fn, warning
2320 @cindex warning for overloaded virtual fn
2321 Warn when a function declaration hides virtual functions from a
2322 base class. For example, in:
2329 struct B: public A @{
2334 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2342 will fail to compile.
2344 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2345 @opindex Wno-pmf-conversions
2346 @opindex Wpmf-conversions
2347 Disable the diagnostic for converting a bound pointer to member function
2350 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2351 @opindex Wsign-promo
2352 @opindex Wno-sign-promo
2353 Warn when overload resolution chooses a promotion from unsigned or
2354 enumerated type to a signed type, over a conversion to an unsigned type of
2355 the same size. Previous versions of G++ would try to preserve
2356 unsignedness, but the standard mandates the current behavior.
2361 A& operator = (int);
2371 In this example, G++ will synthesize a default @samp{A& operator =
2372 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2375 @node Objective-C and Objective-C++ Dialect Options
2376 @section Options Controlling Objective-C and Objective-C++ Dialects
2378 @cindex compiler options, Objective-C and Objective-C++
2379 @cindex Objective-C and Objective-C++ options, command line
2380 @cindex options, Objective-C and Objective-C++
2381 (NOTE: This manual does not describe the Objective-C and Objective-C++
2382 languages themselves. See @xref{Standards,,Language Standards
2383 Supported by GCC}, for references.)
2385 This section describes the command-line options that are only meaningful
2386 for Objective-C and Objective-C++ programs, but you can also use most of
2387 the language-independent GNU compiler options.
2388 For example, you might compile a file @code{some_class.m} like this:
2391 gcc -g -fgnu-runtime -O -c some_class.m
2395 In this example, @option{-fgnu-runtime} is an option meant only for
2396 Objective-C and Objective-C++ programs; you can use the other options with
2397 any language supported by GCC@.
2399 Note that since Objective-C is an extension of the C language, Objective-C
2400 compilations may also use options specific to the C front-end (e.g.,
2401 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2402 C++-specific options (e.g., @option{-Wabi}).
2404 Here is a list of options that are @emph{only} for compiling Objective-C
2405 and Objective-C++ programs:
2408 @item -fconstant-string-class=@var{class-name}
2409 @opindex fconstant-string-class
2410 Use @var{class-name} as the name of the class to instantiate for each
2411 literal string specified with the syntax @code{@@"@dots{}"}. The default
2412 class name is @code{NXConstantString} if the GNU runtime is being used, and
2413 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2414 @option{-fconstant-cfstrings} option, if also present, will override the
2415 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2416 to be laid out as constant CoreFoundation strings.
2419 @opindex fgnu-runtime
2420 Generate object code compatible with the standard GNU Objective-C
2421 runtime. This is the default for most types of systems.
2423 @item -fnext-runtime
2424 @opindex fnext-runtime
2425 Generate output compatible with the NeXT runtime. This is the default
2426 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2427 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2430 @item -fno-nil-receivers
2431 @opindex fno-nil-receivers
2432 Assume that all Objective-C message dispatches (e.g.,
2433 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2434 is not @code{nil}. This allows for more efficient entry points in the runtime
2435 to be used. Currently, this option is only available in conjunction with
2436 the NeXT runtime on Mac OS X 10.3 and later.
2438 @item -fobjc-call-cxx-cdtors
2439 @opindex fobjc-call-cxx-cdtors
2440 For each Objective-C class, check if any of its instance variables is a
2441 C++ object with a non-trivial default constructor. If so, synthesize a
2442 special @code{- (id) .cxx_construct} instance method that will run
2443 non-trivial default constructors on any such instance variables, in order,
2444 and then return @code{self}. Similarly, check if any instance variable
2445 is a C++ object with a non-trivial destructor, and if so, synthesize a
2446 special @code{- (void) .cxx_destruct} method that will run
2447 all such default destructors, in reverse order.
2449 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2450 thusly generated will only operate on instance variables declared in the
2451 current Objective-C class, and not those inherited from superclasses. It
2452 is the responsibility of the Objective-C runtime to invoke all such methods
2453 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2454 will be invoked by the runtime immediately after a new object
2455 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2456 be invoked immediately before the runtime deallocates an object instance.
2458 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2459 support for invoking the @code{- (id) .cxx_construct} and
2460 @code{- (void) .cxx_destruct} methods.
2462 @item -fobjc-direct-dispatch
2463 @opindex fobjc-direct-dispatch
2464 Allow fast jumps to the message dispatcher. On Darwin this is
2465 accomplished via the comm page.
2467 @item -fobjc-exceptions
2468 @opindex fobjc-exceptions
2469 Enable syntactic support for structured exception handling in Objective-C,
2470 similar to what is offered by C++ and Java. This option is
2471 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2480 @@catch (AnObjCClass *exc) @{
2487 @@catch (AnotherClass *exc) @{
2490 @@catch (id allOthers) @{
2500 The @code{@@throw} statement may appear anywhere in an Objective-C or
2501 Objective-C++ program; when used inside of a @code{@@catch} block, the
2502 @code{@@throw} may appear without an argument (as shown above), in which case
2503 the object caught by the @code{@@catch} will be rethrown.
2505 Note that only (pointers to) Objective-C objects may be thrown and
2506 caught using this scheme. When an object is thrown, it will be caught
2507 by the nearest @code{@@catch} clause capable of handling objects of that type,
2508 analogously to how @code{catch} blocks work in C++ and Java. A
2509 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2510 any and all Objective-C exceptions not caught by previous @code{@@catch}
2513 The @code{@@finally} clause, if present, will be executed upon exit from the
2514 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2515 regardless of whether any exceptions are thrown, caught or rethrown
2516 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2517 of the @code{finally} clause in Java.
2519 There are several caveats to using the new exception mechanism:
2523 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2524 idioms provided by the @code{NSException} class, the new
2525 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2526 systems, due to additional functionality needed in the (NeXT) Objective-C
2530 As mentioned above, the new exceptions do not support handling
2531 types other than Objective-C objects. Furthermore, when used from
2532 Objective-C++, the Objective-C exception model does not interoperate with C++
2533 exceptions at this time. This means you cannot @code{@@throw} an exception
2534 from Objective-C and @code{catch} it in C++, or vice versa
2535 (i.e., @code{throw @dots{} @@catch}).
2538 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2539 blocks for thread-safe execution:
2542 @@synchronized (ObjCClass *guard) @{
2547 Upon entering the @code{@@synchronized} block, a thread of execution shall
2548 first check whether a lock has been placed on the corresponding @code{guard}
2549 object by another thread. If it has, the current thread shall wait until
2550 the other thread relinquishes its lock. Once @code{guard} becomes available,
2551 the current thread will place its own lock on it, execute the code contained in
2552 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2553 making @code{guard} available to other threads).
2555 Unlike Java, Objective-C does not allow for entire methods to be marked
2556 @code{@@synchronized}. Note that throwing exceptions out of
2557 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2558 to be unlocked properly.
2562 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2564 @item -freplace-objc-classes
2565 @opindex freplace-objc-classes
2566 Emit a special marker instructing @command{ld(1)} not to statically link in
2567 the resulting object file, and allow @command{dyld(1)} to load it in at
2568 run time instead. This is used in conjunction with the Fix-and-Continue
2569 debugging mode, where the object file in question may be recompiled and
2570 dynamically reloaded in the course of program execution, without the need
2571 to restart the program itself. Currently, Fix-and-Continue functionality
2572 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2577 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2578 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2579 compile time) with static class references that get initialized at load time,
2580 which improves run-time performance. Specifying the @option{-fzero-link} flag
2581 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2582 to be retained. This is useful in Zero-Link debugging mode, since it allows
2583 for individual class implementations to be modified during program execution.
2587 Dump interface declarations for all classes seen in the source file to a
2588 file named @file{@var{sourcename}.decl}.
2590 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2591 @opindex Wassign-intercept
2592 @opindex Wno-assign-intercept
2593 Warn whenever an Objective-C assignment is being intercepted by the
2596 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2597 @opindex Wno-protocol
2599 If a class is declared to implement a protocol, a warning is issued for
2600 every method in the protocol that is not implemented by the class. The
2601 default behavior is to issue a warning for every method not explicitly
2602 implemented in the class, even if a method implementation is inherited
2603 from the superclass. If you use the @option{-Wno-protocol} option, then
2604 methods inherited from the superclass are considered to be implemented,
2605 and no warning is issued for them.
2607 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2609 @opindex Wno-selector
2610 Warn if multiple methods of different types for the same selector are
2611 found during compilation. The check is performed on the list of methods
2612 in the final stage of compilation. Additionally, a check is performed
2613 for each selector appearing in a @code{@@selector(@dots{})}
2614 expression, and a corresponding method for that selector has been found
2615 during compilation. Because these checks scan the method table only at
2616 the end of compilation, these warnings are not produced if the final
2617 stage of compilation is not reached, for example because an error is
2618 found during compilation, or because the @option{-fsyntax-only} option is
2621 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2622 @opindex Wstrict-selector-match
2623 @opindex Wno-strict-selector-match
2624 Warn if multiple methods with differing argument and/or return types are
2625 found for a given selector when attempting to send a message using this
2626 selector to a receiver of type @code{id} or @code{Class}. When this flag
2627 is off (which is the default behavior), the compiler will omit such warnings
2628 if any differences found are confined to types which share the same size
2631 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2632 @opindex Wundeclared-selector
2633 @opindex Wno-undeclared-selector
2634 Warn if a @code{@@selector(@dots{})} expression referring to an
2635 undeclared selector is found. A selector is considered undeclared if no
2636 method with that name has been declared before the
2637 @code{@@selector(@dots{})} expression, either explicitly in an
2638 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2639 an @code{@@implementation} section. This option always performs its
2640 checks as soon as a @code{@@selector(@dots{})} expression is found,
2641 while @option{-Wselector} only performs its checks in the final stage of
2642 compilation. This also enforces the coding style convention
2643 that methods and selectors must be declared before being used.
2645 @item -print-objc-runtime-info
2646 @opindex print-objc-runtime-info
2647 Generate C header describing the largest structure that is passed by
2652 @node Language Independent Options
2653 @section Options to Control Diagnostic Messages Formatting
2654 @cindex options to control diagnostics formatting
2655 @cindex diagnostic messages
2656 @cindex message formatting
2658 Traditionally, diagnostic messages have been formatted irrespective of
2659 the output device's aspect (e.g.@: its width, @dots{}). The options described
2660 below can be used to control the diagnostic messages formatting
2661 algorithm, e.g.@: how many characters per line, how often source location
2662 information should be reported. Right now, only the C++ front end can
2663 honor these options. However it is expected, in the near future, that
2664 the remaining front ends would be able to digest them correctly.
2667 @item -fmessage-length=@var{n}
2668 @opindex fmessage-length
2669 Try to format error messages so that they fit on lines of about @var{n}
2670 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2671 the front ends supported by GCC@. If @var{n} is zero, then no
2672 line-wrapping will be done; each error message will appear on a single
2675 @opindex fdiagnostics-show-location
2676 @item -fdiagnostics-show-location=once
2677 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2678 reporter to emit @emph{once} source location information; that is, in
2679 case the message is too long to fit on a single physical line and has to
2680 be wrapped, the source location won't be emitted (as prefix) again,
2681 over and over, in subsequent continuation lines. This is the default
2684 @item -fdiagnostics-show-location=every-line
2685 Only meaningful in line-wrapping mode. Instructs the diagnostic
2686 messages reporter to emit the same source location information (as
2687 prefix) for physical lines that result from the process of breaking
2688 a message which is too long to fit on a single line.
2690 @item -fdiagnostics-show-option
2691 @opindex fdiagnostics-show-option
2692 This option instructs the diagnostic machinery to add text to each
2693 diagnostic emitted, which indicates which command line option directly
2694 controls that diagnostic, when such an option is known to the
2695 diagnostic machinery.
2697 @item -Wcoverage-mismatch
2698 @opindex Wcoverage-mismatch
2699 Warn if feedback profiles do not match when using the
2700 @option{-fprofile-use} option.
2701 If a source file was changed between @option{-fprofile-gen} and
2702 @option{-fprofile-use}, the files with the profile feedback can fail
2703 to match the source file and GCC can not use the profile feedback
2704 information. By default, GCC emits an error message in this case.
2705 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2706 error. GCC does not use appropriate feedback profiles, so using this
2707 option can result in poorly optimized code. This option is useful
2708 only in the case of very minor changes such as bug fixes to an
2713 @node Warning Options
2714 @section Options to Request or Suppress Warnings
2715 @cindex options to control warnings
2716 @cindex warning messages
2717 @cindex messages, warning
2718 @cindex suppressing warnings
2720 Warnings are diagnostic messages that report constructions which
2721 are not inherently erroneous but which are risky or suggest there
2722 may have been an error.
2724 The following language-independent options do not enable specific
2725 warnings but control the kinds of diagnostics produced by GCC.
2728 @cindex syntax checking
2730 @opindex fsyntax-only
2731 Check the code for syntax errors, but don't do anything beyond that.
2735 Inhibit all warning messages.
2740 Make all warnings into errors.
2745 Make the specified warning into an error. The specifier for a warning
2746 is appended, for example @option{-Werror=switch} turns the warnings
2747 controlled by @option{-Wswitch} into errors. This switch takes a
2748 negative form, to be used to negate @option{-Werror} for specific
2749 warnings, for example @option{-Wno-error=switch} makes
2750 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2751 is in effect. You can use the @option{-fdiagnostics-show-option}
2752 option to have each controllable warning amended with the option which
2753 controls it, to determine what to use with this option.
2755 Note that specifying @option{-Werror=}@var{foo} automatically implies
2756 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2759 @item -Wfatal-errors
2760 @opindex Wfatal-errors
2761 @opindex Wno-fatal-errors
2762 This option causes the compiler to abort compilation on the first error
2763 occurred rather than trying to keep going and printing further error
2768 You can request many specific warnings with options beginning
2769 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2770 implicit declarations. Each of these specific warning options also
2771 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2772 example, @option{-Wno-implicit}. This manual lists only one of the
2773 two forms, whichever is not the default. For further,
2774 language-specific options also refer to @ref{C++ Dialect Options} and
2775 @ref{Objective-C and Objective-C++ Dialect Options}.
2780 Issue all the warnings demanded by strict ISO C and ISO C++;
2781 reject all programs that use forbidden extensions, and some other
2782 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2783 version of the ISO C standard specified by any @option{-std} option used.
2785 Valid ISO C and ISO C++ programs should compile properly with or without
2786 this option (though a rare few will require @option{-ansi} or a
2787 @option{-std} option specifying the required version of ISO C)@. However,
2788 without this option, certain GNU extensions and traditional C and C++
2789 features are supported as well. With this option, they are rejected.
2791 @option{-pedantic} does not cause warning messages for use of the
2792 alternate keywords whose names begin and end with @samp{__}. Pedantic
2793 warnings are also disabled in the expression that follows
2794 @code{__extension__}. However, only system header files should use
2795 these escape routes; application programs should avoid them.
2796 @xref{Alternate Keywords}.
2798 Some users try to use @option{-pedantic} to check programs for strict ISO
2799 C conformance. They soon find that it does not do quite what they want:
2800 it finds some non-ISO practices, but not all---only those for which
2801 ISO C @emph{requires} a diagnostic, and some others for which
2802 diagnostics have been added.
2804 A feature to report any failure to conform to ISO C might be useful in
2805 some instances, but would require considerable additional work and would
2806 be quite different from @option{-pedantic}. We don't have plans to
2807 support such a feature in the near future.
2809 Where the standard specified with @option{-std} represents a GNU
2810 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2811 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2812 extended dialect is based. Warnings from @option{-pedantic} are given
2813 where they are required by the base standard. (It would not make sense
2814 for such warnings to be given only for features not in the specified GNU
2815 C dialect, since by definition the GNU dialects of C include all
2816 features the compiler supports with the given option, and there would be
2817 nothing to warn about.)
2819 @item -pedantic-errors
2820 @opindex pedantic-errors
2821 Like @option{-pedantic}, except that errors are produced rather than
2827 This enables all the warnings about constructions that some users
2828 consider questionable, and that are easy to avoid (or modify to
2829 prevent the warning), even in conjunction with macros. This also
2830 enables some language-specific warnings described in @ref{C++ Dialect
2831 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2833 @option{-Wall} turns on the following warning flags:
2835 @gccoptlist{-Waddress @gol
2836 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2838 -Wchar-subscripts @gol
2839 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2841 -Wimplicit-function-declaration @gol
2844 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2845 -Wmissing-braces @gol
2851 -Wsequence-point @gol
2852 -Wsign-compare @r{(only in C++)} @gol
2853 -Wstrict-aliasing @gol
2854 -Wstrict-overflow=1 @gol
2857 -Wuninitialized @gol
2858 -Wunknown-pragmas @gol
2859 -Wunused-function @gol
2862 -Wunused-variable @gol
2863 -Wvolatile-register-var @gol
2866 Note that some warning flags are not implied by @option{-Wall}. Some of
2867 them warn about constructions that users generally do not consider
2868 questionable, but which occasionally you might wish to check for;
2869 others warn about constructions that are necessary or hard to avoid in
2870 some cases, and there is no simple way to modify the code to suppress
2871 the warning. Some of them are enabled by @option{-Wextra} but many of
2872 them must be enabled individually.
2878 This enables some extra warning flags that are not enabled by
2879 @option{-Wall}. (This option used to be called @option{-W}. The older
2880 name is still supported, but the newer name is more descriptive.)
2882 @gccoptlist{-Wclobbered @gol
2884 -Wignored-qualifiers @gol
2885 -Wmissing-field-initializers @gol
2886 -Wmissing-parameter-type @r{(C only)} @gol
2887 -Wold-style-declaration @r{(C only)} @gol
2888 -Woverride-init @gol
2891 -Wuninitialized @gol
2892 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2895 The option @option{-Wextra} also prints warning messages for the
2901 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2902 @samp{>}, or @samp{>=}.
2905 (C++ only) An enumerator and a non-enumerator both appear in a
2906 conditional expression.
2909 (C++ only) Ambiguous virtual bases.
2912 (C++ only) Subscripting an array which has been declared @samp{register}.
2915 (C++ only) Taking the address of a variable which has been declared
2919 (C++ only) A base class is not initialized in a derived class' copy
2924 @item -Wchar-subscripts
2925 @opindex Wchar-subscripts
2926 @opindex Wno-char-subscripts
2927 Warn if an array subscript has type @code{char}. This is a common cause
2928 of error, as programmers often forget that this type is signed on some
2930 This warning is enabled by @option{-Wall}.
2934 @opindex Wno-comment
2935 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2936 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2937 This warning is enabled by @option{-Wall}.
2942 @opindex ffreestanding
2943 @opindex fno-builtin
2944 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2945 the arguments supplied have types appropriate to the format string
2946 specified, and that the conversions specified in the format string make
2947 sense. This includes standard functions, and others specified by format
2948 attributes (@pxref{Function Attributes}), in the @code{printf},
2949 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2950 not in the C standard) families (or other target-specific families).
2951 Which functions are checked without format attributes having been
2952 specified depends on the standard version selected, and such checks of
2953 functions without the attribute specified are disabled by
2954 @option{-ffreestanding} or @option{-fno-builtin}.
2956 The formats are checked against the format features supported by GNU
2957 libc version 2.2. These include all ISO C90 and C99 features, as well
2958 as features from the Single Unix Specification and some BSD and GNU
2959 extensions. Other library implementations may not support all these
2960 features; GCC does not support warning about features that go beyond a
2961 particular library's limitations. However, if @option{-pedantic} is used
2962 with @option{-Wformat}, warnings will be given about format features not
2963 in the selected standard version (but not for @code{strfmon} formats,
2964 since those are not in any version of the C standard). @xref{C Dialect
2965 Options,,Options Controlling C Dialect}.
2967 Since @option{-Wformat} also checks for null format arguments for
2968 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2970 @option{-Wformat} is included in @option{-Wall}. For more control over some
2971 aspects of format checking, the options @option{-Wformat-y2k},
2972 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2973 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2974 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2977 @opindex Wformat-y2k
2978 @opindex Wno-format-y2k
2979 If @option{-Wformat} is specified, also warn about @code{strftime}
2980 formats which may yield only a two-digit year.
2982 @item -Wno-format-contains-nul
2983 @opindex Wno-format-contains-nul
2984 @opindex Wformat-contains-nul
2985 If @option{-Wformat} is specified, do not warn about format strings that
2988 @item -Wno-format-extra-args
2989 @opindex Wno-format-extra-args
2990 @opindex Wformat-extra-args
2991 If @option{-Wformat} is specified, do not warn about excess arguments to a
2992 @code{printf} or @code{scanf} format function. The C standard specifies
2993 that such arguments are ignored.
2995 Where the unused arguments lie between used arguments that are
2996 specified with @samp{$} operand number specifications, normally
2997 warnings are still given, since the implementation could not know what
2998 type to pass to @code{va_arg} to skip the unused arguments. However,
2999 in the case of @code{scanf} formats, this option will suppress the
3000 warning if the unused arguments are all pointers, since the Single
3001 Unix Specification says that such unused arguments are allowed.
3003 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3004 @opindex Wno-format-zero-length
3005 @opindex Wformat-zero-length
3006 If @option{-Wformat} is specified, do not warn about zero-length formats.
3007 The C standard specifies that zero-length formats are allowed.
3009 @item -Wformat-nonliteral
3010 @opindex Wformat-nonliteral
3011 @opindex Wno-format-nonliteral
3012 If @option{-Wformat} is specified, also warn if the format string is not a
3013 string literal and so cannot be checked, unless the format function
3014 takes its format arguments as a @code{va_list}.
3016 @item -Wformat-security
3017 @opindex Wformat-security
3018 @opindex Wno-format-security
3019 If @option{-Wformat} is specified, also warn about uses of format
3020 functions that represent possible security problems. At present, this
3021 warns about calls to @code{printf} and @code{scanf} functions where the
3022 format string is not a string literal and there are no format arguments,
3023 as in @code{printf (foo);}. This may be a security hole if the format
3024 string came from untrusted input and contains @samp{%n}. (This is
3025 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3026 in future warnings may be added to @option{-Wformat-security} that are not
3027 included in @option{-Wformat-nonliteral}.)
3031 @opindex Wno-format=2
3032 Enable @option{-Wformat} plus format checks not included in
3033 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3034 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3036 @item -Wnonnull @r{(C and Objective-C only)}
3038 @opindex Wno-nonnull
3039 Warn about passing a null pointer for arguments marked as
3040 requiring a non-null value by the @code{nonnull} function attribute.
3042 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3043 can be disabled with the @option{-Wno-nonnull} option.
3045 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3047 @opindex Wno-init-self
3048 Warn about uninitialized variables which are initialized with themselves.
3049 Note this option can only be used with the @option{-Wuninitialized} option.
3051 For example, GCC will warn about @code{i} being uninitialized in the
3052 following snippet only when @option{-Winit-self} has been specified:
3063 @item -Wimplicit-int @r{(C and Objective-C only)}
3064 @opindex Wimplicit-int
3065 @opindex Wno-implicit-int
3066 Warn when a declaration does not specify a type.
3067 This warning is enabled by @option{-Wall}.
3069 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3070 @opindex Wimplicit-function-declaration
3071 @opindex Wno-implicit-function-declaration
3072 Give a warning whenever a function is used before being declared. In
3073 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3074 enabled by default and it is made into an error by
3075 @option{-pedantic-errors}. This warning is also enabled by
3080 @opindex Wno-implicit
3081 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3082 This warning is enabled by @option{-Wall}.
3084 @item -Wignored-qualifiers @r{(C and C++ only)}
3085 @opindex Wignored-qualifiers
3086 @opindex Wno-ignored-qualifiers
3087 Warn if the return type of a function has a type qualifier
3088 such as @code{const}. For ISO C such a type qualifier has no effect,
3089 since the value returned by a function is not an lvalue.
3090 For C++, the warning is only emitted for scalar types or @code{void}.
3091 ISO C prohibits qualified @code{void} return types on function
3092 definitions, so such return types always receive a warning
3093 even without this option.
3095 This warning is also enabled by @option{-Wextra}.
3100 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3101 a function with external linkage, returning int, taking either zero
3102 arguments, two, or three arguments of appropriate types. This warning
3103 is enabled by default in C++ and is enabled by either @option{-Wall}
3104 or @option{-pedantic}.
3106 @item -Wmissing-braces
3107 @opindex Wmissing-braces
3108 @opindex Wno-missing-braces
3109 Warn if an aggregate or union initializer is not fully bracketed. In
3110 the following example, the initializer for @samp{a} is not fully
3111 bracketed, but that for @samp{b} is fully bracketed.
3114 int a[2][2] = @{ 0, 1, 2, 3 @};
3115 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3118 This warning is enabled by @option{-Wall}.
3120 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3121 @opindex Wmissing-include-dirs
3122 @opindex Wno-missing-include-dirs
3123 Warn if a user-supplied include directory does not exist.
3126 @opindex Wparentheses
3127 @opindex Wno-parentheses
3128 Warn if parentheses are omitted in certain contexts, such
3129 as when there is an assignment in a context where a truth value
3130 is expected, or when operators are nested whose precedence people
3131 often get confused about.
3133 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3134 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3135 interpretation from that of ordinary mathematical notation.
3137 Also warn about constructions where there may be confusion to which
3138 @code{if} statement an @code{else} branch belongs. Here is an example of
3153 In C/C++, every @code{else} branch belongs to the innermost possible
3154 @code{if} statement, which in this example is @code{if (b)}. This is
3155 often not what the programmer expected, as illustrated in the above
3156 example by indentation the programmer chose. When there is the
3157 potential for this confusion, GCC will issue a warning when this flag
3158 is specified. To eliminate the warning, add explicit braces around
3159 the innermost @code{if} statement so there is no way the @code{else}
3160 could belong to the enclosing @code{if}. The resulting code would
3177 This warning is enabled by @option{-Wall}.
3179 @item -Wsequence-point
3180 @opindex Wsequence-point
3181 @opindex Wno-sequence-point
3182 Warn about code that may have undefined semantics because of violations
3183 of sequence point rules in the C and C++ standards.
3185 The C and C++ standards defines the order in which expressions in a C/C++
3186 program are evaluated in terms of @dfn{sequence points}, which represent
3187 a partial ordering between the execution of parts of the program: those
3188 executed before the sequence point, and those executed after it. These
3189 occur after the evaluation of a full expression (one which is not part
3190 of a larger expression), after the evaluation of the first operand of a
3191 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3192 function is called (but after the evaluation of its arguments and the
3193 expression denoting the called function), and in certain other places.
3194 Other than as expressed by the sequence point rules, the order of
3195 evaluation of subexpressions of an expression is not specified. All
3196 these rules describe only a partial order rather than a total order,
3197 since, for example, if two functions are called within one expression
3198 with no sequence point between them, the order in which the functions
3199 are called is not specified. However, the standards committee have
3200 ruled that function calls do not overlap.
3202 It is not specified when between sequence points modifications to the
3203 values of objects take effect. Programs whose behavior depends on this
3204 have undefined behavior; the C and C++ standards specify that ``Between
3205 the previous and next sequence point an object shall have its stored
3206 value modified at most once by the evaluation of an expression.
3207 Furthermore, the prior value shall be read only to determine the value
3208 to be stored.''. If a program breaks these rules, the results on any
3209 particular implementation are entirely unpredictable.
3211 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3212 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3213 diagnosed by this option, and it may give an occasional false positive
3214 result, but in general it has been found fairly effective at detecting
3215 this sort of problem in programs.
3217 The standard is worded confusingly, therefore there is some debate
3218 over the precise meaning of the sequence point rules in subtle cases.
3219 Links to discussions of the problem, including proposed formal
3220 definitions, may be found on the GCC readings page, at
3221 @w{@uref{http://gcc.gnu.org/readings.html}}.
3223 This warning is enabled by @option{-Wall} for C and C++.
3226 @opindex Wreturn-type
3227 @opindex Wno-return-type
3228 Warn whenever a function is defined with a return-type that defaults
3229 to @code{int}. Also warn about any @code{return} statement with no
3230 return-value in a function whose return-type is not @code{void}
3231 (falling off the end of the function body is considered returning
3232 without a value), and about a @code{return} statement with an
3233 expression in a function whose return-type is @code{void}.
3235 For C++, a function without return type always produces a diagnostic
3236 message, even when @option{-Wno-return-type} is specified. The only
3237 exceptions are @samp{main} and functions defined in system headers.
3239 This warning is enabled by @option{-Wall}.
3244 Warn whenever a @code{switch} statement has an index of enumerated type
3245 and lacks a @code{case} for one or more of the named codes of that
3246 enumeration. (The presence of a @code{default} label prevents this
3247 warning.) @code{case} labels outside the enumeration range also
3248 provoke warnings when this option is used (even if there is a
3249 @code{default} label).
3250 This warning is enabled by @option{-Wall}.
3252 @item -Wswitch-default
3253 @opindex Wswitch-default
3254 @opindex Wno-switch-default
3255 Warn whenever a @code{switch} statement does not have a @code{default}
3259 @opindex Wswitch-enum
3260 @opindex Wno-switch-enum
3261 Warn whenever a @code{switch} statement has an index of enumerated type
3262 and lacks a @code{case} for one or more of the named codes of that
3263 enumeration. @code{case} labels outside the enumeration range also
3264 provoke warnings when this option is used. The only difference
3265 between @option{-Wswitch} and this option is that this option gives a
3266 warning about an omitted enumeration code even if there is a
3267 @code{default} label.
3269 @item -Wsync-nand @r{(C and C++ only)}
3271 @opindex Wno-sync-nand
3272 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3273 built-in functions are used. These functions changed semantics in GCC 4.4.
3277 @opindex Wno-trigraphs
3278 Warn if any trigraphs are encountered that might change the meaning of
3279 the program (trigraphs within comments are not warned about).
3280 This warning is enabled by @option{-Wall}.
3282 @item -Wunused-function
3283 @opindex Wunused-function
3284 @opindex Wno-unused-function
3285 Warn whenever a static function is declared but not defined or a
3286 non-inline static function is unused.
3287 This warning is enabled by @option{-Wall}.
3289 @item -Wunused-label
3290 @opindex Wunused-label
3291 @opindex Wno-unused-label
3292 Warn whenever a label is declared but not used.
3293 This warning is enabled by @option{-Wall}.
3295 To suppress this warning use the @samp{unused} attribute
3296 (@pxref{Variable Attributes}).
3298 @item -Wunused-parameter
3299 @opindex Wunused-parameter
3300 @opindex Wno-unused-parameter
3301 Warn whenever a function parameter is unused aside from its declaration.
3303 To suppress this warning use the @samp{unused} attribute
3304 (@pxref{Variable Attributes}).
3306 @item -Wno-unused-result
3307 @opindex Wunused-result
3308 @opindex Wno-unused-result
3309 Do not warn if a caller of a function marked with attribute
3310 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3311 its return value. The default is @option{-Wunused-result}.
3313 @item -Wunused-variable
3314 @opindex Wunused-variable
3315 @opindex Wno-unused-variable
3316 Warn whenever a local variable or non-constant static variable is unused
3317 aside from its declaration.
3318 This warning is enabled by @option{-Wall}.
3320 To suppress this warning use the @samp{unused} attribute
3321 (@pxref{Variable Attributes}).
3323 @item -Wunused-value
3324 @opindex Wunused-value
3325 @opindex Wno-unused-value
3326 Warn whenever a statement computes a result that is explicitly not
3327 used. To suppress this warning cast the unused expression to
3328 @samp{void}. This includes an expression-statement or the left-hand
3329 side of a comma expression that contains no side effects. For example,
3330 an expression such as @samp{x[i,j]} will cause a warning, while
3331 @samp{x[(void)i,j]} will not.
3333 This warning is enabled by @option{-Wall}.
3338 All the above @option{-Wunused} options combined.
3340 In order to get a warning about an unused function parameter, you must
3341 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3342 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3344 @item -Wuninitialized
3345 @opindex Wuninitialized
3346 @opindex Wno-uninitialized
3347 Warn if an automatic variable is used without first being initialized
3348 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3349 warn if a non-static reference or non-static @samp{const} member
3350 appears in a class without constructors.
3352 If you want to warn about code which uses the uninitialized value of the
3353 variable in its own initializer, use the @option{-Winit-self} option.
3355 These warnings occur for individual uninitialized or clobbered
3356 elements of structure, union or array variables as well as for
3357 variables which are uninitialized or clobbered as a whole. They do
3358 not occur for variables or elements declared @code{volatile}. Because
3359 these warnings depend on optimization, the exact variables or elements
3360 for which there are warnings will depend on the precise optimization
3361 options and version of GCC used.
3363 Note that there may be no warning about a variable that is used only
3364 to compute a value that itself is never used, because such
3365 computations may be deleted by data flow analysis before the warnings
3368 These warnings are made optional because GCC is not smart
3369 enough to see all the reasons why the code might be correct
3370 despite appearing to have an error. Here is one example of how
3391 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3392 always initialized, but GCC doesn't know this. Here is
3393 another common case:
3398 if (change_y) save_y = y, y = new_y;
3400 if (change_y) y = save_y;
3405 This has no bug because @code{save_y} is used only if it is set.
3407 @cindex @code{longjmp} warnings
3408 This option also warns when a non-volatile automatic variable might be
3409 changed by a call to @code{longjmp}. These warnings as well are possible
3410 only in optimizing compilation.
3412 The compiler sees only the calls to @code{setjmp}. It cannot know
3413 where @code{longjmp} will be called; in fact, a signal handler could
3414 call it at any point in the code. As a result, you may get a warning
3415 even when there is in fact no problem because @code{longjmp} cannot
3416 in fact be called at the place which would cause a problem.
3418 Some spurious warnings can be avoided if you declare all the functions
3419 you use that never return as @code{noreturn}. @xref{Function
3422 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3424 @item -Wunknown-pragmas
3425 @opindex Wunknown-pragmas
3426 @opindex Wno-unknown-pragmas
3427 @cindex warning for unknown pragmas
3428 @cindex unknown pragmas, warning
3429 @cindex pragmas, warning of unknown
3430 Warn when a #pragma directive is encountered which is not understood by
3431 GCC@. If this command line option is used, warnings will even be issued
3432 for unknown pragmas in system header files. This is not the case if
3433 the warnings were only enabled by the @option{-Wall} command line option.
3436 @opindex Wno-pragmas
3438 Do not warn about misuses of pragmas, such as incorrect parameters,
3439 invalid syntax, or conflicts between pragmas. See also
3440 @samp{-Wunknown-pragmas}.
3442 @item -Wstrict-aliasing
3443 @opindex Wstrict-aliasing
3444 @opindex Wno-strict-aliasing
3445 This option is only active when @option{-fstrict-aliasing} is active.
3446 It warns about code which might break the strict aliasing rules that the
3447 compiler is using for optimization. The warning does not catch all
3448 cases, but does attempt to catch the more common pitfalls. It is
3449 included in @option{-Wall}.
3450 It is equivalent to @option{-Wstrict-aliasing=3}
3452 @item -Wstrict-aliasing=n
3453 @opindex Wstrict-aliasing=n
3454 @opindex Wno-strict-aliasing=n
3455 This option is only active when @option{-fstrict-aliasing} is active.
3456 It warns about code which might break the strict aliasing rules that the
3457 compiler is using for optimization.
3458 Higher levels correspond to higher accuracy (fewer false positives).
3459 Higher levels also correspond to more effort, similar to the way -O works.
3460 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3463 Level 1: Most aggressive, quick, least accurate.
3464 Possibly useful when higher levels
3465 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3466 false negatives. However, it has many false positives.
3467 Warns for all pointer conversions between possibly incompatible types,
3468 even if never dereferenced. Runs in the frontend only.
3470 Level 2: Aggressive, quick, not too precise.
3471 May still have many false positives (not as many as level 1 though),
3472 and few false negatives (but possibly more than level 1).
3473 Unlike level 1, it only warns when an address is taken. Warns about
3474 incomplete types. Runs in the frontend only.
3476 Level 3 (default for @option{-Wstrict-aliasing}):
3477 Should have very few false positives and few false
3478 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3479 Takes care of the common punn+dereference pattern in the frontend:
3480 @code{*(int*)&some_float}.
3481 If optimization is enabled, it also runs in the backend, where it deals
3482 with multiple statement cases using flow-sensitive points-to information.
3483 Only warns when the converted pointer is dereferenced.
3484 Does not warn about incomplete types.
3486 @item -Wstrict-overflow
3487 @itemx -Wstrict-overflow=@var{n}
3488 @opindex Wstrict-overflow
3489 @opindex Wno-strict-overflow
3490 This option is only active when @option{-fstrict-overflow} is active.
3491 It warns about cases where the compiler optimizes based on the
3492 assumption that signed overflow does not occur. Note that it does not
3493 warn about all cases where the code might overflow: it only warns
3494 about cases where the compiler implements some optimization. Thus
3495 this warning depends on the optimization level.
3497 An optimization which assumes that signed overflow does not occur is
3498 perfectly safe if the values of the variables involved are such that
3499 overflow never does, in fact, occur. Therefore this warning can
3500 easily give a false positive: a warning about code which is not
3501 actually a problem. To help focus on important issues, several
3502 warning levels are defined. No warnings are issued for the use of
3503 undefined signed overflow when estimating how many iterations a loop
3504 will require, in particular when determining whether a loop will be
3508 @item -Wstrict-overflow=1
3509 Warn about cases which are both questionable and easy to avoid. For
3510 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3511 compiler will simplify this to @code{1}. This level of
3512 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3513 are not, and must be explicitly requested.
3515 @item -Wstrict-overflow=2
3516 Also warn about other cases where a comparison is simplified to a
3517 constant. For example: @code{abs (x) >= 0}. This can only be
3518 simplified when @option{-fstrict-overflow} is in effect, because
3519 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3520 zero. @option{-Wstrict-overflow} (with no level) is the same as
3521 @option{-Wstrict-overflow=2}.
3523 @item -Wstrict-overflow=3
3524 Also warn about other cases where a comparison is simplified. For
3525 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3527 @item -Wstrict-overflow=4
3528 Also warn about other simplifications not covered by the above cases.
3529 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3531 @item -Wstrict-overflow=5
3532 Also warn about cases where the compiler reduces the magnitude of a
3533 constant involved in a comparison. For example: @code{x + 2 > y} will
3534 be simplified to @code{x + 1 >= y}. This is reported only at the
3535 highest warning level because this simplification applies to many
3536 comparisons, so this warning level will give a very large number of
3540 @item -Warray-bounds
3541 @opindex Wno-array-bounds
3542 @opindex Warray-bounds
3543 This option is only active when @option{-ftree-vrp} is active
3544 (default for -O2 and above). It warns about subscripts to arrays
3545 that are always out of bounds. This warning is enabled by @option{-Wall}.
3547 @item -Wno-div-by-zero
3548 @opindex Wno-div-by-zero
3549 @opindex Wdiv-by-zero
3550 Do not warn about compile-time integer division by zero. Floating point
3551 division by zero is not warned about, as it can be a legitimate way of
3552 obtaining infinities and NaNs.
3554 @item -Wsystem-headers
3555 @opindex Wsystem-headers
3556 @opindex Wno-system-headers
3557 @cindex warnings from system headers
3558 @cindex system headers, warnings from
3559 Print warning messages for constructs found in system header files.
3560 Warnings from system headers are normally suppressed, on the assumption
3561 that they usually do not indicate real problems and would only make the
3562 compiler output harder to read. Using this command line option tells
3563 GCC to emit warnings from system headers as if they occurred in user
3564 code. However, note that using @option{-Wall} in conjunction with this
3565 option will @emph{not} warn about unknown pragmas in system
3566 headers---for that, @option{-Wunknown-pragmas} must also be used.
3569 @opindex Wfloat-equal
3570 @opindex Wno-float-equal
3571 Warn if floating point values are used in equality comparisons.
3573 The idea behind this is that sometimes it is convenient (for the
3574 programmer) to consider floating-point values as approximations to
3575 infinitely precise real numbers. If you are doing this, then you need
3576 to compute (by analyzing the code, or in some other way) the maximum or
3577 likely maximum error that the computation introduces, and allow for it
3578 when performing comparisons (and when producing output, but that's a
3579 different problem). In particular, instead of testing for equality, you
3580 would check to see whether the two values have ranges that overlap; and
3581 this is done with the relational operators, so equality comparisons are
3584 @item -Wtraditional @r{(C and Objective-C only)}
3585 @opindex Wtraditional
3586 @opindex Wno-traditional
3587 Warn about certain constructs that behave differently in traditional and
3588 ISO C@. Also warn about ISO C constructs that have no traditional C
3589 equivalent, and/or problematic constructs which should be avoided.
3593 Macro parameters that appear within string literals in the macro body.
3594 In traditional C macro replacement takes place within string literals,
3595 but does not in ISO C@.
3598 In traditional C, some preprocessor directives did not exist.
3599 Traditional preprocessors would only consider a line to be a directive
3600 if the @samp{#} appeared in column 1 on the line. Therefore
3601 @option{-Wtraditional} warns about directives that traditional C
3602 understands but would ignore because the @samp{#} does not appear as the
3603 first character on the line. It also suggests you hide directives like
3604 @samp{#pragma} not understood by traditional C by indenting them. Some
3605 traditional implementations would not recognize @samp{#elif}, so it
3606 suggests avoiding it altogether.
3609 A function-like macro that appears without arguments.
3612 The unary plus operator.
3615 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3616 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3617 constants.) Note, these suffixes appear in macros defined in the system
3618 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3619 Use of these macros in user code might normally lead to spurious
3620 warnings, however GCC's integrated preprocessor has enough context to
3621 avoid warning in these cases.
3624 A function declared external in one block and then used after the end of
3628 A @code{switch} statement has an operand of type @code{long}.
3631 A non-@code{static} function declaration follows a @code{static} one.
3632 This construct is not accepted by some traditional C compilers.
3635 The ISO type of an integer constant has a different width or
3636 signedness from its traditional type. This warning is only issued if
3637 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3638 typically represent bit patterns, are not warned about.
3641 Usage of ISO string concatenation is detected.
3644 Initialization of automatic aggregates.
3647 Identifier conflicts with labels. Traditional C lacks a separate
3648 namespace for labels.
3651 Initialization of unions. If the initializer is zero, the warning is
3652 omitted. This is done under the assumption that the zero initializer in
3653 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3654 initializer warnings and relies on default initialization to zero in the
3658 Conversions by prototypes between fixed/floating point values and vice
3659 versa. The absence of these prototypes when compiling with traditional
3660 C would cause serious problems. This is a subset of the possible
3661 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3664 Use of ISO C style function definitions. This warning intentionally is
3665 @emph{not} issued for prototype declarations or variadic functions
3666 because these ISO C features will appear in your code when using
3667 libiberty's traditional C compatibility macros, @code{PARAMS} and
3668 @code{VPARAMS}. This warning is also bypassed for nested functions
3669 because that feature is already a GCC extension and thus not relevant to
3670 traditional C compatibility.
3673 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3674 @opindex Wtraditional-conversion
3675 @opindex Wno-traditional-conversion
3676 Warn if a prototype causes a type conversion that is different from what
3677 would happen to the same argument in the absence of a prototype. This
3678 includes conversions of fixed point to floating and vice versa, and
3679 conversions changing the width or signedness of a fixed point argument
3680 except when the same as the default promotion.
3682 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3683 @opindex Wdeclaration-after-statement
3684 @opindex Wno-declaration-after-statement
3685 Warn when a declaration is found after a statement in a block. This
3686 construct, known from C++, was introduced with ISO C99 and is by default
3687 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3688 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3693 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3695 @item -Wno-endif-labels
3696 @opindex Wno-endif-labels
3697 @opindex Wendif-labels
3698 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3703 Warn whenever a local variable shadows another local variable, parameter or
3704 global variable or whenever a built-in function is shadowed.
3706 @item -Wlarger-than=@var{len}
3707 @opindex Wlarger-than=@var{len}
3708 @opindex Wlarger-than-@var{len}
3709 Warn whenever an object of larger than @var{len} bytes is defined.
3711 @item -Wframe-larger-than=@var{len}
3712 @opindex Wframe-larger-than
3713 Warn if the size of a function frame is larger than @var{len} bytes.
3714 The computation done to determine the stack frame size is approximate
3715 and not conservative.
3716 The actual requirements may be somewhat greater than @var{len}
3717 even if you do not get a warning. In addition, any space allocated
3718 via @code{alloca}, variable-length arrays, or related constructs
3719 is not included by the compiler when determining
3720 whether or not to issue a warning.
3722 @item -Wunsafe-loop-optimizations
3723 @opindex Wunsafe-loop-optimizations
3724 @opindex Wno-unsafe-loop-optimizations
3725 Warn if the loop cannot be optimized because the compiler could not
3726 assume anything on the bounds of the loop indices. With
3727 @option{-funsafe-loop-optimizations} warn if the compiler made
3730 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3731 @opindex Wno-pedantic-ms-format
3732 @opindex Wpedantic-ms-format
3733 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3734 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3735 depending on the MS runtime, when you are using the options @option{-Wformat}
3736 and @option{-pedantic} without gnu-extensions.
3738 @item -Wpointer-arith
3739 @opindex Wpointer-arith
3740 @opindex Wno-pointer-arith
3741 Warn about anything that depends on the ``size of'' a function type or
3742 of @code{void}. GNU C assigns these types a size of 1, for
3743 convenience in calculations with @code{void *} pointers and pointers
3744 to functions. In C++, warn also when an arithmetic operation involves
3745 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3748 @opindex Wtype-limits
3749 @opindex Wno-type-limits
3750 Warn if a comparison is always true or always false due to the limited
3751 range of the data type, but do not warn for constant expressions. For
3752 example, warn if an unsigned variable is compared against zero with
3753 @samp{<} or @samp{>=}. This warning is also enabled by
3756 @item -Wbad-function-cast @r{(C and Objective-C only)}
3757 @opindex Wbad-function-cast
3758 @opindex Wno-bad-function-cast
3759 Warn whenever a function call is cast to a non-matching type.
3760 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3762 @item -Wc++-compat @r{(C and Objective-C only)}
3763 Warn about ISO C constructs that are outside of the common subset of
3764 ISO C and ISO C++, e.g.@: request for implicit conversion from
3765 @code{void *} to a pointer to non-@code{void} type.
3767 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3768 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3769 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3770 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3774 @opindex Wno-cast-qual
3775 Warn whenever a pointer is cast so as to remove a type qualifier from
3776 the target type. For example, warn if a @code{const char *} is cast
3777 to an ordinary @code{char *}.
3779 Also warn when making a cast which introduces a type qualifier in an
3780 unsafe way. For example, casting @code{char **} to @code{const char **}
3781 is unsafe, as in this example:
3784 /* p is char ** value. */
3785 const char **q = (const char **) p;
3786 /* Assignment of readonly string to const char * is OK. */
3788 /* Now char** pointer points to read-only memory. */
3793 @opindex Wcast-align
3794 @opindex Wno-cast-align
3795 Warn whenever a pointer is cast such that the required alignment of the
3796 target is increased. For example, warn if a @code{char *} is cast to
3797 an @code{int *} on machines where integers can only be accessed at
3798 two- or four-byte boundaries.
3800 @item -Wwrite-strings
3801 @opindex Wwrite-strings
3802 @opindex Wno-write-strings
3803 When compiling C, give string constants the type @code{const
3804 char[@var{length}]} so that copying the address of one into a
3805 non-@code{const} @code{char *} pointer will get a warning. These
3806 warnings will help you find at compile time code that can try to write
3807 into a string constant, but only if you have been very careful about
3808 using @code{const} in declarations and prototypes. Otherwise, it will
3809 just be a nuisance. This is why we did not make @option{-Wall} request
3812 When compiling C++, warn about the deprecated conversion from string
3813 literals to @code{char *}. This warning is enabled by default for C++
3818 @opindex Wno-clobbered
3819 Warn for variables that might be changed by @samp{longjmp} or
3820 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3823 @opindex Wconversion
3824 @opindex Wno-conversion
3825 Warn for implicit conversions that may alter a value. This includes
3826 conversions between real and integer, like @code{abs (x)} when
3827 @code{x} is @code{double}; conversions between signed and unsigned,
3828 like @code{unsigned ui = -1}; and conversions to smaller types, like
3829 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3830 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3831 changed by the conversion like in @code{abs (2.0)}. Warnings about
3832 conversions between signed and unsigned integers can be disabled by
3833 using @option{-Wno-sign-conversion}.
3835 For C++, also warn for conversions between @code{NULL} and non-pointer
3836 types; confusing overload resolution for user-defined conversions; and
3837 conversions that will never use a type conversion operator:
3838 conversions to @code{void}, the same type, a base class or a reference
3839 to them. Warnings about conversions between signed and unsigned
3840 integers are disabled by default in C++ unless
3841 @option{-Wsign-conversion} is explicitly enabled.
3844 @opindex Wempty-body
3845 @opindex Wno-empty-body
3846 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3847 while} statement. This warning is also enabled by @option{-Wextra}.
3849 @item -Wenum-compare
3850 @opindex Wenum-compare
3851 @opindex Wno-enum-compare
3852 Warn about a comparison between values of different enum types. In C++
3853 this warning is enabled by default. In C this warning is enabled by
3856 @item -Wjump-misses-init @r{(C, Objective-C only)}
3857 @opindex Wjump-misses-init
3858 @opindex Wno-jump-misses-init
3859 Warn if a @code{goto} statement or a @code{switch} statement jumps
3860 forward across the initialization of a variable, or jumps backward to a
3861 label after the variable has been initialized. This only warns about
3862 variables which are initialized when they are declared. This warning is
3863 only supported for C and Objective C; in C++ this sort of branch is an
3866 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
3867 can be disabled with the @option{-Wno-jump-misses-init} option.
3869 @item -Wsign-compare
3870 @opindex Wsign-compare
3871 @opindex Wno-sign-compare
3872 @cindex warning for comparison of signed and unsigned values
3873 @cindex comparison of signed and unsigned values, warning
3874 @cindex signed and unsigned values, comparison warning
3875 Warn when a comparison between signed and unsigned values could produce
3876 an incorrect result when the signed value is converted to unsigned.
3877 This warning is also enabled by @option{-Wextra}; to get the other warnings
3878 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3880 @item -Wsign-conversion
3881 @opindex Wsign-conversion
3882 @opindex Wno-sign-conversion
3883 Warn for implicit conversions that may change the sign of an integer
3884 value, like assigning a signed integer expression to an unsigned
3885 integer variable. An explicit cast silences the warning. In C, this
3886 option is enabled also by @option{-Wconversion}.
3890 @opindex Wno-address
3891 Warn about suspicious uses of memory addresses. These include using
3892 the address of a function in a conditional expression, such as
3893 @code{void func(void); if (func)}, and comparisons against the memory
3894 address of a string literal, such as @code{if (x == "abc")}. Such
3895 uses typically indicate a programmer error: the address of a function
3896 always evaluates to true, so their use in a conditional usually
3897 indicate that the programmer forgot the parentheses in a function
3898 call; and comparisons against string literals result in unspecified
3899 behavior and are not portable in C, so they usually indicate that the
3900 programmer intended to use @code{strcmp}. This warning is enabled by
3904 @opindex Wlogical-op
3905 @opindex Wno-logical-op
3906 Warn about suspicious uses of logical operators in expressions.
3907 This includes using logical operators in contexts where a
3908 bit-wise operator is likely to be expected.
3910 @item -Waggregate-return
3911 @opindex Waggregate-return
3912 @opindex Wno-aggregate-return
3913 Warn if any functions that return structures or unions are defined or
3914 called. (In languages where you can return an array, this also elicits
3917 @item -Wno-attributes
3918 @opindex Wno-attributes
3919 @opindex Wattributes
3920 Do not warn if an unexpected @code{__attribute__} is used, such as
3921 unrecognized attributes, function attributes applied to variables,
3922 etc. This will not stop errors for incorrect use of supported
3925 @item -Wno-builtin-macro-redefined
3926 @opindex Wno-builtin-macro-redefined
3927 @opindex Wbuiltin-macro-redefined
3928 Do not warn if certain built-in macros are redefined. This suppresses
3929 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3930 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3932 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3933 @opindex Wstrict-prototypes
3934 @opindex Wno-strict-prototypes
3935 Warn if a function is declared or defined without specifying the
3936 argument types. (An old-style function definition is permitted without
3937 a warning if preceded by a declaration which specifies the argument
3940 @item -Wold-style-declaration @r{(C and Objective-C only)}
3941 @opindex Wold-style-declaration
3942 @opindex Wno-old-style-declaration
3943 Warn for obsolescent usages, according to the C Standard, in a
3944 declaration. For example, warn if storage-class specifiers like
3945 @code{static} are not the first things in a declaration. This warning
3946 is also enabled by @option{-Wextra}.
3948 @item -Wold-style-definition @r{(C and Objective-C only)}
3949 @opindex Wold-style-definition
3950 @opindex Wno-old-style-definition
3951 Warn if an old-style function definition is used. A warning is given
3952 even if there is a previous prototype.
3954 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3955 @opindex Wmissing-parameter-type
3956 @opindex Wno-missing-parameter-type
3957 A function parameter is declared without a type specifier in K&R-style
3964 This warning is also enabled by @option{-Wextra}.
3966 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3967 @opindex Wmissing-prototypes
3968 @opindex Wno-missing-prototypes
3969 Warn if a global function is defined without a previous prototype
3970 declaration. This warning is issued even if the definition itself
3971 provides a prototype. The aim is to detect global functions that fail
3972 to be declared in header files.
3974 @item -Wmissing-declarations
3975 @opindex Wmissing-declarations
3976 @opindex Wno-missing-declarations
3977 Warn if a global function is defined without a previous declaration.
3978 Do so even if the definition itself provides a prototype.
3979 Use this option to detect global functions that are not declared in
3980 header files. In C++, no warnings are issued for function templates,
3981 or for inline functions, or for functions in anonymous namespaces.
3983 @item -Wmissing-field-initializers
3984 @opindex Wmissing-field-initializers
3985 @opindex Wno-missing-field-initializers
3989 Warn if a structure's initializer has some fields missing. For
3990 example, the following code would cause such a warning, because
3991 @code{x.h} is implicitly zero:
3994 struct s @{ int f, g, h; @};
3995 struct s x = @{ 3, 4 @};
3998 This option does not warn about designated initializers, so the following
3999 modification would not trigger a warning:
4002 struct s @{ int f, g, h; @};
4003 struct s x = @{ .f = 3, .g = 4 @};
4006 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4007 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4009 @item -Wmissing-noreturn
4010 @opindex Wmissing-noreturn
4011 @opindex Wno-missing-noreturn
4012 Warn about functions which might be candidates for attribute @code{noreturn}.
4013 Note these are only possible candidates, not absolute ones. Care should
4014 be taken to manually verify functions actually do not ever return before
4015 adding the @code{noreturn} attribute, otherwise subtle code generation
4016 bugs could be introduced. You will not get a warning for @code{main} in
4017 hosted C environments.
4019 @item -Wmissing-format-attribute
4020 @opindex Wmissing-format-attribute
4021 @opindex Wno-missing-format-attribute
4024 Warn about function pointers which might be candidates for @code{format}
4025 attributes. Note these are only possible candidates, not absolute ones.
4026 GCC will guess that function pointers with @code{format} attributes that
4027 are used in assignment, initialization, parameter passing or return
4028 statements should have a corresponding @code{format} attribute in the
4029 resulting type. I.e.@: the left-hand side of the assignment or
4030 initialization, the type of the parameter variable, or the return type
4031 of the containing function respectively should also have a @code{format}
4032 attribute to avoid the warning.
4034 GCC will also warn about function definitions which might be
4035 candidates for @code{format} attributes. Again, these are only
4036 possible candidates. GCC will guess that @code{format} attributes
4037 might be appropriate for any function that calls a function like
4038 @code{vprintf} or @code{vscanf}, but this might not always be the
4039 case, and some functions for which @code{format} attributes are
4040 appropriate may not be detected.
4042 @item -Wno-multichar
4043 @opindex Wno-multichar
4045 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4046 Usually they indicate a typo in the user's code, as they have
4047 implementation-defined values, and should not be used in portable code.
4049 @item -Wnormalized=<none|id|nfc|nfkc>
4050 @opindex Wnormalized=
4053 @cindex character set, input normalization
4054 In ISO C and ISO C++, two identifiers are different if they are
4055 different sequences of characters. However, sometimes when characters
4056 outside the basic ASCII character set are used, you can have two
4057 different character sequences that look the same. To avoid confusion,
4058 the ISO 10646 standard sets out some @dfn{normalization rules} which
4059 when applied ensure that two sequences that look the same are turned into
4060 the same sequence. GCC can warn you if you are using identifiers which
4061 have not been normalized; this option controls that warning.
4063 There are four levels of warning that GCC supports. The default is
4064 @option{-Wnormalized=nfc}, which warns about any identifier which is
4065 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4066 recommended form for most uses.
4068 Unfortunately, there are some characters which ISO C and ISO C++ allow
4069 in identifiers that when turned into NFC aren't allowable as
4070 identifiers. That is, there's no way to use these symbols in portable
4071 ISO C or C++ and have all your identifiers in NFC@.
4072 @option{-Wnormalized=id} suppresses the warning for these characters.
4073 It is hoped that future versions of the standards involved will correct
4074 this, which is why this option is not the default.
4076 You can switch the warning off for all characters by writing
4077 @option{-Wnormalized=none}. You would only want to do this if you
4078 were using some other normalization scheme (like ``D''), because
4079 otherwise you can easily create bugs that are literally impossible to see.
4081 Some characters in ISO 10646 have distinct meanings but look identical
4082 in some fonts or display methodologies, especially once formatting has
4083 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4084 LETTER N'', will display just like a regular @code{n} which has been
4085 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4086 normalization scheme to convert all these into a standard form as
4087 well, and GCC will warn if your code is not in NFKC if you use
4088 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4089 about every identifier that contains the letter O because it might be
4090 confused with the digit 0, and so is not the default, but may be
4091 useful as a local coding convention if the programming environment is
4092 unable to be fixed to display these characters distinctly.
4094 @item -Wno-deprecated
4095 @opindex Wno-deprecated
4096 @opindex Wdeprecated
4097 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4099 @item -Wno-deprecated-declarations
4100 @opindex Wno-deprecated-declarations
4101 @opindex Wdeprecated-declarations
4102 Do not warn about uses of functions (@pxref{Function Attributes}),
4103 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4104 Attributes}) marked as deprecated by using the @code{deprecated}
4108 @opindex Wno-overflow
4110 Do not warn about compile-time overflow in constant expressions.
4112 @item -Woverride-init @r{(C and Objective-C only)}
4113 @opindex Woverride-init
4114 @opindex Wno-override-init
4118 Warn if an initialized field without side effects is overridden when
4119 using designated initializers (@pxref{Designated Inits, , Designated
4122 This warning is included in @option{-Wextra}. To get other
4123 @option{-Wextra} warnings without this one, use @samp{-Wextra
4124 -Wno-override-init}.
4129 Warn if a structure is given the packed attribute, but the packed
4130 attribute has no effect on the layout or size of the structure.
4131 Such structures may be mis-aligned for little benefit. For
4132 instance, in this code, the variable @code{f.x} in @code{struct bar}
4133 will be misaligned even though @code{struct bar} does not itself
4134 have the packed attribute:
4141 @} __attribute__((packed));
4149 @item -Wpacked-bitfield-compat
4150 @opindex Wpacked-bitfield-compat
4151 @opindex Wno-packed-bitfield-compat
4152 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4153 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4154 the change can lead to differences in the structure layout. GCC
4155 informs you when the offset of such a field has changed in GCC 4.4.
4156 For example there is no longer a 4-bit padding between field @code{a}
4157 and @code{b} in this structure:
4164 @} __attribute__ ((packed));
4167 This warning is enabled by default. Use
4168 @option{-Wno-packed-bitfield-compat} to disable this warning.
4173 Warn if padding is included in a structure, either to align an element
4174 of the structure or to align the whole structure. Sometimes when this
4175 happens it is possible to rearrange the fields of the structure to
4176 reduce the padding and so make the structure smaller.
4178 @item -Wredundant-decls
4179 @opindex Wredundant-decls
4180 @opindex Wno-redundant-decls
4181 Warn if anything is declared more than once in the same scope, even in
4182 cases where multiple declaration is valid and changes nothing.
4184 @item -Wnested-externs @r{(C and Objective-C only)}
4185 @opindex Wnested-externs
4186 @opindex Wno-nested-externs
4187 Warn if an @code{extern} declaration is encountered within a function.
4189 @item -Wunreachable-code
4190 @opindex Wunreachable-code
4191 @opindex Wno-unreachable-code
4192 Warn if the compiler detects that code will never be executed.
4194 This option is intended to warn when the compiler detects that at
4195 least a whole line of source code will never be executed, because
4196 some condition is never satisfied or because it is after a
4197 procedure that never returns.
4199 It is possible for this option to produce a warning even though there
4200 are circumstances under which part of the affected line can be executed,
4201 so care should be taken when removing apparently-unreachable code.
4203 For instance, when a function is inlined, a warning may mean that the
4204 line is unreachable in only one inlined copy of the function.
4206 This option is not made part of @option{-Wall} because in a debugging
4207 version of a program there is often substantial code which checks
4208 correct functioning of the program and is, hopefully, unreachable
4209 because the program does work. Another common use of unreachable
4210 code is to provide behavior which is selectable at compile-time.
4215 Warn if a function can not be inlined and it was declared as inline.
4216 Even with this option, the compiler will not warn about failures to
4217 inline functions declared in system headers.
4219 The compiler uses a variety of heuristics to determine whether or not
4220 to inline a function. For example, the compiler takes into account
4221 the size of the function being inlined and the amount of inlining
4222 that has already been done in the current function. Therefore,
4223 seemingly insignificant changes in the source program can cause the
4224 warnings produced by @option{-Winline} to appear or disappear.
4226 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4227 @opindex Wno-invalid-offsetof
4228 @opindex Winvalid-offsetof
4229 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4230 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4231 to a non-POD type is undefined. In existing C++ implementations,
4232 however, @samp{offsetof} typically gives meaningful results even when
4233 applied to certain kinds of non-POD types. (Such as a simple
4234 @samp{struct} that fails to be a POD type only by virtue of having a
4235 constructor.) This flag is for users who are aware that they are
4236 writing nonportable code and who have deliberately chosen to ignore the
4239 The restrictions on @samp{offsetof} may be relaxed in a future version
4240 of the C++ standard.
4242 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4243 @opindex Wno-int-to-pointer-cast
4244 @opindex Wint-to-pointer-cast
4245 Suppress warnings from casts to pointer type of an integer of a
4248 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4249 @opindex Wno-pointer-to-int-cast
4250 @opindex Wpointer-to-int-cast
4251 Suppress warnings from casts from a pointer to an integer type of a
4255 @opindex Winvalid-pch
4256 @opindex Wno-invalid-pch
4257 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4258 the search path but can't be used.
4262 @opindex Wno-long-long
4263 Warn if @samp{long long} type is used. This is enabled by either
4264 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4265 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4267 @item -Wvariadic-macros
4268 @opindex Wvariadic-macros
4269 @opindex Wno-variadic-macros
4270 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4271 alternate syntax when in pedantic ISO C99 mode. This is default.
4272 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4277 Warn if variable length array is used in the code.
4278 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4279 the variable length array.
4281 @item -Wvolatile-register-var
4282 @opindex Wvolatile-register-var
4283 @opindex Wno-volatile-register-var
4284 Warn if a register variable is declared volatile. The volatile
4285 modifier does not inhibit all optimizations that may eliminate reads
4286 and/or writes to register variables. This warning is enabled by
4289 @item -Wdisabled-optimization
4290 @opindex Wdisabled-optimization
4291 @opindex Wno-disabled-optimization
4292 Warn if a requested optimization pass is disabled. This warning does
4293 not generally indicate that there is anything wrong with your code; it
4294 merely indicates that GCC's optimizers were unable to handle the code
4295 effectively. Often, the problem is that your code is too big or too
4296 complex; GCC will refuse to optimize programs when the optimization
4297 itself is likely to take inordinate amounts of time.
4299 @item -Wpointer-sign @r{(C and Objective-C only)}
4300 @opindex Wpointer-sign
4301 @opindex Wno-pointer-sign
4302 Warn for pointer argument passing or assignment with different signedness.
4303 This option is only supported for C and Objective-C@. It is implied by
4304 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4305 @option{-Wno-pointer-sign}.
4307 @item -Wstack-protector
4308 @opindex Wstack-protector
4309 @opindex Wno-stack-protector
4310 This option is only active when @option{-fstack-protector} is active. It
4311 warns about functions that will not be protected against stack smashing.
4314 @opindex Wno-mudflap
4315 Suppress warnings about constructs that cannot be instrumented by
4318 @item -Woverlength-strings
4319 @opindex Woverlength-strings
4320 @opindex Wno-overlength-strings
4321 Warn about string constants which are longer than the ``minimum
4322 maximum'' length specified in the C standard. Modern compilers
4323 generally allow string constants which are much longer than the
4324 standard's minimum limit, but very portable programs should avoid
4325 using longer strings.
4327 The limit applies @emph{after} string constant concatenation, and does
4328 not count the trailing NUL@. In C89, the limit was 509 characters; in
4329 C99, it was raised to 4095. C++98 does not specify a normative
4330 minimum maximum, so we do not diagnose overlength strings in C++@.
4332 This option is implied by @option{-pedantic}, and can be disabled with
4333 @option{-Wno-overlength-strings}.
4335 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4336 @opindex Wunsuffixed-float-constants
4338 GCC will issue a warning for any floating constant that does not have
4339 a suffix. When used together with @option{-Wsystem-headers} it will
4340 warn about such constants in system header files. This can be useful
4341 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4342 from the decimal floating-point extension to C99.
4345 @node Debugging Options
4346 @section Options for Debugging Your Program or GCC
4347 @cindex options, debugging
4348 @cindex debugging information options
4350 GCC has various special options that are used for debugging
4351 either your program or GCC:
4356 Produce debugging information in the operating system's native format
4357 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4360 On most systems that use stabs format, @option{-g} enables use of extra
4361 debugging information that only GDB can use; this extra information
4362 makes debugging work better in GDB but will probably make other debuggers
4364 refuse to read the program. If you want to control for certain whether
4365 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4366 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4368 GCC allows you to use @option{-g} with
4369 @option{-O}. The shortcuts taken by optimized code may occasionally
4370 produce surprising results: some variables you declared may not exist
4371 at all; flow of control may briefly move where you did not expect it;
4372 some statements may not be executed because they compute constant
4373 results or their values were already at hand; some statements may
4374 execute in different places because they were moved out of loops.
4376 Nevertheless it proves possible to debug optimized output. This makes
4377 it reasonable to use the optimizer for programs that might have bugs.
4379 The following options are useful when GCC is generated with the
4380 capability for more than one debugging format.
4384 Produce debugging information for use by GDB@. This means to use the
4385 most expressive format available (DWARF 2, stabs, or the native format
4386 if neither of those are supported), including GDB extensions if at all
4391 Produce debugging information in stabs format (if that is supported),
4392 without GDB extensions. This is the format used by DBX on most BSD
4393 systems. On MIPS, Alpha and System V Release 4 systems this option
4394 produces stabs debugging output which is not understood by DBX or SDB@.
4395 On System V Release 4 systems this option requires the GNU assembler.
4397 @item -feliminate-unused-debug-symbols
4398 @opindex feliminate-unused-debug-symbols
4399 Produce debugging information in stabs format (if that is supported),
4400 for only symbols that are actually used.
4402 @item -femit-class-debug-always
4403 Instead of emitting debugging information for a C++ class in only one
4404 object file, emit it in all object files using the class. This option
4405 should be used only with debuggers that are unable to handle the way GCC
4406 normally emits debugging information for classes because using this
4407 option will increase the size of debugging information by as much as a
4412 Produce debugging information in stabs format (if that is supported),
4413 using GNU extensions understood only by the GNU debugger (GDB)@. The
4414 use of these extensions is likely to make other debuggers crash or
4415 refuse to read the program.
4419 Produce debugging information in COFF format (if that is supported).
4420 This is the format used by SDB on most System V systems prior to
4425 Produce debugging information in XCOFF format (if that is supported).
4426 This is the format used by the DBX debugger on IBM RS/6000 systems.
4430 Produce debugging information in XCOFF format (if that is supported),
4431 using GNU extensions understood only by the GNU debugger (GDB)@. The
4432 use of these extensions is likely to make other debuggers crash or
4433 refuse to read the program, and may cause assemblers other than the GNU
4434 assembler (GAS) to fail with an error.
4436 @item -gdwarf-@var{version}
4437 @opindex gdwarf-@var{version}
4438 Produce debugging information in DWARF format (if that is
4439 supported). This is the format used by DBX on IRIX 6. The value
4440 of @var{version} may be either 2, 3 or 4; the default version is 2.
4442 Note that with DWARF version 2 some ports require, and will always
4443 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4445 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4446 for maximum benefit.
4448 @item -gstrict-dwarf
4449 @opindex gstrict-dwarf
4450 Disallow using extensions of later DWARF standard version than selected
4451 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4452 DWARF extensions from later standard versions is allowed.
4454 @item -gno-strict-dwarf
4455 @opindex gno-strict-dwarf
4456 Allow using extensions of later DWARF standard version than selected with
4457 @option{-gdwarf-@var{version}}.
4461 Produce debugging information in VMS debug format (if that is
4462 supported). This is the format used by DEBUG on VMS systems.
4465 @itemx -ggdb@var{level}
4466 @itemx -gstabs@var{level}
4467 @itemx -gcoff@var{level}
4468 @itemx -gxcoff@var{level}
4469 @itemx -gvms@var{level}
4470 Request debugging information and also use @var{level} to specify how
4471 much information. The default level is 2.
4473 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4476 Level 1 produces minimal information, enough for making backtraces in
4477 parts of the program that you don't plan to debug. This includes
4478 descriptions of functions and external variables, but no information
4479 about local variables and no line numbers.
4481 Level 3 includes extra information, such as all the macro definitions
4482 present in the program. Some debuggers support macro expansion when
4483 you use @option{-g3}.
4485 @option{-gdwarf-2} does not accept a concatenated debug level, because
4486 GCC used to support an option @option{-gdwarf} that meant to generate
4487 debug information in version 1 of the DWARF format (which is very
4488 different from version 2), and it would have been too confusing. That
4489 debug format is long obsolete, but the option cannot be changed now.
4490 Instead use an additional @option{-g@var{level}} option to change the
4491 debug level for DWARF.
4495 Turn off generation of debug info, if leaving out this option would have
4496 generated it, or turn it on at level 2 otherwise. The position of this
4497 argument in the command line does not matter, it takes effect after all
4498 other options are processed, and it does so only once, no matter how
4499 many times it is given. This is mainly intended to be used with
4500 @option{-fcompare-debug}.
4502 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4503 @opindex fdump-final-insns
4504 Dump the final internal representation (RTL) to @var{file}. If the
4505 optional argument is omitted (or if @var{file} is @code{.}), the name
4506 of the dump file will be determined by appending @code{.gkd} to the
4507 compilation output file name.
4509 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4510 @opindex fcompare-debug
4511 @opindex fno-compare-debug
4512 If no error occurs during compilation, run the compiler a second time,
4513 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4514 passed to the second compilation. Dump the final internal
4515 representation in both compilations, and print an error if they differ.
4517 If the equal sign is omitted, the default @option{-gtoggle} is used.
4519 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4520 and nonzero, implicitly enables @option{-fcompare-debug}. If
4521 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4522 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4525 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4526 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4527 of the final representation and the second compilation, preventing even
4528 @env{GCC_COMPARE_DEBUG} from taking effect.
4530 To verify full coverage during @option{-fcompare-debug} testing, set
4531 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4532 which GCC will reject as an invalid option in any actual compilation
4533 (rather than preprocessing, assembly or linking). To get just a
4534 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4535 not overridden} will do.
4537 @item -fcompare-debug-second
4538 @opindex fcompare-debug-second
4539 This option is implicitly passed to the compiler for the second
4540 compilation requested by @option{-fcompare-debug}, along with options to
4541 silence warnings, and omitting other options that would cause
4542 side-effect compiler outputs to files or to the standard output. Dump
4543 files and preserved temporary files are renamed so as to contain the
4544 @code{.gk} additional extension during the second compilation, to avoid
4545 overwriting those generated by the first.
4547 When this option is passed to the compiler driver, it causes the
4548 @emph{first} compilation to be skipped, which makes it useful for little
4549 other than debugging the compiler proper.
4551 @item -feliminate-dwarf2-dups
4552 @opindex feliminate-dwarf2-dups
4553 Compress DWARF2 debugging information by eliminating duplicated
4554 information about each symbol. This option only makes sense when
4555 generating DWARF2 debugging information with @option{-gdwarf-2}.
4557 @item -femit-struct-debug-baseonly
4558 Emit debug information for struct-like types
4559 only when the base name of the compilation source file
4560 matches the base name of file in which the struct was defined.
4562 This option substantially reduces the size of debugging information,
4563 but at significant potential loss in type information to the debugger.
4564 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4565 See @option{-femit-struct-debug-detailed} for more detailed control.
4567 This option works only with DWARF 2.
4569 @item -femit-struct-debug-reduced
4570 Emit debug information for struct-like types
4571 only when the base name of the compilation source file
4572 matches the base name of file in which the type was defined,
4573 unless the struct is a template or defined in a system header.
4575 This option significantly reduces the size of debugging information,
4576 with some potential loss in type information to the debugger.
4577 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4578 See @option{-femit-struct-debug-detailed} for more detailed control.
4580 This option works only with DWARF 2.
4582 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4583 Specify the struct-like types
4584 for which the compiler will generate debug information.
4585 The intent is to reduce duplicate struct debug information
4586 between different object files within the same program.
4588 This option is a detailed version of
4589 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4590 which will serve for most needs.
4592 A specification has the syntax
4593 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4595 The optional first word limits the specification to
4596 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4597 A struct type is used directly when it is the type of a variable, member.
4598 Indirect uses arise through pointers to structs.
4599 That is, when use of an incomplete struct would be legal, the use is indirect.
4601 @samp{struct one direct; struct two * indirect;}.
4603 The optional second word limits the specification to
4604 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4605 Generic structs are a bit complicated to explain.
4606 For C++, these are non-explicit specializations of template classes,
4607 or non-template classes within the above.
4608 Other programming languages have generics,
4609 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4611 The third word specifies the source files for those
4612 structs for which the compiler will emit debug information.
4613 The values @samp{none} and @samp{any} have the normal meaning.
4614 The value @samp{base} means that
4615 the base of name of the file in which the type declaration appears
4616 must match the base of the name of the main compilation file.
4617 In practice, this means that
4618 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4619 but types declared in other header will not.
4620 The value @samp{sys} means those types satisfying @samp{base}
4621 or declared in system or compiler headers.
4623 You may need to experiment to determine the best settings for your application.
4625 The default is @samp{-femit-struct-debug-detailed=all}.
4627 This option works only with DWARF 2.
4629 @item -fenable-icf-debug
4630 @opindex fenable-icf-debug
4631 Generate additional debug information to support identical code folding (ICF).
4632 This option only works with DWARF version 2 or higher.
4634 @item -fno-merge-debug-strings
4635 @opindex fmerge-debug-strings
4636 @opindex fno-merge-debug-strings
4637 Direct the linker to not merge together strings in the debugging
4638 information which are identical in different object files. Merging is
4639 not supported by all assemblers or linkers. Merging decreases the size
4640 of the debug information in the output file at the cost of increasing
4641 link processing time. Merging is enabled by default.
4643 @item -fdebug-prefix-map=@var{old}=@var{new}
4644 @opindex fdebug-prefix-map
4645 When compiling files in directory @file{@var{old}}, record debugging
4646 information describing them as in @file{@var{new}} instead.
4648 @item -fno-dwarf2-cfi-asm
4649 @opindex fdwarf2-cfi-asm
4650 @opindex fno-dwarf2-cfi-asm
4651 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4652 instead of using GAS @code{.cfi_*} directives.
4654 @cindex @command{prof}
4657 Generate extra code to write profile information suitable for the
4658 analysis program @command{prof}. You must use this option when compiling
4659 the source files you want data about, and you must also use it when
4662 @cindex @command{gprof}
4665 Generate extra code to write profile information suitable for the
4666 analysis program @command{gprof}. You must use this option when compiling
4667 the source files you want data about, and you must also use it when
4672 Makes the compiler print out each function name as it is compiled, and
4673 print some statistics about each pass when it finishes.
4676 @opindex ftime-report
4677 Makes the compiler print some statistics about the time consumed by each
4678 pass when it finishes.
4681 @opindex fmem-report
4682 Makes the compiler print some statistics about permanent memory
4683 allocation when it finishes.
4685 @item -fpre-ipa-mem-report
4686 @opindex fpre-ipa-mem-report
4687 @item -fpost-ipa-mem-report
4688 @opindex fpost-ipa-mem-report
4689 Makes the compiler print some statistics about permanent memory
4690 allocation before or after interprocedural optimization.
4692 @item -fprofile-arcs
4693 @opindex fprofile-arcs
4694 Add code so that program flow @dfn{arcs} are instrumented. During
4695 execution the program records how many times each branch and call is
4696 executed and how many times it is taken or returns. When the compiled
4697 program exits it saves this data to a file called
4698 @file{@var{auxname}.gcda} for each source file. The data may be used for
4699 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4700 test coverage analysis (@option{-ftest-coverage}). Each object file's
4701 @var{auxname} is generated from the name of the output file, if
4702 explicitly specified and it is not the final executable, otherwise it is
4703 the basename of the source file. In both cases any suffix is removed
4704 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4705 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4706 @xref{Cross-profiling}.
4708 @cindex @command{gcov}
4712 This option is used to compile and link code instrumented for coverage
4713 analysis. The option is a synonym for @option{-fprofile-arcs}
4714 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4715 linking). See the documentation for those options for more details.
4720 Compile the source files with @option{-fprofile-arcs} plus optimization
4721 and code generation options. For test coverage analysis, use the
4722 additional @option{-ftest-coverage} option. You do not need to profile
4723 every source file in a program.
4726 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4727 (the latter implies the former).
4730 Run the program on a representative workload to generate the arc profile
4731 information. This may be repeated any number of times. You can run
4732 concurrent instances of your program, and provided that the file system
4733 supports locking, the data files will be correctly updated. Also
4734 @code{fork} calls are detected and correctly handled (double counting
4738 For profile-directed optimizations, compile the source files again with
4739 the same optimization and code generation options plus
4740 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4741 Control Optimization}).
4744 For test coverage analysis, use @command{gcov} to produce human readable
4745 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4746 @command{gcov} documentation for further information.
4750 With @option{-fprofile-arcs}, for each function of your program GCC
4751 creates a program flow graph, then finds a spanning tree for the graph.
4752 Only arcs that are not on the spanning tree have to be instrumented: the
4753 compiler adds code to count the number of times that these arcs are
4754 executed. When an arc is the only exit or only entrance to a block, the
4755 instrumentation code can be added to the block; otherwise, a new basic
4756 block must be created to hold the instrumentation code.
4759 @item -ftest-coverage
4760 @opindex ftest-coverage
4761 Produce a notes file that the @command{gcov} code-coverage utility
4762 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4763 show program coverage. Each source file's note file is called
4764 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4765 above for a description of @var{auxname} and instructions on how to
4766 generate test coverage data. Coverage data will match the source files
4767 more closely, if you do not optimize.
4769 @item -fdbg-cnt-list
4770 @opindex fdbg-cnt-list
4771 Print the name and the counter upperbound for all debug counters.
4773 @item -fdbg-cnt=@var{counter-value-list}
4775 Set the internal debug counter upperbound. @var{counter-value-list}
4776 is a comma-separated list of @var{name}:@var{value} pairs
4777 which sets the upperbound of each debug counter @var{name} to @var{value}.
4778 All debug counters have the initial upperbound of @var{UINT_MAX},
4779 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4780 e.g. With -fdbg-cnt=dce:10,tail_call:0
4781 dbg_cnt(dce) will return true only for first 10 invocations
4782 and dbg_cnt(tail_call) will return false always.
4784 @item -d@var{letters}
4785 @itemx -fdump-rtl-@var{pass}
4787 Says to make debugging dumps during compilation at times specified by
4788 @var{letters}. This is used for debugging the RTL-based passes of the
4789 compiler. The file names for most of the dumps are made by appending
4790 a pass number and a word to the @var{dumpname}, and the files are
4791 created in the directory of the output file. @var{dumpname} is
4792 generated from the name of the output file, if explicitly specified
4793 and it is not an executable, otherwise it is the basename of the
4794 source file. These switches may have different effects when
4795 @option{-E} is used for preprocessing.
4797 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4798 @option{-d} option @var{letters}. Here are the possible
4799 letters for use in @var{pass} and @var{letters}, and their meanings:
4803 @item -fdump-rtl-alignments
4804 @opindex fdump-rtl-alignments
4805 Dump after branch alignments have been computed.
4807 @item -fdump-rtl-asmcons
4808 @opindex fdump-rtl-asmcons
4809 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4811 @item -fdump-rtl-auto_inc_dec
4812 @opindex fdump-rtl-auto_inc_dec
4813 Dump after auto-inc-dec discovery. This pass is only run on
4814 architectures that have auto inc or auto dec instructions.
4816 @item -fdump-rtl-barriers
4817 @opindex fdump-rtl-barriers
4818 Dump after cleaning up the barrier instructions.
4820 @item -fdump-rtl-bbpart
4821 @opindex fdump-rtl-bbpart
4822 Dump after partitioning hot and cold basic blocks.
4824 @item -fdump-rtl-bbro
4825 @opindex fdump-rtl-bbro
4826 Dump after block reordering.
4828 @item -fdump-rtl-btl1
4829 @itemx -fdump-rtl-btl2
4830 @opindex fdump-rtl-btl2
4831 @opindex fdump-rtl-btl2
4832 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4833 after the two branch
4834 target load optimization passes.
4836 @item -fdump-rtl-bypass
4837 @opindex fdump-rtl-bypass
4838 Dump after jump bypassing and control flow optimizations.
4840 @item -fdump-rtl-combine
4841 @opindex fdump-rtl-combine
4842 Dump after the RTL instruction combination pass.
4844 @item -fdump-rtl-compgotos
4845 @opindex fdump-rtl-compgotos
4846 Dump after duplicating the computed gotos.
4848 @item -fdump-rtl-ce1
4849 @itemx -fdump-rtl-ce2
4850 @itemx -fdump-rtl-ce3
4851 @opindex fdump-rtl-ce1
4852 @opindex fdump-rtl-ce2
4853 @opindex fdump-rtl-ce3
4854 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4855 @option{-fdump-rtl-ce3} enable dumping after the three
4856 if conversion passes.
4858 @itemx -fdump-rtl-cprop_hardreg
4859 @opindex fdump-rtl-cprop_hardreg
4860 Dump after hard register copy propagation.
4862 @itemx -fdump-rtl-csa
4863 @opindex fdump-rtl-csa
4864 Dump after combining stack adjustments.
4866 @item -fdump-rtl-cse1
4867 @itemx -fdump-rtl-cse2
4868 @opindex fdump-rtl-cse1
4869 @opindex fdump-rtl-cse2
4870 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4871 the two common sub-expression elimination passes.
4873 @itemx -fdump-rtl-dce
4874 @opindex fdump-rtl-dce
4875 Dump after the standalone dead code elimination passes.
4877 @itemx -fdump-rtl-dbr
4878 @opindex fdump-rtl-dbr
4879 Dump after delayed branch scheduling.
4881 @item -fdump-rtl-dce1
4882 @itemx -fdump-rtl-dce2
4883 @opindex fdump-rtl-dce1
4884 @opindex fdump-rtl-dce2
4885 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4886 the two dead store elimination passes.
4889 @opindex fdump-rtl-eh
4890 Dump after finalization of EH handling code.
4892 @item -fdump-rtl-eh_ranges
4893 @opindex fdump-rtl-eh_ranges
4894 Dump after conversion of EH handling range regions.
4896 @item -fdump-rtl-expand
4897 @opindex fdump-rtl-expand
4898 Dump after RTL generation.
4900 @item -fdump-rtl-fwprop1
4901 @itemx -fdump-rtl-fwprop2
4902 @opindex fdump-rtl-fwprop1
4903 @opindex fdump-rtl-fwprop2
4904 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4905 dumping after the two forward propagation passes.
4907 @item -fdump-rtl-gcse1
4908 @itemx -fdump-rtl-gcse2
4909 @opindex fdump-rtl-gcse1
4910 @opindex fdump-rtl-gcse2
4911 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4912 after global common subexpression elimination.
4914 @item -fdump-rtl-init-regs
4915 @opindex fdump-rtl-init-regs
4916 Dump after the initialization of the registers.
4918 @item -fdump-rtl-initvals
4919 @opindex fdump-rtl-initvals
4920 Dump after the computation of the initial value sets.
4922 @itemx -fdump-rtl-into_cfglayout
4923 @opindex fdump-rtl-into_cfglayout
4924 Dump after converting to cfglayout mode.
4926 @item -fdump-rtl-ira
4927 @opindex fdump-rtl-ira
4928 Dump after iterated register allocation.
4930 @item -fdump-rtl-jump
4931 @opindex fdump-rtl-jump
4932 Dump after the second jump optimization.
4934 @item -fdump-rtl-loop2
4935 @opindex fdump-rtl-loop2
4936 @option{-fdump-rtl-loop2} enables dumping after the rtl
4937 loop optimization passes.
4939 @item -fdump-rtl-mach
4940 @opindex fdump-rtl-mach
4941 Dump after performing the machine dependent reorganization pass, if that
4944 @item -fdump-rtl-mode_sw
4945 @opindex fdump-rtl-mode_sw
4946 Dump after removing redundant mode switches.
4948 @item -fdump-rtl-rnreg
4949 @opindex fdump-rtl-rnreg
4950 Dump after register renumbering.
4952 @itemx -fdump-rtl-outof_cfglayout
4953 @opindex fdump-rtl-outof_cfglayout
4954 Dump after converting from cfglayout mode.
4956 @item -fdump-rtl-peephole2
4957 @opindex fdump-rtl-peephole2
4958 Dump after the peephole pass.
4960 @item -fdump-rtl-postreload
4961 @opindex fdump-rtl-postreload
4962 Dump after post-reload optimizations.
4964 @itemx -fdump-rtl-pro_and_epilogue
4965 @opindex fdump-rtl-pro_and_epilogue
4966 Dump after generating the function pro and epilogues.
4968 @item -fdump-rtl-regmove
4969 @opindex fdump-rtl-regmove
4970 Dump after the register move pass.
4972 @item -fdump-rtl-sched1
4973 @itemx -fdump-rtl-sched2
4974 @opindex fdump-rtl-sched1
4975 @opindex fdump-rtl-sched2
4976 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4977 after the basic block scheduling passes.
4979 @item -fdump-rtl-see
4980 @opindex fdump-rtl-see
4981 Dump after sign extension elimination.
4983 @item -fdump-rtl-seqabstr
4984 @opindex fdump-rtl-seqabstr
4985 Dump after common sequence discovery.
4987 @item -fdump-rtl-shorten
4988 @opindex fdump-rtl-shorten
4989 Dump after shortening branches.
4991 @item -fdump-rtl-sibling
4992 @opindex fdump-rtl-sibling
4993 Dump after sibling call optimizations.
4995 @item -fdump-rtl-split1
4996 @itemx -fdump-rtl-split2
4997 @itemx -fdump-rtl-split3
4998 @itemx -fdump-rtl-split4
4999 @itemx -fdump-rtl-split5
5000 @opindex fdump-rtl-split1
5001 @opindex fdump-rtl-split2
5002 @opindex fdump-rtl-split3
5003 @opindex fdump-rtl-split4
5004 @opindex fdump-rtl-split5
5005 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5006 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5007 @option{-fdump-rtl-split5} enable dumping after five rounds of
5008 instruction splitting.
5010 @item -fdump-rtl-sms
5011 @opindex fdump-rtl-sms
5012 Dump after modulo scheduling. This pass is only run on some
5015 @item -fdump-rtl-stack
5016 @opindex fdump-rtl-stack
5017 Dump after conversion from GCC's "flat register file" registers to the
5018 x87's stack-like registers. This pass is only run on x86 variants.
5020 @item -fdump-rtl-subreg1
5021 @itemx -fdump-rtl-subreg2
5022 @opindex fdump-rtl-subreg1
5023 @opindex fdump-rtl-subreg2
5024 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5025 the two subreg expansion passes.
5027 @item -fdump-rtl-unshare
5028 @opindex fdump-rtl-unshare
5029 Dump after all rtl has been unshared.
5031 @item -fdump-rtl-vartrack
5032 @opindex fdump-rtl-vartrack
5033 Dump after variable tracking.
5035 @item -fdump-rtl-vregs
5036 @opindex fdump-rtl-vregs
5037 Dump after converting virtual registers to hard registers.
5039 @item -fdump-rtl-web
5040 @opindex fdump-rtl-web
5041 Dump after live range splitting.
5043 @item -fdump-rtl-regclass
5044 @itemx -fdump-rtl-subregs_of_mode_init
5045 @itemx -fdump-rtl-subregs_of_mode_finish
5046 @itemx -fdump-rtl-dfinit
5047 @itemx -fdump-rtl-dfinish
5048 @opindex fdump-rtl-regclass
5049 @opindex fdump-rtl-subregs_of_mode_init
5050 @opindex fdump-rtl-subregs_of_mode_finish
5051 @opindex fdump-rtl-dfinit
5052 @opindex fdump-rtl-dfinish
5053 These dumps are defined but always produce empty files.
5055 @item -fdump-rtl-all
5056 @opindex fdump-rtl-all
5057 Produce all the dumps listed above.
5061 Annotate the assembler output with miscellaneous debugging information.
5065 Dump all macro definitions, at the end of preprocessing, in addition to
5070 Produce a core dump whenever an error occurs.
5074 Print statistics on memory usage, at the end of the run, to
5079 Annotate the assembler output with a comment indicating which
5080 pattern and alternative was used. The length of each instruction is
5085 Dump the RTL in the assembler output as a comment before each instruction.
5086 Also turns on @option{-dp} annotation.
5090 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5091 dump a representation of the control flow graph suitable for viewing with VCG
5092 to @file{@var{file}.@var{pass}.vcg}.
5096 Just generate RTL for a function instead of compiling it. Usually used
5097 with @option{-fdump-rtl-expand}.
5101 Dump debugging information during parsing, to standard error.
5105 @opindex fdump-noaddr
5106 When doing debugging dumps, suppress address output. This makes it more
5107 feasible to use diff on debugging dumps for compiler invocations with
5108 different compiler binaries and/or different
5109 text / bss / data / heap / stack / dso start locations.
5111 @item -fdump-unnumbered
5112 @opindex fdump-unnumbered
5113 When doing debugging dumps, suppress instruction numbers and address output.
5114 This makes it more feasible to use diff on debugging dumps for compiler
5115 invocations with different options, in particular with and without
5118 @item -fdump-unnumbered-links
5119 @opindex fdump-unnumbered-links
5120 When doing debugging dumps (see @option{-d} option above), suppress
5121 instruction numbers for the links to the previous and next instructions
5124 @item -fdump-translation-unit @r{(C++ only)}
5125 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5126 @opindex fdump-translation-unit
5127 Dump a representation of the tree structure for the entire translation
5128 unit to a file. The file name is made by appending @file{.tu} to the
5129 source file name, and the file is created in the same directory as the
5130 output file. If the @samp{-@var{options}} form is used, @var{options}
5131 controls the details of the dump as described for the
5132 @option{-fdump-tree} options.
5134 @item -fdump-class-hierarchy @r{(C++ only)}
5135 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5136 @opindex fdump-class-hierarchy
5137 Dump a representation of each class's hierarchy and virtual function
5138 table layout to a file. The file name is made by appending
5139 @file{.class} to the source file name, and the file is created in the
5140 same directory as the output file. If the @samp{-@var{options}} form
5141 is used, @var{options} controls the details of the dump as described
5142 for the @option{-fdump-tree} options.
5144 @item -fdump-ipa-@var{switch}
5146 Control the dumping at various stages of inter-procedural analysis
5147 language tree to a file. The file name is generated by appending a
5148 switch specific suffix to the source file name, and the file is created
5149 in the same directory as the output file. The following dumps are
5154 Enables all inter-procedural analysis dumps.
5157 Dumps information about call-graph optimization, unused function removal,
5158 and inlining decisions.
5161 Dump after function inlining.
5165 @item -fdump-statistics-@var{option}
5166 @opindex fdump-statistics
5167 Enable and control dumping of pass statistics in a separate file. The
5168 file name is generated by appending a suffix ending in
5169 @samp{.statistics} to the source file name, and the file is created in
5170 the same directory as the output file. If the @samp{-@var{option}}
5171 form is used, @samp{-stats} will cause counters to be summed over the
5172 whole compilation unit while @samp{-details} will dump every event as
5173 the passes generate them. The default with no option is to sum
5174 counters for each function compiled.
5176 @item -fdump-tree-@var{switch}
5177 @itemx -fdump-tree-@var{switch}-@var{options}
5179 Control the dumping at various stages of processing the intermediate
5180 language tree to a file. The file name is generated by appending a
5181 switch specific suffix to the source file name, and the file is
5182 created in the same directory as the output file. If the
5183 @samp{-@var{options}} form is used, @var{options} is a list of
5184 @samp{-} separated options that control the details of the dump. Not
5185 all options are applicable to all dumps, those which are not
5186 meaningful will be ignored. The following options are available
5190 Print the address of each node. Usually this is not meaningful as it
5191 changes according to the environment and source file. Its primary use
5192 is for tying up a dump file with a debug environment.
5194 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5195 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5196 use working backward from mangled names in the assembly file.
5198 Inhibit dumping of members of a scope or body of a function merely
5199 because that scope has been reached. Only dump such items when they
5200 are directly reachable by some other path. When dumping pretty-printed
5201 trees, this option inhibits dumping the bodies of control structures.
5203 Print a raw representation of the tree. By default, trees are
5204 pretty-printed into a C-like representation.
5206 Enable more detailed dumps (not honored by every dump option).
5208 Enable dumping various statistics about the pass (not honored by every dump
5211 Enable showing basic block boundaries (disabled in raw dumps).
5213 Enable showing virtual operands for every statement.
5215 Enable showing line numbers for statements.
5217 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5219 Enable showing the tree dump for each statement.
5221 Enable showing the EH region number holding each statement.
5223 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5224 and @option{lineno}.
5227 The following tree dumps are possible:
5231 @opindex fdump-tree-original
5232 Dump before any tree based optimization, to @file{@var{file}.original}.
5235 @opindex fdump-tree-optimized
5236 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5239 @opindex fdump-tree-gimple
5240 Dump each function before and after the gimplification pass to a file. The
5241 file name is made by appending @file{.gimple} to the source file name.
5244 @opindex fdump-tree-cfg
5245 Dump the control flow graph of each function to a file. The file name is
5246 made by appending @file{.cfg} to the source file name.
5249 @opindex fdump-tree-vcg
5250 Dump the control flow graph of each function to a file in VCG format. The
5251 file name is made by appending @file{.vcg} to the source file name. Note
5252 that if the file contains more than one function, the generated file cannot
5253 be used directly by VCG@. You will need to cut and paste each function's
5254 graph into its own separate file first.
5257 @opindex fdump-tree-ch
5258 Dump each function after copying loop headers. The file name is made by
5259 appending @file{.ch} to the source file name.
5262 @opindex fdump-tree-ssa
5263 Dump SSA related information to a file. The file name is made by appending
5264 @file{.ssa} to the source file name.
5267 @opindex fdump-tree-alias
5268 Dump aliasing information for each function. The file name is made by
5269 appending @file{.alias} to the source file name.
5272 @opindex fdump-tree-ccp
5273 Dump each function after CCP@. The file name is made by appending
5274 @file{.ccp} to the source file name.
5277 @opindex fdump-tree-storeccp
5278 Dump each function after STORE-CCP@. The file name is made by appending
5279 @file{.storeccp} to the source file name.
5282 @opindex fdump-tree-pre
5283 Dump trees after partial redundancy elimination. The file name is made
5284 by appending @file{.pre} to the source file name.
5287 @opindex fdump-tree-fre
5288 Dump trees after full redundancy elimination. The file name is made
5289 by appending @file{.fre} to the source file name.
5292 @opindex fdump-tree-copyprop
5293 Dump trees after copy propagation. The file name is made
5294 by appending @file{.copyprop} to the source file name.
5296 @item store_copyprop
5297 @opindex fdump-tree-store_copyprop
5298 Dump trees after store copy-propagation. The file name is made
5299 by appending @file{.store_copyprop} to the source file name.
5302 @opindex fdump-tree-dce
5303 Dump each function after dead code elimination. The file name is made by
5304 appending @file{.dce} to the source file name.
5307 @opindex fdump-tree-mudflap
5308 Dump each function after adding mudflap instrumentation. The file name is
5309 made by appending @file{.mudflap} to the source file name.
5312 @opindex fdump-tree-sra
5313 Dump each function after performing scalar replacement of aggregates. The
5314 file name is made by appending @file{.sra} to the source file name.
5317 @opindex fdump-tree-sink
5318 Dump each function after performing code sinking. The file name is made
5319 by appending @file{.sink} to the source file name.
5322 @opindex fdump-tree-dom
5323 Dump each function after applying dominator tree optimizations. The file
5324 name is made by appending @file{.dom} to the source file name.
5327 @opindex fdump-tree-dse
5328 Dump each function after applying dead store elimination. The file
5329 name is made by appending @file{.dse} to the source file name.
5332 @opindex fdump-tree-phiopt
5333 Dump each function after optimizing PHI nodes into straightline code. The file
5334 name is made by appending @file{.phiopt} to the source file name.
5337 @opindex fdump-tree-forwprop
5338 Dump each function after forward propagating single use variables. The file
5339 name is made by appending @file{.forwprop} to the source file name.
5342 @opindex fdump-tree-copyrename
5343 Dump each function after applying the copy rename optimization. The file
5344 name is made by appending @file{.copyrename} to the source file name.
5347 @opindex fdump-tree-nrv
5348 Dump each function after applying the named return value optimization on
5349 generic trees. The file name is made by appending @file{.nrv} to the source
5353 @opindex fdump-tree-vect
5354 Dump each function after applying vectorization of loops. The file name is
5355 made by appending @file{.vect} to the source file name.
5358 @opindex fdump-tree-vrp
5359 Dump each function after Value Range Propagation (VRP). The file name
5360 is made by appending @file{.vrp} to the source file name.
5363 @opindex fdump-tree-all
5364 Enable all the available tree dumps with the flags provided in this option.
5367 @item -ftree-vectorizer-verbose=@var{n}
5368 @opindex ftree-vectorizer-verbose
5369 This option controls the amount of debugging output the vectorizer prints.
5370 This information is written to standard error, unless
5371 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5372 in which case it is output to the usual dump listing file, @file{.vect}.
5373 For @var{n}=0 no diagnostic information is reported.
5374 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5375 and the total number of loops that got vectorized.
5376 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5377 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5378 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5379 level that @option{-fdump-tree-vect-stats} uses.
5380 Higher verbosity levels mean either more information dumped for each
5381 reported loop, or same amount of information reported for more loops:
5382 If @var{n}=3, alignment related information is added to the reports.
5383 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5384 memory access-patterns) is added to the reports.
5385 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5386 that did not pass the first analysis phase (i.e., may not be countable, or
5387 may have complicated control-flow).
5388 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5389 For @var{n}=7, all the information the vectorizer generates during its
5390 analysis and transformation is reported. This is the same verbosity level
5391 that @option{-fdump-tree-vect-details} uses.
5393 @item -frandom-seed=@var{string}
5394 @opindex frandom-seed
5395 This option provides a seed that GCC uses when it would otherwise use
5396 random numbers. It is used to generate certain symbol names
5397 that have to be different in every compiled file. It is also used to
5398 place unique stamps in coverage data files and the object files that
5399 produce them. You can use the @option{-frandom-seed} option to produce
5400 reproducibly identical object files.
5402 The @var{string} should be different for every file you compile.
5404 @item -fsched-verbose=@var{n}
5405 @opindex fsched-verbose
5406 On targets that use instruction scheduling, this option controls the
5407 amount of debugging output the scheduler prints. This information is
5408 written to standard error, unless @option{-fdump-rtl-sched1} or
5409 @option{-fdump-rtl-sched2} is specified, in which case it is output
5410 to the usual dump listing file, @file{.sched} or @file{.sched2}
5411 respectively. However for @var{n} greater than nine, the output is
5412 always printed to standard error.
5414 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5415 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5416 For @var{n} greater than one, it also output basic block probabilities,
5417 detailed ready list information and unit/insn info. For @var{n} greater
5418 than two, it includes RTL at abort point, control-flow and regions info.
5419 And for @var{n} over four, @option{-fsched-verbose} also includes
5423 @itemx -save-temps=cwd
5425 Store the usual ``temporary'' intermediate files permanently; place them
5426 in the current directory and name them based on the source file. Thus,
5427 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5428 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5429 preprocessed @file{foo.i} output file even though the compiler now
5430 normally uses an integrated preprocessor.
5432 When used in combination with the @option{-x} command line option,
5433 @option{-save-temps} is sensible enough to avoid over writing an
5434 input source file with the same extension as an intermediate file.
5435 The corresponding intermediate file may be obtained by renaming the
5436 source file before using @option{-save-temps}.
5438 If you invoke GCC in parallel, compiling several different source
5439 files that share a common base name in different subdirectories or the
5440 same source file compiled for multiple output destinations, it is
5441 likely that the different parallel compilers will interfere with each
5442 other, and overwrite the temporary files. For instance:
5445 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5446 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5449 may result in @file{foo.i} and @file{foo.o} being written to
5450 simultaneously by both compilers.
5452 @item -save-temps=obj
5453 @opindex save-temps=obj
5454 Store the usual ``temporary'' intermediate files permanently. If the
5455 @option{-o} option is used, the temporary files are based on the
5456 object file. If the @option{-o} option is not used, the
5457 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5462 gcc -save-temps=obj -c foo.c
5463 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5464 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5467 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5468 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5469 @file{dir2/yfoobar.o}.
5471 @item -time@r{[}=@var{file}@r{]}
5473 Report the CPU time taken by each subprocess in the compilation
5474 sequence. For C source files, this is the compiler proper and assembler
5475 (plus the linker if linking is done).
5477 Without the specification of an output file, the output looks like this:
5484 The first number on each line is the ``user time'', that is time spent
5485 executing the program itself. The second number is ``system time'',
5486 time spent executing operating system routines on behalf of the program.
5487 Both numbers are in seconds.
5489 With the specification of an output file, the output is appended to the
5490 named file, and it looks like this:
5493 0.12 0.01 cc1 @var{options}
5494 0.00 0.01 as @var{options}
5497 The ``user time'' and the ``system time'' are moved before the program
5498 name, and the options passed to the program are displayed, so that one
5499 can later tell what file was being compiled, and with which options.
5501 @item -fvar-tracking
5502 @opindex fvar-tracking
5503 Run variable tracking pass. It computes where variables are stored at each
5504 position in code. Better debugging information is then generated
5505 (if the debugging information format supports this information).
5507 It is enabled by default when compiling with optimization (@option{-Os},
5508 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5509 the debug info format supports it.
5511 @item -fvar-tracking-assignments
5512 @opindex fvar-tracking-assignments
5513 @opindex fno-var-tracking-assignments
5514 Annotate assignments to user variables early in the compilation and
5515 attempt to carry the annotations over throughout the compilation all the
5516 way to the end, in an attempt to improve debug information while
5517 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5519 It can be enabled even if var-tracking is disabled, in which case
5520 annotations will be created and maintained, but discarded at the end.
5522 @item -fvar-tracking-assignments-toggle
5523 @opindex fvar-tracking-assignments-toggle
5524 @opindex fno-var-tracking-assignments-toggle
5525 Toggle @option{-fvar-tracking-assignments}, in the same way that
5526 @option{-gtoggle} toggles @option{-g}.
5528 @item -print-file-name=@var{library}
5529 @opindex print-file-name
5530 Print the full absolute name of the library file @var{library} that
5531 would be used when linking---and don't do anything else. With this
5532 option, GCC does not compile or link anything; it just prints the
5535 @item -print-multi-directory
5536 @opindex print-multi-directory
5537 Print the directory name corresponding to the multilib selected by any
5538 other switches present in the command line. This directory is supposed
5539 to exist in @env{GCC_EXEC_PREFIX}.
5541 @item -print-multi-lib
5542 @opindex print-multi-lib
5543 Print the mapping from multilib directory names to compiler switches
5544 that enable them. The directory name is separated from the switches by
5545 @samp{;}, and each switch starts with an @samp{@@} instead of the
5546 @samp{-}, without spaces between multiple switches. This is supposed to
5547 ease shell-processing.
5549 @item -print-multi-os-directory
5550 @opindex print-multi-os-directory
5551 Print the path to OS libraries for the selected
5552 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5553 present in the @file{lib} subdirectory and no multilibs are used, this is
5554 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5555 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5556 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5557 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5559 @item -print-prog-name=@var{program}
5560 @opindex print-prog-name
5561 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5563 @item -print-libgcc-file-name
5564 @opindex print-libgcc-file-name
5565 Same as @option{-print-file-name=libgcc.a}.
5567 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5568 but you do want to link with @file{libgcc.a}. You can do
5571 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5574 @item -print-search-dirs
5575 @opindex print-search-dirs
5576 Print the name of the configured installation directory and a list of
5577 program and library directories @command{gcc} will search---and don't do anything else.
5579 This is useful when @command{gcc} prints the error message
5580 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5581 To resolve this you either need to put @file{cpp0} and the other compiler
5582 components where @command{gcc} expects to find them, or you can set the environment
5583 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5584 Don't forget the trailing @samp{/}.
5585 @xref{Environment Variables}.
5587 @item -print-sysroot
5588 @opindex print-sysroot
5589 Print the target sysroot directory that will be used during
5590 compilation. This is the target sysroot specified either at configure
5591 time or using the @option{--sysroot} option, possibly with an extra
5592 suffix that depends on compilation options. If no target sysroot is
5593 specified, the option prints nothing.
5595 @item -print-sysroot-headers-suffix
5596 @opindex print-sysroot-headers-suffix
5597 Print the suffix added to the target sysroot when searching for
5598 headers, or give an error if the compiler is not configured with such
5599 a suffix---and don't do anything else.
5602 @opindex dumpmachine
5603 Print the compiler's target machine (for example,
5604 @samp{i686-pc-linux-gnu})---and don't do anything else.
5607 @opindex dumpversion
5608 Print the compiler version (for example, @samp{3.0})---and don't do
5613 Print the compiler's built-in specs---and don't do anything else. (This
5614 is used when GCC itself is being built.) @xref{Spec Files}.
5616 @item -feliminate-unused-debug-types
5617 @opindex feliminate-unused-debug-types
5618 Normally, when producing DWARF2 output, GCC will emit debugging
5619 information for all types declared in a compilation
5620 unit, regardless of whether or not they are actually used
5621 in that compilation unit. Sometimes this is useful, such as
5622 if, in the debugger, you want to cast a value to a type that is
5623 not actually used in your program (but is declared). More often,
5624 however, this results in a significant amount of wasted space.
5625 With this option, GCC will avoid producing debug symbol output
5626 for types that are nowhere used in the source file being compiled.
5629 @node Optimize Options
5630 @section Options That Control Optimization
5631 @cindex optimize options
5632 @cindex options, optimization
5634 These options control various sorts of optimizations.
5636 Without any optimization option, the compiler's goal is to reduce the
5637 cost of compilation and to make debugging produce the expected
5638 results. Statements are independent: if you stop the program with a
5639 breakpoint between statements, you can then assign a new value to any
5640 variable or change the program counter to any other statement in the
5641 function and get exactly the results you would expect from the source
5644 Turning on optimization flags makes the compiler attempt to improve
5645 the performance and/or code size at the expense of compilation time
5646 and possibly the ability to debug the program.
5648 The compiler performs optimization based on the knowledge it has of the
5649 program. Compiling multiple files at once to a single output file mode allows
5650 the compiler to use information gained from all of the files when compiling
5653 Not all optimizations are controlled directly by a flag. Only
5654 optimizations that have a flag are listed in this section.
5656 Most of the optimizations are not enabled if a @option{-O} level is not set on
5657 the command line, even if individual optimization flags are specified.
5659 Depending on the target and how GCC was configured, a slightly different
5660 set of optimizations may be enabled at each @option{-O} level than
5661 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5662 to find out the exact set of optimizations that are enabled at each level.
5663 @xref{Overall Options}, for examples.
5670 Optimize. Optimizing compilation takes somewhat more time, and a lot
5671 more memory for a large function.
5673 With @option{-O}, the compiler tries to reduce code size and execution
5674 time, without performing any optimizations that take a great deal of
5677 @option{-O} turns on the following optimization flags:
5680 -fcprop-registers @gol
5683 -fdelayed-branch @gol
5685 -fguess-branch-probability @gol
5686 -fif-conversion2 @gol
5687 -fif-conversion @gol
5688 -fipa-pure-const @gol
5689 -fipa-reference @gol
5691 -fsplit-wide-types @gol
5692 -ftree-builtin-call-dce @gol
5695 -ftree-copyrename @gol
5697 -ftree-dominator-opts @gol
5699 -ftree-forwprop @gol
5707 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5708 where doing so does not interfere with debugging.
5712 Optimize even more. GCC performs nearly all supported optimizations
5713 that do not involve a space-speed tradeoff.
5714 As compared to @option{-O}, this option increases both compilation time
5715 and the performance of the generated code.
5717 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5718 also turns on the following optimization flags:
5719 @gccoptlist{-fthread-jumps @gol
5720 -falign-functions -falign-jumps @gol
5721 -falign-loops -falign-labels @gol
5724 -fcse-follow-jumps -fcse-skip-blocks @gol
5725 -fdelete-null-pointer-checks @gol
5726 -fexpensive-optimizations @gol
5727 -fgcse -fgcse-lm @gol
5728 -finline-small-functions @gol
5729 -findirect-inlining @gol
5731 -foptimize-sibling-calls @gol
5734 -freorder-blocks -freorder-functions @gol
5735 -frerun-cse-after-loop @gol
5736 -fsched-interblock -fsched-spec @gol
5737 -fschedule-insns -fschedule-insns2 @gol
5738 -fstrict-aliasing -fstrict-overflow @gol
5739 -ftree-switch-conversion @gol
5743 Please note the warning under @option{-fgcse} about
5744 invoking @option{-O2} on programs that use computed gotos.
5748 Optimize yet more. @option{-O3} turns on all optimizations specified
5749 by @option{-O2} and also turns on the @option{-finline-functions},
5750 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5751 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5755 Reduce compilation time and make debugging produce the expected
5756 results. This is the default.
5760 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5761 do not typically increase code size. It also performs further
5762 optimizations designed to reduce code size.
5764 @option{-Os} disables the following optimization flags:
5765 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5766 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5767 -fprefetch-loop-arrays -ftree-vect-loop-version}
5769 If you use multiple @option{-O} options, with or without level numbers,
5770 the last such option is the one that is effective.
5773 Options of the form @option{-f@var{flag}} specify machine-independent
5774 flags. Most flags have both positive and negative forms; the negative
5775 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5776 below, only one of the forms is listed---the one you typically will
5777 use. You can figure out the other form by either removing @samp{no-}
5780 The following options control specific optimizations. They are either
5781 activated by @option{-O} options or are related to ones that are. You
5782 can use the following flags in the rare cases when ``fine-tuning'' of
5783 optimizations to be performed is desired.
5786 @item -fno-default-inline
5787 @opindex fno-default-inline
5788 Do not make member functions inline by default merely because they are
5789 defined inside the class scope (C++ only). Otherwise, when you specify
5790 @w{@option{-O}}, member functions defined inside class scope are compiled
5791 inline by default; i.e., you don't need to add @samp{inline} in front of
5792 the member function name.
5794 @item -fno-defer-pop
5795 @opindex fno-defer-pop
5796 Always pop the arguments to each function call as soon as that function
5797 returns. For machines which must pop arguments after a function call,
5798 the compiler normally lets arguments accumulate on the stack for several
5799 function calls and pops them all at once.
5801 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5803 @item -fforward-propagate
5804 @opindex fforward-propagate
5805 Perform a forward propagation pass on RTL@. The pass tries to combine two
5806 instructions and checks if the result can be simplified. If loop unrolling
5807 is active, two passes are performed and the second is scheduled after
5810 This option is enabled by default at optimization levels @option{-O},
5811 @option{-O2}, @option{-O3}, @option{-Os}.
5813 @item -fomit-frame-pointer
5814 @opindex fomit-frame-pointer
5815 Don't keep the frame pointer in a register for functions that
5816 don't need one. This avoids the instructions to save, set up and
5817 restore frame pointers; it also makes an extra register available
5818 in many functions. @strong{It also makes debugging impossible on
5821 On some machines, such as the VAX, this flag has no effect, because
5822 the standard calling sequence automatically handles the frame pointer
5823 and nothing is saved by pretending it doesn't exist. The
5824 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5825 whether a target machine supports this flag. @xref{Registers,,Register
5826 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5828 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5830 @item -foptimize-sibling-calls
5831 @opindex foptimize-sibling-calls
5832 Optimize sibling and tail recursive calls.
5834 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5838 Don't pay attention to the @code{inline} keyword. Normally this option
5839 is used to keep the compiler from expanding any functions inline.
5840 Note that if you are not optimizing, no functions can be expanded inline.
5842 @item -finline-small-functions
5843 @opindex finline-small-functions
5844 Integrate functions into their callers when their body is smaller than expected
5845 function call code (so overall size of program gets smaller). The compiler
5846 heuristically decides which functions are simple enough to be worth integrating
5849 Enabled at level @option{-O2}.
5851 @item -findirect-inlining
5852 @opindex findirect-inlining
5853 Inline also indirect calls that are discovered to be known at compile
5854 time thanks to previous inlining. This option has any effect only
5855 when inlining itself is turned on by the @option{-finline-functions}
5856 or @option{-finline-small-functions} options.
5858 Enabled at level @option{-O2}.
5860 @item -finline-functions
5861 @opindex finline-functions
5862 Integrate all simple functions into their callers. The compiler
5863 heuristically decides which functions are simple enough to be worth
5864 integrating in this way.
5866 If all calls to a given function are integrated, and the function is
5867 declared @code{static}, then the function is normally not output as
5868 assembler code in its own right.
5870 Enabled at level @option{-O3}.
5872 @item -finline-functions-called-once
5873 @opindex finline-functions-called-once
5874 Consider all @code{static} functions called once for inlining into their
5875 caller even if they are not marked @code{inline}. If a call to a given
5876 function is integrated, then the function is not output as assembler code
5879 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5881 @item -fearly-inlining
5882 @opindex fearly-inlining
5883 Inline functions marked by @code{always_inline} and functions whose body seems
5884 smaller than the function call overhead early before doing
5885 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5886 makes profiling significantly cheaper and usually inlining faster on programs
5887 having large chains of nested wrapper functions.
5893 Perform interprocedural scalar replacement of aggregates, removal of
5894 unused parameters and replacement of parameters passed by reference
5895 by parameters passed by value.
5897 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
5899 @item -finline-limit=@var{n}
5900 @opindex finline-limit
5901 By default, GCC limits the size of functions that can be inlined. This flag
5902 allows coarse control of this limit. @var{n} is the size of functions that
5903 can be inlined in number of pseudo instructions.
5905 Inlining is actually controlled by a number of parameters, which may be
5906 specified individually by using @option{--param @var{name}=@var{value}}.
5907 The @option{-finline-limit=@var{n}} option sets some of these parameters
5911 @item max-inline-insns-single
5912 is set to @var{n}/2.
5913 @item max-inline-insns-auto
5914 is set to @var{n}/2.
5917 See below for a documentation of the individual
5918 parameters controlling inlining and for the defaults of these parameters.
5920 @emph{Note:} there may be no value to @option{-finline-limit} that results
5921 in default behavior.
5923 @emph{Note:} pseudo instruction represents, in this particular context, an
5924 abstract measurement of function's size. In no way does it represent a count
5925 of assembly instructions and as such its exact meaning might change from one
5926 release to an another.
5928 @item -fkeep-inline-functions
5929 @opindex fkeep-inline-functions
5930 In C, emit @code{static} functions that are declared @code{inline}
5931 into the object file, even if the function has been inlined into all
5932 of its callers. This switch does not affect functions using the
5933 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5934 inline functions into the object file.
5936 @item -fkeep-static-consts
5937 @opindex fkeep-static-consts
5938 Emit variables declared @code{static const} when optimization isn't turned
5939 on, even if the variables aren't referenced.
5941 GCC enables this option by default. If you want to force the compiler to
5942 check if the variable was referenced, regardless of whether or not
5943 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5945 @item -fmerge-constants
5946 @opindex fmerge-constants
5947 Attempt to merge identical constants (string constants and floating point
5948 constants) across compilation units.
5950 This option is the default for optimized compilation if the assembler and
5951 linker support it. Use @option{-fno-merge-constants} to inhibit this
5954 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5956 @item -fmerge-all-constants
5957 @opindex fmerge-all-constants
5958 Attempt to merge identical constants and identical variables.
5960 This option implies @option{-fmerge-constants}. In addition to
5961 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5962 arrays or initialized constant variables with integral or floating point
5963 types. Languages like C or C++ require each variable, including multiple
5964 instances of the same variable in recursive calls, to have distinct locations,
5965 so using this option will result in non-conforming
5968 @item -fmodulo-sched
5969 @opindex fmodulo-sched
5970 Perform swing modulo scheduling immediately before the first scheduling
5971 pass. This pass looks at innermost loops and reorders their
5972 instructions by overlapping different iterations.
5974 @item -fmodulo-sched-allow-regmoves
5975 @opindex fmodulo-sched-allow-regmoves
5976 Perform more aggressive SMS based modulo scheduling with register moves
5977 allowed. By setting this flag certain anti-dependences edges will be
5978 deleted which will trigger the generation of reg-moves based on the
5979 life-range analysis. This option is effective only with
5980 @option{-fmodulo-sched} enabled.
5982 @item -fno-branch-count-reg
5983 @opindex fno-branch-count-reg
5984 Do not use ``decrement and branch'' instructions on a count register,
5985 but instead generate a sequence of instructions that decrement a
5986 register, compare it against zero, then branch based upon the result.
5987 This option is only meaningful on architectures that support such
5988 instructions, which include x86, PowerPC, IA-64 and S/390.
5990 The default is @option{-fbranch-count-reg}.
5992 @item -fno-function-cse
5993 @opindex fno-function-cse
5994 Do not put function addresses in registers; make each instruction that
5995 calls a constant function contain the function's address explicitly.
5997 This option results in less efficient code, but some strange hacks
5998 that alter the assembler output may be confused by the optimizations
5999 performed when this option is not used.
6001 The default is @option{-ffunction-cse}
6003 @item -fno-zero-initialized-in-bss
6004 @opindex fno-zero-initialized-in-bss
6005 If the target supports a BSS section, GCC by default puts variables that
6006 are initialized to zero into BSS@. This can save space in the resulting
6009 This option turns off this behavior because some programs explicitly
6010 rely on variables going to the data section. E.g., so that the
6011 resulting executable can find the beginning of that section and/or make
6012 assumptions based on that.
6014 The default is @option{-fzero-initialized-in-bss}.
6016 @item -fmudflap -fmudflapth -fmudflapir
6020 @cindex bounds checking
6022 For front-ends that support it (C and C++), instrument all risky
6023 pointer/array dereferencing operations, some standard library
6024 string/heap functions, and some other associated constructs with
6025 range/validity tests. Modules so instrumented should be immune to
6026 buffer overflows, invalid heap use, and some other classes of C/C++
6027 programming errors. The instrumentation relies on a separate runtime
6028 library (@file{libmudflap}), which will be linked into a program if
6029 @option{-fmudflap} is given at link time. Run-time behavior of the
6030 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6031 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6034 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6035 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6036 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6037 instrumentation should ignore pointer reads. This produces less
6038 instrumentation (and therefore faster execution) and still provides
6039 some protection against outright memory corrupting writes, but allows
6040 erroneously read data to propagate within a program.
6042 @item -fthread-jumps
6043 @opindex fthread-jumps
6044 Perform optimizations where we check to see if a jump branches to a
6045 location where another comparison subsumed by the first is found. If
6046 so, the first branch is redirected to either the destination of the
6047 second branch or a point immediately following it, depending on whether
6048 the condition is known to be true or false.
6050 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6052 @item -fsplit-wide-types
6053 @opindex fsplit-wide-types
6054 When using a type that occupies multiple registers, such as @code{long
6055 long} on a 32-bit system, split the registers apart and allocate them
6056 independently. This normally generates better code for those types,
6057 but may make debugging more difficult.
6059 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6062 @item -fcse-follow-jumps
6063 @opindex fcse-follow-jumps
6064 In common subexpression elimination (CSE), scan through jump instructions
6065 when the target of the jump is not reached by any other path. For
6066 example, when CSE encounters an @code{if} statement with an
6067 @code{else} clause, CSE will follow the jump when the condition
6070 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6072 @item -fcse-skip-blocks
6073 @opindex fcse-skip-blocks
6074 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6075 follow jumps which conditionally skip over blocks. When CSE
6076 encounters a simple @code{if} statement with no else clause,
6077 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6078 body of the @code{if}.
6080 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6082 @item -frerun-cse-after-loop
6083 @opindex frerun-cse-after-loop
6084 Re-run common subexpression elimination after loop optimizations has been
6087 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6091 Perform a global common subexpression elimination pass.
6092 This pass also performs global constant and copy propagation.
6094 @emph{Note:} When compiling a program using computed gotos, a GCC
6095 extension, you may get better runtime performance if you disable
6096 the global common subexpression elimination pass by adding
6097 @option{-fno-gcse} to the command line.
6099 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6103 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6104 attempt to move loads which are only killed by stores into themselves. This
6105 allows a loop containing a load/store sequence to be changed to a load outside
6106 the loop, and a copy/store within the loop.
6108 Enabled by default when gcse is enabled.
6112 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6113 global common subexpression elimination. This pass will attempt to move
6114 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6115 loops containing a load/store sequence can be changed to a load before
6116 the loop and a store after the loop.
6118 Not enabled at any optimization level.
6122 When @option{-fgcse-las} is enabled, the global common subexpression
6123 elimination pass eliminates redundant loads that come after stores to the
6124 same memory location (both partial and full redundancies).
6126 Not enabled at any optimization level.
6128 @item -fgcse-after-reload
6129 @opindex fgcse-after-reload
6130 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6131 pass is performed after reload. The purpose of this pass is to cleanup
6134 @item -funsafe-loop-optimizations
6135 @opindex funsafe-loop-optimizations
6136 If given, the loop optimizer will assume that loop indices do not
6137 overflow, and that the loops with nontrivial exit condition are not
6138 infinite. This enables a wider range of loop optimizations even if
6139 the loop optimizer itself cannot prove that these assumptions are valid.
6140 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6141 if it finds this kind of loop.
6143 @item -fcrossjumping
6144 @opindex fcrossjumping
6145 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6146 resulting code may or may not perform better than without cross-jumping.
6148 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6150 @item -fauto-inc-dec
6151 @opindex fauto-inc-dec
6152 Combine increments or decrements of addresses with memory accesses.
6153 This pass is always skipped on architectures that do not have
6154 instructions to support this. Enabled by default at @option{-O} and
6155 higher on architectures that support this.
6159 Perform dead code elimination (DCE) on RTL@.
6160 Enabled by default at @option{-O} and higher.
6164 Perform dead store elimination (DSE) on RTL@.
6165 Enabled by default at @option{-O} and higher.
6167 @item -fif-conversion
6168 @opindex fif-conversion
6169 Attempt to transform conditional jumps into branch-less equivalents. This
6170 include use of conditional moves, min, max, set flags and abs instructions, and
6171 some tricks doable by standard arithmetics. The use of conditional execution
6172 on chips where it is available is controlled by @code{if-conversion2}.
6174 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6176 @item -fif-conversion2
6177 @opindex fif-conversion2
6178 Use conditional execution (where available) to transform conditional jumps into
6179 branch-less equivalents.
6181 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6183 @item -fdelete-null-pointer-checks
6184 @opindex fdelete-null-pointer-checks
6185 Assume that programs cannot safely dereference null pointers, and that
6186 no code or data element resides there. This enables simple constant
6187 folding optimizations at all optimization levels. In addition, other
6188 optimization passes in GCC use this flag to control global dataflow
6189 analyses that eliminate useless checks for null pointers; these assume
6190 that if a pointer is checked after it has already been dereferenced,
6193 Note however that in some environments this assumption is not true.
6194 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6195 for programs which depend on that behavior.
6197 Some targets, especially embedded ones, disable this option at all levels.
6198 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6199 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6200 are enabled independently at different optimization levels.
6202 @item -fexpensive-optimizations
6203 @opindex fexpensive-optimizations
6204 Perform a number of minor optimizations that are relatively expensive.
6206 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6208 @item -foptimize-register-move
6210 @opindex foptimize-register-move
6212 Attempt to reassign register numbers in move instructions and as
6213 operands of other simple instructions in order to maximize the amount of
6214 register tying. This is especially helpful on machines with two-operand
6217 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6220 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6222 @item -fira-algorithm=@var{algorithm}
6223 Use specified coloring algorithm for the integrated register
6224 allocator. The @var{algorithm} argument should be @code{priority} or
6225 @code{CB}. The first algorithm specifies Chow's priority coloring,
6226 the second one specifies Chaitin-Briggs coloring. The second
6227 algorithm can be unimplemented for some architectures. If it is
6228 implemented, it is the default because Chaitin-Briggs coloring as a
6229 rule generates a better code.
6231 @item -fira-region=@var{region}
6232 Use specified regions for the integrated register allocator. The
6233 @var{region} argument should be one of @code{all}, @code{mixed}, or
6234 @code{one}. The first value means using all loops as register
6235 allocation regions, the second value which is the default means using
6236 all loops except for loops with small register pressure as the
6237 regions, and third one means using all function as a single region.
6238 The first value can give best result for machines with small size and
6239 irregular register set, the third one results in faster and generates
6240 decent code and the smallest size code, and the default value usually
6241 give the best results in most cases and for most architectures.
6243 @item -fira-coalesce
6244 @opindex fira-coalesce
6245 Do optimistic register coalescing. This option might be profitable for
6246 architectures with big regular register files.
6248 @item -fira-loop-pressure
6249 @opindex fira-loop-pressure
6250 Use IRA to evaluate register pressure in loops for decision to move
6251 loop invariants. Usage of this option usually results in generation
6252 of faster and smaller code on machines with big register files (>= 32
6253 registers) but it can slow compiler down.
6255 This option is enabled at level @option{-O3} for some targets.
6257 @item -fno-ira-share-save-slots
6258 @opindex fno-ira-share-save-slots
6259 Switch off sharing stack slots used for saving call used hard
6260 registers living through a call. Each hard register will get a
6261 separate stack slot and as a result function stack frame will be
6264 @item -fno-ira-share-spill-slots
6265 @opindex fno-ira-share-spill-slots
6266 Switch off sharing stack slots allocated for pseudo-registers. Each
6267 pseudo-register which did not get a hard register will get a separate
6268 stack slot and as a result function stack frame will be bigger.
6270 @item -fira-verbose=@var{n}
6271 @opindex fira-verbose
6272 Set up how verbose dump file for the integrated register allocator
6273 will be. Default value is 5. If the value is greater or equal to 10,
6274 the dump file will be stderr as if the value were @var{n} minus 10.
6276 @item -fdelayed-branch
6277 @opindex fdelayed-branch
6278 If supported for the target machine, attempt to reorder instructions
6279 to exploit instruction slots available after delayed branch
6282 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6284 @item -fschedule-insns
6285 @opindex fschedule-insns
6286 If supported for the target machine, attempt to reorder instructions to
6287 eliminate execution stalls due to required data being unavailable. This
6288 helps machines that have slow floating point or memory load instructions
6289 by allowing other instructions to be issued until the result of the load
6290 or floating point instruction is required.
6292 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6294 @item -fschedule-insns2
6295 @opindex fschedule-insns2
6296 Similar to @option{-fschedule-insns}, but requests an additional pass of
6297 instruction scheduling after register allocation has been done. This is
6298 especially useful on machines with a relatively small number of
6299 registers and where memory load instructions take more than one cycle.
6301 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6303 @item -fno-sched-interblock
6304 @opindex fno-sched-interblock
6305 Don't schedule instructions across basic blocks. This is normally
6306 enabled by default when scheduling before register allocation, i.e.@:
6307 with @option{-fschedule-insns} or at @option{-O2} or higher.
6309 @item -fno-sched-spec
6310 @opindex fno-sched-spec
6311 Don't allow speculative motion of non-load instructions. This is normally
6312 enabled by default when scheduling before register allocation, i.e.@:
6313 with @option{-fschedule-insns} or at @option{-O2} or higher.
6315 @item -fsched-pressure
6316 @opindex fsched-pressure
6317 Enable register pressure sensitive insn scheduling before the register
6318 allocation. This only makes sense when scheduling before register
6319 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6320 @option{-O2} or higher. Usage of this option can improve the
6321 generated code and decrease its size by preventing register pressure
6322 increase above the number of available hard registers and as a
6323 consequence register spills in the register allocation.
6325 @item -fsched-spec-load
6326 @opindex fsched-spec-load
6327 Allow speculative motion of some load instructions. This only makes
6328 sense when scheduling before register allocation, i.e.@: with
6329 @option{-fschedule-insns} or at @option{-O2} or higher.
6331 @item -fsched-spec-load-dangerous
6332 @opindex fsched-spec-load-dangerous
6333 Allow speculative motion of more load instructions. This only makes
6334 sense when scheduling before register allocation, i.e.@: with
6335 @option{-fschedule-insns} or at @option{-O2} or higher.
6337 @item -fsched-stalled-insns
6338 @itemx -fsched-stalled-insns=@var{n}
6339 @opindex fsched-stalled-insns
6340 Define how many insns (if any) can be moved prematurely from the queue
6341 of stalled insns into the ready list, during the second scheduling pass.
6342 @option{-fno-sched-stalled-insns} means that no insns will be moved
6343 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6344 on how many queued insns can be moved prematurely.
6345 @option{-fsched-stalled-insns} without a value is equivalent to
6346 @option{-fsched-stalled-insns=1}.
6348 @item -fsched-stalled-insns-dep
6349 @itemx -fsched-stalled-insns-dep=@var{n}
6350 @opindex fsched-stalled-insns-dep
6351 Define how many insn groups (cycles) will be examined for a dependency
6352 on a stalled insn that is candidate for premature removal from the queue
6353 of stalled insns. This has an effect only during the second scheduling pass,
6354 and only if @option{-fsched-stalled-insns} is used.
6355 @option{-fno-sched-stalled-insns-dep} is equivalent to
6356 @option{-fsched-stalled-insns-dep=0}.
6357 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6358 @option{-fsched-stalled-insns-dep=1}.
6360 @item -fsched2-use-superblocks
6361 @opindex fsched2-use-superblocks
6362 When scheduling after register allocation, do use superblock scheduling
6363 algorithm. Superblock scheduling allows motion across basic block boundaries
6364 resulting on faster schedules. This option is experimental, as not all machine
6365 descriptions used by GCC model the CPU closely enough to avoid unreliable
6366 results from the algorithm.
6368 This only makes sense when scheduling after register allocation, i.e.@: with
6369 @option{-fschedule-insns2} or at @option{-O2} or higher.
6371 @item -fsched-group-heuristic
6372 @opindex fsched-group-heuristic
6373 Enable the group heuristic in the scheduler. This heuristic favors
6374 the instruction that belongs to a schedule group. This is enabled
6375 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6376 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6378 @item -fsched-critical-path-heuristic
6379 @opindex fsched-critical-path-heuristic
6380 Enable the critical-path heuristic in the scheduler. This heuristic favors
6381 instructions on the critical path. This is enabled by default when
6382 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6383 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6385 @item -fsched-spec-insn-heuristic
6386 @opindex fsched-spec-insn-heuristic
6387 Enable the speculative instruction heuristic in the scheduler. This
6388 heuristic favors speculative instructions with greater dependency weakness.
6389 This is enabled by default when scheduling is enabled, i.e.@:
6390 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6391 or at @option{-O2} or higher.
6393 @item -fsched-rank-heuristic
6394 @opindex fsched-rank-heuristic
6395 Enable the rank heuristic in the scheduler. This heuristic favors
6396 the instruction belonging to a basic block with greater size or frequency.
6397 This is enabled by default when scheduling is enabled, i.e.@:
6398 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6399 at @option{-O2} or higher.
6401 @item -fsched-last-insn-heuristic
6402 @opindex fsched-last-insn-heuristic
6403 Enable the last-instruction heuristic in the scheduler. This heuristic
6404 favors the instruction that is less dependent on the last instruction
6405 scheduled. This is enabled by default when scheduling is enabled,
6406 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6407 at @option{-O2} or higher.
6409 @item -fsched-dep-count-heuristic
6410 @opindex fsched-dep-count-heuristic
6411 Enable the dependent-count heuristic in the scheduler. This heuristic
6412 favors the instruction that has more instructions depending on it.
6413 This is enabled by default when scheduling is enabled, i.e.@:
6414 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6415 at @option{-O2} or higher.
6417 @item -fsched2-use-traces
6418 @opindex fsched2-use-traces
6419 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
6420 allocation and additionally perform code duplication in order to increase the
6421 size of superblocks using tracer pass. See @option{-ftracer} for details on
6424 This mode should produce faster but significantly longer programs. Also
6425 without @option{-fbranch-probabilities} the traces constructed may not
6426 match the reality and hurt the performance. This only makes
6427 sense when scheduling after register allocation, i.e.@: with
6428 @option{-fschedule-insns2} or at @option{-O2} or higher.
6430 @item -freschedule-modulo-scheduled-loops
6431 @opindex freschedule-modulo-scheduled-loops
6432 The modulo scheduling comes before the traditional scheduling, if a loop
6433 was modulo scheduled we may want to prevent the later scheduling passes
6434 from changing its schedule, we use this option to control that.
6436 @item -fselective-scheduling
6437 @opindex fselective-scheduling
6438 Schedule instructions using selective scheduling algorithm. Selective
6439 scheduling runs instead of the first scheduler pass.
6441 @item -fselective-scheduling2
6442 @opindex fselective-scheduling2
6443 Schedule instructions using selective scheduling algorithm. Selective
6444 scheduling runs instead of the second scheduler pass.
6446 @item -fsel-sched-pipelining
6447 @opindex fsel-sched-pipelining
6448 Enable software pipelining of innermost loops during selective scheduling.
6449 This option has no effect until one of @option{-fselective-scheduling} or
6450 @option{-fselective-scheduling2} is turned on.
6452 @item -fsel-sched-pipelining-outer-loops
6453 @opindex fsel-sched-pipelining-outer-loops
6454 When pipelining loops during selective scheduling, also pipeline outer loops.
6455 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6457 @item -fcaller-saves
6458 @opindex fcaller-saves
6459 Enable values to be allocated in registers that will be clobbered by
6460 function calls, by emitting extra instructions to save and restore the
6461 registers around such calls. Such allocation is done only when it
6462 seems to result in better code than would otherwise be produced.
6464 This option is always enabled by default on certain machines, usually
6465 those which have no call-preserved registers to use instead.
6467 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6469 @item -fconserve-stack
6470 @opindex fconserve-stack
6471 Attempt to minimize stack usage. The compiler will attempt to use less
6472 stack space, even if that makes the program slower. This option
6473 implies setting the @option{large-stack-frame} parameter to 100
6474 and the @option{large-stack-frame-growth} parameter to 400.
6476 @item -ftree-reassoc
6477 @opindex ftree-reassoc
6478 Perform reassociation on trees. This flag is enabled by default
6479 at @option{-O} and higher.
6483 Perform partial redundancy elimination (PRE) on trees. This flag is
6484 enabled by default at @option{-O2} and @option{-O3}.
6486 @item -ftree-forwprop
6487 @opindex ftree-forwprop
6488 Perform forward propagation on trees. This flag is enabled by default
6489 at @option{-O} and higher.
6493 Perform full redundancy elimination (FRE) on trees. The difference
6494 between FRE and PRE is that FRE only considers expressions
6495 that are computed on all paths leading to the redundant computation.
6496 This analysis is faster than PRE, though it exposes fewer redundancies.
6497 This flag is enabled by default at @option{-O} and higher.
6499 @item -ftree-phiprop
6500 @opindex ftree-phiprop
6501 Perform hoisting of loads from conditional pointers on trees. This
6502 pass is enabled by default at @option{-O} and higher.
6504 @item -ftree-copy-prop
6505 @opindex ftree-copy-prop
6506 Perform copy propagation on trees. This pass eliminates unnecessary
6507 copy operations. This flag is enabled by default at @option{-O} and
6510 @item -fipa-pure-const
6511 @opindex fipa-pure-const
6512 Discover which functions are pure or constant.
6513 Enabled by default at @option{-O} and higher.
6515 @item -fipa-reference
6516 @opindex fipa-reference
6517 Discover which static variables do not escape cannot escape the
6519 Enabled by default at @option{-O} and higher.
6521 @item -fipa-struct-reorg
6522 @opindex fipa-struct-reorg
6523 Perform structure reorganization optimization, that change C-like structures
6524 layout in order to better utilize spatial locality. This transformation is
6525 affective for programs containing arrays of structures. Available in two
6526 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6527 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6528 to provide the safety of this transformation. It works only in whole program
6529 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6530 enabled. Structures considered @samp{cold} by this transformation are not
6531 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6533 With this flag, the program debug info reflects a new structure layout.
6537 Perform interprocedural pointer analysis. This option is experimental
6538 and does not affect generated code.
6542 Perform interprocedural constant propagation.
6543 This optimization analyzes the program to determine when values passed
6544 to functions are constants and then optimizes accordingly.
6545 This optimization can substantially increase performance
6546 if the application has constants passed to functions.
6547 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6549 @item -fipa-cp-clone
6550 @opindex fipa-cp-clone
6551 Perform function cloning to make interprocedural constant propagation stronger.
6552 When enabled, interprocedural constant propagation will perform function cloning
6553 when externally visible function can be called with constant arguments.
6554 Because this optimization can create multiple copies of functions,
6555 it may significantly increase code size
6556 (see @option{--param ipcp-unit-growth=@var{value}}).
6557 This flag is enabled by default at @option{-O3}.
6559 @item -fipa-matrix-reorg
6560 @opindex fipa-matrix-reorg
6561 Perform matrix flattening and transposing.
6562 Matrix flattening tries to replace an @math{m}-dimensional matrix
6563 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6564 This reduces the level of indirection needed for accessing the elements
6565 of the matrix. The second optimization is matrix transposing that
6566 attempts to change the order of the matrix's dimensions in order to
6567 improve cache locality.
6568 Both optimizations need the @option{-fwhole-program} flag.
6569 Transposing is enabled only if profiling information is available.
6573 Perform forward store motion on trees. This flag is
6574 enabled by default at @option{-O} and higher.
6578 Perform sparse conditional constant propagation (CCP) on trees. This
6579 pass only operates on local scalar variables and is enabled by default
6580 at @option{-O} and higher.
6582 @item -ftree-switch-conversion
6583 Perform conversion of simple initializations in a switch to
6584 initializations from a scalar array. This flag is enabled by default
6585 at @option{-O2} and higher.
6589 Perform dead code elimination (DCE) on trees. This flag is enabled by
6590 default at @option{-O} and higher.
6592 @item -ftree-builtin-call-dce
6593 @opindex ftree-builtin-call-dce
6594 Perform conditional dead code elimination (DCE) for calls to builtin functions
6595 that may set @code{errno} but are otherwise side-effect free. This flag is
6596 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6599 @item -ftree-dominator-opts
6600 @opindex ftree-dominator-opts
6601 Perform a variety of simple scalar cleanups (constant/copy
6602 propagation, redundancy elimination, range propagation and expression
6603 simplification) based on a dominator tree traversal. This also
6604 performs jump threading (to reduce jumps to jumps). This flag is
6605 enabled by default at @option{-O} and higher.
6609 Perform dead store elimination (DSE) on trees. A dead store is a store into
6610 a memory location which will later be overwritten by another store without
6611 any intervening loads. In this case the earlier store can be deleted. This
6612 flag is enabled by default at @option{-O} and higher.
6616 Perform loop header copying on trees. This is beneficial since it increases
6617 effectiveness of code motion optimizations. It also saves one jump. This flag
6618 is enabled by default at @option{-O} and higher. It is not enabled
6619 for @option{-Os}, since it usually increases code size.
6621 @item -ftree-loop-optimize
6622 @opindex ftree-loop-optimize
6623 Perform loop optimizations on trees. This flag is enabled by default
6624 at @option{-O} and higher.
6626 @item -ftree-loop-linear
6627 @opindex ftree-loop-linear
6628 Perform linear loop transformations on tree. This flag can improve cache
6629 performance and allow further loop optimizations to take place.
6631 @item -floop-interchange
6632 Perform loop interchange transformations on loops. Interchanging two
6633 nested loops switches the inner and outer loops. For example, given a
6638 A(J, I) = A(J, I) * C
6642 loop interchange will transform the loop as if the user had written:
6646 A(J, I) = A(J, I) * C
6650 which can be beneficial when @code{N} is larger than the caches,
6651 because in Fortran, the elements of an array are stored in memory
6652 contiguously by column, and the original loop iterates over rows,
6653 potentially creating at each access a cache miss. This optimization
6654 applies to all the languages supported by GCC and is not limited to
6655 Fortran. To use this code transformation, GCC has to be configured
6656 with @option{--with-ppl} and @option{--with-cloog} to enable the
6657 Graphite loop transformation infrastructure.
6659 @item -floop-strip-mine
6660 Perform loop strip mining transformations on loops. Strip mining
6661 splits a loop into two nested loops. The outer loop has strides
6662 equal to the strip size and the inner loop has strides of the
6663 original loop within a strip. For example, given a loop like:
6669 loop strip mining will transform the loop as if the user had written:
6672 DO I = II, min (II + 3, N)
6677 This optimization applies to all the languages supported by GCC and is
6678 not limited to Fortran. To use this code transformation, GCC has to
6679 be configured with @option{--with-ppl} and @option{--with-cloog} to
6680 enable the Graphite loop transformation infrastructure.
6683 Perform loop blocking transformations on loops. Blocking strip mines
6684 each loop in the loop nest such that the memory accesses of the
6685 element loops fit inside caches. For example, given a loop like:
6689 A(J, I) = B(I) + C(J)
6693 loop blocking will transform the loop as if the user had written:
6697 DO I = II, min (II + 63, N)
6698 DO J = JJ, min (JJ + 63, M)
6699 A(J, I) = B(I) + C(J)
6705 which can be beneficial when @code{M} is larger than the caches,
6706 because the innermost loop will iterate over a smaller amount of data
6707 that can be kept in the caches. This optimization applies to all the
6708 languages supported by GCC and is not limited to Fortran. To use this
6709 code transformation, GCC has to be configured with @option{--with-ppl}
6710 and @option{--with-cloog} to enable the Graphite loop transformation
6713 @item -fgraphite-identity
6714 @opindex fgraphite-identity
6715 Enable the identity transformation for graphite. For every SCoP we generate
6716 the polyhedral representation and transform it back to gimple. Using
6717 @option{-fgraphite-identity} we can check the costs or benefits of the
6718 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6719 are also performed by the code generator CLooG, like index splitting and
6720 dead code elimination in loops.
6722 @item -floop-parallelize-all
6723 Use the Graphite data dependence analysis to identify loops that can
6724 be parallelized. Parallelize all the loops that can be analyzed to
6725 not contain loop carried dependences without checking that it is
6726 profitable to parallelize the loops.
6728 @item -fcheck-data-deps
6729 @opindex fcheck-data-deps
6730 Compare the results of several data dependence analyzers. This option
6731 is used for debugging the data dependence analyzers.
6733 @item -ftree-loop-distribution
6734 Perform loop distribution. This flag can improve cache performance on
6735 big loop bodies and allow further loop optimizations, like
6736 parallelization or vectorization, to take place. For example, the loop
6753 @item -ftree-loop-im
6754 @opindex ftree-loop-im
6755 Perform loop invariant motion on trees. This pass moves only invariants that
6756 would be hard to handle at RTL level (function calls, operations that expand to
6757 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6758 operands of conditions that are invariant out of the loop, so that we can use
6759 just trivial invariantness analysis in loop unswitching. The pass also includes
6762 @item -ftree-loop-ivcanon
6763 @opindex ftree-loop-ivcanon
6764 Create a canonical counter for number of iterations in the loop for that
6765 determining number of iterations requires complicated analysis. Later
6766 optimizations then may determine the number easily. Useful especially
6767 in connection with unrolling.
6771 Perform induction variable optimizations (strength reduction, induction
6772 variable merging and induction variable elimination) on trees.
6774 @item -ftree-parallelize-loops=n
6775 @opindex ftree-parallelize-loops
6776 Parallelize loops, i.e., split their iteration space to run in n threads.
6777 This is only possible for loops whose iterations are independent
6778 and can be arbitrarily reordered. The optimization is only
6779 profitable on multiprocessor machines, for loops that are CPU-intensive,
6780 rather than constrained e.g.@: by memory bandwidth. This option
6781 implies @option{-pthread}, and thus is only supported on targets
6782 that have support for @option{-pthread}.
6786 Perform function-local points-to analysis on trees. This flag is
6787 enabled by default at @option{-O} and higher.
6791 Perform scalar replacement of aggregates. This pass replaces structure
6792 references with scalars to prevent committing structures to memory too
6793 early. This flag is enabled by default at @option{-O} and higher.
6795 @item -ftree-copyrename
6796 @opindex ftree-copyrename
6797 Perform copy renaming on trees. This pass attempts to rename compiler
6798 temporaries to other variables at copy locations, usually resulting in
6799 variable names which more closely resemble the original variables. This flag
6800 is enabled by default at @option{-O} and higher.
6804 Perform temporary expression replacement during the SSA->normal phase. Single
6805 use/single def temporaries are replaced at their use location with their
6806 defining expression. This results in non-GIMPLE code, but gives the expanders
6807 much more complex trees to work on resulting in better RTL generation. This is
6808 enabled by default at @option{-O} and higher.
6810 @item -ftree-vectorize
6811 @opindex ftree-vectorize
6812 Perform loop vectorization on trees. This flag is enabled by default at
6815 @item -ftree-vect-loop-version
6816 @opindex ftree-vect-loop-version
6817 Perform loop versioning when doing loop vectorization on trees. When a loop
6818 appears to be vectorizable except that data alignment or data dependence cannot
6819 be determined at compile time then vectorized and non-vectorized versions of
6820 the loop are generated along with runtime checks for alignment or dependence
6821 to control which version is executed. This option is enabled by default
6822 except at level @option{-Os} where it is disabled.
6824 @item -fvect-cost-model
6825 @opindex fvect-cost-model
6826 Enable cost model for vectorization.
6830 Perform Value Range Propagation on trees. This is similar to the
6831 constant propagation pass, but instead of values, ranges of values are
6832 propagated. This allows the optimizers to remove unnecessary range
6833 checks like array bound checks and null pointer checks. This is
6834 enabled by default at @option{-O2} and higher. Null pointer check
6835 elimination is only done if @option{-fdelete-null-pointer-checks} is
6840 Perform tail duplication to enlarge superblock size. This transformation
6841 simplifies the control flow of the function allowing other optimizations to do
6844 @item -funroll-loops
6845 @opindex funroll-loops
6846 Unroll loops whose number of iterations can be determined at compile
6847 time or upon entry to the loop. @option{-funroll-loops} implies
6848 @option{-frerun-cse-after-loop}. This option makes code larger,
6849 and may or may not make it run faster.
6851 @item -funroll-all-loops
6852 @opindex funroll-all-loops
6853 Unroll all loops, even if their number of iterations is uncertain when
6854 the loop is entered. This usually makes programs run more slowly.
6855 @option{-funroll-all-loops} implies the same options as
6856 @option{-funroll-loops},
6858 @item -fsplit-ivs-in-unroller
6859 @opindex fsplit-ivs-in-unroller
6860 Enables expressing of values of induction variables in later iterations
6861 of the unrolled loop using the value in the first iteration. This breaks
6862 long dependency chains, thus improving efficiency of the scheduling passes.
6864 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6865 same effect. However in cases the loop body is more complicated than
6866 a single basic block, this is not reliable. It also does not work at all
6867 on some of the architectures due to restrictions in the CSE pass.
6869 This optimization is enabled by default.
6871 @item -fvariable-expansion-in-unroller
6872 @opindex fvariable-expansion-in-unroller
6873 With this option, the compiler will create multiple copies of some
6874 local variables when unrolling a loop which can result in superior code.
6876 @item -fpredictive-commoning
6877 @opindex fpredictive-commoning
6878 Perform predictive commoning optimization, i.e., reusing computations
6879 (especially memory loads and stores) performed in previous
6880 iterations of loops.
6882 This option is enabled at level @option{-O3}.
6884 @item -fprefetch-loop-arrays
6885 @opindex fprefetch-loop-arrays
6886 If supported by the target machine, generate instructions to prefetch
6887 memory to improve the performance of loops that access large arrays.
6889 This option may generate better or worse code; results are highly
6890 dependent on the structure of loops within the source code.
6892 Disabled at level @option{-Os}.
6895 @itemx -fno-peephole2
6896 @opindex fno-peephole
6897 @opindex fno-peephole2
6898 Disable any machine-specific peephole optimizations. The difference
6899 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6900 are implemented in the compiler; some targets use one, some use the
6901 other, a few use both.
6903 @option{-fpeephole} is enabled by default.
6904 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6906 @item -fno-guess-branch-probability
6907 @opindex fno-guess-branch-probability
6908 Do not guess branch probabilities using heuristics.
6910 GCC will use heuristics to guess branch probabilities if they are
6911 not provided by profiling feedback (@option{-fprofile-arcs}). These
6912 heuristics are based on the control flow graph. If some branch probabilities
6913 are specified by @samp{__builtin_expect}, then the heuristics will be
6914 used to guess branch probabilities for the rest of the control flow graph,
6915 taking the @samp{__builtin_expect} info into account. The interactions
6916 between the heuristics and @samp{__builtin_expect} can be complex, and in
6917 some cases, it may be useful to disable the heuristics so that the effects
6918 of @samp{__builtin_expect} are easier to understand.
6920 The default is @option{-fguess-branch-probability} at levels
6921 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6923 @item -freorder-blocks
6924 @opindex freorder-blocks
6925 Reorder basic blocks in the compiled function in order to reduce number of
6926 taken branches and improve code locality.
6928 Enabled at levels @option{-O2}, @option{-O3}.
6930 @item -freorder-blocks-and-partition
6931 @opindex freorder-blocks-and-partition
6932 In addition to reordering basic blocks in the compiled function, in order
6933 to reduce number of taken branches, partitions hot and cold basic blocks
6934 into separate sections of the assembly and .o files, to improve
6935 paging and cache locality performance.
6937 This optimization is automatically turned off in the presence of
6938 exception handling, for linkonce sections, for functions with a user-defined
6939 section attribute and on any architecture that does not support named
6942 @item -freorder-functions
6943 @opindex freorder-functions
6944 Reorder functions in the object file in order to
6945 improve code locality. This is implemented by using special
6946 subsections @code{.text.hot} for most frequently executed functions and
6947 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6948 the linker so object file format must support named sections and linker must
6949 place them in a reasonable way.
6951 Also profile feedback must be available in to make this option effective. See
6952 @option{-fprofile-arcs} for details.
6954 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6956 @item -fstrict-aliasing
6957 @opindex fstrict-aliasing
6958 Allow the compiler to assume the strictest aliasing rules applicable to
6959 the language being compiled. For C (and C++), this activates
6960 optimizations based on the type of expressions. In particular, an
6961 object of one type is assumed never to reside at the same address as an
6962 object of a different type, unless the types are almost the same. For
6963 example, an @code{unsigned int} can alias an @code{int}, but not a
6964 @code{void*} or a @code{double}. A character type may alias any other
6967 @anchor{Type-punning}Pay special attention to code like this:
6980 The practice of reading from a different union member than the one most
6981 recently written to (called ``type-punning'') is common. Even with
6982 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6983 is accessed through the union type. So, the code above will work as
6984 expected. @xref{Structures unions enumerations and bit-fields
6985 implementation}. However, this code might not:
6996 Similarly, access by taking the address, casting the resulting pointer
6997 and dereferencing the result has undefined behavior, even if the cast
6998 uses a union type, e.g.:
7002 return ((union a_union *) &d)->i;
7006 The @option{-fstrict-aliasing} option is enabled at levels
7007 @option{-O2}, @option{-O3}, @option{-Os}.
7009 @item -fstrict-overflow
7010 @opindex fstrict-overflow
7011 Allow the compiler to assume strict signed overflow rules, depending
7012 on the language being compiled. For C (and C++) this means that
7013 overflow when doing arithmetic with signed numbers is undefined, which
7014 means that the compiler may assume that it will not happen. This
7015 permits various optimizations. For example, the compiler will assume
7016 that an expression like @code{i + 10 > i} will always be true for
7017 signed @code{i}. This assumption is only valid if signed overflow is
7018 undefined, as the expression is false if @code{i + 10} overflows when
7019 using twos complement arithmetic. When this option is in effect any
7020 attempt to determine whether an operation on signed numbers will
7021 overflow must be written carefully to not actually involve overflow.
7023 This option also allows the compiler to assume strict pointer
7024 semantics: given a pointer to an object, if adding an offset to that
7025 pointer does not produce a pointer to the same object, the addition is
7026 undefined. This permits the compiler to conclude that @code{p + u >
7027 p} is always true for a pointer @code{p} and unsigned integer
7028 @code{u}. This assumption is only valid because pointer wraparound is
7029 undefined, as the expression is false if @code{p + u} overflows using
7030 twos complement arithmetic.
7032 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7033 that integer signed overflow is fully defined: it wraps. When
7034 @option{-fwrapv} is used, there is no difference between
7035 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7036 integers. With @option{-fwrapv} certain types of overflow are
7037 permitted. For example, if the compiler gets an overflow when doing
7038 arithmetic on constants, the overflowed value can still be used with
7039 @option{-fwrapv}, but not otherwise.
7041 The @option{-fstrict-overflow} option is enabled at levels
7042 @option{-O2}, @option{-O3}, @option{-Os}.
7044 @item -falign-functions
7045 @itemx -falign-functions=@var{n}
7046 @opindex falign-functions
7047 Align the start of functions to the next power-of-two greater than
7048 @var{n}, skipping up to @var{n} bytes. For instance,
7049 @option{-falign-functions=32} aligns functions to the next 32-byte
7050 boundary, but @option{-falign-functions=24} would align to the next
7051 32-byte boundary only if this can be done by skipping 23 bytes or less.
7053 @option{-fno-align-functions} and @option{-falign-functions=1} are
7054 equivalent and mean that functions will not be aligned.
7056 Some assemblers only support this flag when @var{n} is a power of two;
7057 in that case, it is rounded up.
7059 If @var{n} is not specified or is zero, use a machine-dependent default.
7061 Enabled at levels @option{-O2}, @option{-O3}.
7063 @item -falign-labels
7064 @itemx -falign-labels=@var{n}
7065 @opindex falign-labels
7066 Align all branch targets to a power-of-two boundary, skipping up to
7067 @var{n} bytes like @option{-falign-functions}. This option can easily
7068 make code slower, because it must insert dummy operations for when the
7069 branch target is reached in the usual flow of the code.
7071 @option{-fno-align-labels} and @option{-falign-labels=1} are
7072 equivalent and mean that labels will not be aligned.
7074 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7075 are greater than this value, then their values are used instead.
7077 If @var{n} is not specified or is zero, use a machine-dependent default
7078 which is very likely to be @samp{1}, meaning no alignment.
7080 Enabled at levels @option{-O2}, @option{-O3}.
7083 @itemx -falign-loops=@var{n}
7084 @opindex falign-loops
7085 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7086 like @option{-falign-functions}. The hope is that the loop will be
7087 executed many times, which will make up for any execution of the dummy
7090 @option{-fno-align-loops} and @option{-falign-loops=1} are
7091 equivalent and mean that loops will not be aligned.
7093 If @var{n} is not specified or is zero, use a machine-dependent default.
7095 Enabled at levels @option{-O2}, @option{-O3}.
7098 @itemx -falign-jumps=@var{n}
7099 @opindex falign-jumps
7100 Align branch targets to a power-of-two boundary, for branch targets
7101 where the targets can only be reached by jumping, skipping up to @var{n}
7102 bytes like @option{-falign-functions}. In this case, no dummy operations
7105 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7106 equivalent and mean that loops will not be aligned.
7108 If @var{n} is not specified or is zero, use a machine-dependent default.
7110 Enabled at levels @option{-O2}, @option{-O3}.
7112 @item -funit-at-a-time
7113 @opindex funit-at-a-time
7114 This option is left for compatibility reasons. @option{-funit-at-a-time}
7115 has no effect, while @option{-fno-unit-at-a-time} implies
7116 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7120 @item -fno-toplevel-reorder
7121 @opindex fno-toplevel-reorder
7122 Do not reorder top-level functions, variables, and @code{asm}
7123 statements. Output them in the same order that they appear in the
7124 input file. When this option is used, unreferenced static variables
7125 will not be removed. This option is intended to support existing code
7126 which relies on a particular ordering. For new code, it is better to
7129 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7130 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7135 Constructs webs as commonly used for register allocation purposes and assign
7136 each web individual pseudo register. This allows the register allocation pass
7137 to operate on pseudos directly, but also strengthens several other optimization
7138 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7139 however, make debugging impossible, since variables will no longer stay in a
7142 Enabled by default with @option{-funroll-loops}.
7144 @item -fwhole-program
7145 @opindex fwhole-program
7146 Assume that the current compilation unit represents the whole program being
7147 compiled. All public functions and variables with the exception of @code{main}
7148 and those merged by attribute @code{externally_visible} become static functions
7149 and in effect are optimized more aggressively by interprocedural optimizers.
7150 While this option is equivalent to proper use of the @code{static} keyword for
7151 programs consisting of a single file, in combination with option
7152 @option{-combine}, @option{-flto} or @option{-fwhopr} this flag can be used to
7153 compile many smaller scale programs since the functions and variables become
7154 local for the whole combined compilation unit, not for the single source file
7157 This option implies @option{-fwhole-file} for Fortran programs.
7161 This option runs the standard link-time optimizer. When invoked
7162 with source code, it generates GIMPLE (one of GCC's internal
7163 representations) and writes it to special ELF sections in the object
7164 file. When the object files are linked together, all the function
7165 bodies are read from these ELF sections and instantiated as if they
7166 had been part of the same translation unit.
7168 To use the link-timer optimizer, @option{-flto} needs to be specified at
7169 compile time and during the final link. For example,
7172 gcc -c -O2 -flto foo.c
7173 gcc -c -O2 -flto bar.c
7174 gcc -o myprog -flto -O2 foo.o bar.o
7177 The first two invocations to GCC will save a bytecode representation
7178 of GIMPLE into special ELF sections inside @file{foo.o} and
7179 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7180 @file{foo.o} and @file{bar.o}, merge the two files into a single
7181 internal image, and compile the result as usual. Since both
7182 @file{foo.o} and @file{bar.o} are merged into a single image, this
7183 causes all the inter-procedural analyses and optimizations in GCC to
7184 work across the two files as if they were a single one. This means,
7185 for example, that the inliner will be able to inline functions in
7186 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7188 Another (simpler) way to enable link-time optimization is,
7191 gcc -o myprog -flto -O2 foo.c bar.c
7194 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7195 merge them together into a single GIMPLE representation and optimize
7196 them as usual to produce @file{myprog}.
7198 The only important thing to keep in mind is that to enable link-time
7199 optimizations the @option{-flto} flag needs to be passed to both the
7200 compile and the link commands.
7202 Note that when a file is compiled with @option{-flto}, the generated
7203 object file will be larger than a regular object file because it will
7204 contain GIMPLE bytecodes and the usual final code. This means that
7205 object files with LTO information can be linked as a normal object
7206 file. So, in the previous example, if the final link is done with
7209 gcc -o myprog foo.o bar.o
7212 The only difference will be that no inter-procedural optimizations
7213 will be applied to produce @file{myprog}. The two object files
7214 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7217 Additionally, the optimization flags used to compile individual files
7218 are not necessarily related to those used at link-time. For instance,
7221 gcc -c -O0 -flto foo.c
7222 gcc -c -O0 -flto bar.c
7223 gcc -o myprog -flto -O3 foo.o bar.o
7226 This will produce individual object files with unoptimized assembler
7227 code, but the resulting binary @file{myprog} will be optimized at
7228 @option{-O3}. Now, if the final binary is generated without
7229 @option{-flto}, then @file{myprog} will not be optimized.
7231 When producing the final binary with @option{-flto}, GCC will only
7232 apply link-time optimizations to those files that contain bytecode.
7233 Therefore, you can mix and match object files and libraries with
7234 GIMPLE bytecodes and final object code. GCC will automatically select
7235 which files to optimize in LTO mode and which files to link without
7238 There are some code generation flags that GCC will preserve when
7239 generating bytecodes, as they need to be used during the final link
7240 stage. Currently, the following options are saved into the GIMPLE
7241 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7242 @option{-m} target flags.
7244 At link time, these options are read-in and reapplied. Note that the
7245 current implementation makes no attempt at recognizing conflicting
7246 values for these options. If two or more files have a conflicting
7247 value (e.g., one file is compiled with @option{-fPIC} and another
7248 isn't), the compiler will simply use the last value read from the
7249 bytecode files. It is recommended, then, that all the files
7250 participating in the same link be compiled with the same options.
7252 Another feature of LTO is that it is possible to apply interprocedural
7253 optimizations on files written in different languages. This requires
7254 some support in the language front end. Currently, the C, C++ and
7255 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7256 something like this should work
7261 gfortran -c -flto baz.f90
7262 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7265 Notice that the final link is done with @command{g++} to get the C++
7266 runtime libraries and @option{-lgfortran} is added to get the Fortran
7267 runtime libraries. In general, when mixing languages in LTO mode, you
7268 should use the same link command used when mixing languages in a
7269 regular (non-LTO) compilation. This means that if your build process
7270 was mixing languages before, all you need to add is @option{-flto} to
7271 all the compile and link commands.
7273 If object files containing GIMPLE bytecode are stored in a library
7274 archive, say @file{libfoo.a}, it is possible to extract and use them
7275 in an LTO link if you are using @command{gold} as the linker (which,
7276 in turn requires GCC to be configured with @option{--enable-gold}).
7277 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7281 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7284 With the linker plugin enabled, @command{gold} will extract the needed
7285 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7286 to make them part of the aggregated GIMPLE image to be optimized.
7288 If you are not using @command{gold} and/or do not specify
7289 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7290 will be extracted and linked as usual, but they will not participate
7291 in the LTO optimization process.
7293 Link time optimizations do not require the presence of the whole
7294 program to operate. If the program does not require any symbols to
7295 be exported, it is possible to combine @option{-flto} and
7296 @option{-fwhopr} with @option{-fwhole-program} to allow the
7297 interprocedural optimizers to use more aggressive assumptions which
7298 may lead to improved optimization opportunities.
7300 Regarding portability: the current implementation of LTO makes no
7301 attempt at generating bytecode that can be ported between different
7302 types of hosts. The bytecode files are versioned and there is a
7303 strict version check, so bytecode files generated in one version of
7304 GCC will not work with an older/newer version of GCC.
7306 This option is disabled by default.
7310 This option is identical in functionality to @option{-flto} but it
7311 differs in how the final link stage is executed. Instead of loading
7312 all the function bodies in memory, the callgraph is analyzed and
7313 optimization decisions are made (whole program analysis or WPA). Once
7314 optimization decisions are made, the callgraph is partitioned and the
7315 different sections are compiled separately (local transformations or
7316 LTRANS)@. This process allows optimizations on very large programs
7317 that otherwise would not fit in memory. This option enables
7318 @option{-fwpa} and @option{-fltrans} automatically.
7320 Disabled by default.
7324 This is an internal option used by GCC when compiling with
7325 @option{-fwhopr}. You should never need to use it.
7327 This option runs the link-time optimizer in the whole-program-analysis
7328 (WPA) mode, which reads in summary information from all inputs and
7329 performs a whole-program analysis based on summary information only.
7330 It generates object files for subsequent runs of the link-time
7331 optimizer where individual object files are optimized using both
7332 summary information from the WPA mode and the actual function bodies.
7333 It then drives the LTRANS phase.
7335 Disabled by default.
7339 This is an internal option used by GCC when compiling with
7340 @option{-fwhopr}. You should never need to use it.
7342 This option runs the link-time optimizer in the local-transformation (LTRANS)
7343 mode, which reads in output from a previous run of the LTO in WPA mode.
7344 In the LTRANS mode, LTO optimizes an object and produces the final assembly.
7346 Disabled by default.
7348 @item -fltrans-output-list=@var{file}
7349 @opindex fltrans-output-list
7350 This is an internal option used by GCC when compiling with
7351 @option{-fwhopr}. You should never need to use it.
7353 This option specifies a file to which the names of LTRANS output files are
7354 written. This option is only meaningful in conjunction with @option{-fwpa}.
7356 Disabled by default.
7358 @item -flto-compression-level=@var{n}
7359 This option specifies the level of compression used for intermediate
7360 language written to LTO object files, and is only meaningful in
7361 conjunction with LTO mode (@option{-fwhopr}, @option{-flto}). Valid
7362 values are 0 (no compression) to 9 (maximum compression). Values
7363 outside this range are clamped to either 0 or 9. If the option is not
7364 given, a default balanced compression setting is used.
7367 Prints a report with internal details on the workings of the link-time
7368 optimizer. The contents of this report vary from version to version,
7369 it is meant to be useful to GCC developers when processing object
7370 files in LTO mode (via @option{-fwhopr} or @option{-flto}).
7372 Disabled by default.
7374 @item -fuse-linker-plugin
7375 Enables the extraction of objects with GIMPLE bytecode information
7376 from library archives. This option relies on features available only
7377 in @command{gold}, so to use this you must configure GCC with
7378 @option{--enable-gold}. See @option{-flto} for a description on the
7379 effect of this flag and how to use it.
7381 Disabled by default.
7383 @item -fcprop-registers
7384 @opindex fcprop-registers
7385 After register allocation and post-register allocation instruction splitting,
7386 we perform a copy-propagation pass to try to reduce scheduling dependencies
7387 and occasionally eliminate the copy.
7389 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7391 @item -fprofile-correction
7392 @opindex fprofile-correction
7393 Profiles collected using an instrumented binary for multi-threaded programs may
7394 be inconsistent due to missed counter updates. When this option is specified,
7395 GCC will use heuristics to correct or smooth out such inconsistencies. By
7396 default, GCC will emit an error message when an inconsistent profile is detected.
7398 @item -fprofile-dir=@var{path}
7399 @opindex fprofile-dir
7401 Set the directory to search the profile data files in to @var{path}.
7402 This option affects only the profile data generated by
7403 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7404 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7405 and its related options.
7406 By default, GCC will use the current directory as @var{path}
7407 thus the profile data file will appear in the same directory as the object file.
7409 @item -fprofile-generate
7410 @itemx -fprofile-generate=@var{path}
7411 @opindex fprofile-generate
7413 Enable options usually used for instrumenting application to produce
7414 profile useful for later recompilation with profile feedback based
7415 optimization. You must use @option{-fprofile-generate} both when
7416 compiling and when linking your program.
7418 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7420 If @var{path} is specified, GCC will look at the @var{path} to find
7421 the profile feedback data files. See @option{-fprofile-dir}.
7424 @itemx -fprofile-use=@var{path}
7425 @opindex fprofile-use
7426 Enable profile feedback directed optimizations, and optimizations
7427 generally profitable only with profile feedback available.
7429 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7430 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7432 By default, GCC emits an error message if the feedback profiles do not
7433 match the source code. This error can be turned into a warning by using
7434 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7437 If @var{path} is specified, GCC will look at the @var{path} to find
7438 the profile feedback data files. See @option{-fprofile-dir}.
7441 The following options control compiler behavior regarding floating
7442 point arithmetic. These options trade off between speed and
7443 correctness. All must be specifically enabled.
7447 @opindex ffloat-store
7448 Do not store floating point variables in registers, and inhibit other
7449 options that might change whether a floating point value is taken from a
7452 @cindex floating point precision
7453 This option prevents undesirable excess precision on machines such as
7454 the 68000 where the floating registers (of the 68881) keep more
7455 precision than a @code{double} is supposed to have. Similarly for the
7456 x86 architecture. For most programs, the excess precision does only
7457 good, but a few programs rely on the precise definition of IEEE floating
7458 point. Use @option{-ffloat-store} for such programs, after modifying
7459 them to store all pertinent intermediate computations into variables.
7461 @item -fexcess-precision=@var{style}
7462 @opindex fexcess-precision
7463 This option allows further control over excess precision on machines
7464 where floating-point registers have more precision than the IEEE
7465 @code{float} and @code{double} types and the processor does not
7466 support operations rounding to those types. By default,
7467 @option{-fexcess-precision=fast} is in effect; this means that
7468 operations are carried out in the precision of the registers and that
7469 it is unpredictable when rounding to the types specified in the source
7470 code takes place. When compiling C, if
7471 @option{-fexcess-precision=standard} is specified then excess
7472 precision will follow the rules specified in ISO C99; in particular,
7473 both casts and assignments cause values to be rounded to their
7474 semantic types (whereas @option{-ffloat-store} only affects
7475 assignments). This option is enabled by default for C if a strict
7476 conformance option such as @option{-std=c99} is used.
7479 @option{-fexcess-precision=standard} is not implemented for languages
7480 other than C, and has no effect if
7481 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7482 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7483 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7484 semantics apply without excess precision, and in the latter, rounding
7489 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7490 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7491 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7493 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7495 This option is not turned on by any @option{-O} option since
7496 it can result in incorrect output for programs which depend on
7497 an exact implementation of IEEE or ISO rules/specifications for
7498 math functions. It may, however, yield faster code for programs
7499 that do not require the guarantees of these specifications.
7501 @item -fno-math-errno
7502 @opindex fno-math-errno
7503 Do not set ERRNO after calling math functions that are executed
7504 with a single instruction, e.g., sqrt. A program that relies on
7505 IEEE exceptions for math error handling may want to use this flag
7506 for speed while maintaining IEEE arithmetic compatibility.
7508 This option is not turned on by any @option{-O} option since
7509 it can result in incorrect output for programs which depend on
7510 an exact implementation of IEEE or ISO rules/specifications for
7511 math functions. It may, however, yield faster code for programs
7512 that do not require the guarantees of these specifications.
7514 The default is @option{-fmath-errno}.
7516 On Darwin systems, the math library never sets @code{errno}. There is
7517 therefore no reason for the compiler to consider the possibility that
7518 it might, and @option{-fno-math-errno} is the default.
7520 @item -funsafe-math-optimizations
7521 @opindex funsafe-math-optimizations
7523 Allow optimizations for floating-point arithmetic that (a) assume
7524 that arguments and results are valid and (b) may violate IEEE or
7525 ANSI standards. When used at link-time, it may include libraries
7526 or startup files that change the default FPU control word or other
7527 similar optimizations.
7529 This option is not turned on by any @option{-O} option since
7530 it can result in incorrect output for programs which depend on
7531 an exact implementation of IEEE or ISO rules/specifications for
7532 math functions. It may, however, yield faster code for programs
7533 that do not require the guarantees of these specifications.
7534 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7535 @option{-fassociative-math} and @option{-freciprocal-math}.
7537 The default is @option{-fno-unsafe-math-optimizations}.
7539 @item -fassociative-math
7540 @opindex fassociative-math
7542 Allow re-association of operands in series of floating-point operations.
7543 This violates the ISO C and C++ language standard by possibly changing
7544 computation result. NOTE: re-ordering may change the sign of zero as
7545 well as ignore NaNs and inhibit or create underflow or overflow (and
7546 thus cannot be used on a code which relies on rounding behavior like
7547 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7548 and thus may not be used when ordered comparisons are required.
7549 This option requires that both @option{-fno-signed-zeros} and
7550 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7551 much sense with @option{-frounding-math}.
7553 The default is @option{-fno-associative-math}.
7555 @item -freciprocal-math
7556 @opindex freciprocal-math
7558 Allow the reciprocal of a value to be used instead of dividing by
7559 the value if this enables optimizations. For example @code{x / y}
7560 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7561 is subject to common subexpression elimination. Note that this loses
7562 precision and increases the number of flops operating on the value.
7564 The default is @option{-fno-reciprocal-math}.
7566 @item -ffinite-math-only
7567 @opindex ffinite-math-only
7568 Allow optimizations for floating-point arithmetic that assume
7569 that arguments and results are not NaNs or +-Infs.
7571 This option is not turned on by any @option{-O} option since
7572 it can result in incorrect output for programs which depend on
7573 an exact implementation of IEEE or ISO rules/specifications for
7574 math functions. It may, however, yield faster code for programs
7575 that do not require the guarantees of these specifications.
7577 The default is @option{-fno-finite-math-only}.
7579 @item -fno-signed-zeros
7580 @opindex fno-signed-zeros
7581 Allow optimizations for floating point arithmetic that ignore the
7582 signedness of zero. IEEE arithmetic specifies the behavior of
7583 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7584 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7585 This option implies that the sign of a zero result isn't significant.
7587 The default is @option{-fsigned-zeros}.
7589 @item -fno-trapping-math
7590 @opindex fno-trapping-math
7591 Compile code assuming that floating-point operations cannot generate
7592 user-visible traps. These traps include division by zero, overflow,
7593 underflow, inexact result and invalid operation. This option requires
7594 that @option{-fno-signaling-nans} be in effect. Setting this option may
7595 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7597 This option should never be turned on by any @option{-O} option since
7598 it can result in incorrect output for programs which depend on
7599 an exact implementation of IEEE or ISO rules/specifications for
7602 The default is @option{-ftrapping-math}.
7604 @item -frounding-math
7605 @opindex frounding-math
7606 Disable transformations and optimizations that assume default floating
7607 point rounding behavior. This is round-to-zero for all floating point
7608 to integer conversions, and round-to-nearest for all other arithmetic
7609 truncations. This option should be specified for programs that change
7610 the FP rounding mode dynamically, or that may be executed with a
7611 non-default rounding mode. This option disables constant folding of
7612 floating point expressions at compile-time (which may be affected by
7613 rounding mode) and arithmetic transformations that are unsafe in the
7614 presence of sign-dependent rounding modes.
7616 The default is @option{-fno-rounding-math}.
7618 This option is experimental and does not currently guarantee to
7619 disable all GCC optimizations that are affected by rounding mode.
7620 Future versions of GCC may provide finer control of this setting
7621 using C99's @code{FENV_ACCESS} pragma. This command line option
7622 will be used to specify the default state for @code{FENV_ACCESS}.
7624 @item -fsignaling-nans
7625 @opindex fsignaling-nans
7626 Compile code assuming that IEEE signaling NaNs may generate user-visible
7627 traps during floating-point operations. Setting this option disables
7628 optimizations that may change the number of exceptions visible with
7629 signaling NaNs. This option implies @option{-ftrapping-math}.
7631 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7634 The default is @option{-fno-signaling-nans}.
7636 This option is experimental and does not currently guarantee to
7637 disable all GCC optimizations that affect signaling NaN behavior.
7639 @item -fsingle-precision-constant
7640 @opindex fsingle-precision-constant
7641 Treat floating point constant as single precision constant instead of
7642 implicitly converting it to double precision constant.
7644 @item -fcx-limited-range
7645 @opindex fcx-limited-range
7646 When enabled, this option states that a range reduction step is not
7647 needed when performing complex division. Also, there is no checking
7648 whether the result of a complex multiplication or division is @code{NaN
7649 + I*NaN}, with an attempt to rescue the situation in that case. The
7650 default is @option{-fno-cx-limited-range}, but is enabled by
7651 @option{-ffast-math}.
7653 This option controls the default setting of the ISO C99
7654 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7657 @item -fcx-fortran-rules
7658 @opindex fcx-fortran-rules
7659 Complex multiplication and division follow Fortran rules. Range
7660 reduction is done as part of complex division, but there is no checking
7661 whether the result of a complex multiplication or division is @code{NaN
7662 + I*NaN}, with an attempt to rescue the situation in that case.
7664 The default is @option{-fno-cx-fortran-rules}.
7668 The following options control optimizations that may improve
7669 performance, but are not enabled by any @option{-O} options. This
7670 section includes experimental options that may produce broken code.
7673 @item -fbranch-probabilities
7674 @opindex fbranch-probabilities
7675 After running a program compiled with @option{-fprofile-arcs}
7676 (@pxref{Debugging Options,, Options for Debugging Your Program or
7677 @command{gcc}}), you can compile it a second time using
7678 @option{-fbranch-probabilities}, to improve optimizations based on
7679 the number of times each branch was taken. When the program
7680 compiled with @option{-fprofile-arcs} exits it saves arc execution
7681 counts to a file called @file{@var{sourcename}.gcda} for each source
7682 file. The information in this data file is very dependent on the
7683 structure of the generated code, so you must use the same source code
7684 and the same optimization options for both compilations.
7686 With @option{-fbranch-probabilities}, GCC puts a
7687 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7688 These can be used to improve optimization. Currently, they are only
7689 used in one place: in @file{reorg.c}, instead of guessing which path a
7690 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7691 exactly determine which path is taken more often.
7693 @item -fprofile-values
7694 @opindex fprofile-values
7695 If combined with @option{-fprofile-arcs}, it adds code so that some
7696 data about values of expressions in the program is gathered.
7698 With @option{-fbranch-probabilities}, it reads back the data gathered
7699 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7700 notes to instructions for their later usage in optimizations.
7702 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7706 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7707 a code to gather information about values of expressions.
7709 With @option{-fbranch-probabilities}, it reads back the data gathered
7710 and actually performs the optimizations based on them.
7711 Currently the optimizations include specialization of division operation
7712 using the knowledge about the value of the denominator.
7714 @item -frename-registers
7715 @opindex frename-registers
7716 Attempt to avoid false dependencies in scheduled code by making use
7717 of registers left over after register allocation. This optimization
7718 will most benefit processors with lots of registers. Depending on the
7719 debug information format adopted by the target, however, it can
7720 make debugging impossible, since variables will no longer stay in
7721 a ``home register''.
7723 Enabled by default with @option{-funroll-loops}.
7727 Perform tail duplication to enlarge superblock size. This transformation
7728 simplifies the control flow of the function allowing other optimizations to do
7731 Enabled with @option{-fprofile-use}.
7733 @item -funroll-loops
7734 @opindex funroll-loops
7735 Unroll loops whose number of iterations can be determined at compile time or
7736 upon entry to the loop. @option{-funroll-loops} implies
7737 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7738 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7739 small constant number of iterations). This option makes code larger, and may
7740 or may not make it run faster.
7742 Enabled with @option{-fprofile-use}.
7744 @item -funroll-all-loops
7745 @opindex funroll-all-loops
7746 Unroll all loops, even if their number of iterations is uncertain when
7747 the loop is entered. This usually makes programs run more slowly.
7748 @option{-funroll-all-loops} implies the same options as
7749 @option{-funroll-loops}.
7752 @opindex fpeel-loops
7753 Peels the loops for that there is enough information that they do not
7754 roll much (from profile feedback). It also turns on complete loop peeling
7755 (i.e.@: complete removal of loops with small constant number of iterations).
7757 Enabled with @option{-fprofile-use}.
7759 @item -fmove-loop-invariants
7760 @opindex fmove-loop-invariants
7761 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7762 at level @option{-O1}
7764 @item -funswitch-loops
7765 @opindex funswitch-loops
7766 Move branches with loop invariant conditions out of the loop, with duplicates
7767 of the loop on both branches (modified according to result of the condition).
7769 @item -ffunction-sections
7770 @itemx -fdata-sections
7771 @opindex ffunction-sections
7772 @opindex fdata-sections
7773 Place each function or data item into its own section in the output
7774 file if the target supports arbitrary sections. The name of the
7775 function or the name of the data item determines the section's name
7778 Use these options on systems where the linker can perform optimizations
7779 to improve locality of reference in the instruction space. Most systems
7780 using the ELF object format and SPARC processors running Solaris 2 have
7781 linkers with such optimizations. AIX may have these optimizations in
7784 Only use these options when there are significant benefits from doing
7785 so. When you specify these options, the assembler and linker will
7786 create larger object and executable files and will also be slower.
7787 You will not be able to use @code{gprof} on all systems if you
7788 specify this option and you may have problems with debugging if
7789 you specify both this option and @option{-g}.
7791 @item -fbranch-target-load-optimize
7792 @opindex fbranch-target-load-optimize
7793 Perform branch target register load optimization before prologue / epilogue
7795 The use of target registers can typically be exposed only during reload,
7796 thus hoisting loads out of loops and doing inter-block scheduling needs
7797 a separate optimization pass.
7799 @item -fbranch-target-load-optimize2
7800 @opindex fbranch-target-load-optimize2
7801 Perform branch target register load optimization after prologue / epilogue
7804 @item -fbtr-bb-exclusive
7805 @opindex fbtr-bb-exclusive
7806 When performing branch target register load optimization, don't reuse
7807 branch target registers in within any basic block.
7809 @item -fstack-protector
7810 @opindex fstack-protector
7811 Emit extra code to check for buffer overflows, such as stack smashing
7812 attacks. This is done by adding a guard variable to functions with
7813 vulnerable objects. This includes functions that call alloca, and
7814 functions with buffers larger than 8 bytes. The guards are initialized
7815 when a function is entered and then checked when the function exits.
7816 If a guard check fails, an error message is printed and the program exits.
7818 @item -fstack-protector-all
7819 @opindex fstack-protector-all
7820 Like @option{-fstack-protector} except that all functions are protected.
7822 @item -fsection-anchors
7823 @opindex fsection-anchors
7824 Try to reduce the number of symbolic address calculations by using
7825 shared ``anchor'' symbols to address nearby objects. This transformation
7826 can help to reduce the number of GOT entries and GOT accesses on some
7829 For example, the implementation of the following function @code{foo}:
7833 int foo (void) @{ return a + b + c; @}
7836 would usually calculate the addresses of all three variables, but if you
7837 compile it with @option{-fsection-anchors}, it will access the variables
7838 from a common anchor point instead. The effect is similar to the
7839 following pseudocode (which isn't valid C):
7844 register int *xr = &x;
7845 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7849 Not all targets support this option.
7851 @item --param @var{name}=@var{value}
7853 In some places, GCC uses various constants to control the amount of
7854 optimization that is done. For example, GCC will not inline functions
7855 that contain more that a certain number of instructions. You can
7856 control some of these constants on the command-line using the
7857 @option{--param} option.
7859 The names of specific parameters, and the meaning of the values, are
7860 tied to the internals of the compiler, and are subject to change
7861 without notice in future releases.
7863 In each case, the @var{value} is an integer. The allowable choices for
7864 @var{name} are given in the following table:
7867 @item struct-reorg-cold-struct-ratio
7868 The threshold ratio (as a percentage) between a structure frequency
7869 and the frequency of the hottest structure in the program. This parameter
7870 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7871 We say that if the ratio of a structure frequency, calculated by profiling,
7872 to the hottest structure frequency in the program is less than this
7873 parameter, then structure reorganization is not applied to this structure.
7876 @item predictable-branch-cost-outcome
7877 When branch is predicted to be taken with probability lower than this threshold
7878 (in percent), then it is considered well predictable. The default is 10.
7880 @item max-crossjump-edges
7881 The maximum number of incoming edges to consider for crossjumping.
7882 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7883 the number of edges incoming to each block. Increasing values mean
7884 more aggressive optimization, making the compile time increase with
7885 probably small improvement in executable size.
7887 @item min-crossjump-insns
7888 The minimum number of instructions which must be matched at the end
7889 of two blocks before crossjumping will be performed on them. This
7890 value is ignored in the case where all instructions in the block being
7891 crossjumped from are matched. The default value is 5.
7893 @item max-grow-copy-bb-insns
7894 The maximum code size expansion factor when copying basic blocks
7895 instead of jumping. The expansion is relative to a jump instruction.
7896 The default value is 8.
7898 @item max-goto-duplication-insns
7899 The maximum number of instructions to duplicate to a block that jumps
7900 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7901 passes, GCC factors computed gotos early in the compilation process,
7902 and unfactors them as late as possible. Only computed jumps at the
7903 end of a basic blocks with no more than max-goto-duplication-insns are
7904 unfactored. The default value is 8.
7906 @item max-delay-slot-insn-search
7907 The maximum number of instructions to consider when looking for an
7908 instruction to fill a delay slot. If more than this arbitrary number of
7909 instructions is searched, the time savings from filling the delay slot
7910 will be minimal so stop searching. Increasing values mean more
7911 aggressive optimization, making the compile time increase with probably
7912 small improvement in executable run time.
7914 @item max-delay-slot-live-search
7915 When trying to fill delay slots, the maximum number of instructions to
7916 consider when searching for a block with valid live register
7917 information. Increasing this arbitrarily chosen value means more
7918 aggressive optimization, increasing the compile time. This parameter
7919 should be removed when the delay slot code is rewritten to maintain the
7922 @item max-gcse-memory
7923 The approximate maximum amount of memory that will be allocated in
7924 order to perform the global common subexpression elimination
7925 optimization. If more memory than specified is required, the
7926 optimization will not be done.
7928 @item max-pending-list-length
7929 The maximum number of pending dependencies scheduling will allow
7930 before flushing the current state and starting over. Large functions
7931 with few branches or calls can create excessively large lists which
7932 needlessly consume memory and resources.
7934 @item max-inline-insns-single
7935 Several parameters control the tree inliner used in gcc.
7936 This number sets the maximum number of instructions (counted in GCC's
7937 internal representation) in a single function that the tree inliner
7938 will consider for inlining. This only affects functions declared
7939 inline and methods implemented in a class declaration (C++).
7940 The default value is 300.
7942 @item max-inline-insns-auto
7943 When you use @option{-finline-functions} (included in @option{-O3}),
7944 a lot of functions that would otherwise not be considered for inlining
7945 by the compiler will be investigated. To those functions, a different
7946 (more restrictive) limit compared to functions declared inline can
7948 The default value is 50.
7950 @item large-function-insns
7951 The limit specifying really large functions. For functions larger than this
7952 limit after inlining, inlining is constrained by
7953 @option{--param large-function-growth}. This parameter is useful primarily
7954 to avoid extreme compilation time caused by non-linear algorithms used by the
7956 The default value is 2700.
7958 @item large-function-growth
7959 Specifies maximal growth of large function caused by inlining in percents.
7960 The default value is 100 which limits large function growth to 2.0 times
7963 @item large-unit-insns
7964 The limit specifying large translation unit. Growth caused by inlining of
7965 units larger than this limit is limited by @option{--param inline-unit-growth}.
7966 For small units this might be too tight (consider unit consisting of function A
7967 that is inline and B that just calls A three time. If B is small relative to
7968 A, the growth of unit is 300\% and yet such inlining is very sane. For very
7969 large units consisting of small inlineable functions however the overall unit
7970 growth limit is needed to avoid exponential explosion of code size. Thus for
7971 smaller units, the size is increased to @option{--param large-unit-insns}
7972 before applying @option{--param inline-unit-growth}. The default is 10000
7974 @item inline-unit-growth
7975 Specifies maximal overall growth of the compilation unit caused by inlining.
7976 The default value is 30 which limits unit growth to 1.3 times the original
7979 @item ipcp-unit-growth
7980 Specifies maximal overall growth of the compilation unit caused by
7981 interprocedural constant propagation. The default value is 10 which limits
7982 unit growth to 1.1 times the original size.
7984 @item large-stack-frame
7985 The limit specifying large stack frames. While inlining the algorithm is trying
7986 to not grow past this limit too much. Default value is 256 bytes.
7988 @item large-stack-frame-growth
7989 Specifies maximal growth of large stack frames caused by inlining in percents.
7990 The default value is 1000 which limits large stack frame growth to 11 times
7993 @item max-inline-insns-recursive
7994 @itemx max-inline-insns-recursive-auto
7995 Specifies maximum number of instructions out-of-line copy of self recursive inline
7996 function can grow into by performing recursive inlining.
7998 For functions declared inline @option{--param max-inline-insns-recursive} is
7999 taken into account. For function not declared inline, recursive inlining
8000 happens only when @option{-finline-functions} (included in @option{-O3}) is
8001 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8002 default value is 450.
8004 @item max-inline-recursive-depth
8005 @itemx max-inline-recursive-depth-auto
8006 Specifies maximum recursion depth used by the recursive inlining.
8008 For functions declared inline @option{--param max-inline-recursive-depth} is
8009 taken into account. For function not declared inline, recursive inlining
8010 happens only when @option{-finline-functions} (included in @option{-O3}) is
8011 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8014 @item min-inline-recursive-probability
8015 Recursive inlining is profitable only for function having deep recursion
8016 in average and can hurt for function having little recursion depth by
8017 increasing the prologue size or complexity of function body to other
8020 When profile feedback is available (see @option{-fprofile-generate}) the actual
8021 recursion depth can be guessed from probability that function will recurse via
8022 given call expression. This parameter limits inlining only to call expression
8023 whose probability exceeds given threshold (in percents). The default value is
8026 @item early-inlining-insns
8027 Specify growth that early inliner can make. In effect it increases amount of
8028 inlining for code having large abstraction penalty. The default value is 8.
8030 @item max-early-inliner-iterations
8031 @itemx max-early-inliner-iterations
8032 Limit of iterations of early inliner. This basically bounds number of nested
8033 indirect calls early inliner can resolve. Deeper chains are still handled by
8036 @item min-vect-loop-bound
8037 The minimum number of iterations under which a loop will not get vectorized
8038 when @option{-ftree-vectorize} is used. The number of iterations after
8039 vectorization needs to be greater than the value specified by this option
8040 to allow vectorization. The default value is 0.
8042 @item max-unrolled-insns
8043 The maximum number of instructions that a loop should have if that loop
8044 is unrolled, and if the loop is unrolled, it determines how many times
8045 the loop code is unrolled.
8047 @item max-average-unrolled-insns
8048 The maximum number of instructions biased by probabilities of their execution
8049 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8050 it determines how many times the loop code is unrolled.
8052 @item max-unroll-times
8053 The maximum number of unrollings of a single loop.
8055 @item max-peeled-insns
8056 The maximum number of instructions that a loop should have if that loop
8057 is peeled, and if the loop is peeled, it determines how many times
8058 the loop code is peeled.
8060 @item max-peel-times
8061 The maximum number of peelings of a single loop.
8063 @item max-completely-peeled-insns
8064 The maximum number of insns of a completely peeled loop.
8066 @item max-completely-peel-times
8067 The maximum number of iterations of a loop to be suitable for complete peeling.
8069 @item max-unswitch-insns
8070 The maximum number of insns of an unswitched loop.
8072 @item max-unswitch-level
8073 The maximum number of branches unswitched in a single loop.
8076 The minimum cost of an expensive expression in the loop invariant motion.
8078 @item iv-consider-all-candidates-bound
8079 Bound on number of candidates for induction variables below that
8080 all candidates are considered for each use in induction variable
8081 optimizations. Only the most relevant candidates are considered
8082 if there are more candidates, to avoid quadratic time complexity.
8084 @item iv-max-considered-uses
8085 The induction variable optimizations give up on loops that contain more
8086 induction variable uses.
8088 @item iv-always-prune-cand-set-bound
8089 If number of candidates in the set is smaller than this value,
8090 we always try to remove unnecessary ivs from the set during its
8091 optimization when a new iv is added to the set.
8093 @item scev-max-expr-size
8094 Bound on size of expressions used in the scalar evolutions analyzer.
8095 Large expressions slow the analyzer.
8097 @item omega-max-vars
8098 The maximum number of variables in an Omega constraint system.
8099 The default value is 128.
8101 @item omega-max-geqs
8102 The maximum number of inequalities in an Omega constraint system.
8103 The default value is 256.
8106 The maximum number of equalities in an Omega constraint system.
8107 The default value is 128.
8109 @item omega-max-wild-cards
8110 The maximum number of wildcard variables that the Omega solver will
8111 be able to insert. The default value is 18.
8113 @item omega-hash-table-size
8114 The size of the hash table in the Omega solver. The default value is
8117 @item omega-max-keys
8118 The maximal number of keys used by the Omega solver. The default
8121 @item omega-eliminate-redundant-constraints
8122 When set to 1, use expensive methods to eliminate all redundant
8123 constraints. The default value is 0.
8125 @item vect-max-version-for-alignment-checks
8126 The maximum number of runtime checks that can be performed when
8127 doing loop versioning for alignment in the vectorizer. See option
8128 ftree-vect-loop-version for more information.
8130 @item vect-max-version-for-alias-checks
8131 The maximum number of runtime checks that can be performed when
8132 doing loop versioning for alias in the vectorizer. See option
8133 ftree-vect-loop-version for more information.
8135 @item max-iterations-to-track
8137 The maximum number of iterations of a loop the brute force algorithm
8138 for analysis of # of iterations of the loop tries to evaluate.
8140 @item hot-bb-count-fraction
8141 Select fraction of the maximal count of repetitions of basic block in program
8142 given basic block needs to have to be considered hot.
8144 @item hot-bb-frequency-fraction
8145 Select fraction of the maximal frequency of executions of basic block in
8146 function given basic block needs to have to be considered hot
8148 @item max-predicted-iterations
8149 The maximum number of loop iterations we predict statically. This is useful
8150 in cases where function contain single loop with known bound and other loop
8151 with unknown. We predict the known number of iterations correctly, while
8152 the unknown number of iterations average to roughly 10. This means that the
8153 loop without bounds would appear artificially cold relative to the other one.
8155 @item align-threshold
8157 Select fraction of the maximal frequency of executions of basic block in
8158 function given basic block will get aligned.
8160 @item align-loop-iterations
8162 A loop expected to iterate at lest the selected number of iterations will get
8165 @item tracer-dynamic-coverage
8166 @itemx tracer-dynamic-coverage-feedback
8168 This value is used to limit superblock formation once the given percentage of
8169 executed instructions is covered. This limits unnecessary code size
8172 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8173 feedback is available. The real profiles (as opposed to statically estimated
8174 ones) are much less balanced allowing the threshold to be larger value.
8176 @item tracer-max-code-growth
8177 Stop tail duplication once code growth has reached given percentage. This is
8178 rather hokey argument, as most of the duplicates will be eliminated later in
8179 cross jumping, so it may be set to much higher values than is the desired code
8182 @item tracer-min-branch-ratio
8184 Stop reverse growth when the reverse probability of best edge is less than this
8185 threshold (in percent).
8187 @item tracer-min-branch-ratio
8188 @itemx tracer-min-branch-ratio-feedback
8190 Stop forward growth if the best edge do have probability lower than this
8193 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8194 compilation for profile feedback and one for compilation without. The value
8195 for compilation with profile feedback needs to be more conservative (higher) in
8196 order to make tracer effective.
8198 @item max-cse-path-length
8200 Maximum number of basic blocks on path that cse considers. The default is 10.
8203 The maximum instructions CSE process before flushing. The default is 1000.
8205 @item ggc-min-expand
8207 GCC uses a garbage collector to manage its own memory allocation. This
8208 parameter specifies the minimum percentage by which the garbage
8209 collector's heap should be allowed to expand between collections.
8210 Tuning this may improve compilation speed; it has no effect on code
8213 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8214 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8215 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8216 GCC is not able to calculate RAM on a particular platform, the lower
8217 bound of 30% is used. Setting this parameter and
8218 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8219 every opportunity. This is extremely slow, but can be useful for
8222 @item ggc-min-heapsize
8224 Minimum size of the garbage collector's heap before it begins bothering
8225 to collect garbage. The first collection occurs after the heap expands
8226 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8227 tuning this may improve compilation speed, and has no effect on code
8230 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8231 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8232 with a lower bound of 4096 (four megabytes) and an upper bound of
8233 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8234 particular platform, the lower bound is used. Setting this parameter
8235 very large effectively disables garbage collection. Setting this
8236 parameter and @option{ggc-min-expand} to zero causes a full collection
8237 to occur at every opportunity.
8239 @item max-reload-search-insns
8240 The maximum number of instruction reload should look backward for equivalent
8241 register. Increasing values mean more aggressive optimization, making the
8242 compile time increase with probably slightly better performance. The default
8245 @item max-cselib-memory-locations
8246 The maximum number of memory locations cselib should take into account.
8247 Increasing values mean more aggressive optimization, making the compile time
8248 increase with probably slightly better performance. The default value is 500.
8250 @item reorder-blocks-duplicate
8251 @itemx reorder-blocks-duplicate-feedback
8253 Used by basic block reordering pass to decide whether to use unconditional
8254 branch or duplicate the code on its destination. Code is duplicated when its
8255 estimated size is smaller than this value multiplied by the estimated size of
8256 unconditional jump in the hot spots of the program.
8258 The @option{reorder-block-duplicate-feedback} is used only when profile
8259 feedback is available and may be set to higher values than
8260 @option{reorder-block-duplicate} since information about the hot spots is more
8263 @item max-sched-ready-insns
8264 The maximum number of instructions ready to be issued the scheduler should
8265 consider at any given time during the first scheduling pass. Increasing
8266 values mean more thorough searches, making the compilation time increase
8267 with probably little benefit. The default value is 100.
8269 @item max-sched-region-blocks
8270 The maximum number of blocks in a region to be considered for
8271 interblock scheduling. The default value is 10.
8273 @item max-pipeline-region-blocks
8274 The maximum number of blocks in a region to be considered for
8275 pipelining in the selective scheduler. The default value is 15.
8277 @item max-sched-region-insns
8278 The maximum number of insns in a region to be considered for
8279 interblock scheduling. The default value is 100.
8281 @item max-pipeline-region-insns
8282 The maximum number of insns in a region to be considered for
8283 pipelining in the selective scheduler. The default value is 200.
8286 The minimum probability (in percents) of reaching a source block
8287 for interblock speculative scheduling. The default value is 40.
8289 @item max-sched-extend-regions-iters
8290 The maximum number of iterations through CFG to extend regions.
8291 0 - disable region extension,
8292 N - do at most N iterations.
8293 The default value is 0.
8295 @item max-sched-insn-conflict-delay
8296 The maximum conflict delay for an insn to be considered for speculative motion.
8297 The default value is 3.
8299 @item sched-spec-prob-cutoff
8300 The minimal probability of speculation success (in percents), so that
8301 speculative insn will be scheduled.
8302 The default value is 40.
8304 @item sched-mem-true-dep-cost
8305 Minimal distance (in CPU cycles) between store and load targeting same
8306 memory locations. The default value is 1.
8308 @item selsched-max-lookahead
8309 The maximum size of the lookahead window of selective scheduling. It is a
8310 depth of search for available instructions.
8311 The default value is 50.
8313 @item selsched-max-sched-times
8314 The maximum number of times that an instruction will be scheduled during
8315 selective scheduling. This is the limit on the number of iterations
8316 through which the instruction may be pipelined. The default value is 2.
8318 @item selsched-max-insns-to-rename
8319 The maximum number of best instructions in the ready list that are considered
8320 for renaming in the selective scheduler. The default value is 2.
8322 @item max-last-value-rtl
8323 The maximum size measured as number of RTLs that can be recorded in an expression
8324 in combiner for a pseudo register as last known value of that register. The default
8327 @item integer-share-limit
8328 Small integer constants can use a shared data structure, reducing the
8329 compiler's memory usage and increasing its speed. This sets the maximum
8330 value of a shared integer constant. The default value is 256.
8332 @item min-virtual-mappings
8333 Specifies the minimum number of virtual mappings in the incremental
8334 SSA updater that should be registered to trigger the virtual mappings
8335 heuristic defined by virtual-mappings-ratio. The default value is
8338 @item virtual-mappings-ratio
8339 If the number of virtual mappings is virtual-mappings-ratio bigger
8340 than the number of virtual symbols to be updated, then the incremental
8341 SSA updater switches to a full update for those symbols. The default
8344 @item ssp-buffer-size
8345 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8346 protection when @option{-fstack-protection} is used.
8348 @item max-jump-thread-duplication-stmts
8349 Maximum number of statements allowed in a block that needs to be
8350 duplicated when threading jumps.
8352 @item max-fields-for-field-sensitive
8353 Maximum number of fields in a structure we will treat in
8354 a field sensitive manner during pointer analysis. The default is zero
8355 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8357 @item prefetch-latency
8358 Estimate on average number of instructions that are executed before
8359 prefetch finishes. The distance we prefetch ahead is proportional
8360 to this constant. Increasing this number may also lead to less
8361 streams being prefetched (see @option{simultaneous-prefetches}).
8363 @item simultaneous-prefetches
8364 Maximum number of prefetches that can run at the same time.
8366 @item l1-cache-line-size
8367 The size of cache line in L1 cache, in bytes.
8370 The size of L1 cache, in kilobytes.
8373 The size of L2 cache, in kilobytes.
8375 @item min-insn-to-prefetch-ratio
8376 The minimum ratio between the number of instructions and the
8377 number of prefetches to enable prefetching in a loop with an
8380 @item prefetch-min-insn-to-mem-ratio
8381 The minimum ratio between the number of instructions and the
8382 number of memory references to enable prefetching in a loop.
8384 @item use-canonical-types
8385 Whether the compiler should use the ``canonical'' type system. By
8386 default, this should always be 1, which uses a more efficient internal
8387 mechanism for comparing types in C++ and Objective-C++. However, if
8388 bugs in the canonical type system are causing compilation failures,
8389 set this value to 0 to disable canonical types.
8391 @item switch-conversion-max-branch-ratio
8392 Switch initialization conversion will refuse to create arrays that are
8393 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8394 branches in the switch.
8396 @item max-partial-antic-length
8397 Maximum length of the partial antic set computed during the tree
8398 partial redundancy elimination optimization (@option{-ftree-pre}) when
8399 optimizing at @option{-O3} and above. For some sorts of source code
8400 the enhanced partial redundancy elimination optimization can run away,
8401 consuming all of the memory available on the host machine. This
8402 parameter sets a limit on the length of the sets that are computed,
8403 which prevents the runaway behavior. Setting a value of 0 for
8404 this parameter will allow an unlimited set length.
8406 @item sccvn-max-scc-size
8407 Maximum size of a strongly connected component (SCC) during SCCVN
8408 processing. If this limit is hit, SCCVN processing for the whole
8409 function will not be done and optimizations depending on it will
8410 be disabled. The default maximum SCC size is 10000.
8412 @item ira-max-loops-num
8413 IRA uses a regional register allocation by default. If a function
8414 contains loops more than number given by the parameter, only at most
8415 given number of the most frequently executed loops will form regions
8416 for the regional register allocation. The default value of the
8419 @item ira-max-conflict-table-size
8420 Although IRA uses a sophisticated algorithm of compression conflict
8421 table, the table can be still big for huge functions. If the conflict
8422 table for a function could be more than size in MB given by the
8423 parameter, the conflict table is not built and faster, simpler, and
8424 lower quality register allocation algorithm will be used. The
8425 algorithm do not use pseudo-register conflicts. The default value of
8426 the parameter is 2000.
8428 @item ira-loop-reserved-regs
8429 IRA can be used to evaluate more accurate register pressure in loops
8430 for decision to move loop invariants (see @option{-O3}). The number
8431 of available registers reserved for some other purposes is described
8432 by this parameter. The default value of the parameter is 2 which is
8433 minimal number of registers needed for execution of typical
8434 instruction. This value is the best found from numerous experiments.
8436 @item loop-invariant-max-bbs-in-loop
8437 Loop invariant motion can be very expensive, both in compile time and
8438 in amount of needed compile time memory, with very large loops. Loops
8439 with more basic blocks than this parameter won't have loop invariant
8440 motion optimization performed on them. The default value of the
8441 parameter is 1000 for -O1 and 10000 for -O2 and above.
8443 @item min-nondebug-insn-uid
8444 Use uids starting at this parameter for nondebug insns. The range below
8445 the parameter is reserved exclusively for debug insns created by
8446 @option{-fvar-tracking-assignments}, but debug insns may get
8447 (non-overlapping) uids above it if the reserved range is exhausted.
8449 @item ipa-sra-ptr-growth-factor
8450 IPA-SRA will replace a pointer to an aggregate with one or more new
8451 parameters only when their cumulative size is less or equal to
8452 @option{ipa-sra-ptr-growth-factor} times the size of the original
8458 @node Preprocessor Options
8459 @section Options Controlling the Preprocessor
8460 @cindex preprocessor options
8461 @cindex options, preprocessor
8463 These options control the C preprocessor, which is run on each C source
8464 file before actual compilation.
8466 If you use the @option{-E} option, nothing is done except preprocessing.
8467 Some of these options make sense only together with @option{-E} because
8468 they cause the preprocessor output to be unsuitable for actual
8472 @item -Wp,@var{option}
8474 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8475 and pass @var{option} directly through to the preprocessor. If
8476 @var{option} contains commas, it is split into multiple options at the
8477 commas. However, many options are modified, translated or interpreted
8478 by the compiler driver before being passed to the preprocessor, and
8479 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8480 interface is undocumented and subject to change, so whenever possible
8481 you should avoid using @option{-Wp} and let the driver handle the
8484 @item -Xpreprocessor @var{option}
8485 @opindex Xpreprocessor
8486 Pass @var{option} as an option to the preprocessor. You can use this to
8487 supply system-specific preprocessor options which GCC does not know how to
8490 If you want to pass an option that takes an argument, you must use
8491 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8494 @include cppopts.texi
8496 @node Assembler Options
8497 @section Passing Options to the Assembler
8499 @c prevent bad page break with this line
8500 You can pass options to the assembler.
8503 @item -Wa,@var{option}
8505 Pass @var{option} as an option to the assembler. If @var{option}
8506 contains commas, it is split into multiple options at the commas.
8508 @item -Xassembler @var{option}
8510 Pass @var{option} as an option to the assembler. You can use this to
8511 supply system-specific assembler options which GCC does not know how to
8514 If you want to pass an option that takes an argument, you must use
8515 @option{-Xassembler} twice, once for the option and once for the argument.
8520 @section Options for Linking
8521 @cindex link options
8522 @cindex options, linking
8524 These options come into play when the compiler links object files into
8525 an executable output file. They are meaningless if the compiler is
8526 not doing a link step.
8530 @item @var{object-file-name}
8531 A file name that does not end in a special recognized suffix is
8532 considered to name an object file or library. (Object files are
8533 distinguished from libraries by the linker according to the file
8534 contents.) If linking is done, these object files are used as input
8543 If any of these options is used, then the linker is not run, and
8544 object file names should not be used as arguments. @xref{Overall
8548 @item -l@var{library}
8549 @itemx -l @var{library}
8551 Search the library named @var{library} when linking. (The second
8552 alternative with the library as a separate argument is only for
8553 POSIX compliance and is not recommended.)
8555 It makes a difference where in the command you write this option; the
8556 linker searches and processes libraries and object files in the order they
8557 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8558 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8559 to functions in @samp{z}, those functions may not be loaded.
8561 The linker searches a standard list of directories for the library,
8562 which is actually a file named @file{lib@var{library}.a}. The linker
8563 then uses this file as if it had been specified precisely by name.
8565 The directories searched include several standard system directories
8566 plus any that you specify with @option{-L}.
8568 Normally the files found this way are library files---archive files
8569 whose members are object files. The linker handles an archive file by
8570 scanning through it for members which define symbols that have so far
8571 been referenced but not defined. But if the file that is found is an
8572 ordinary object file, it is linked in the usual fashion. The only
8573 difference between using an @option{-l} option and specifying a file name
8574 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8575 and searches several directories.
8579 You need this special case of the @option{-l} option in order to
8580 link an Objective-C or Objective-C++ program.
8583 @opindex nostartfiles
8584 Do not use the standard system startup files when linking.
8585 The standard system libraries are used normally, unless @option{-nostdlib}
8586 or @option{-nodefaultlibs} is used.
8588 @item -nodefaultlibs
8589 @opindex nodefaultlibs
8590 Do not use the standard system libraries when linking.
8591 Only the libraries you specify will be passed to the linker, options
8592 specifying linkage of the system libraries, such as @code{-static-libgcc}
8593 or @code{-shared-libgcc}, will be ignored.
8594 The standard startup files are used normally, unless @option{-nostartfiles}
8595 is used. The compiler may generate calls to @code{memcmp},
8596 @code{memset}, @code{memcpy} and @code{memmove}.
8597 These entries are usually resolved by entries in
8598 libc. These entry points should be supplied through some other
8599 mechanism when this option is specified.
8603 Do not use the standard system startup files or libraries when linking.
8604 No startup files and only the libraries you specify will be passed to
8605 the linker, options specifying linkage of the system libraries, such as
8606 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8607 The compiler may generate calls to @code{memcmp}, @code{memset},
8608 @code{memcpy} and @code{memmove}.
8609 These entries are usually resolved by entries in
8610 libc. These entry points should be supplied through some other
8611 mechanism when this option is specified.
8613 @cindex @option{-lgcc}, use with @option{-nostdlib}
8614 @cindex @option{-nostdlib} and unresolved references
8615 @cindex unresolved references and @option{-nostdlib}
8616 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8617 @cindex @option{-nodefaultlibs} and unresolved references
8618 @cindex unresolved references and @option{-nodefaultlibs}
8619 One of the standard libraries bypassed by @option{-nostdlib} and
8620 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8621 that GCC uses to overcome shortcomings of particular machines, or special
8622 needs for some languages.
8623 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8624 Collection (GCC) Internals},
8625 for more discussion of @file{libgcc.a}.)
8626 In most cases, you need @file{libgcc.a} even when you want to avoid
8627 other standard libraries. In other words, when you specify @option{-nostdlib}
8628 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8629 This ensures that you have no unresolved references to internal GCC
8630 library subroutines. (For example, @samp{__main}, used to ensure C++
8631 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8632 GNU Compiler Collection (GCC) Internals}.)
8636 Produce a position independent executable on targets which support it.
8637 For predictable results, you must also specify the same set of options
8638 that were used to generate code (@option{-fpie}, @option{-fPIE},
8639 or model suboptions) when you specify this option.
8643 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8644 that support it. This instructs the linker to add all symbols, not
8645 only used ones, to the dynamic symbol table. This option is needed
8646 for some uses of @code{dlopen} or to allow obtaining backtraces
8647 from within a program.
8651 Remove all symbol table and relocation information from the executable.
8655 On systems that support dynamic linking, this prevents linking with the shared
8656 libraries. On other systems, this option has no effect.
8660 Produce a shared object which can then be linked with other objects to
8661 form an executable. Not all systems support this option. For predictable
8662 results, you must also specify the same set of options that were used to
8663 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8664 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8665 needs to build supplementary stub code for constructors to work. On
8666 multi-libbed systems, @samp{gcc -shared} must select the correct support
8667 libraries to link against. Failing to supply the correct flags may lead
8668 to subtle defects. Supplying them in cases where they are not necessary
8671 @item -shared-libgcc
8672 @itemx -static-libgcc
8673 @opindex shared-libgcc
8674 @opindex static-libgcc
8675 On systems that provide @file{libgcc} as a shared library, these options
8676 force the use of either the shared or static version respectively.
8677 If no shared version of @file{libgcc} was built when the compiler was
8678 configured, these options have no effect.
8680 There are several situations in which an application should use the
8681 shared @file{libgcc} instead of the static version. The most common
8682 of these is when the application wishes to throw and catch exceptions
8683 across different shared libraries. In that case, each of the libraries
8684 as well as the application itself should use the shared @file{libgcc}.
8686 Therefore, the G++ and GCJ drivers automatically add
8687 @option{-shared-libgcc} whenever you build a shared library or a main
8688 executable, because C++ and Java programs typically use exceptions, so
8689 this is the right thing to do.
8691 If, instead, you use the GCC driver to create shared libraries, you may
8692 find that they will not always be linked with the shared @file{libgcc}.
8693 If GCC finds, at its configuration time, that you have a non-GNU linker
8694 or a GNU linker that does not support option @option{--eh-frame-hdr},
8695 it will link the shared version of @file{libgcc} into shared libraries
8696 by default. Otherwise, it will take advantage of the linker and optimize
8697 away the linking with the shared version of @file{libgcc}, linking with
8698 the static version of libgcc by default. This allows exceptions to
8699 propagate through such shared libraries, without incurring relocation
8700 costs at library load time.
8702 However, if a library or main executable is supposed to throw or catch
8703 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8704 for the languages used in the program, or using the option
8705 @option{-shared-libgcc}, such that it is linked with the shared
8708 @item -static-libstdc++
8709 When the @command{g++} program is used to link a C++ program, it will
8710 normally automatically link against @option{libstdc++}. If
8711 @file{libstdc++} is available as a shared library, and the
8712 @option{-static} option is not used, then this will link against the
8713 shared version of @file{libstdc++}. That is normally fine. However, it
8714 is sometimes useful to freeze the version of @file{libstdc++} used by
8715 the program without going all the way to a fully static link. The
8716 @option{-static-libstdc++} option directs the @command{g++} driver to
8717 link @file{libstdc++} statically, without necessarily linking other
8718 libraries statically.
8722 Bind references to global symbols when building a shared object. Warn
8723 about any unresolved references (unless overridden by the link editor
8724 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8727 @item -T @var{script}
8729 @cindex linker script
8730 Use @var{script} as the linker script. This option is supported by most
8731 systems using the GNU linker. On some targets, such as bare-board
8732 targets without an operating system, the @option{-T} option may be required
8733 when linking to avoid references to undefined symbols.
8735 @item -Xlinker @var{option}
8737 Pass @var{option} as an option to the linker. You can use this to
8738 supply system-specific linker options which GCC does not know how to
8741 If you want to pass an option that takes a separate argument, you must use
8742 @option{-Xlinker} twice, once for the option and once for the argument.
8743 For example, to pass @option{-assert definitions}, you must write
8744 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8745 @option{-Xlinker "-assert definitions"}, because this passes the entire
8746 string as a single argument, which is not what the linker expects.
8748 When using the GNU linker, it is usually more convenient to pass
8749 arguments to linker options using the @option{@var{option}=@var{value}}
8750 syntax than as separate arguments. For example, you can specify
8751 @samp{-Xlinker -Map=output.map} rather than
8752 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8753 this syntax for command-line options.
8755 @item -Wl,@var{option}
8757 Pass @var{option} as an option to the linker. If @var{option} contains
8758 commas, it is split into multiple options at the commas. You can use this
8759 syntax to pass an argument to the option.
8760 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8761 linker. When using the GNU linker, you can also get the same effect with
8762 @samp{-Wl,-Map=output.map}.
8764 @item -u @var{symbol}
8766 Pretend the symbol @var{symbol} is undefined, to force linking of
8767 library modules to define it. You can use @option{-u} multiple times with
8768 different symbols to force loading of additional library modules.
8771 @node Directory Options
8772 @section Options for Directory Search
8773 @cindex directory options
8774 @cindex options, directory search
8777 These options specify directories to search for header files, for
8778 libraries and for parts of the compiler:
8783 Add the directory @var{dir} to the head of the list of directories to be
8784 searched for header files. This can be used to override a system header
8785 file, substituting your own version, since these directories are
8786 searched before the system header file directories. However, you should
8787 not use this option to add directories that contain vendor-supplied
8788 system header files (use @option{-isystem} for that). If you use more than
8789 one @option{-I} option, the directories are scanned in left-to-right
8790 order; the standard system directories come after.
8792 If a standard system include directory, or a directory specified with
8793 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8794 option will be ignored. The directory will still be searched but as a
8795 system directory at its normal position in the system include chain.
8796 This is to ensure that GCC's procedure to fix buggy system headers and
8797 the ordering for the include_next directive are not inadvertently changed.
8798 If you really need to change the search order for system directories,
8799 use the @option{-nostdinc} and/or @option{-isystem} options.
8801 @item -iquote@var{dir}
8803 Add the directory @var{dir} to the head of the list of directories to
8804 be searched for header files only for the case of @samp{#include
8805 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8806 otherwise just like @option{-I}.
8810 Add directory @var{dir} to the list of directories to be searched
8813 @item -B@var{prefix}
8815 This option specifies where to find the executables, libraries,
8816 include files, and data files of the compiler itself.
8818 The compiler driver program runs one or more of the subprograms
8819 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8820 @var{prefix} as a prefix for each program it tries to run, both with and
8821 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8823 For each subprogram to be run, the compiler driver first tries the
8824 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8825 was not specified, the driver tries two standard prefixes, which are
8826 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8827 those results in a file name that is found, the unmodified program
8828 name is searched for using the directories specified in your
8829 @env{PATH} environment variable.
8831 The compiler will check to see if the path provided by the @option{-B}
8832 refers to a directory, and if necessary it will add a directory
8833 separator character at the end of the path.
8835 @option{-B} prefixes that effectively specify directory names also apply
8836 to libraries in the linker, because the compiler translates these
8837 options into @option{-L} options for the linker. They also apply to
8838 includes files in the preprocessor, because the compiler translates these
8839 options into @option{-isystem} options for the preprocessor. In this case,
8840 the compiler appends @samp{include} to the prefix.
8842 The run-time support file @file{libgcc.a} can also be searched for using
8843 the @option{-B} prefix, if needed. If it is not found there, the two
8844 standard prefixes above are tried, and that is all. The file is left
8845 out of the link if it is not found by those means.
8847 Another way to specify a prefix much like the @option{-B} prefix is to use
8848 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8851 As a special kludge, if the path provided by @option{-B} is
8852 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8853 9, then it will be replaced by @file{[dir/]include}. This is to help
8854 with boot-strapping the compiler.
8856 @item -specs=@var{file}
8858 Process @var{file} after the compiler reads in the standard @file{specs}
8859 file, in order to override the defaults that the @file{gcc} driver
8860 program uses when determining what switches to pass to @file{cc1},
8861 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8862 @option{-specs=@var{file}} can be specified on the command line, and they
8863 are processed in order, from left to right.
8865 @item --sysroot=@var{dir}
8867 Use @var{dir} as the logical root directory for headers and libraries.
8868 For example, if the compiler would normally search for headers in
8869 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8870 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8872 If you use both this option and the @option{-isysroot} option, then
8873 the @option{--sysroot} option will apply to libraries, but the
8874 @option{-isysroot} option will apply to header files.
8876 The GNU linker (beginning with version 2.16) has the necessary support
8877 for this option. If your linker does not support this option, the
8878 header file aspect of @option{--sysroot} will still work, but the
8879 library aspect will not.
8883 This option has been deprecated. Please use @option{-iquote} instead for
8884 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8885 Any directories you specify with @option{-I} options before the @option{-I-}
8886 option are searched only for the case of @samp{#include "@var{file}"};
8887 they are not searched for @samp{#include <@var{file}>}.
8889 If additional directories are specified with @option{-I} options after
8890 the @option{-I-}, these directories are searched for all @samp{#include}
8891 directives. (Ordinarily @emph{all} @option{-I} directories are used
8894 In addition, the @option{-I-} option inhibits the use of the current
8895 directory (where the current input file came from) as the first search
8896 directory for @samp{#include "@var{file}"}. There is no way to
8897 override this effect of @option{-I-}. With @option{-I.} you can specify
8898 searching the directory which was current when the compiler was
8899 invoked. That is not exactly the same as what the preprocessor does
8900 by default, but it is often satisfactory.
8902 @option{-I-} does not inhibit the use of the standard system directories
8903 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8910 @section Specifying subprocesses and the switches to pass to them
8913 @command{gcc} is a driver program. It performs its job by invoking a
8914 sequence of other programs to do the work of compiling, assembling and
8915 linking. GCC interprets its command-line parameters and uses these to
8916 deduce which programs it should invoke, and which command-line options
8917 it ought to place on their command lines. This behavior is controlled
8918 by @dfn{spec strings}. In most cases there is one spec string for each
8919 program that GCC can invoke, but a few programs have multiple spec
8920 strings to control their behavior. The spec strings built into GCC can
8921 be overridden by using the @option{-specs=} command-line switch to specify
8924 @dfn{Spec files} are plaintext files that are used to construct spec
8925 strings. They consist of a sequence of directives separated by blank
8926 lines. The type of directive is determined by the first non-whitespace
8927 character on the line and it can be one of the following:
8930 @item %@var{command}
8931 Issues a @var{command} to the spec file processor. The commands that can
8935 @item %include <@var{file}>
8937 Search for @var{file} and insert its text at the current point in the
8940 @item %include_noerr <@var{file}>
8941 @cindex %include_noerr
8942 Just like @samp{%include}, but do not generate an error message if the include
8943 file cannot be found.
8945 @item %rename @var{old_name} @var{new_name}
8947 Rename the spec string @var{old_name} to @var{new_name}.
8951 @item *[@var{spec_name}]:
8952 This tells the compiler to create, override or delete the named spec
8953 string. All lines after this directive up to the next directive or
8954 blank line are considered to be the text for the spec string. If this
8955 results in an empty string then the spec will be deleted. (Or, if the
8956 spec did not exist, then nothing will happened.) Otherwise, if the spec
8957 does not currently exist a new spec will be created. If the spec does
8958 exist then its contents will be overridden by the text of this
8959 directive, unless the first character of that text is the @samp{+}
8960 character, in which case the text will be appended to the spec.
8962 @item [@var{suffix}]:
8963 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
8964 and up to the next directive or blank line are considered to make up the
8965 spec string for the indicated suffix. When the compiler encounters an
8966 input file with the named suffix, it will processes the spec string in
8967 order to work out how to compile that file. For example:
8974 This says that any input file whose name ends in @samp{.ZZ} should be
8975 passed to the program @samp{z-compile}, which should be invoked with the
8976 command-line switch @option{-input} and with the result of performing the
8977 @samp{%i} substitution. (See below.)
8979 As an alternative to providing a spec string, the text that follows a
8980 suffix directive can be one of the following:
8983 @item @@@var{language}
8984 This says that the suffix is an alias for a known @var{language}. This is
8985 similar to using the @option{-x} command-line switch to GCC to specify a
8986 language explicitly. For example:
8993 Says that .ZZ files are, in fact, C++ source files.
8996 This causes an error messages saying:
8999 @var{name} compiler not installed on this system.
9003 GCC already has an extensive list of suffixes built into it.
9004 This directive will add an entry to the end of the list of suffixes, but
9005 since the list is searched from the end backwards, it is effectively
9006 possible to override earlier entries using this technique.
9010 GCC has the following spec strings built into it. Spec files can
9011 override these strings or create their own. Note that individual
9012 targets can also add their own spec strings to this list.
9015 asm Options to pass to the assembler
9016 asm_final Options to pass to the assembler post-processor
9017 cpp Options to pass to the C preprocessor
9018 cc1 Options to pass to the C compiler
9019 cc1plus Options to pass to the C++ compiler
9020 endfile Object files to include at the end of the link
9021 link Options to pass to the linker
9022 lib Libraries to include on the command line to the linker
9023 libgcc Decides which GCC support library to pass to the linker
9024 linker Sets the name of the linker
9025 predefines Defines to be passed to the C preprocessor
9026 signed_char Defines to pass to CPP to say whether @code{char} is signed
9028 startfile Object files to include at the start of the link
9031 Here is a small example of a spec file:
9037 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9040 This example renames the spec called @samp{lib} to @samp{old_lib} and
9041 then overrides the previous definition of @samp{lib} with a new one.
9042 The new definition adds in some extra command-line options before
9043 including the text of the old definition.
9045 @dfn{Spec strings} are a list of command-line options to be passed to their
9046 corresponding program. In addition, the spec strings can contain
9047 @samp{%}-prefixed sequences to substitute variable text or to
9048 conditionally insert text into the command line. Using these constructs
9049 it is possible to generate quite complex command lines.
9051 Here is a table of all defined @samp{%}-sequences for spec
9052 strings. Note that spaces are not generated automatically around the
9053 results of expanding these sequences. Therefore you can concatenate them
9054 together or combine them with constant text in a single argument.
9058 Substitute one @samp{%} into the program name or argument.
9061 Substitute the name of the input file being processed.
9064 Substitute the basename of the input file being processed.
9065 This is the substring up to (and not including) the last period
9066 and not including the directory.
9069 This is the same as @samp{%b}, but include the file suffix (text after
9073 Marks the argument containing or following the @samp{%d} as a
9074 temporary file name, so that that file will be deleted if GCC exits
9075 successfully. Unlike @samp{%g}, this contributes no text to the
9078 @item %g@var{suffix}
9079 Substitute a file name that has suffix @var{suffix} and is chosen
9080 once per compilation, and mark the argument in the same way as
9081 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9082 name is now chosen in a way that is hard to predict even when previously
9083 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9084 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9085 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9086 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9087 was simply substituted with a file name chosen once per compilation,
9088 without regard to any appended suffix (which was therefore treated
9089 just like ordinary text), making such attacks more likely to succeed.
9091 @item %u@var{suffix}
9092 Like @samp{%g}, but generates a new temporary file name even if
9093 @samp{%u@var{suffix}} was already seen.
9095 @item %U@var{suffix}
9096 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9097 new one if there is no such last file name. In the absence of any
9098 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9099 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9100 would involve the generation of two distinct file names, one
9101 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9102 simply substituted with a file name chosen for the previous @samp{%u},
9103 without regard to any appended suffix.
9105 @item %j@var{suffix}
9106 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9107 writable, and if save-temps is off; otherwise, substitute the name
9108 of a temporary file, just like @samp{%u}. This temporary file is not
9109 meant for communication between processes, but rather as a junk
9112 @item %|@var{suffix}
9113 @itemx %m@var{suffix}
9114 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9115 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9116 all. These are the two most common ways to instruct a program that it
9117 should read from standard input or write to standard output. If you
9118 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9119 construct: see for example @file{f/lang-specs.h}.
9121 @item %.@var{SUFFIX}
9122 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9123 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9124 terminated by the next space or %.
9127 Marks the argument containing or following the @samp{%w} as the
9128 designated output file of this compilation. This puts the argument
9129 into the sequence of arguments that @samp{%o} will substitute later.
9132 Substitutes the names of all the output files, with spaces
9133 automatically placed around them. You should write spaces
9134 around the @samp{%o} as well or the results are undefined.
9135 @samp{%o} is for use in the specs for running the linker.
9136 Input files whose names have no recognized suffix are not compiled
9137 at all, but they are included among the output files, so they will
9141 Substitutes the suffix for object files. Note that this is
9142 handled specially when it immediately follows @samp{%g, %u, or %U},
9143 because of the need for those to form complete file names. The
9144 handling is such that @samp{%O} is treated exactly as if it had already
9145 been substituted, except that @samp{%g, %u, and %U} do not currently
9146 support additional @var{suffix} characters following @samp{%O} as they would
9147 following, for example, @samp{.o}.
9150 Substitutes the standard macro predefinitions for the
9151 current target machine. Use this when running @code{cpp}.
9154 Like @samp{%p}, but puts @samp{__} before and after the name of each
9155 predefined macro, except for macros that start with @samp{__} or with
9156 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9160 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9161 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9162 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9163 and @option{-imultilib} as necessary.
9166 Current argument is the name of a library or startup file of some sort.
9167 Search for that file in a standard list of directories and substitute
9168 the full name found. The current working directory is included in the
9169 list of directories scanned.
9172 Current argument is the name of a linker script. Search for that file
9173 in the current list of directories to scan for libraries. If the file
9174 is located insert a @option{--script} option into the command line
9175 followed by the full path name found. If the file is not found then
9176 generate an error message. Note: the current working directory is not
9180 Print @var{str} as an error message. @var{str} is terminated by a newline.
9181 Use this when inconsistent options are detected.
9184 Substitute the contents of spec string @var{name} at this point.
9187 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9189 @item %x@{@var{option}@}
9190 Accumulate an option for @samp{%X}.
9193 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9197 Output the accumulated assembler options specified by @option{-Wa}.
9200 Output the accumulated preprocessor options specified by @option{-Wp}.
9203 Process the @code{asm} spec. This is used to compute the
9204 switches to be passed to the assembler.
9207 Process the @code{asm_final} spec. This is a spec string for
9208 passing switches to an assembler post-processor, if such a program is
9212 Process the @code{link} spec. This is the spec for computing the
9213 command line passed to the linker. Typically it will make use of the
9214 @samp{%L %G %S %D and %E} sequences.
9217 Dump out a @option{-L} option for each directory that GCC believes might
9218 contain startup files. If the target supports multilibs then the
9219 current multilib directory will be prepended to each of these paths.
9222 Process the @code{lib} spec. This is a spec string for deciding which
9223 libraries should be included on the command line to the linker.
9226 Process the @code{libgcc} spec. This is a spec string for deciding
9227 which GCC support library should be included on the command line to the linker.
9230 Process the @code{startfile} spec. This is a spec for deciding which
9231 object files should be the first ones passed to the linker. Typically
9232 this might be a file named @file{crt0.o}.
9235 Process the @code{endfile} spec. This is a spec string that specifies
9236 the last object files that will be passed to the linker.
9239 Process the @code{cpp} spec. This is used to construct the arguments
9240 to be passed to the C preprocessor.
9243 Process the @code{cc1} spec. This is used to construct the options to be
9244 passed to the actual C compiler (@samp{cc1}).
9247 Process the @code{cc1plus} spec. This is used to construct the options to be
9248 passed to the actual C++ compiler (@samp{cc1plus}).
9251 Substitute the variable part of a matched option. See below.
9252 Note that each comma in the substituted string is replaced by
9256 Remove all occurrences of @code{-S} from the command line. Note---this
9257 command is position dependent. @samp{%} commands in the spec string
9258 before this one will see @code{-S}, @samp{%} commands in the spec string
9259 after this one will not.
9261 @item %:@var{function}(@var{args})
9262 Call the named function @var{function}, passing it @var{args}.
9263 @var{args} is first processed as a nested spec string, then split
9264 into an argument vector in the usual fashion. The function returns
9265 a string which is processed as if it had appeared literally as part
9266 of the current spec.
9268 The following built-in spec functions are provided:
9272 The @code{getenv} spec function takes two arguments: an environment
9273 variable name and a string. If the environment variable is not
9274 defined, a fatal error is issued. Otherwise, the return value is the
9275 value of the environment variable concatenated with the string. For
9276 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9279 %:getenv(TOPDIR /include)
9282 expands to @file{/path/to/top/include}.
9284 @item @code{if-exists}
9285 The @code{if-exists} spec function takes one argument, an absolute
9286 pathname to a file. If the file exists, @code{if-exists} returns the
9287 pathname. Here is a small example of its usage:
9291 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9294 @item @code{if-exists-else}
9295 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9296 spec function, except that it takes two arguments. The first argument is
9297 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9298 returns the pathname. If it does not exist, it returns the second argument.
9299 This way, @code{if-exists-else} can be used to select one file or another,
9300 based on the existence of the first. Here is a small example of its usage:
9304 crt0%O%s %:if-exists(crti%O%s) \
9305 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9308 @item @code{replace-outfile}
9309 The @code{replace-outfile} spec function takes two arguments. It looks for the
9310 first argument in the outfiles array and replaces it with the second argument. Here
9311 is a small example of its usage:
9314 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9317 @item @code{print-asm-header}
9318 The @code{print-asm-header} function takes no arguments and simply
9319 prints a banner like:
9325 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9328 It is used to separate compiler options from assembler options
9329 in the @option{--target-help} output.
9333 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9334 If that switch was not specified, this substitutes nothing. Note that
9335 the leading dash is omitted when specifying this option, and it is
9336 automatically inserted if the substitution is performed. Thus the spec
9337 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9338 and would output the command line option @option{-foo}.
9340 @item %W@{@code{S}@}
9341 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9344 @item %@{@code{S}*@}
9345 Substitutes all the switches specified to GCC whose names start
9346 with @code{-S}, but which also take an argument. This is used for
9347 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9348 GCC considers @option{-o foo} as being
9349 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9350 text, including the space. Thus two arguments would be generated.
9352 @item %@{@code{S}*&@code{T}*@}
9353 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9354 (the order of @code{S} and @code{T} in the spec is not significant).
9355 There can be any number of ampersand-separated variables; for each the
9356 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9358 @item %@{@code{S}:@code{X}@}
9359 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9361 @item %@{!@code{S}:@code{X}@}
9362 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9364 @item %@{@code{S}*:@code{X}@}
9365 Substitutes @code{X} if one or more switches whose names start with
9366 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9367 once, no matter how many such switches appeared. However, if @code{%*}
9368 appears somewhere in @code{X}, then @code{X} will be substituted once
9369 for each matching switch, with the @code{%*} replaced by the part of
9370 that switch that matched the @code{*}.
9372 @item %@{.@code{S}:@code{X}@}
9373 Substitutes @code{X}, if processing a file with suffix @code{S}.
9375 @item %@{!.@code{S}:@code{X}@}
9376 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9378 @item %@{,@code{S}:@code{X}@}
9379 Substitutes @code{X}, if processing a file for language @code{S}.
9381 @item %@{!,@code{S}:@code{X}@}
9382 Substitutes @code{X}, if not processing a file for language @code{S}.
9384 @item %@{@code{S}|@code{P}:@code{X}@}
9385 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9386 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9387 @code{*} sequences as well, although they have a stronger binding than
9388 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9389 alternatives must be starred, and only the first matching alternative
9392 For example, a spec string like this:
9395 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9398 will output the following command-line options from the following input
9399 command-line options:
9404 -d fred.c -foo -baz -boggle
9405 -d jim.d -bar -baz -boggle
9408 @item %@{S:X; T:Y; :D@}
9410 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9411 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9412 be as many clauses as you need. This may be combined with @code{.},
9413 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9418 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9419 construct may contain other nested @samp{%} constructs or spaces, or
9420 even newlines. They are processed as usual, as described above.
9421 Trailing white space in @code{X} is ignored. White space may also
9422 appear anywhere on the left side of the colon in these constructs,
9423 except between @code{.} or @code{*} and the corresponding word.
9425 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9426 handled specifically in these constructs. If another value of
9427 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9428 @option{-W} switch is found later in the command line, the earlier
9429 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9430 just one letter, which passes all matching options.
9432 The character @samp{|} at the beginning of the predicate text is used to
9433 indicate that a command should be piped to the following command, but
9434 only if @option{-pipe} is specified.
9436 It is built into GCC which switches take arguments and which do not.
9437 (You might think it would be useful to generalize this to allow each
9438 compiler's spec to say which switches take arguments. But this cannot
9439 be done in a consistent fashion. GCC cannot even decide which input
9440 files have been specified without knowing which switches take arguments,
9441 and it must know which input files to compile in order to tell which
9444 GCC also knows implicitly that arguments starting in @option{-l} are to be
9445 treated as compiler output files, and passed to the linker in their
9446 proper position among the other output files.
9448 @c man begin OPTIONS
9450 @node Target Options
9451 @section Specifying Target Machine and Compiler Version
9452 @cindex target options
9453 @cindex cross compiling
9454 @cindex specifying machine version
9455 @cindex specifying compiler version and target machine
9456 @cindex compiler version, specifying
9457 @cindex target machine, specifying
9459 The usual way to run GCC is to run the executable called @file{gcc}, or
9460 @file{<machine>-gcc} when cross-compiling, or
9461 @file{<machine>-gcc-<version>} to run a version other than the one that
9462 was installed last. Sometimes this is inconvenient, so GCC provides
9463 options that will switch to another cross-compiler or version.
9466 @item -b @var{machine}
9468 The argument @var{machine} specifies the target machine for compilation.
9470 The value to use for @var{machine} is the same as was specified as the
9471 machine type when configuring GCC as a cross-compiler. For
9472 example, if a cross-compiler was configured with @samp{configure
9473 arm-elf}, meaning to compile for an arm processor with elf binaries,
9474 then you would specify @option{-b arm-elf} to run that cross compiler.
9475 Because there are other options beginning with @option{-b}, the
9476 configuration must contain a hyphen, or @option{-b} alone should be one
9477 argument followed by the configuration in the next argument.
9479 @item -V @var{version}
9481 The argument @var{version} specifies which version of GCC to run.
9482 This is useful when multiple versions are installed. For example,
9483 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
9486 The @option{-V} and @option{-b} options work by running the
9487 @file{<machine>-gcc-<version>} executable, so there's no real reason to
9488 use them if you can just run that directly.
9490 @node Submodel Options
9491 @section Hardware Models and Configurations
9492 @cindex submodel options
9493 @cindex specifying hardware config
9494 @cindex hardware models and configurations, specifying
9495 @cindex machine dependent options
9497 Earlier we discussed the standard option @option{-b} which chooses among
9498 different installed compilers for completely different target
9499 machines, such as VAX vs.@: 68000 vs.@: 80386.
9501 In addition, each of these target machine types can have its own
9502 special options, starting with @samp{-m}, to choose among various
9503 hardware models or configurations---for example, 68010 vs 68020,
9504 floating coprocessor or none. A single installed version of the
9505 compiler can compile for any model or configuration, according to the
9508 Some configurations of the compiler also support additional special
9509 options, usually for compatibility with other compilers on the same
9512 @c This list is ordered alphanumerically by subsection name.
9513 @c It should be the same order and spelling as these options are listed
9514 @c in Machine Dependent Options
9520 * Blackfin Options::
9524 * DEC Alpha Options::
9525 * DEC Alpha/VMS Options::
9528 * GNU/Linux Options::
9531 * i386 and x86-64 Options::
9532 * i386 and x86-64 Windows Options::
9534 * IA-64/VMS Options::
9545 * picoChip Options::
9547 * RS/6000 and PowerPC Options::
9549 * S/390 and zSeries Options::
9554 * System V Options::
9559 * Xstormy16 Options::
9565 @subsection ARC Options
9568 These options are defined for ARC implementations:
9573 Compile code for little endian mode. This is the default.
9577 Compile code for big endian mode.
9580 @opindex mmangle-cpu
9581 Prepend the name of the cpu to all public symbol names.
9582 In multiple-processor systems, there are many ARC variants with different
9583 instruction and register set characteristics. This flag prevents code
9584 compiled for one cpu to be linked with code compiled for another.
9585 No facility exists for handling variants that are ``almost identical''.
9586 This is an all or nothing option.
9588 @item -mcpu=@var{cpu}
9590 Compile code for ARC variant @var{cpu}.
9591 Which variants are supported depend on the configuration.
9592 All variants support @option{-mcpu=base}, this is the default.
9594 @item -mtext=@var{text-section}
9595 @itemx -mdata=@var{data-section}
9596 @itemx -mrodata=@var{readonly-data-section}
9600 Put functions, data, and readonly data in @var{text-section},
9601 @var{data-section}, and @var{readonly-data-section} respectively
9602 by default. This can be overridden with the @code{section} attribute.
9603 @xref{Variable Attributes}.
9605 @item -mfix-cortex-m3-ldrd
9606 @opindex mfix-cortex-m3-ldrd
9607 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
9608 with overlapping destination and base registers are used. This option avoids
9609 generating these instructions. This option is enabled by default when
9610 @option{-mcpu=cortex-m3} is specified.
9615 @subsection ARM Options
9618 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9622 @item -mabi=@var{name}
9624 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9625 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9628 @opindex mapcs-frame
9629 Generate a stack frame that is compliant with the ARM Procedure Call
9630 Standard for all functions, even if this is not strictly necessary for
9631 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9632 with this option will cause the stack frames not to be generated for
9633 leaf functions. The default is @option{-mno-apcs-frame}.
9637 This is a synonym for @option{-mapcs-frame}.
9640 @c not currently implemented
9641 @item -mapcs-stack-check
9642 @opindex mapcs-stack-check
9643 Generate code to check the amount of stack space available upon entry to
9644 every function (that actually uses some stack space). If there is
9645 insufficient space available then either the function
9646 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9647 called, depending upon the amount of stack space required. The run time
9648 system is required to provide these functions. The default is
9649 @option{-mno-apcs-stack-check}, since this produces smaller code.
9651 @c not currently implemented
9653 @opindex mapcs-float
9654 Pass floating point arguments using the float point registers. This is
9655 one of the variants of the APCS@. This option is recommended if the
9656 target hardware has a floating point unit or if a lot of floating point
9657 arithmetic is going to be performed by the code. The default is
9658 @option{-mno-apcs-float}, since integer only code is slightly increased in
9659 size if @option{-mapcs-float} is used.
9661 @c not currently implemented
9662 @item -mapcs-reentrant
9663 @opindex mapcs-reentrant
9664 Generate reentrant, position independent code. The default is
9665 @option{-mno-apcs-reentrant}.
9668 @item -mthumb-interwork
9669 @opindex mthumb-interwork
9670 Generate code which supports calling between the ARM and Thumb
9671 instruction sets. Without this option the two instruction sets cannot
9672 be reliably used inside one program. The default is
9673 @option{-mno-thumb-interwork}, since slightly larger code is generated
9674 when @option{-mthumb-interwork} is specified.
9676 @item -mno-sched-prolog
9677 @opindex mno-sched-prolog
9678 Prevent the reordering of instructions in the function prolog, or the
9679 merging of those instruction with the instructions in the function's
9680 body. This means that all functions will start with a recognizable set
9681 of instructions (or in fact one of a choice from a small set of
9682 different function prologues), and this information can be used to
9683 locate the start if functions inside an executable piece of code. The
9684 default is @option{-msched-prolog}.
9686 @item -mfloat-abi=@var{name}
9688 Specifies which floating-point ABI to use. Permissible values
9689 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9691 Specifying @samp{soft} causes GCC to generate output containing
9692 library calls for floating-point operations.
9693 @samp{softfp} allows the generation of code using hardware floating-point
9694 instructions, but still uses the soft-float calling conventions.
9695 @samp{hard} allows generation of floating-point instructions
9696 and uses FPU-specific calling conventions.
9698 The default depends on the specific target configuration. Note that
9699 the hard-float and soft-float ABIs are not link-compatible; you must
9700 compile your entire program with the same ABI, and link with a
9701 compatible set of libraries.
9704 @opindex mhard-float
9705 Equivalent to @option{-mfloat-abi=hard}.
9708 @opindex msoft-float
9709 Equivalent to @option{-mfloat-abi=soft}.
9711 @item -mlittle-endian
9712 @opindex mlittle-endian
9713 Generate code for a processor running in little-endian mode. This is
9714 the default for all standard configurations.
9717 @opindex mbig-endian
9718 Generate code for a processor running in big-endian mode; the default is
9719 to compile code for a little-endian processor.
9721 @item -mwords-little-endian
9722 @opindex mwords-little-endian
9723 This option only applies when generating code for big-endian processors.
9724 Generate code for a little-endian word order but a big-endian byte
9725 order. That is, a byte order of the form @samp{32107654}. Note: this
9726 option should only be used if you require compatibility with code for
9727 big-endian ARM processors generated by versions of the compiler prior to
9730 @item -mcpu=@var{name}
9732 This specifies the name of the target ARM processor. GCC uses this name
9733 to determine what kind of instructions it can emit when generating
9734 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9735 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9736 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9737 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9738 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9740 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9741 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9742 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9743 @samp{strongarm1110},
9744 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9745 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9746 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9747 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9748 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9749 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9750 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9751 @samp{cortex-a8}, @samp{cortex-a9},
9752 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9755 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9757 @item -mtune=@var{name}
9759 This option is very similar to the @option{-mcpu=} option, except that
9760 instead of specifying the actual target processor type, and hence
9761 restricting which instructions can be used, it specifies that GCC should
9762 tune the performance of the code as if the target were of the type
9763 specified in this option, but still choosing the instructions that it
9764 will generate based on the cpu specified by a @option{-mcpu=} option.
9765 For some ARM implementations better performance can be obtained by using
9768 @item -march=@var{name}
9770 This specifies the name of the target ARM architecture. GCC uses this
9771 name to determine what kind of instructions it can emit when generating
9772 assembly code. This option can be used in conjunction with or instead
9773 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9774 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9775 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9776 @samp{armv6}, @samp{armv6j},
9777 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9778 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9779 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9781 @item -mfpu=@var{name}
9782 @itemx -mfpe=@var{number}
9783 @itemx -mfp=@var{number}
9787 This specifies what floating point hardware (or hardware emulation) is
9788 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9789 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16},
9790 @samp{neon}, and @samp{neon-fp16}. @option{-mfp} and @option{-mfpe}
9791 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9792 with older versions of GCC@.
9794 If @option{-msoft-float} is specified this specifies the format of
9795 floating point values.
9797 @item -mfp16-format=@var{name}
9798 @opindex mfp16-format
9799 Specify the format of the @code{__fp16} half-precision floating-point type.
9800 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
9801 the default is @samp{none}, in which case the @code{__fp16} type is not
9802 defined. @xref{Half-Precision}, for more information.
9804 @item -mstructure-size-boundary=@var{n}
9805 @opindex mstructure-size-boundary
9806 The size of all structures and unions will be rounded up to a multiple
9807 of the number of bits set by this option. Permissible values are 8, 32
9808 and 64. The default value varies for different toolchains. For the COFF
9809 targeted toolchain the default value is 8. A value of 64 is only allowed
9810 if the underlying ABI supports it.
9812 Specifying the larger number can produce faster, more efficient code, but
9813 can also increase the size of the program. Different values are potentially
9814 incompatible. Code compiled with one value cannot necessarily expect to
9815 work with code or libraries compiled with another value, if they exchange
9816 information using structures or unions.
9818 @item -mabort-on-noreturn
9819 @opindex mabort-on-noreturn
9820 Generate a call to the function @code{abort} at the end of a
9821 @code{noreturn} function. It will be executed if the function tries to
9825 @itemx -mno-long-calls
9826 @opindex mlong-calls
9827 @opindex mno-long-calls
9828 Tells the compiler to perform function calls by first loading the
9829 address of the function into a register and then performing a subroutine
9830 call on this register. This switch is needed if the target function
9831 will lie outside of the 64 megabyte addressing range of the offset based
9832 version of subroutine call instruction.
9834 Even if this switch is enabled, not all function calls will be turned
9835 into long calls. The heuristic is that static functions, functions
9836 which have the @samp{short-call} attribute, functions that are inside
9837 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9838 definitions have already been compiled within the current compilation
9839 unit, will not be turned into long calls. The exception to this rule is
9840 that weak function definitions, functions with the @samp{long-call}
9841 attribute or the @samp{section} attribute, and functions that are within
9842 the scope of a @samp{#pragma long_calls} directive, will always be
9843 turned into long calls.
9845 This feature is not enabled by default. Specifying
9846 @option{-mno-long-calls} will restore the default behavior, as will
9847 placing the function calls within the scope of a @samp{#pragma
9848 long_calls_off} directive. Note these switches have no effect on how
9849 the compiler generates code to handle function calls via function
9852 @item -msingle-pic-base
9853 @opindex msingle-pic-base
9854 Treat the register used for PIC addressing as read-only, rather than
9855 loading it in the prologue for each function. The run-time system is
9856 responsible for initializing this register with an appropriate value
9857 before execution begins.
9859 @item -mpic-register=@var{reg}
9860 @opindex mpic-register
9861 Specify the register to be used for PIC addressing. The default is R10
9862 unless stack-checking is enabled, when R9 is used.
9864 @item -mcirrus-fix-invalid-insns
9865 @opindex mcirrus-fix-invalid-insns
9866 @opindex mno-cirrus-fix-invalid-insns
9867 Insert NOPs into the instruction stream to in order to work around
9868 problems with invalid Maverick instruction combinations. This option
9869 is only valid if the @option{-mcpu=ep9312} option has been used to
9870 enable generation of instructions for the Cirrus Maverick floating
9871 point co-processor. This option is not enabled by default, since the
9872 problem is only present in older Maverick implementations. The default
9873 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9876 @item -mpoke-function-name
9877 @opindex mpoke-function-name
9878 Write the name of each function into the text section, directly
9879 preceding the function prologue. The generated code is similar to this:
9883 .ascii "arm_poke_function_name", 0
9886 .word 0xff000000 + (t1 - t0)
9887 arm_poke_function_name
9889 stmfd sp!, @{fp, ip, lr, pc@}
9893 When performing a stack backtrace, code can inspect the value of
9894 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9895 location @code{pc - 12} and the top 8 bits are set, then we know that
9896 there is a function name embedded immediately preceding this location
9897 and has length @code{((pc[-3]) & 0xff000000)}.
9901 Generate code for the Thumb instruction set. The default is to
9902 use the 32-bit ARM instruction set.
9903 This option automatically enables either 16-bit Thumb-1 or
9904 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9905 and @option{-march=@var{name}} options. This option is not passed to the
9906 assembler. If you want to force assembler files to be interpreted as Thumb code,
9907 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
9908 option directly to the assembler by prefixing it with @option{-Wa}.
9911 @opindex mtpcs-frame
9912 Generate a stack frame that is compliant with the Thumb Procedure Call
9913 Standard for all non-leaf functions. (A leaf function is one that does
9914 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9916 @item -mtpcs-leaf-frame
9917 @opindex mtpcs-leaf-frame
9918 Generate a stack frame that is compliant with the Thumb Procedure Call
9919 Standard for all leaf functions. (A leaf function is one that does
9920 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9922 @item -mcallee-super-interworking
9923 @opindex mcallee-super-interworking
9924 Gives all externally visible functions in the file being compiled an ARM
9925 instruction set header which switches to Thumb mode before executing the
9926 rest of the function. This allows these functions to be called from
9927 non-interworking code. This option is not valid in AAPCS configurations
9928 because interworking is enabled by default.
9930 @item -mcaller-super-interworking
9931 @opindex mcaller-super-interworking
9932 Allows calls via function pointers (including virtual functions) to
9933 execute correctly regardless of whether the target code has been
9934 compiled for interworking or not. There is a small overhead in the cost
9935 of executing a function pointer if this option is enabled. This option
9936 is not valid in AAPCS configurations because interworking is enabled
9939 @item -mtp=@var{name}
9941 Specify the access model for the thread local storage pointer. The valid
9942 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9943 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9944 (supported in the arm6k architecture), and @option{auto}, which uses the
9945 best available method for the selected processor. The default setting is
9948 @item -mword-relocations
9949 @opindex mword-relocations
9950 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9951 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9952 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9958 @subsection AVR Options
9961 These options are defined for AVR implementations:
9964 @item -mmcu=@var{mcu}
9966 Specify ATMEL AVR instruction set or MCU type.
9968 Instruction set avr1 is for the minimal AVR core, not supported by the C
9969 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9970 attiny11, attiny12, attiny15, attiny28).
9972 Instruction set avr2 (default) is for the classic AVR core with up to
9973 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9974 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9975 at90c8534, at90s8535).
9977 Instruction set avr3 is for the classic AVR core with up to 128K program
9978 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9980 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9981 memory space (MCU types: atmega8, atmega83, atmega85).
9983 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9984 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9985 atmega64, atmega128, at43usb355, at94k).
9987 @item -mno-interrupts
9988 @opindex mno-interrupts
9989 Generated code is not compatible with hardware interrupts.
9990 Code size will be smaller.
9992 @item -mcall-prologues
9993 @opindex mcall-prologues
9994 Functions prologues/epilogues expanded as call to appropriate
9995 subroutines. Code size will be smaller.
9998 @opindex mtiny-stack
9999 Change only the low 8 bits of the stack pointer.
10003 Assume int to be 8 bit integer. This affects the sizes of all types: A
10004 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10005 and long long will be 4 bytes. Please note that this option does not
10006 comply to the C standards, but it will provide you with smaller code
10010 @node Blackfin Options
10011 @subsection Blackfin Options
10012 @cindex Blackfin Options
10015 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10017 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10018 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10019 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10020 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10021 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10022 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10023 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10025 The optional @var{sirevision} specifies the silicon revision of the target
10026 Blackfin processor. Any workarounds available for the targeted silicon revision
10027 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10028 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10029 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10030 hexadecimal digits representing the major and minor numbers in the silicon
10031 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10032 is not defined. If @var{sirevision} is @samp{any}, the
10033 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10034 If this optional @var{sirevision} is not used, GCC assumes the latest known
10035 silicon revision of the targeted Blackfin processor.
10037 Support for @samp{bf561} is incomplete. For @samp{bf561},
10038 Only the processor macro is defined.
10039 Without this option, @samp{bf532} is used as the processor by default.
10040 The corresponding predefined processor macros for @var{cpu} is to
10041 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10042 provided by libgloss to be linked in if @option{-msim} is not given.
10046 Specifies that the program will be run on the simulator. This causes
10047 the simulator BSP provided by libgloss to be linked in. This option
10048 has effect only for @samp{bfin-elf} toolchain.
10049 Certain other options, such as @option{-mid-shared-library} and
10050 @option{-mfdpic}, imply @option{-msim}.
10052 @item -momit-leaf-frame-pointer
10053 @opindex momit-leaf-frame-pointer
10054 Don't keep the frame pointer in a register for leaf functions. This
10055 avoids the instructions to save, set up and restore frame pointers and
10056 makes an extra register available in leaf functions. The option
10057 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10058 which might make debugging harder.
10060 @item -mspecld-anomaly
10061 @opindex mspecld-anomaly
10062 When enabled, the compiler will ensure that the generated code does not
10063 contain speculative loads after jump instructions. If this option is used,
10064 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10066 @item -mno-specld-anomaly
10067 @opindex mno-specld-anomaly
10068 Don't generate extra code to prevent speculative loads from occurring.
10070 @item -mcsync-anomaly
10071 @opindex mcsync-anomaly
10072 When enabled, the compiler will ensure that the generated code does not
10073 contain CSYNC or SSYNC instructions too soon after conditional branches.
10074 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10076 @item -mno-csync-anomaly
10077 @opindex mno-csync-anomaly
10078 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10079 occurring too soon after a conditional branch.
10083 When enabled, the compiler is free to take advantage of the knowledge that
10084 the entire program fits into the low 64k of memory.
10087 @opindex mno-low-64k
10088 Assume that the program is arbitrarily large. This is the default.
10090 @item -mstack-check-l1
10091 @opindex mstack-check-l1
10092 Do stack checking using information placed into L1 scratchpad memory by the
10095 @item -mid-shared-library
10096 @opindex mid-shared-library
10097 Generate code that supports shared libraries via the library ID method.
10098 This allows for execute in place and shared libraries in an environment
10099 without virtual memory management. This option implies @option{-fPIC}.
10100 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10102 @item -mno-id-shared-library
10103 @opindex mno-id-shared-library
10104 Generate code that doesn't assume ID based shared libraries are being used.
10105 This is the default.
10107 @item -mleaf-id-shared-library
10108 @opindex mleaf-id-shared-library
10109 Generate code that supports shared libraries via the library ID method,
10110 but assumes that this library or executable won't link against any other
10111 ID shared libraries. That allows the compiler to use faster code for jumps
10114 @item -mno-leaf-id-shared-library
10115 @opindex mno-leaf-id-shared-library
10116 Do not assume that the code being compiled won't link against any ID shared
10117 libraries. Slower code will be generated for jump and call insns.
10119 @item -mshared-library-id=n
10120 @opindex mshared-library-id
10121 Specified the identification number of the ID based shared library being
10122 compiled. Specifying a value of 0 will generate more compact code, specifying
10123 other values will force the allocation of that number to the current
10124 library but is no more space or time efficient than omitting this option.
10128 Generate code that allows the data segment to be located in a different
10129 area of memory from the text segment. This allows for execute in place in
10130 an environment without virtual memory management by eliminating relocations
10131 against the text section.
10133 @item -mno-sep-data
10134 @opindex mno-sep-data
10135 Generate code that assumes that the data segment follows the text segment.
10136 This is the default.
10139 @itemx -mno-long-calls
10140 @opindex mlong-calls
10141 @opindex mno-long-calls
10142 Tells the compiler to perform function calls by first loading the
10143 address of the function into a register and then performing a subroutine
10144 call on this register. This switch is needed if the target function
10145 will lie outside of the 24 bit addressing range of the offset based
10146 version of subroutine call instruction.
10148 This feature is not enabled by default. Specifying
10149 @option{-mno-long-calls} will restore the default behavior. Note these
10150 switches have no effect on how the compiler generates code to handle
10151 function calls via function pointers.
10155 Link with the fast floating-point library. This library relaxes some of
10156 the IEEE floating-point standard's rules for checking inputs against
10157 Not-a-Number (NAN), in the interest of performance.
10160 @opindex minline-plt
10161 Enable inlining of PLT entries in function calls to functions that are
10162 not known to bind locally. It has no effect without @option{-mfdpic}.
10165 @opindex mmulticore
10166 Build standalone application for multicore Blackfin processor. Proper
10167 start files and link scripts will be used to support multicore.
10168 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10169 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10170 @option{-mcorea} or @option{-mcoreb}. If it's used without
10171 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10172 programming model is used. In this model, the main function of Core B
10173 should be named as coreb_main. If it's used with @option{-mcorea} or
10174 @option{-mcoreb}, one application per core programming model is used.
10175 If this option is not used, single core application programming
10180 Build standalone application for Core A of BF561 when using
10181 one application per core programming model. Proper start files
10182 and link scripts will be used to support Core A. This option
10183 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10187 Build standalone application for Core B of BF561 when using
10188 one application per core programming model. Proper start files
10189 and link scripts will be used to support Core B. This option
10190 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10191 should be used instead of main. It must be used with
10192 @option{-mmulticore}.
10196 Build standalone application for SDRAM. Proper start files and
10197 link scripts will be used to put the application into SDRAM.
10198 Loader should initialize SDRAM before loading the application
10199 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10203 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10204 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10205 are enabled; for standalone applications the default is off.
10209 @subsection CRIS Options
10210 @cindex CRIS Options
10212 These options are defined specifically for the CRIS ports.
10215 @item -march=@var{architecture-type}
10216 @itemx -mcpu=@var{architecture-type}
10219 Generate code for the specified architecture. The choices for
10220 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10221 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10222 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10225 @item -mtune=@var{architecture-type}
10227 Tune to @var{architecture-type} everything applicable about the generated
10228 code, except for the ABI and the set of available instructions. The
10229 choices for @var{architecture-type} are the same as for
10230 @option{-march=@var{architecture-type}}.
10232 @item -mmax-stack-frame=@var{n}
10233 @opindex mmax-stack-frame
10234 Warn when the stack frame of a function exceeds @var{n} bytes.
10240 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10241 @option{-march=v3} and @option{-march=v8} respectively.
10243 @item -mmul-bug-workaround
10244 @itemx -mno-mul-bug-workaround
10245 @opindex mmul-bug-workaround
10246 @opindex mno-mul-bug-workaround
10247 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10248 models where it applies. This option is active by default.
10252 Enable CRIS-specific verbose debug-related information in the assembly
10253 code. This option also has the effect to turn off the @samp{#NO_APP}
10254 formatted-code indicator to the assembler at the beginning of the
10259 Do not use condition-code results from previous instruction; always emit
10260 compare and test instructions before use of condition codes.
10262 @item -mno-side-effects
10263 @opindex mno-side-effects
10264 Do not emit instructions with side-effects in addressing modes other than
10267 @item -mstack-align
10268 @itemx -mno-stack-align
10269 @itemx -mdata-align
10270 @itemx -mno-data-align
10271 @itemx -mconst-align
10272 @itemx -mno-const-align
10273 @opindex mstack-align
10274 @opindex mno-stack-align
10275 @opindex mdata-align
10276 @opindex mno-data-align
10277 @opindex mconst-align
10278 @opindex mno-const-align
10279 These options (no-options) arranges (eliminate arrangements) for the
10280 stack-frame, individual data and constants to be aligned for the maximum
10281 single data access size for the chosen CPU model. The default is to
10282 arrange for 32-bit alignment. ABI details such as structure layout are
10283 not affected by these options.
10291 Similar to the stack- data- and const-align options above, these options
10292 arrange for stack-frame, writable data and constants to all be 32-bit,
10293 16-bit or 8-bit aligned. The default is 32-bit alignment.
10295 @item -mno-prologue-epilogue
10296 @itemx -mprologue-epilogue
10297 @opindex mno-prologue-epilogue
10298 @opindex mprologue-epilogue
10299 With @option{-mno-prologue-epilogue}, the normal function prologue and
10300 epilogue that sets up the stack-frame are omitted and no return
10301 instructions or return sequences are generated in the code. Use this
10302 option only together with visual inspection of the compiled code: no
10303 warnings or errors are generated when call-saved registers must be saved,
10304 or storage for local variable needs to be allocated.
10308 @opindex mno-gotplt
10310 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10311 instruction sequences that load addresses for functions from the PLT part
10312 of the GOT rather than (traditional on other architectures) calls to the
10313 PLT@. The default is @option{-mgotplt}.
10317 Legacy no-op option only recognized with the cris-axis-elf and
10318 cris-axis-linux-gnu targets.
10322 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10326 This option, recognized for the cris-axis-elf arranges
10327 to link with input-output functions from a simulator library. Code,
10328 initialized data and zero-initialized data are allocated consecutively.
10332 Like @option{-sim}, but pass linker options to locate initialized data at
10333 0x40000000 and zero-initialized data at 0x80000000.
10337 @subsection CRX Options
10338 @cindex CRX Options
10340 These options are defined specifically for the CRX ports.
10346 Enable the use of multiply-accumulate instructions. Disabled by default.
10349 @opindex mpush-args
10350 Push instructions will be used to pass outgoing arguments when functions
10351 are called. Enabled by default.
10354 @node Darwin Options
10355 @subsection Darwin Options
10356 @cindex Darwin options
10358 These options are defined for all architectures running the Darwin operating
10361 FSF GCC on Darwin does not create ``fat'' object files; it will create
10362 an object file for the single architecture that it was built to
10363 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10364 @option{-arch} options are used; it does so by running the compiler or
10365 linker multiple times and joining the results together with
10368 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10369 @samp{i686}) is determined by the flags that specify the ISA
10370 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10371 @option{-force_cpusubtype_ALL} option can be used to override this.
10373 The Darwin tools vary in their behavior when presented with an ISA
10374 mismatch. The assembler, @file{as}, will only permit instructions to
10375 be used that are valid for the subtype of the file it is generating,
10376 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10377 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10378 and print an error if asked to create a shared library with a less
10379 restrictive subtype than its input files (for instance, trying to put
10380 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10381 for executables, @file{ld}, will quietly give the executable the most
10382 restrictive subtype of any of its input files.
10387 Add the framework directory @var{dir} to the head of the list of
10388 directories to be searched for header files. These directories are
10389 interleaved with those specified by @option{-I} options and are
10390 scanned in a left-to-right order.
10392 A framework directory is a directory with frameworks in it. A
10393 framework is a directory with a @samp{"Headers"} and/or
10394 @samp{"PrivateHeaders"} directory contained directly in it that ends
10395 in @samp{".framework"}. The name of a framework is the name of this
10396 directory excluding the @samp{".framework"}. Headers associated with
10397 the framework are found in one of those two directories, with
10398 @samp{"Headers"} being searched first. A subframework is a framework
10399 directory that is in a framework's @samp{"Frameworks"} directory.
10400 Includes of subframework headers can only appear in a header of a
10401 framework that contains the subframework, or in a sibling subframework
10402 header. Two subframeworks are siblings if they occur in the same
10403 framework. A subframework should not have the same name as a
10404 framework, a warning will be issued if this is violated. Currently a
10405 subframework cannot have subframeworks, in the future, the mechanism
10406 may be extended to support this. The standard frameworks can be found
10407 in @samp{"/System/Library/Frameworks"} and
10408 @samp{"/Library/Frameworks"}. An example include looks like
10409 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10410 the name of the framework and header.h is found in the
10411 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10413 @item -iframework@var{dir}
10414 @opindex iframework
10415 Like @option{-F} except the directory is a treated as a system
10416 directory. The main difference between this @option{-iframework} and
10417 @option{-F} is that with @option{-iframework} the compiler does not
10418 warn about constructs contained within header files found via
10419 @var{dir}. This option is valid only for the C family of languages.
10423 Emit debugging information for symbols that are used. For STABS
10424 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10425 This is by default ON@.
10429 Emit debugging information for all symbols and types.
10431 @item -mmacosx-version-min=@var{version}
10432 The earliest version of MacOS X that this executable will run on
10433 is @var{version}. Typical values of @var{version} include @code{10.1},
10434 @code{10.2}, and @code{10.3.9}.
10436 If the compiler was built to use the system's headers by default,
10437 then the default for this option is the system version on which the
10438 compiler is running, otherwise the default is to make choices which
10439 are compatible with as many systems and code bases as possible.
10443 Enable kernel development mode. The @option{-mkernel} option sets
10444 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10445 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10446 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10447 applicable. This mode also sets @option{-mno-altivec},
10448 @option{-msoft-float}, @option{-fno-builtin} and
10449 @option{-mlong-branch} for PowerPC targets.
10451 @item -mone-byte-bool
10452 @opindex mone-byte-bool
10453 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10454 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10455 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10456 option has no effect on x86.
10458 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10459 to generate code that is not binary compatible with code generated
10460 without that switch. Using this switch may require recompiling all
10461 other modules in a program, including system libraries. Use this
10462 switch to conform to a non-default data model.
10464 @item -mfix-and-continue
10465 @itemx -ffix-and-continue
10466 @itemx -findirect-data
10467 @opindex mfix-and-continue
10468 @opindex ffix-and-continue
10469 @opindex findirect-data
10470 Generate code suitable for fast turn around development. Needed to
10471 enable gdb to dynamically load @code{.o} files into already running
10472 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10473 are provided for backwards compatibility.
10477 Loads all members of static archive libraries.
10478 See man ld(1) for more information.
10480 @item -arch_errors_fatal
10481 @opindex arch_errors_fatal
10482 Cause the errors having to do with files that have the wrong architecture
10485 @item -bind_at_load
10486 @opindex bind_at_load
10487 Causes the output file to be marked such that the dynamic linker will
10488 bind all undefined references when the file is loaded or launched.
10492 Produce a Mach-o bundle format file.
10493 See man ld(1) for more information.
10495 @item -bundle_loader @var{executable}
10496 @opindex bundle_loader
10497 This option specifies the @var{executable} that will be loading the build
10498 output file being linked. See man ld(1) for more information.
10501 @opindex dynamiclib
10502 When passed this option, GCC will produce a dynamic library instead of
10503 an executable when linking, using the Darwin @file{libtool} command.
10505 @item -force_cpusubtype_ALL
10506 @opindex force_cpusubtype_ALL
10507 This causes GCC's output file to have the @var{ALL} subtype, instead of
10508 one controlled by the @option{-mcpu} or @option{-march} option.
10510 @item -allowable_client @var{client_name}
10511 @itemx -client_name
10512 @itemx -compatibility_version
10513 @itemx -current_version
10515 @itemx -dependency-file
10517 @itemx -dylinker_install_name
10519 @itemx -exported_symbols_list
10521 @itemx -flat_namespace
10522 @itemx -force_flat_namespace
10523 @itemx -headerpad_max_install_names
10526 @itemx -install_name
10527 @itemx -keep_private_externs
10528 @itemx -multi_module
10529 @itemx -multiply_defined
10530 @itemx -multiply_defined_unused
10532 @itemx -no_dead_strip_inits_and_terms
10533 @itemx -nofixprebinding
10534 @itemx -nomultidefs
10536 @itemx -noseglinkedit
10537 @itemx -pagezero_size
10539 @itemx -prebind_all_twolevel_modules
10540 @itemx -private_bundle
10541 @itemx -read_only_relocs
10543 @itemx -sectobjectsymbols
10547 @itemx -sectobjectsymbols
10550 @itemx -segs_read_only_addr
10551 @itemx -segs_read_write_addr
10552 @itemx -seg_addr_table
10553 @itemx -seg_addr_table_filename
10554 @itemx -seglinkedit
10556 @itemx -segs_read_only_addr
10557 @itemx -segs_read_write_addr
10558 @itemx -single_module
10560 @itemx -sub_library
10561 @itemx -sub_umbrella
10562 @itemx -twolevel_namespace
10565 @itemx -unexported_symbols_list
10566 @itemx -weak_reference_mismatches
10567 @itemx -whatsloaded
10568 @opindex allowable_client
10569 @opindex client_name
10570 @opindex compatibility_version
10571 @opindex current_version
10572 @opindex dead_strip
10573 @opindex dependency-file
10574 @opindex dylib_file
10575 @opindex dylinker_install_name
10577 @opindex exported_symbols_list
10579 @opindex flat_namespace
10580 @opindex force_flat_namespace
10581 @opindex headerpad_max_install_names
10582 @opindex image_base
10584 @opindex install_name
10585 @opindex keep_private_externs
10586 @opindex multi_module
10587 @opindex multiply_defined
10588 @opindex multiply_defined_unused
10589 @opindex noall_load
10590 @opindex no_dead_strip_inits_and_terms
10591 @opindex nofixprebinding
10592 @opindex nomultidefs
10594 @opindex noseglinkedit
10595 @opindex pagezero_size
10597 @opindex prebind_all_twolevel_modules
10598 @opindex private_bundle
10599 @opindex read_only_relocs
10601 @opindex sectobjectsymbols
10604 @opindex sectcreate
10605 @opindex sectobjectsymbols
10608 @opindex segs_read_only_addr
10609 @opindex segs_read_write_addr
10610 @opindex seg_addr_table
10611 @opindex seg_addr_table_filename
10612 @opindex seglinkedit
10614 @opindex segs_read_only_addr
10615 @opindex segs_read_write_addr
10616 @opindex single_module
10618 @opindex sub_library
10619 @opindex sub_umbrella
10620 @opindex twolevel_namespace
10623 @opindex unexported_symbols_list
10624 @opindex weak_reference_mismatches
10625 @opindex whatsloaded
10626 These options are passed to the Darwin linker. The Darwin linker man page
10627 describes them in detail.
10630 @node DEC Alpha Options
10631 @subsection DEC Alpha Options
10633 These @samp{-m} options are defined for the DEC Alpha implementations:
10636 @item -mno-soft-float
10637 @itemx -msoft-float
10638 @opindex mno-soft-float
10639 @opindex msoft-float
10640 Use (do not use) the hardware floating-point instructions for
10641 floating-point operations. When @option{-msoft-float} is specified,
10642 functions in @file{libgcc.a} will be used to perform floating-point
10643 operations. Unless they are replaced by routines that emulate the
10644 floating-point operations, or compiled in such a way as to call such
10645 emulations routines, these routines will issue floating-point
10646 operations. If you are compiling for an Alpha without floating-point
10647 operations, you must ensure that the library is built so as not to call
10650 Note that Alpha implementations without floating-point operations are
10651 required to have floating-point registers.
10654 @itemx -mno-fp-regs
10656 @opindex mno-fp-regs
10657 Generate code that uses (does not use) the floating-point register set.
10658 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10659 register set is not used, floating point operands are passed in integer
10660 registers as if they were integers and floating-point results are passed
10661 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10662 so any function with a floating-point argument or return value called by code
10663 compiled with @option{-mno-fp-regs} must also be compiled with that
10666 A typical use of this option is building a kernel that does not use,
10667 and hence need not save and restore, any floating-point registers.
10671 The Alpha architecture implements floating-point hardware optimized for
10672 maximum performance. It is mostly compliant with the IEEE floating
10673 point standard. However, for full compliance, software assistance is
10674 required. This option generates code fully IEEE compliant code
10675 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10676 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10677 defined during compilation. The resulting code is less efficient but is
10678 able to correctly support denormalized numbers and exceptional IEEE
10679 values such as not-a-number and plus/minus infinity. Other Alpha
10680 compilers call this option @option{-ieee_with_no_inexact}.
10682 @item -mieee-with-inexact
10683 @opindex mieee-with-inexact
10684 This is like @option{-mieee} except the generated code also maintains
10685 the IEEE @var{inexact-flag}. Turning on this option causes the
10686 generated code to implement fully-compliant IEEE math. In addition to
10687 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10688 macro. On some Alpha implementations the resulting code may execute
10689 significantly slower than the code generated by default. Since there is
10690 very little code that depends on the @var{inexact-flag}, you should
10691 normally not specify this option. Other Alpha compilers call this
10692 option @option{-ieee_with_inexact}.
10694 @item -mfp-trap-mode=@var{trap-mode}
10695 @opindex mfp-trap-mode
10696 This option controls what floating-point related traps are enabled.
10697 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10698 The trap mode can be set to one of four values:
10702 This is the default (normal) setting. The only traps that are enabled
10703 are the ones that cannot be disabled in software (e.g., division by zero
10707 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10711 Like @samp{u}, but the instructions are marked to be safe for software
10712 completion (see Alpha architecture manual for details).
10715 Like @samp{su}, but inexact traps are enabled as well.
10718 @item -mfp-rounding-mode=@var{rounding-mode}
10719 @opindex mfp-rounding-mode
10720 Selects the IEEE rounding mode. Other Alpha compilers call this option
10721 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10726 Normal IEEE rounding mode. Floating point numbers are rounded towards
10727 the nearest machine number or towards the even machine number in case
10731 Round towards minus infinity.
10734 Chopped rounding mode. Floating point numbers are rounded towards zero.
10737 Dynamic rounding mode. A field in the floating point control register
10738 (@var{fpcr}, see Alpha architecture reference manual) controls the
10739 rounding mode in effect. The C library initializes this register for
10740 rounding towards plus infinity. Thus, unless your program modifies the
10741 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10744 @item -mtrap-precision=@var{trap-precision}
10745 @opindex mtrap-precision
10746 In the Alpha architecture, floating point traps are imprecise. This
10747 means without software assistance it is impossible to recover from a
10748 floating trap and program execution normally needs to be terminated.
10749 GCC can generate code that can assist operating system trap handlers
10750 in determining the exact location that caused a floating point trap.
10751 Depending on the requirements of an application, different levels of
10752 precisions can be selected:
10756 Program precision. This option is the default and means a trap handler
10757 can only identify which program caused a floating point exception.
10760 Function precision. The trap handler can determine the function that
10761 caused a floating point exception.
10764 Instruction precision. The trap handler can determine the exact
10765 instruction that caused a floating point exception.
10768 Other Alpha compilers provide the equivalent options called
10769 @option{-scope_safe} and @option{-resumption_safe}.
10771 @item -mieee-conformant
10772 @opindex mieee-conformant
10773 This option marks the generated code as IEEE conformant. You must not
10774 use this option unless you also specify @option{-mtrap-precision=i} and either
10775 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10776 is to emit the line @samp{.eflag 48} in the function prologue of the
10777 generated assembly file. Under DEC Unix, this has the effect that
10778 IEEE-conformant math library routines will be linked in.
10780 @item -mbuild-constants
10781 @opindex mbuild-constants
10782 Normally GCC examines a 32- or 64-bit integer constant to
10783 see if it can construct it from smaller constants in two or three
10784 instructions. If it cannot, it will output the constant as a literal and
10785 generate code to load it from the data segment at runtime.
10787 Use this option to require GCC to construct @emph{all} integer constants
10788 using code, even if it takes more instructions (the maximum is six).
10790 You would typically use this option to build a shared library dynamic
10791 loader. Itself a shared library, it must relocate itself in memory
10792 before it can find the variables and constants in its own data segment.
10798 Select whether to generate code to be assembled by the vendor-supplied
10799 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10817 Indicate whether GCC should generate code to use the optional BWX,
10818 CIX, FIX and MAX instruction sets. The default is to use the instruction
10819 sets supported by the CPU type specified via @option{-mcpu=} option or that
10820 of the CPU on which GCC was built if none was specified.
10823 @itemx -mfloat-ieee
10824 @opindex mfloat-vax
10825 @opindex mfloat-ieee
10826 Generate code that uses (does not use) VAX F and G floating point
10827 arithmetic instead of IEEE single and double precision.
10829 @item -mexplicit-relocs
10830 @itemx -mno-explicit-relocs
10831 @opindex mexplicit-relocs
10832 @opindex mno-explicit-relocs
10833 Older Alpha assemblers provided no way to generate symbol relocations
10834 except via assembler macros. Use of these macros does not allow
10835 optimal instruction scheduling. GNU binutils as of version 2.12
10836 supports a new syntax that allows the compiler to explicitly mark
10837 which relocations should apply to which instructions. This option
10838 is mostly useful for debugging, as GCC detects the capabilities of
10839 the assembler when it is built and sets the default accordingly.
10842 @itemx -mlarge-data
10843 @opindex msmall-data
10844 @opindex mlarge-data
10845 When @option{-mexplicit-relocs} is in effect, static data is
10846 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10847 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10848 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10849 16-bit relocations off of the @code{$gp} register. This limits the
10850 size of the small data area to 64KB, but allows the variables to be
10851 directly accessed via a single instruction.
10853 The default is @option{-mlarge-data}. With this option the data area
10854 is limited to just below 2GB@. Programs that require more than 2GB of
10855 data must use @code{malloc} or @code{mmap} to allocate the data in the
10856 heap instead of in the program's data segment.
10858 When generating code for shared libraries, @option{-fpic} implies
10859 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10862 @itemx -mlarge-text
10863 @opindex msmall-text
10864 @opindex mlarge-text
10865 When @option{-msmall-text} is used, the compiler assumes that the
10866 code of the entire program (or shared library) fits in 4MB, and is
10867 thus reachable with a branch instruction. When @option{-msmall-data}
10868 is used, the compiler can assume that all local symbols share the
10869 same @code{$gp} value, and thus reduce the number of instructions
10870 required for a function call from 4 to 1.
10872 The default is @option{-mlarge-text}.
10874 @item -mcpu=@var{cpu_type}
10876 Set the instruction set and instruction scheduling parameters for
10877 machine type @var{cpu_type}. You can specify either the @samp{EV}
10878 style name or the corresponding chip number. GCC supports scheduling
10879 parameters for the EV4, EV5 and EV6 family of processors and will
10880 choose the default values for the instruction set from the processor
10881 you specify. If you do not specify a processor type, GCC will default
10882 to the processor on which the compiler was built.
10884 Supported values for @var{cpu_type} are
10890 Schedules as an EV4 and has no instruction set extensions.
10894 Schedules as an EV5 and has no instruction set extensions.
10898 Schedules as an EV5 and supports the BWX extension.
10903 Schedules as an EV5 and supports the BWX and MAX extensions.
10907 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10911 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10914 Native Linux/GNU toolchains also support the value @samp{native},
10915 which selects the best architecture option for the host processor.
10916 @option{-mcpu=native} has no effect if GCC does not recognize
10919 @item -mtune=@var{cpu_type}
10921 Set only the instruction scheduling parameters for machine type
10922 @var{cpu_type}. The instruction set is not changed.
10924 Native Linux/GNU toolchains also support the value @samp{native},
10925 which selects the best architecture option for the host processor.
10926 @option{-mtune=native} has no effect if GCC does not recognize
10929 @item -mmemory-latency=@var{time}
10930 @opindex mmemory-latency
10931 Sets the latency the scheduler should assume for typical memory
10932 references as seen by the application. This number is highly
10933 dependent on the memory access patterns used by the application
10934 and the size of the external cache on the machine.
10936 Valid options for @var{time} are
10940 A decimal number representing clock cycles.
10946 The compiler contains estimates of the number of clock cycles for
10947 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10948 (also called Dcache, Scache, and Bcache), as well as to main memory.
10949 Note that L3 is only valid for EV5.
10954 @node DEC Alpha/VMS Options
10955 @subsection DEC Alpha/VMS Options
10957 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10960 @item -mvms-return-codes
10961 @opindex mvms-return-codes
10962 Return VMS condition codes from main. The default is to return POSIX
10963 style condition (e.g.@: error) codes.
10965 @item -mdebug-main=@var{prefix}
10966 @opindex mdebug-main=@var{prefix}
10967 Flag the first routine whose name starts with @var{prefix} as the main
10968 routine for the debugger.
10972 Default to 64bit memory allocation routines.
10976 @subsection FR30 Options
10977 @cindex FR30 Options
10979 These options are defined specifically for the FR30 port.
10983 @item -msmall-model
10984 @opindex msmall-model
10985 Use the small address space model. This can produce smaller code, but
10986 it does assume that all symbolic values and addresses will fit into a
10991 Assume that run-time support has been provided and so there is no need
10992 to include the simulator library (@file{libsim.a}) on the linker
10998 @subsection FRV Options
10999 @cindex FRV Options
11005 Only use the first 32 general purpose registers.
11010 Use all 64 general purpose registers.
11015 Use only the first 32 floating point registers.
11020 Use all 64 floating point registers
11023 @opindex mhard-float
11025 Use hardware instructions for floating point operations.
11028 @opindex msoft-float
11030 Use library routines for floating point operations.
11035 Dynamically allocate condition code registers.
11040 Do not try to dynamically allocate condition code registers, only
11041 use @code{icc0} and @code{fcc0}.
11046 Change ABI to use double word insns.
11051 Do not use double word instructions.
11056 Use floating point double instructions.
11059 @opindex mno-double
11061 Do not use floating point double instructions.
11066 Use media instructions.
11071 Do not use media instructions.
11076 Use multiply and add/subtract instructions.
11079 @opindex mno-muladd
11081 Do not use multiply and add/subtract instructions.
11086 Select the FDPIC ABI, that uses function descriptors to represent
11087 pointers to functions. Without any PIC/PIE-related options, it
11088 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11089 assumes GOT entries and small data are within a 12-bit range from the
11090 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11091 are computed with 32 bits.
11092 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11095 @opindex minline-plt
11097 Enable inlining of PLT entries in function calls to functions that are
11098 not known to bind locally. It has no effect without @option{-mfdpic}.
11099 It's enabled by default if optimizing for speed and compiling for
11100 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11101 optimization option such as @option{-O3} or above is present in the
11107 Assume a large TLS segment when generating thread-local code.
11112 Do not assume a large TLS segment when generating thread-local code.
11117 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11118 that is known to be in read-only sections. It's enabled by default,
11119 except for @option{-fpic} or @option{-fpie}: even though it may help
11120 make the global offset table smaller, it trades 1 instruction for 4.
11121 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11122 one of which may be shared by multiple symbols, and it avoids the need
11123 for a GOT entry for the referenced symbol, so it's more likely to be a
11124 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11126 @item -multilib-library-pic
11127 @opindex multilib-library-pic
11129 Link with the (library, not FD) pic libraries. It's implied by
11130 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11131 @option{-fpic} without @option{-mfdpic}. You should never have to use
11135 @opindex mlinked-fp
11137 Follow the EABI requirement of always creating a frame pointer whenever
11138 a stack frame is allocated. This option is enabled by default and can
11139 be disabled with @option{-mno-linked-fp}.
11142 @opindex mlong-calls
11144 Use indirect addressing to call functions outside the current
11145 compilation unit. This allows the functions to be placed anywhere
11146 within the 32-bit address space.
11148 @item -malign-labels
11149 @opindex malign-labels
11151 Try to align labels to an 8-byte boundary by inserting nops into the
11152 previous packet. This option only has an effect when VLIW packing
11153 is enabled. It doesn't create new packets; it merely adds nops to
11156 @item -mlibrary-pic
11157 @opindex mlibrary-pic
11159 Generate position-independent EABI code.
11164 Use only the first four media accumulator registers.
11169 Use all eight media accumulator registers.
11174 Pack VLIW instructions.
11179 Do not pack VLIW instructions.
11182 @opindex mno-eflags
11184 Do not mark ABI switches in e_flags.
11187 @opindex mcond-move
11189 Enable the use of conditional-move instructions (default).
11191 This switch is mainly for debugging the compiler and will likely be removed
11192 in a future version.
11194 @item -mno-cond-move
11195 @opindex mno-cond-move
11197 Disable the use of conditional-move instructions.
11199 This switch is mainly for debugging the compiler and will likely be removed
11200 in a future version.
11205 Enable the use of conditional set instructions (default).
11207 This switch is mainly for debugging the compiler and will likely be removed
11208 in a future version.
11213 Disable the use of conditional set instructions.
11215 This switch is mainly for debugging the compiler and will likely be removed
11216 in a future version.
11219 @opindex mcond-exec
11221 Enable the use of conditional execution (default).
11223 This switch is mainly for debugging the compiler and will likely be removed
11224 in a future version.
11226 @item -mno-cond-exec
11227 @opindex mno-cond-exec
11229 Disable the use of conditional execution.
11231 This switch is mainly for debugging the compiler and will likely be removed
11232 in a future version.
11234 @item -mvliw-branch
11235 @opindex mvliw-branch
11237 Run a pass to pack branches into VLIW instructions (default).
11239 This switch is mainly for debugging the compiler and will likely be removed
11240 in a future version.
11242 @item -mno-vliw-branch
11243 @opindex mno-vliw-branch
11245 Do not run a pass to pack branches into VLIW instructions.
11247 This switch is mainly for debugging the compiler and will likely be removed
11248 in a future version.
11250 @item -mmulti-cond-exec
11251 @opindex mmulti-cond-exec
11253 Enable optimization of @code{&&} and @code{||} in conditional execution
11256 This switch is mainly for debugging the compiler and will likely be removed
11257 in a future version.
11259 @item -mno-multi-cond-exec
11260 @opindex mno-multi-cond-exec
11262 Disable optimization of @code{&&} and @code{||} in conditional execution.
11264 This switch is mainly for debugging the compiler and will likely be removed
11265 in a future version.
11267 @item -mnested-cond-exec
11268 @opindex mnested-cond-exec
11270 Enable nested conditional execution optimizations (default).
11272 This switch is mainly for debugging the compiler and will likely be removed
11273 in a future version.
11275 @item -mno-nested-cond-exec
11276 @opindex mno-nested-cond-exec
11278 Disable nested conditional execution optimizations.
11280 This switch is mainly for debugging the compiler and will likely be removed
11281 in a future version.
11283 @item -moptimize-membar
11284 @opindex moptimize-membar
11286 This switch removes redundant @code{membar} instructions from the
11287 compiler generated code. It is enabled by default.
11289 @item -mno-optimize-membar
11290 @opindex mno-optimize-membar
11292 This switch disables the automatic removal of redundant @code{membar}
11293 instructions from the generated code.
11295 @item -mtomcat-stats
11296 @opindex mtomcat-stats
11298 Cause gas to print out tomcat statistics.
11300 @item -mcpu=@var{cpu}
11303 Select the processor type for which to generate code. Possible values are
11304 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11305 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11309 @node GNU/Linux Options
11310 @subsection GNU/Linux Options
11312 These @samp{-m} options are defined for GNU/Linux targets:
11317 Use the GNU C library instead of uClibc. This is the default except
11318 on @samp{*-*-linux-*uclibc*} targets.
11322 Use uClibc instead of the GNU C library. This is the default on
11323 @samp{*-*-linux-*uclibc*} targets.
11326 @node H8/300 Options
11327 @subsection H8/300 Options
11329 These @samp{-m} options are defined for the H8/300 implementations:
11334 Shorten some address references at link time, when possible; uses the
11335 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11336 ld, Using ld}, for a fuller description.
11340 Generate code for the H8/300H@.
11344 Generate code for the H8S@.
11348 Generate code for the H8S and H8/300H in the normal mode. This switch
11349 must be used either with @option{-mh} or @option{-ms}.
11353 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11357 Make @code{int} data 32 bits by default.
11360 @opindex malign-300
11361 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11362 The default for the H8/300H and H8S is to align longs and floats on 4
11364 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11365 This option has no effect on the H8/300.
11369 @subsection HPPA Options
11370 @cindex HPPA Options
11372 These @samp{-m} options are defined for the HPPA family of computers:
11375 @item -march=@var{architecture-type}
11377 Generate code for the specified architecture. The choices for
11378 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11379 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11380 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11381 architecture option for your machine. Code compiled for lower numbered
11382 architectures will run on higher numbered architectures, but not the
11385 @item -mpa-risc-1-0
11386 @itemx -mpa-risc-1-1
11387 @itemx -mpa-risc-2-0
11388 @opindex mpa-risc-1-0
11389 @opindex mpa-risc-1-1
11390 @opindex mpa-risc-2-0
11391 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11394 @opindex mbig-switch
11395 Generate code suitable for big switch tables. Use this option only if
11396 the assembler/linker complain about out of range branches within a switch
11399 @item -mjump-in-delay
11400 @opindex mjump-in-delay
11401 Fill delay slots of function calls with unconditional jump instructions
11402 by modifying the return pointer for the function call to be the target
11403 of the conditional jump.
11405 @item -mdisable-fpregs
11406 @opindex mdisable-fpregs
11407 Prevent floating point registers from being used in any manner. This is
11408 necessary for compiling kernels which perform lazy context switching of
11409 floating point registers. If you use this option and attempt to perform
11410 floating point operations, the compiler will abort.
11412 @item -mdisable-indexing
11413 @opindex mdisable-indexing
11414 Prevent the compiler from using indexing address modes. This avoids some
11415 rather obscure problems when compiling MIG generated code under MACH@.
11417 @item -mno-space-regs
11418 @opindex mno-space-regs
11419 Generate code that assumes the target has no space registers. This allows
11420 GCC to generate faster indirect calls and use unscaled index address modes.
11422 Such code is suitable for level 0 PA systems and kernels.
11424 @item -mfast-indirect-calls
11425 @opindex mfast-indirect-calls
11426 Generate code that assumes calls never cross space boundaries. This
11427 allows GCC to emit code which performs faster indirect calls.
11429 This option will not work in the presence of shared libraries or nested
11432 @item -mfixed-range=@var{register-range}
11433 @opindex mfixed-range
11434 Generate code treating the given register range as fixed registers.
11435 A fixed register is one that the register allocator can not use. This is
11436 useful when compiling kernel code. A register range is specified as
11437 two registers separated by a dash. Multiple register ranges can be
11438 specified separated by a comma.
11440 @item -mlong-load-store
11441 @opindex mlong-load-store
11442 Generate 3-instruction load and store sequences as sometimes required by
11443 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11446 @item -mportable-runtime
11447 @opindex mportable-runtime
11448 Use the portable calling conventions proposed by HP for ELF systems.
11452 Enable the use of assembler directives only GAS understands.
11454 @item -mschedule=@var{cpu-type}
11456 Schedule code according to the constraints for the machine type
11457 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11458 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11459 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11460 proper scheduling option for your machine. The default scheduling is
11464 @opindex mlinker-opt
11465 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11466 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11467 linkers in which they give bogus error messages when linking some programs.
11470 @opindex msoft-float
11471 Generate output containing library calls for floating point.
11472 @strong{Warning:} the requisite libraries are not available for all HPPA
11473 targets. Normally the facilities of the machine's usual C compiler are
11474 used, but this cannot be done directly in cross-compilation. You must make
11475 your own arrangements to provide suitable library functions for
11478 @option{-msoft-float} changes the calling convention in the output file;
11479 therefore, it is only useful if you compile @emph{all} of a program with
11480 this option. In particular, you need to compile @file{libgcc.a}, the
11481 library that comes with GCC, with @option{-msoft-float} in order for
11486 Generate the predefine, @code{_SIO}, for server IO@. The default is
11487 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11488 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11489 options are available under HP-UX and HI-UX@.
11493 Use GNU ld specific options. This passes @option{-shared} to ld when
11494 building a shared library. It is the default when GCC is configured,
11495 explicitly or implicitly, with the GNU linker. This option does not
11496 have any affect on which ld is called, it only changes what parameters
11497 are passed to that ld. The ld that is called is determined by the
11498 @option{--with-ld} configure option, GCC's program search path, and
11499 finally by the user's @env{PATH}. The linker used by GCC can be printed
11500 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11501 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11505 Use HP ld specific options. This passes @option{-b} to ld when building
11506 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11507 links. It is the default when GCC is configured, explicitly or
11508 implicitly, with the HP linker. This option does not have any affect on
11509 which ld is called, it only changes what parameters are passed to that
11510 ld. The ld that is called is determined by the @option{--with-ld}
11511 configure option, GCC's program search path, and finally by the user's
11512 @env{PATH}. The linker used by GCC can be printed using @samp{which
11513 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11514 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11517 @opindex mno-long-calls
11518 Generate code that uses long call sequences. This ensures that a call
11519 is always able to reach linker generated stubs. The default is to generate
11520 long calls only when the distance from the call site to the beginning
11521 of the function or translation unit, as the case may be, exceeds a
11522 predefined limit set by the branch type being used. The limits for
11523 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11524 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11527 Distances are measured from the beginning of functions when using the
11528 @option{-ffunction-sections} option, or when using the @option{-mgas}
11529 and @option{-mno-portable-runtime} options together under HP-UX with
11532 It is normally not desirable to use this option as it will degrade
11533 performance. However, it may be useful in large applications,
11534 particularly when partial linking is used to build the application.
11536 The types of long calls used depends on the capabilities of the
11537 assembler and linker, and the type of code being generated. The
11538 impact on systems that support long absolute calls, and long pic
11539 symbol-difference or pc-relative calls should be relatively small.
11540 However, an indirect call is used on 32-bit ELF systems in pic code
11541 and it is quite long.
11543 @item -munix=@var{unix-std}
11545 Generate compiler predefines and select a startfile for the specified
11546 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11547 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11548 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11549 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11550 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11553 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11554 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11555 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11556 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11557 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11558 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11560 It is @emph{important} to note that this option changes the interfaces
11561 for various library routines. It also affects the operational behavior
11562 of the C library. Thus, @emph{extreme} care is needed in using this
11565 Library code that is intended to operate with more than one UNIX
11566 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11567 as appropriate. Most GNU software doesn't provide this capability.
11571 Suppress the generation of link options to search libdld.sl when the
11572 @option{-static} option is specified on HP-UX 10 and later.
11576 The HP-UX implementation of setlocale in libc has a dependency on
11577 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11578 when the @option{-static} option is specified, special link options
11579 are needed to resolve this dependency.
11581 On HP-UX 10 and later, the GCC driver adds the necessary options to
11582 link with libdld.sl when the @option{-static} option is specified.
11583 This causes the resulting binary to be dynamic. On the 64-bit port,
11584 the linkers generate dynamic binaries by default in any case. The
11585 @option{-nolibdld} option can be used to prevent the GCC driver from
11586 adding these link options.
11590 Add support for multithreading with the @dfn{dce thread} library
11591 under HP-UX@. This option sets flags for both the preprocessor and
11595 @node i386 and x86-64 Options
11596 @subsection Intel 386 and AMD x86-64 Options
11597 @cindex i386 Options
11598 @cindex x86-64 Options
11599 @cindex Intel 386 Options
11600 @cindex AMD x86-64 Options
11602 These @samp{-m} options are defined for the i386 and x86-64 family of
11606 @item -mtune=@var{cpu-type}
11608 Tune to @var{cpu-type} everything applicable about the generated code, except
11609 for the ABI and the set of available instructions. The choices for
11610 @var{cpu-type} are:
11613 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11614 If you know the CPU on which your code will run, then you should use
11615 the corresponding @option{-mtune} option instead of
11616 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11617 of your application will have, then you should use this option.
11619 As new processors are deployed in the marketplace, the behavior of this
11620 option will change. Therefore, if you upgrade to a newer version of
11621 GCC, the code generated option will change to reflect the processors
11622 that were most common when that version of GCC was released.
11624 There is no @option{-march=generic} option because @option{-march}
11625 indicates the instruction set the compiler can use, and there is no
11626 generic instruction set applicable to all processors. In contrast,
11627 @option{-mtune} indicates the processor (or, in this case, collection of
11628 processors) for which the code is optimized.
11630 This selects the CPU to tune for at compilation time by determining
11631 the processor type of the compiling machine. Using @option{-mtune=native}
11632 will produce code optimized for the local machine under the constraints
11633 of the selected instruction set. Using @option{-march=native} will
11634 enable all instruction subsets supported by the local machine (hence
11635 the result might not run on different machines).
11637 Original Intel's i386 CPU@.
11639 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11640 @item i586, pentium
11641 Intel Pentium CPU with no MMX support.
11643 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11645 Intel PentiumPro CPU@.
11647 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11648 instruction set will be used, so the code will run on all i686 family chips.
11650 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11651 @item pentium3, pentium3m
11652 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11655 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11656 support. Used by Centrino notebooks.
11657 @item pentium4, pentium4m
11658 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11660 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11663 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11664 SSE2 and SSE3 instruction set support.
11666 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11667 instruction set support.
11669 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11670 instruction set support.
11672 AMD K6 CPU with MMX instruction set support.
11674 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
11675 @item athlon, athlon-tbird
11676 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
11678 @item athlon-4, athlon-xp, athlon-mp
11679 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
11680 instruction set support.
11681 @item k8, opteron, athlon64, athlon-fx
11682 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11683 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
11684 @item k8-sse3, opteron-sse3, athlon64-sse3
11685 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11686 @item amdfam10, barcelona
11687 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11688 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
11689 instruction set extensions.)
11691 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11694 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
11695 instruction set support.
11697 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
11698 implemented for this chip.)
11700 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11701 implemented for this chip.)
11703 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
11706 While picking a specific @var{cpu-type} will schedule things appropriately
11707 for that particular chip, the compiler will not generate any code that
11708 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11711 @item -march=@var{cpu-type}
11713 Generate instructions for the machine type @var{cpu-type}. The choices
11714 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11715 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11717 @item -mcpu=@var{cpu-type}
11719 A deprecated synonym for @option{-mtune}.
11721 @item -mfpmath=@var{unit}
11723 Generate floating point arithmetics for selected unit @var{unit}. The choices
11724 for @var{unit} are:
11728 Use the standard 387 floating point coprocessor present majority of chips and
11729 emulated otherwise. Code compiled with this option will run almost everywhere.
11730 The temporary results are computed in 80bit precision instead of precision
11731 specified by the type resulting in slightly different results compared to most
11732 of other chips. See @option{-ffloat-store} for more detailed description.
11734 This is the default choice for i386 compiler.
11737 Use scalar floating point instructions present in the SSE instruction set.
11738 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11739 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11740 instruction set supports only single precision arithmetics, thus the double and
11741 extended precision arithmetics is still done using 387. Later version, present
11742 only in Pentium4 and the future AMD x86-64 chips supports double precision
11745 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11746 or @option{-msse2} switches to enable SSE extensions and make this option
11747 effective. For the x86-64 compiler, these extensions are enabled by default.
11749 The resulting code should be considerably faster in the majority of cases and avoid
11750 the numerical instability problems of 387 code, but may break some existing
11751 code that expects temporaries to be 80bit.
11753 This is the default choice for the x86-64 compiler.
11758 Attempt to utilize both instruction sets at once. This effectively double the
11759 amount of available registers and on chips with separate execution units for
11760 387 and SSE the execution resources too. Use this option with care, as it is
11761 still experimental, because the GCC register allocator does not model separate
11762 functional units well resulting in instable performance.
11765 @item -masm=@var{dialect}
11766 @opindex masm=@var{dialect}
11767 Output asm instructions using selected @var{dialect}. Supported
11768 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11769 not support @samp{intel}.
11772 @itemx -mno-ieee-fp
11774 @opindex mno-ieee-fp
11775 Control whether or not the compiler uses IEEE floating point
11776 comparisons. These handle correctly the case where the result of a
11777 comparison is unordered.
11780 @opindex msoft-float
11781 Generate output containing library calls for floating point.
11782 @strong{Warning:} the requisite libraries are not part of GCC@.
11783 Normally the facilities of the machine's usual C compiler are used, but
11784 this can't be done directly in cross-compilation. You must make your
11785 own arrangements to provide suitable library functions for
11788 On machines where a function returns floating point results in the 80387
11789 register stack, some floating point opcodes may be emitted even if
11790 @option{-msoft-float} is used.
11792 @item -mno-fp-ret-in-387
11793 @opindex mno-fp-ret-in-387
11794 Do not use the FPU registers for return values of functions.
11796 The usual calling convention has functions return values of types
11797 @code{float} and @code{double} in an FPU register, even if there
11798 is no FPU@. The idea is that the operating system should emulate
11801 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11802 in ordinary CPU registers instead.
11804 @item -mno-fancy-math-387
11805 @opindex mno-fancy-math-387
11806 Some 387 emulators do not support the @code{sin}, @code{cos} and
11807 @code{sqrt} instructions for the 387. Specify this option to avoid
11808 generating those instructions. This option is the default on FreeBSD,
11809 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11810 indicates that the target cpu will always have an FPU and so the
11811 instruction will not need emulation. As of revision 2.6.1, these
11812 instructions are not generated unless you also use the
11813 @option{-funsafe-math-optimizations} switch.
11815 @item -malign-double
11816 @itemx -mno-align-double
11817 @opindex malign-double
11818 @opindex mno-align-double
11819 Control whether GCC aligns @code{double}, @code{long double}, and
11820 @code{long long} variables on a two word boundary or a one word
11821 boundary. Aligning @code{double} variables on a two word boundary will
11822 produce code that runs somewhat faster on a @samp{Pentium} at the
11823 expense of more memory.
11825 On x86-64, @option{-malign-double} is enabled by default.
11827 @strong{Warning:} if you use the @option{-malign-double} switch,
11828 structures containing the above types will be aligned differently than
11829 the published application binary interface specifications for the 386
11830 and will not be binary compatible with structures in code compiled
11831 without that switch.
11833 @item -m96bit-long-double
11834 @itemx -m128bit-long-double
11835 @opindex m96bit-long-double
11836 @opindex m128bit-long-double
11837 These switches control the size of @code{long double} type. The i386
11838 application binary interface specifies the size to be 96 bits,
11839 so @option{-m96bit-long-double} is the default in 32 bit mode.
11841 Modern architectures (Pentium and newer) would prefer @code{long double}
11842 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11843 conforming to the ABI, this would not be possible. So specifying a
11844 @option{-m128bit-long-double} will align @code{long double}
11845 to a 16 byte boundary by padding the @code{long double} with an additional
11848 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11849 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11851 Notice that neither of these options enable any extra precision over the x87
11852 standard of 80 bits for a @code{long double}.
11854 @strong{Warning:} if you override the default value for your target ABI, the
11855 structures and arrays containing @code{long double} variables will change
11856 their size as well as function calling convention for function taking
11857 @code{long double} will be modified. Hence they will not be binary
11858 compatible with arrays or structures in code compiled without that switch.
11860 @item -mlarge-data-threshold=@var{number}
11861 @opindex mlarge-data-threshold=@var{number}
11862 When @option{-mcmodel=medium} is specified, the data greater than
11863 @var{threshold} are placed in large data section. This value must be the
11864 same across all object linked into the binary and defaults to 65535.
11868 Use a different function-calling convention, in which functions that
11869 take a fixed number of arguments return with the @code{ret} @var{num}
11870 instruction, which pops their arguments while returning. This saves one
11871 instruction in the caller since there is no need to pop the arguments
11874 You can specify that an individual function is called with this calling
11875 sequence with the function attribute @samp{stdcall}. You can also
11876 override the @option{-mrtd} option by using the function attribute
11877 @samp{cdecl}. @xref{Function Attributes}.
11879 @strong{Warning:} this calling convention is incompatible with the one
11880 normally used on Unix, so you cannot use it if you need to call
11881 libraries compiled with the Unix compiler.
11883 Also, you must provide function prototypes for all functions that
11884 take variable numbers of arguments (including @code{printf});
11885 otherwise incorrect code will be generated for calls to those
11888 In addition, seriously incorrect code will result if you call a
11889 function with too many arguments. (Normally, extra arguments are
11890 harmlessly ignored.)
11892 @item -mregparm=@var{num}
11894 Control how many registers are used to pass integer arguments. By
11895 default, no registers are used to pass arguments, and at most 3
11896 registers can be used. You can control this behavior for a specific
11897 function by using the function attribute @samp{regparm}.
11898 @xref{Function Attributes}.
11900 @strong{Warning:} if you use this switch, and
11901 @var{num} is nonzero, then you must build all modules with the same
11902 value, including any libraries. This includes the system libraries and
11906 @opindex msseregparm
11907 Use SSE register passing conventions for float and double arguments
11908 and return values. You can control this behavior for a specific
11909 function by using the function attribute @samp{sseregparm}.
11910 @xref{Function Attributes}.
11912 @strong{Warning:} if you use this switch then you must build all
11913 modules with the same value, including any libraries. This includes
11914 the system libraries and startup modules.
11923 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11924 is specified, the significands of results of floating-point operations are
11925 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11926 significands of results of floating-point operations to 53 bits (double
11927 precision) and @option{-mpc80} rounds the significands of results of
11928 floating-point operations to 64 bits (extended double precision), which is
11929 the default. When this option is used, floating-point operations in higher
11930 precisions are not available to the programmer without setting the FPU
11931 control word explicitly.
11933 Setting the rounding of floating-point operations to less than the default
11934 80 bits can speed some programs by 2% or more. Note that some mathematical
11935 libraries assume that extended precision (80 bit) floating-point operations
11936 are enabled by default; routines in such libraries could suffer significant
11937 loss of accuracy, typically through so-called "catastrophic cancellation",
11938 when this option is used to set the precision to less than extended precision.
11940 @item -mstackrealign
11941 @opindex mstackrealign
11942 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
11943 option will generate an alternate prologue and epilogue that realigns the
11944 runtime stack if necessary. This supports mixing legacy codes that keep
11945 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11946 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
11947 applicable to individual functions.
11949 @item -mpreferred-stack-boundary=@var{num}
11950 @opindex mpreferred-stack-boundary
11951 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11952 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
11953 the default is 4 (16 bytes or 128 bits).
11955 @item -mincoming-stack-boundary=@var{num}
11956 @opindex mincoming-stack-boundary
11957 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11958 boundary. If @option{-mincoming-stack-boundary} is not specified,
11959 the one specified by @option{-mpreferred-stack-boundary} will be used.
11961 On Pentium and PentiumPro, @code{double} and @code{long double} values
11962 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11963 suffer significant run time performance penalties. On Pentium III, the
11964 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11965 properly if it is not 16 byte aligned.
11967 To ensure proper alignment of this values on the stack, the stack boundary
11968 must be as aligned as that required by any value stored on the stack.
11969 Further, every function must be generated such that it keeps the stack
11970 aligned. Thus calling a function compiled with a higher preferred
11971 stack boundary from a function compiled with a lower preferred stack
11972 boundary will most likely misalign the stack. It is recommended that
11973 libraries that use callbacks always use the default setting.
11975 This extra alignment does consume extra stack space, and generally
11976 increases code size. Code that is sensitive to stack space usage, such
11977 as embedded systems and operating system kernels, may want to reduce the
11978 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12018 These switches enable or disable the use of instructions in the MMX,
12019 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, FMA4, ABM or
12020 3DNow!@: extended instruction sets.
12021 These extensions are also available as built-in functions: see
12022 @ref{X86 Built-in Functions}, for details of the functions enabled and
12023 disabled by these switches.
12025 To have SSE/SSE2 instructions generated automatically from floating-point
12026 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12028 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12029 generates new AVX instructions or AVX equivalence for all SSEx instructions
12032 These options will enable GCC to use these extended instructions in
12033 generated code, even without @option{-mfpmath=sse}. Applications which
12034 perform runtime CPU detection must compile separate files for each
12035 supported architecture, using the appropriate flags. In particular,
12036 the file containing the CPU detection code should be compiled without
12041 This option instructs GCC to emit a @code{cld} instruction in the prologue
12042 of functions that use string instructions. String instructions depend on
12043 the DF flag to select between autoincrement or autodecrement mode. While the
12044 ABI specifies the DF flag to be cleared on function entry, some operating
12045 systems violate this specification by not clearing the DF flag in their
12046 exception dispatchers. The exception handler can be invoked with the DF flag
12047 set which leads to wrong direction mode, when string instructions are used.
12048 This option can be enabled by default on 32-bit x86 targets by configuring
12049 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12050 instructions can be suppressed with the @option{-mno-cld} compiler option
12055 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12056 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12057 data types. This is useful for high resolution counters that could be updated
12058 by multiple processors (or cores). This instruction is generated as part of
12059 atomic built-in functions: see @ref{Atomic Builtins} for details.
12063 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12064 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12065 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12066 SAHF are load and store instructions, respectively, for certain status flags.
12067 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12068 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12072 This option will enable GCC to use movbe instruction to implement
12073 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12077 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12078 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12079 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12083 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12084 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12085 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12086 variants) for single precision floating point arguments. These instructions
12087 are generated only when @option{-funsafe-math-optimizations} is enabled
12088 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12089 Note that while the throughput of the sequence is higher than the throughput
12090 of the non-reciprocal instruction, the precision of the sequence can be
12091 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12093 @item -mveclibabi=@var{type}
12094 @opindex mveclibabi
12095 Specifies the ABI type to use for vectorizing intrinsics using an
12096 external library. Supported types are @code{svml} for the Intel short
12097 vector math library and @code{acml} for the AMD math core library style
12098 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12099 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12100 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12101 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12102 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12103 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12104 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12105 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12106 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12107 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12108 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12109 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12110 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12111 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12112 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12113 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12114 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12115 compatible library will have to be specified at link time.
12117 @item -mabi=@var{name}
12119 Generate code for the specified calling convention. Permissible values
12120 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12121 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12122 ABI when targeting Windows. On all other systems, the default is the
12123 SYSV ABI. You can control this behavior for a specific function by
12124 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12125 @xref{Function Attributes}.
12128 @itemx -mno-push-args
12129 @opindex mpush-args
12130 @opindex mno-push-args
12131 Use PUSH operations to store outgoing parameters. This method is shorter
12132 and usually equally fast as method using SUB/MOV operations and is enabled
12133 by default. In some cases disabling it may improve performance because of
12134 improved scheduling and reduced dependencies.
12136 @item -maccumulate-outgoing-args
12137 @opindex maccumulate-outgoing-args
12138 If enabled, the maximum amount of space required for outgoing arguments will be
12139 computed in the function prologue. This is faster on most modern CPUs
12140 because of reduced dependencies, improved scheduling and reduced stack usage
12141 when preferred stack boundary is not equal to 2. The drawback is a notable
12142 increase in code size. This switch implies @option{-mno-push-args}.
12146 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12147 on thread-safe exception handling must compile and link all code with the
12148 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12149 @option{-D_MT}; when linking, it links in a special thread helper library
12150 @option{-lmingwthrd} which cleans up per thread exception handling data.
12152 @item -mno-align-stringops
12153 @opindex mno-align-stringops
12154 Do not align destination of inlined string operations. This switch reduces
12155 code size and improves performance in case the destination is already aligned,
12156 but GCC doesn't know about it.
12158 @item -minline-all-stringops
12159 @opindex minline-all-stringops
12160 By default GCC inlines string operations only when destination is known to be
12161 aligned at least to 4 byte boundary. This enables more inlining, increase code
12162 size, but may improve performance of code that depends on fast memcpy, strlen
12163 and memset for short lengths.
12165 @item -minline-stringops-dynamically
12166 @opindex minline-stringops-dynamically
12167 For string operation of unknown size, inline runtime checks so for small
12168 blocks inline code is used, while for large blocks library call is used.
12170 @item -mstringop-strategy=@var{alg}
12171 @opindex mstringop-strategy=@var{alg}
12172 Overwrite internal decision heuristic about particular algorithm to inline
12173 string operation with. The allowed values are @code{rep_byte},
12174 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12175 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12176 expanding inline loop, @code{libcall} for always expanding library call.
12178 @item -momit-leaf-frame-pointer
12179 @opindex momit-leaf-frame-pointer
12180 Don't keep the frame pointer in a register for leaf functions. This
12181 avoids the instructions to save, set up and restore frame pointers and
12182 makes an extra register available in leaf functions. The option
12183 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12184 which might make debugging harder.
12186 @item -mtls-direct-seg-refs
12187 @itemx -mno-tls-direct-seg-refs
12188 @opindex mtls-direct-seg-refs
12189 Controls whether TLS variables may be accessed with offsets from the
12190 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12191 or whether the thread base pointer must be added. Whether or not this
12192 is legal depends on the operating system, and whether it maps the
12193 segment to cover the entire TLS area.
12195 For systems that use GNU libc, the default is on.
12198 @itemx -mno-sse2avx
12200 Specify that the assembler should encode SSE instructions with VEX
12201 prefix. The option @option{-mavx} turns this on by default.
12204 These @samp{-m} switches are supported in addition to the above
12205 on AMD x86-64 processors in 64-bit environments.
12212 Generate code for a 32-bit or 64-bit environment.
12213 The 32-bit environment sets int, long and pointer to 32 bits and
12214 generates code that runs on any i386 system.
12215 The 64-bit environment sets int to 32 bits and long and pointer
12216 to 64 bits and generates code for AMD's x86-64 architecture. For
12217 darwin only the -m64 option turns off the @option{-fno-pic} and
12218 @option{-mdynamic-no-pic} options.
12220 @item -mno-red-zone
12221 @opindex mno-red-zone
12222 Do not use a so called red zone for x86-64 code. The red zone is mandated
12223 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12224 stack pointer that will not be modified by signal or interrupt handlers
12225 and therefore can be used for temporary data without adjusting the stack
12226 pointer. The flag @option{-mno-red-zone} disables this red zone.
12228 @item -mcmodel=small
12229 @opindex mcmodel=small
12230 Generate code for the small code model: the program and its symbols must
12231 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12232 Programs can be statically or dynamically linked. This is the default
12235 @item -mcmodel=kernel
12236 @opindex mcmodel=kernel
12237 Generate code for the kernel code model. The kernel runs in the
12238 negative 2 GB of the address space.
12239 This model has to be used for Linux kernel code.
12241 @item -mcmodel=medium
12242 @opindex mcmodel=medium
12243 Generate code for the medium model: The program is linked in the lower 2
12244 GB of the address space. Small symbols are also placed there. Symbols
12245 with sizes larger than @option{-mlarge-data-threshold} are put into
12246 large data or bss sections and can be located above 2GB. Programs can
12247 be statically or dynamically linked.
12249 @item -mcmodel=large
12250 @opindex mcmodel=large
12251 Generate code for the large model: This model makes no assumptions
12252 about addresses and sizes of sections.
12255 @node IA-64 Options
12256 @subsection IA-64 Options
12257 @cindex IA-64 Options
12259 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12263 @opindex mbig-endian
12264 Generate code for a big endian target. This is the default for HP-UX@.
12266 @item -mlittle-endian
12267 @opindex mlittle-endian
12268 Generate code for a little endian target. This is the default for AIX5
12274 @opindex mno-gnu-as
12275 Generate (or don't) code for the GNU assembler. This is the default.
12276 @c Also, this is the default if the configure option @option{--with-gnu-as}
12282 @opindex mno-gnu-ld
12283 Generate (or don't) code for the GNU linker. This is the default.
12284 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12289 Generate code that does not use a global pointer register. The result
12290 is not position independent code, and violates the IA-64 ABI@.
12292 @item -mvolatile-asm-stop
12293 @itemx -mno-volatile-asm-stop
12294 @opindex mvolatile-asm-stop
12295 @opindex mno-volatile-asm-stop
12296 Generate (or don't) a stop bit immediately before and after volatile asm
12299 @item -mregister-names
12300 @itemx -mno-register-names
12301 @opindex mregister-names
12302 @opindex mno-register-names
12303 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12304 the stacked registers. This may make assembler output more readable.
12310 Disable (or enable) optimizations that use the small data section. This may
12311 be useful for working around optimizer bugs.
12313 @item -mconstant-gp
12314 @opindex mconstant-gp
12315 Generate code that uses a single constant global pointer value. This is
12316 useful when compiling kernel code.
12320 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12321 This is useful when compiling firmware code.
12323 @item -minline-float-divide-min-latency
12324 @opindex minline-float-divide-min-latency
12325 Generate code for inline divides of floating point values
12326 using the minimum latency algorithm.
12328 @item -minline-float-divide-max-throughput
12329 @opindex minline-float-divide-max-throughput
12330 Generate code for inline divides of floating point values
12331 using the maximum throughput algorithm.
12333 @item -mno-inline-float-divide
12334 @opindex mno-inline-float-divide
12335 Do not generate inline code for divides of floating point values.
12337 @item -minline-int-divide-min-latency
12338 @opindex minline-int-divide-min-latency
12339 Generate code for inline divides of integer values
12340 using the minimum latency algorithm.
12342 @item -minline-int-divide-max-throughput
12343 @opindex minline-int-divide-max-throughput
12344 Generate code for inline divides of integer values
12345 using the maximum throughput algorithm.
12347 @item -mno-inline-int-divide
12348 @opindex mno-inline-int-divide
12349 Do not generate inline code for divides of integer values.
12351 @item -minline-sqrt-min-latency
12352 @opindex minline-sqrt-min-latency
12353 Generate code for inline square roots
12354 using the minimum latency algorithm.
12356 @item -minline-sqrt-max-throughput
12357 @opindex minline-sqrt-max-throughput
12358 Generate code for inline square roots
12359 using the maximum throughput algorithm.
12361 @item -mno-inline-sqrt
12362 @opindex mno-inline-sqrt
12363 Do not generate inline code for sqrt.
12366 @itemx -mno-fused-madd
12367 @opindex mfused-madd
12368 @opindex mno-fused-madd
12369 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12370 instructions. The default is to use these instructions.
12372 @item -mno-dwarf2-asm
12373 @itemx -mdwarf2-asm
12374 @opindex mno-dwarf2-asm
12375 @opindex mdwarf2-asm
12376 Don't (or do) generate assembler code for the DWARF2 line number debugging
12377 info. This may be useful when not using the GNU assembler.
12379 @item -mearly-stop-bits
12380 @itemx -mno-early-stop-bits
12381 @opindex mearly-stop-bits
12382 @opindex mno-early-stop-bits
12383 Allow stop bits to be placed earlier than immediately preceding the
12384 instruction that triggered the stop bit. This can improve instruction
12385 scheduling, but does not always do so.
12387 @item -mfixed-range=@var{register-range}
12388 @opindex mfixed-range
12389 Generate code treating the given register range as fixed registers.
12390 A fixed register is one that the register allocator can not use. This is
12391 useful when compiling kernel code. A register range is specified as
12392 two registers separated by a dash. Multiple register ranges can be
12393 specified separated by a comma.
12395 @item -mtls-size=@var{tls-size}
12397 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12400 @item -mtune=@var{cpu-type}
12402 Tune the instruction scheduling for a particular CPU, Valid values are
12403 itanium, itanium1, merced, itanium2, and mckinley.
12409 Generate code for a 32-bit or 64-bit environment.
12410 The 32-bit environment sets int, long and pointer to 32 bits.
12411 The 64-bit environment sets int to 32 bits and long and pointer
12412 to 64 bits. These are HP-UX specific flags.
12414 @item -mno-sched-br-data-spec
12415 @itemx -msched-br-data-spec
12416 @opindex mno-sched-br-data-spec
12417 @opindex msched-br-data-spec
12418 (Dis/En)able data speculative scheduling before reload.
12419 This will result in generation of the ld.a instructions and
12420 the corresponding check instructions (ld.c / chk.a).
12421 The default is 'disable'.
12423 @item -msched-ar-data-spec
12424 @itemx -mno-sched-ar-data-spec
12425 @opindex msched-ar-data-spec
12426 @opindex mno-sched-ar-data-spec
12427 (En/Dis)able data speculative scheduling after reload.
12428 This will result in generation of the ld.a instructions and
12429 the corresponding check instructions (ld.c / chk.a).
12430 The default is 'enable'.
12432 @item -mno-sched-control-spec
12433 @itemx -msched-control-spec
12434 @opindex mno-sched-control-spec
12435 @opindex msched-control-spec
12436 (Dis/En)able control speculative scheduling. This feature is
12437 available only during region scheduling (i.e.@: before reload).
12438 This will result in generation of the ld.s instructions and
12439 the corresponding check instructions chk.s .
12440 The default is 'disable'.
12442 @item -msched-br-in-data-spec
12443 @itemx -mno-sched-br-in-data-spec
12444 @opindex msched-br-in-data-spec
12445 @opindex mno-sched-br-in-data-spec
12446 (En/Dis)able speculative scheduling of the instructions that
12447 are dependent on the data speculative loads before reload.
12448 This is effective only with @option{-msched-br-data-spec} enabled.
12449 The default is 'enable'.
12451 @item -msched-ar-in-data-spec
12452 @itemx -mno-sched-ar-in-data-spec
12453 @opindex msched-ar-in-data-spec
12454 @opindex mno-sched-ar-in-data-spec
12455 (En/Dis)able speculative scheduling of the instructions that
12456 are dependent on the data speculative loads after reload.
12457 This is effective only with @option{-msched-ar-data-spec} enabled.
12458 The default is 'enable'.
12460 @item -msched-in-control-spec
12461 @itemx -mno-sched-in-control-spec
12462 @opindex msched-in-control-spec
12463 @opindex mno-sched-in-control-spec
12464 (En/Dis)able speculative scheduling of the instructions that
12465 are dependent on the control speculative loads.
12466 This is effective only with @option{-msched-control-spec} enabled.
12467 The default is 'enable'.
12469 @item -mno-sched-prefer-non-data-spec-insns
12470 @itemx -msched-prefer-non-data-spec-insns
12471 @opindex mno-sched-prefer-non-data-spec-insns
12472 @opindex msched-prefer-non-data-spec-insns
12473 If enabled, data speculative instructions will be chosen for schedule
12474 only if there are no other choices at the moment. This will make
12475 the use of the data speculation much more conservative.
12476 The default is 'disable'.
12478 @item -mno-sched-prefer-non-control-spec-insns
12479 @itemx -msched-prefer-non-control-spec-insns
12480 @opindex mno-sched-prefer-non-control-spec-insns
12481 @opindex msched-prefer-non-control-spec-insns
12482 If enabled, control speculative instructions will be chosen for schedule
12483 only if there are no other choices at the moment. This will make
12484 the use of the control speculation much more conservative.
12485 The default is 'disable'.
12487 @item -mno-sched-count-spec-in-critical-path
12488 @itemx -msched-count-spec-in-critical-path
12489 @opindex mno-sched-count-spec-in-critical-path
12490 @opindex msched-count-spec-in-critical-path
12491 If enabled, speculative dependencies will be considered during
12492 computation of the instructions priorities. This will make the use of the
12493 speculation a bit more conservative.
12494 The default is 'disable'.
12496 @item -msched-spec-ldc
12497 @opindex msched-spec-ldc
12498 Use a simple data speculation check. This option is on by default.
12500 @item -msched-control-spec-ldc
12501 @opindex msched-spec-ldc
12502 Use a simple check for control speculation. This option is on by default.
12504 @item -msched-stop-bits-after-every-cycle
12505 @opindex msched-stop-bits-after-every-cycle
12506 Place a stop bit after every cycle when scheduling. This option is on
12509 @item -msched-fp-mem-deps-zero-cost
12510 @opindex msched-fp-mem-deps-zero-cost
12511 Assume that floating-point stores and loads are not likely to cause a conflict
12512 when placed into the same instruction group. This option is disabled by
12515 @item -msel-sched-dont-check-control-spec
12516 @opindex msel-sched-dont-check-control-spec
12517 Generate checks for control speculation in selective scheduling.
12518 This flag is disabled by default.
12520 @item -msched-max-memory-insns=@var{max-insns}
12521 @opindex msched-max-memory-insns
12522 Limit on the number of memory insns per instruction group, giving lower
12523 priority to subsequent memory insns attempting to schedule in the same
12524 instruction group. Frequently useful to prevent cache bank conflicts.
12525 The default value is 1.
12527 @item -msched-max-memory-insns-hard-limit
12528 @opindex msched-max-memory-insns-hard-limit
12529 Disallow more than `msched-max-memory-insns' in instruction group.
12530 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12531 when limit is reached but may still schedule memory operations.
12535 @node IA-64/VMS Options
12536 @subsection IA-64/VMS Options
12538 These @samp{-m} options are defined for the IA-64/VMS implementations:
12541 @item -mvms-return-codes
12542 @opindex mvms-return-codes
12543 Return VMS condition codes from main. The default is to return POSIX
12544 style condition (e.g.@ error) codes.
12546 @item -mdebug-main=@var{prefix}
12547 @opindex mdebug-main=@var{prefix}
12548 Flag the first routine whose name starts with @var{prefix} as the main
12549 routine for the debugger.
12553 Default to 64bit memory allocation routines.
12557 @subsection M32C Options
12558 @cindex M32C options
12561 @item -mcpu=@var{name}
12563 Select the CPU for which code is generated. @var{name} may be one of
12564 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12565 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12566 the M32C/80 series.
12570 Specifies that the program will be run on the simulator. This causes
12571 an alternate runtime library to be linked in which supports, for
12572 example, file I/O@. You must not use this option when generating
12573 programs that will run on real hardware; you must provide your own
12574 runtime library for whatever I/O functions are needed.
12576 @item -memregs=@var{number}
12578 Specifies the number of memory-based pseudo-registers GCC will use
12579 during code generation. These pseudo-registers will be used like real
12580 registers, so there is a tradeoff between GCC's ability to fit the
12581 code into available registers, and the performance penalty of using
12582 memory instead of registers. Note that all modules in a program must
12583 be compiled with the same value for this option. Because of that, you
12584 must not use this option with the default runtime libraries gcc
12589 @node M32R/D Options
12590 @subsection M32R/D Options
12591 @cindex M32R/D options
12593 These @option{-m} options are defined for Renesas M32R/D architectures:
12598 Generate code for the M32R/2@.
12602 Generate code for the M32R/X@.
12606 Generate code for the M32R@. This is the default.
12608 @item -mmodel=small
12609 @opindex mmodel=small
12610 Assume all objects live in the lower 16MB of memory (so that their addresses
12611 can be loaded with the @code{ld24} instruction), and assume all subroutines
12612 are reachable with the @code{bl} instruction.
12613 This is the default.
12615 The addressability of a particular object can be set with the
12616 @code{model} attribute.
12618 @item -mmodel=medium
12619 @opindex mmodel=medium
12620 Assume objects may be anywhere in the 32-bit address space (the compiler
12621 will generate @code{seth/add3} instructions to load their addresses), and
12622 assume all subroutines are reachable with the @code{bl} instruction.
12624 @item -mmodel=large
12625 @opindex mmodel=large
12626 Assume objects may be anywhere in the 32-bit address space (the compiler
12627 will generate @code{seth/add3} instructions to load their addresses), and
12628 assume subroutines may not be reachable with the @code{bl} instruction
12629 (the compiler will generate the much slower @code{seth/add3/jl}
12630 instruction sequence).
12633 @opindex msdata=none
12634 Disable use of the small data area. Variables will be put into
12635 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12636 @code{section} attribute has been specified).
12637 This is the default.
12639 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12640 Objects may be explicitly put in the small data area with the
12641 @code{section} attribute using one of these sections.
12643 @item -msdata=sdata
12644 @opindex msdata=sdata
12645 Put small global and static data in the small data area, but do not
12646 generate special code to reference them.
12649 @opindex msdata=use
12650 Put small global and static data in the small data area, and generate
12651 special instructions to reference them.
12655 @cindex smaller data references
12656 Put global and static objects less than or equal to @var{num} bytes
12657 into the small data or bss sections instead of the normal data or bss
12658 sections. The default value of @var{num} is 8.
12659 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12660 for this option to have any effect.
12662 All modules should be compiled with the same @option{-G @var{num}} value.
12663 Compiling with different values of @var{num} may or may not work; if it
12664 doesn't the linker will give an error message---incorrect code will not be
12669 Makes the M32R specific code in the compiler display some statistics
12670 that might help in debugging programs.
12672 @item -malign-loops
12673 @opindex malign-loops
12674 Align all loops to a 32-byte boundary.
12676 @item -mno-align-loops
12677 @opindex mno-align-loops
12678 Do not enforce a 32-byte alignment for loops. This is the default.
12680 @item -missue-rate=@var{number}
12681 @opindex missue-rate=@var{number}
12682 Issue @var{number} instructions per cycle. @var{number} can only be 1
12685 @item -mbranch-cost=@var{number}
12686 @opindex mbranch-cost=@var{number}
12687 @var{number} can only be 1 or 2. If it is 1 then branches will be
12688 preferred over conditional code, if it is 2, then the opposite will
12691 @item -mflush-trap=@var{number}
12692 @opindex mflush-trap=@var{number}
12693 Specifies the trap number to use to flush the cache. The default is
12694 12. Valid numbers are between 0 and 15 inclusive.
12696 @item -mno-flush-trap
12697 @opindex mno-flush-trap
12698 Specifies that the cache cannot be flushed by using a trap.
12700 @item -mflush-func=@var{name}
12701 @opindex mflush-func=@var{name}
12702 Specifies the name of the operating system function to call to flush
12703 the cache. The default is @emph{_flush_cache}, but a function call
12704 will only be used if a trap is not available.
12706 @item -mno-flush-func
12707 @opindex mno-flush-func
12708 Indicates that there is no OS function for flushing the cache.
12712 @node M680x0 Options
12713 @subsection M680x0 Options
12714 @cindex M680x0 options
12716 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
12717 The default settings depend on which architecture was selected when
12718 the compiler was configured; the defaults for the most common choices
12722 @item -march=@var{arch}
12724 Generate code for a specific M680x0 or ColdFire instruction set
12725 architecture. Permissible values of @var{arch} for M680x0
12726 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
12727 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
12728 architectures are selected according to Freescale's ISA classification
12729 and the permissible values are: @samp{isaa}, @samp{isaaplus},
12730 @samp{isab} and @samp{isac}.
12732 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
12733 code for a ColdFire target. The @var{arch} in this macro is one of the
12734 @option{-march} arguments given above.
12736 When used together, @option{-march} and @option{-mtune} select code
12737 that runs on a family of similar processors but that is optimized
12738 for a particular microarchitecture.
12740 @item -mcpu=@var{cpu}
12742 Generate code for a specific M680x0 or ColdFire processor.
12743 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12744 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12745 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12746 below, which also classifies the CPUs into families:
12748 @multitable @columnfractions 0.20 0.80
12749 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12750 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
12751 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12752 @item @samp{5206e} @tab @samp{5206e}
12753 @item @samp{5208} @tab @samp{5207} @samp{5208}
12754 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12755 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12756 @item @samp{5216} @tab @samp{5214} @samp{5216}
12757 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12758 @item @samp{5225} @tab @samp{5224} @samp{5225}
12759 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
12760 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12761 @item @samp{5249} @tab @samp{5249}
12762 @item @samp{5250} @tab @samp{5250}
12763 @item @samp{5271} @tab @samp{5270} @samp{5271}
12764 @item @samp{5272} @tab @samp{5272}
12765 @item @samp{5275} @tab @samp{5274} @samp{5275}
12766 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12767 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
12768 @item @samp{5307} @tab @samp{5307}
12769 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12770 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12771 @item @samp{5407} @tab @samp{5407}
12772 @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}
12775 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12776 @var{arch} is compatible with @var{cpu}. Other combinations of
12777 @option{-mcpu} and @option{-march} are rejected.
12779 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12780 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12781 where the value of @var{family} is given by the table above.
12783 @item -mtune=@var{tune}
12785 Tune the code for a particular microarchitecture, within the
12786 constraints set by @option{-march} and @option{-mcpu}.
12787 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12788 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12789 and @samp{cpu32}. The ColdFire microarchitectures
12790 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12792 You can also use @option{-mtune=68020-40} for code that needs
12793 to run relatively well on 68020, 68030 and 68040 targets.
12794 @option{-mtune=68020-60} is similar but includes 68060 targets
12795 as well. These two options select the same tuning decisions as
12796 @option{-m68020-40} and @option{-m68020-60} respectively.
12798 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12799 when tuning for 680x0 architecture @var{arch}. It also defines
12800 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12801 option is used. If gcc is tuning for a range of architectures,
12802 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12803 it defines the macros for every architecture in the range.
12805 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12806 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12807 of the arguments given above.
12813 Generate output for a 68000. This is the default
12814 when the compiler is configured for 68000-based systems.
12815 It is equivalent to @option{-march=68000}.
12817 Use this option for microcontrollers with a 68000 or EC000 core,
12818 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12822 Generate output for a 68010. This is the default
12823 when the compiler is configured for 68010-based systems.
12824 It is equivalent to @option{-march=68010}.
12830 Generate output for a 68020. This is the default
12831 when the compiler is configured for 68020-based systems.
12832 It is equivalent to @option{-march=68020}.
12836 Generate output for a 68030. This is the default when the compiler is
12837 configured for 68030-based systems. It is equivalent to
12838 @option{-march=68030}.
12842 Generate output for a 68040. This is the default when the compiler is
12843 configured for 68040-based systems. It is equivalent to
12844 @option{-march=68040}.
12846 This option inhibits the use of 68881/68882 instructions that have to be
12847 emulated by software on the 68040. Use this option if your 68040 does not
12848 have code to emulate those instructions.
12852 Generate output for a 68060. This is the default when the compiler is
12853 configured for 68060-based systems. It is equivalent to
12854 @option{-march=68060}.
12856 This option inhibits the use of 68020 and 68881/68882 instructions that
12857 have to be emulated by software on the 68060. Use this option if your 68060
12858 does not have code to emulate those instructions.
12862 Generate output for a CPU32. This is the default
12863 when the compiler is configured for CPU32-based systems.
12864 It is equivalent to @option{-march=cpu32}.
12866 Use this option for microcontrollers with a
12867 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12868 68336, 68340, 68341, 68349 and 68360.
12872 Generate output for a 520X ColdFire CPU@. This is the default
12873 when the compiler is configured for 520X-based systems.
12874 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12875 in favor of that option.
12877 Use this option for microcontroller with a 5200 core, including
12878 the MCF5202, MCF5203, MCF5204 and MCF5206.
12882 Generate output for a 5206e ColdFire CPU@. The option is now
12883 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12887 Generate output for a member of the ColdFire 528X family.
12888 The option is now deprecated in favor of the equivalent
12889 @option{-mcpu=528x}.
12893 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
12894 in favor of the equivalent @option{-mcpu=5307}.
12898 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
12899 in favor of the equivalent @option{-mcpu=5407}.
12903 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12904 This includes use of hardware floating point instructions.
12905 The option is equivalent to @option{-mcpu=547x}, and is now
12906 deprecated in favor of that option.
12910 Generate output for a 68040, without using any of the new instructions.
12911 This results in code which can run relatively efficiently on either a
12912 68020/68881 or a 68030 or a 68040. The generated code does use the
12913 68881 instructions that are emulated on the 68040.
12915 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12919 Generate output for a 68060, without using any of the new instructions.
12920 This results in code which can run relatively efficiently on either a
12921 68020/68881 or a 68030 or a 68040. The generated code does use the
12922 68881 instructions that are emulated on the 68060.
12924 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12928 @opindex mhard-float
12930 Generate floating-point instructions. This is the default for 68020
12931 and above, and for ColdFire devices that have an FPU@. It defines the
12932 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12933 on ColdFire targets.
12936 @opindex msoft-float
12937 Do not generate floating-point instructions; use library calls instead.
12938 This is the default for 68000, 68010, and 68832 targets. It is also
12939 the default for ColdFire devices that have no FPU.
12945 Generate (do not generate) ColdFire hardware divide and remainder
12946 instructions. If @option{-march} is used without @option{-mcpu},
12947 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12948 architectures. Otherwise, the default is taken from the target CPU
12949 (either the default CPU, or the one specified by @option{-mcpu}). For
12950 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12951 @option{-mcpu=5206e}.
12953 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12957 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12958 Additionally, parameters passed on the stack are also aligned to a
12959 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12963 Do not consider type @code{int} to be 16 bits wide. This is the default.
12966 @itemx -mno-bitfield
12967 @opindex mnobitfield
12968 @opindex mno-bitfield
12969 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
12970 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12974 Do use the bit-field instructions. The @option{-m68020} option implies
12975 @option{-mbitfield}. This is the default if you use a configuration
12976 designed for a 68020.
12980 Use a different function-calling convention, in which functions
12981 that take a fixed number of arguments return with the @code{rtd}
12982 instruction, which pops their arguments while returning. This
12983 saves one instruction in the caller since there is no need to pop
12984 the arguments there.
12986 This calling convention is incompatible with the one normally
12987 used on Unix, so you cannot use it if you need to call libraries
12988 compiled with the Unix compiler.
12990 Also, you must provide function prototypes for all functions that
12991 take variable numbers of arguments (including @code{printf});
12992 otherwise incorrect code will be generated for calls to those
12995 In addition, seriously incorrect code will result if you call a
12996 function with too many arguments. (Normally, extra arguments are
12997 harmlessly ignored.)
12999 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13000 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13004 Do not use the calling conventions selected by @option{-mrtd}.
13005 This is the default.
13008 @itemx -mno-align-int
13009 @opindex malign-int
13010 @opindex mno-align-int
13011 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13012 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13013 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13014 Aligning variables on 32-bit boundaries produces code that runs somewhat
13015 faster on processors with 32-bit busses at the expense of more memory.
13017 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13018 align structures containing the above types differently than
13019 most published application binary interface specifications for the m68k.
13023 Use the pc-relative addressing mode of the 68000 directly, instead of
13024 using a global offset table. At present, this option implies @option{-fpic},
13025 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13026 not presently supported with @option{-mpcrel}, though this could be supported for
13027 68020 and higher processors.
13029 @item -mno-strict-align
13030 @itemx -mstrict-align
13031 @opindex mno-strict-align
13032 @opindex mstrict-align
13033 Do not (do) assume that unaligned memory references will be handled by
13037 Generate code that allows the data segment to be located in a different
13038 area of memory from the text segment. This allows for execute in place in
13039 an environment without virtual memory management. This option implies
13042 @item -mno-sep-data
13043 Generate code that assumes that the data segment follows the text segment.
13044 This is the default.
13046 @item -mid-shared-library
13047 Generate code that supports shared libraries via the library ID method.
13048 This allows for execute in place and shared libraries in an environment
13049 without virtual memory management. This option implies @option{-fPIC}.
13051 @item -mno-id-shared-library
13052 Generate code that doesn't assume ID based shared libraries are being used.
13053 This is the default.
13055 @item -mshared-library-id=n
13056 Specified the identification number of the ID based shared library being
13057 compiled. Specifying a value of 0 will generate more compact code, specifying
13058 other values will force the allocation of that number to the current
13059 library but is no more space or time efficient than omitting this option.
13065 When generating position-independent code for ColdFire, generate code
13066 that works if the GOT has more than 8192 entries. This code is
13067 larger and slower than code generated without this option. On M680x0
13068 processors, this option is not needed; @option{-fPIC} suffices.
13070 GCC normally uses a single instruction to load values from the GOT@.
13071 While this is relatively efficient, it only works if the GOT
13072 is smaller than about 64k. Anything larger causes the linker
13073 to report an error such as:
13075 @cindex relocation truncated to fit (ColdFire)
13077 relocation truncated to fit: R_68K_GOT16O foobar
13080 If this happens, you should recompile your code with @option{-mxgot}.
13081 It should then work with very large GOTs. However, code generated with
13082 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13083 the value of a global symbol.
13085 Note that some linkers, including newer versions of the GNU linker,
13086 can create multiple GOTs and sort GOT entries. If you have such a linker,
13087 you should only need to use @option{-mxgot} when compiling a single
13088 object file that accesses more than 8192 GOT entries. Very few do.
13090 These options have no effect unless GCC is generating
13091 position-independent code.
13095 @node M68hc1x Options
13096 @subsection M68hc1x Options
13097 @cindex M68hc1x options
13099 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13100 microcontrollers. The default values for these options depends on
13101 which style of microcontroller was selected when the compiler was configured;
13102 the defaults for the most common choices are given below.
13109 Generate output for a 68HC11. This is the default
13110 when the compiler is configured for 68HC11-based systems.
13116 Generate output for a 68HC12. This is the default
13117 when the compiler is configured for 68HC12-based systems.
13123 Generate output for a 68HCS12.
13125 @item -mauto-incdec
13126 @opindex mauto-incdec
13127 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13134 Enable the use of 68HC12 min and max instructions.
13137 @itemx -mno-long-calls
13138 @opindex mlong-calls
13139 @opindex mno-long-calls
13140 Treat all calls as being far away (near). If calls are assumed to be
13141 far away, the compiler will use the @code{call} instruction to
13142 call a function and the @code{rtc} instruction for returning.
13146 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13148 @item -msoft-reg-count=@var{count}
13149 @opindex msoft-reg-count
13150 Specify the number of pseudo-soft registers which are used for the
13151 code generation. The maximum number is 32. Using more pseudo-soft
13152 register may or may not result in better code depending on the program.
13153 The default is 4 for 68HC11 and 2 for 68HC12.
13157 @node MCore Options
13158 @subsection MCore Options
13159 @cindex MCore options
13161 These are the @samp{-m} options defined for the Motorola M*Core
13167 @itemx -mno-hardlit
13169 @opindex mno-hardlit
13170 Inline constants into the code stream if it can be done in two
13171 instructions or less.
13177 Use the divide instruction. (Enabled by default).
13179 @item -mrelax-immediate
13180 @itemx -mno-relax-immediate
13181 @opindex mrelax-immediate
13182 @opindex mno-relax-immediate
13183 Allow arbitrary sized immediates in bit operations.
13185 @item -mwide-bitfields
13186 @itemx -mno-wide-bitfields
13187 @opindex mwide-bitfields
13188 @opindex mno-wide-bitfields
13189 Always treat bit-fields as int-sized.
13191 @item -m4byte-functions
13192 @itemx -mno-4byte-functions
13193 @opindex m4byte-functions
13194 @opindex mno-4byte-functions
13195 Force all functions to be aligned to a four byte boundary.
13197 @item -mcallgraph-data
13198 @itemx -mno-callgraph-data
13199 @opindex mcallgraph-data
13200 @opindex mno-callgraph-data
13201 Emit callgraph information.
13204 @itemx -mno-slow-bytes
13205 @opindex mslow-bytes
13206 @opindex mno-slow-bytes
13207 Prefer word access when reading byte quantities.
13209 @item -mlittle-endian
13210 @itemx -mbig-endian
13211 @opindex mlittle-endian
13212 @opindex mbig-endian
13213 Generate code for a little endian target.
13219 Generate code for the 210 processor.
13223 Assume that run-time support has been provided and so omit the
13224 simulator library (@file{libsim.a)} from the linker command line.
13226 @item -mstack-increment=@var{size}
13227 @opindex mstack-increment
13228 Set the maximum amount for a single stack increment operation. Large
13229 values can increase the speed of programs which contain functions
13230 that need a large amount of stack space, but they can also trigger a
13231 segmentation fault if the stack is extended too much. The default
13237 @subsection MeP Options
13238 @cindex MeP options
13244 Enables the @code{abs} instruction, which is the absolute difference
13245 between two registers.
13249 Enables all the optional instructions - average, multiply, divide, bit
13250 operations, leading zero, absolute difference, min/max, clip, and
13256 Enables the @code{ave} instruction, which computes the average of two
13259 @item -mbased=@var{n}
13261 Variables of size @var{n} bytes or smaller will be placed in the
13262 @code{.based} section by default. Based variables use the @code{$tp}
13263 register as a base register, and there is a 128 byte limit to the
13264 @code{.based} section.
13268 Enables the bit operation instructions - bit test (@code{btstm}), set
13269 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13270 test-and-set (@code{tas}).
13272 @item -mc=@var{name}
13274 Selects which section constant data will be placed in. @var{name} may
13275 be @code{tiny}, @code{near}, or @code{far}.
13279 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13280 useful unless you also provide @code{-mminmax}.
13282 @item -mconfig=@var{name}
13284 Selects one of the build-in core configurations. Each MeP chip has
13285 one or more modules in it; each module has a core CPU and a variety of
13286 coprocessors, optional instructions, and peripherals. The
13287 @code{MeP-Integrator} tool, not part of GCC, provides these
13288 configurations through this option; using this option is the same as
13289 using all the corresponding command line options. The default
13290 configuration is @code{default}.
13294 Enables the coprocessor instructions. By default, this is a 32-bit
13295 coprocessor. Note that the coprocessor is normally enabled via the
13296 @code{-mconfig=} option.
13300 Enables the 32-bit coprocessor's instructions.
13304 Enables the 64-bit coprocessor's instructions.
13308 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13312 Causes constant variables to be placed in the @code{.near} section.
13316 Enables the @code{div} and @code{divu} instructions.
13320 Generate big-endian code.
13324 Generate little-endian code.
13326 @item -mio-volatile
13327 @opindex mio-volatile
13328 Tells the compiler that any variable marked with the @code{io}
13329 attribute is to be considered volatile.
13333 Causes variables to be assigned to the @code{.far} section by default.
13337 Enables the @code{leadz} (leading zero) instruction.
13341 Causes variables to be assigned to the @code{.near} section by default.
13345 Enables the @code{min} and @code{max} instructions.
13349 Enables the multiplication and multiply-accumulate instructions.
13353 Disables all the optional instructions enabled by @code{-mall-opts}.
13357 Enables the @code{repeat} and @code{erepeat} instructions, used for
13358 low-overhead looping.
13362 Causes all variables to default to the @code{.tiny} section. Note
13363 that there is a 65536 byte limit to this section. Accesses to these
13364 variables use the @code{%gp} base register.
13368 Enables the saturation instructions. Note that the compiler does not
13369 currently generate these itself, but this option is included for
13370 compatibility with other tools, like @code{as}.
13374 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13378 Link the simulator runtime libraries.
13382 Link the simulator runtime libraries, excluding built-in support
13383 for reset and exception vectors and tables.
13387 Causes all functions to default to the @code{.far} section. Without
13388 this option, functions default to the @code{.near} section.
13390 @item -mtiny=@var{n}
13392 Variables that are @var{n} bytes or smaller will be allocated to the
13393 @code{.tiny} section. These variables use the @code{$gp} base
13394 register. The default for this option is 4, but note that there's a
13395 65536 byte limit to the @code{.tiny} section.
13400 @subsection MIPS Options
13401 @cindex MIPS options
13407 Generate big-endian code.
13411 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13414 @item -march=@var{arch}
13416 Generate code that will run on @var{arch}, which can be the name of a
13417 generic MIPS ISA, or the name of a particular processor.
13419 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13420 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13421 The processor names are:
13422 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13423 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13424 @samp{5kc}, @samp{5kf},
13426 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13427 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13428 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13429 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13430 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13431 @samp{loongson2e}, @samp{loongson2f},
13435 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13436 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13437 @samp{rm7000}, @samp{rm9000},
13438 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13441 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13442 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13444 The special value @samp{from-abi} selects the
13445 most compatible architecture for the selected ABI (that is,
13446 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13448 Native Linux/GNU toolchains also support the value @samp{native},
13449 which selects the best architecture option for the host processor.
13450 @option{-march=native} has no effect if GCC does not recognize
13453 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13454 (for example, @samp{-march=r2k}). Prefixes are optional, and
13455 @samp{vr} may be written @samp{r}.
13457 Names of the form @samp{@var{n}f2_1} refer to processors with
13458 FPUs clocked at half the rate of the core, names of the form
13459 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13460 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13461 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13462 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13463 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13464 accepted as synonyms for @samp{@var{n}f1_1}.
13466 GCC defines two macros based on the value of this option. The first
13467 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13468 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13469 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13470 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13471 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13473 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13474 above. In other words, it will have the full prefix and will not
13475 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13476 the macro names the resolved architecture (either @samp{"mips1"} or
13477 @samp{"mips3"}). It names the default architecture when no
13478 @option{-march} option is given.
13480 @item -mtune=@var{arch}
13482 Optimize for @var{arch}. Among other things, this option controls
13483 the way instructions are scheduled, and the perceived cost of arithmetic
13484 operations. The list of @var{arch} values is the same as for
13487 When this option is not used, GCC will optimize for the processor
13488 specified by @option{-march}. By using @option{-march} and
13489 @option{-mtune} together, it is possible to generate code that will
13490 run on a family of processors, but optimize the code for one
13491 particular member of that family.
13493 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13494 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13495 @samp{-march} ones described above.
13499 Equivalent to @samp{-march=mips1}.
13503 Equivalent to @samp{-march=mips2}.
13507 Equivalent to @samp{-march=mips3}.
13511 Equivalent to @samp{-march=mips4}.
13515 Equivalent to @samp{-march=mips32}.
13519 Equivalent to @samp{-march=mips32r2}.
13523 Equivalent to @samp{-march=mips64}.
13527 Equivalent to @samp{-march=mips64r2}.
13532 @opindex mno-mips16
13533 Generate (do not generate) MIPS16 code. If GCC is targetting a
13534 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13536 MIPS16 code generation can also be controlled on a per-function basis
13537 by means of @code{mips16} and @code{nomips16} attributes.
13538 @xref{Function Attributes}, for more information.
13540 @item -mflip-mips16
13541 @opindex mflip-mips16
13542 Generate MIPS16 code on alternating functions. This option is provided
13543 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13544 not intended for ordinary use in compiling user code.
13546 @item -minterlink-mips16
13547 @itemx -mno-interlink-mips16
13548 @opindex minterlink-mips16
13549 @opindex mno-interlink-mips16
13550 Require (do not require) that non-MIPS16 code be link-compatible with
13553 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13554 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13555 therefore disables direct jumps unless GCC knows that the target of the
13556 jump is not MIPS16.
13568 Generate code for the given ABI@.
13570 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13571 generates 64-bit code when you select a 64-bit architecture, but you
13572 can use @option{-mgp32} to get 32-bit code instead.
13574 For information about the O64 ABI, see
13575 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13577 GCC supports a variant of the o32 ABI in which floating-point registers
13578 are 64 rather than 32 bits wide. You can select this combination with
13579 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13580 and @samp{mfhc1} instructions and is therefore only supported for
13581 MIPS32R2 processors.
13583 The register assignments for arguments and return values remain the
13584 same, but each scalar value is passed in a single 64-bit register
13585 rather than a pair of 32-bit registers. For example, scalar
13586 floating-point values are returned in @samp{$f0} only, not a
13587 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13588 remains the same, but all 64 bits are saved.
13591 @itemx -mno-abicalls
13593 @opindex mno-abicalls
13594 Generate (do not generate) code that is suitable for SVR4-style
13595 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13600 Generate (do not generate) code that is fully position-independent,
13601 and that can therefore be linked into shared libraries. This option
13602 only affects @option{-mabicalls}.
13604 All @option{-mabicalls} code has traditionally been position-independent,
13605 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13606 as an extension, the GNU toolchain allows executables to use absolute
13607 accesses for locally-binding symbols. It can also use shorter GP
13608 initialization sequences and generate direct calls to locally-defined
13609 functions. This mode is selected by @option{-mno-shared}.
13611 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13612 objects that can only be linked by the GNU linker. However, the option
13613 does not affect the ABI of the final executable; it only affects the ABI
13614 of relocatable objects. Using @option{-mno-shared} will generally make
13615 executables both smaller and quicker.
13617 @option{-mshared} is the default.
13623 Assume (do not assume) that the static and dynamic linkers
13624 support PLTs and copy relocations. This option only affects
13625 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13626 has no effect without @samp{-msym32}.
13628 You can make @option{-mplt} the default by configuring
13629 GCC with @option{--with-mips-plt}. The default is
13630 @option{-mno-plt} otherwise.
13636 Lift (do not lift) the usual restrictions on the size of the global
13639 GCC normally uses a single instruction to load values from the GOT@.
13640 While this is relatively efficient, it will only work if the GOT
13641 is smaller than about 64k. Anything larger will cause the linker
13642 to report an error such as:
13644 @cindex relocation truncated to fit (MIPS)
13646 relocation truncated to fit: R_MIPS_GOT16 foobar
13649 If this happens, you should recompile your code with @option{-mxgot}.
13650 It should then work with very large GOTs, although it will also be
13651 less efficient, since it will take three instructions to fetch the
13652 value of a global symbol.
13654 Note that some linkers can create multiple GOTs. If you have such a
13655 linker, you should only need to use @option{-mxgot} when a single object
13656 file accesses more than 64k's worth of GOT entries. Very few do.
13658 These options have no effect unless GCC is generating position
13663 Assume that general-purpose registers are 32 bits wide.
13667 Assume that general-purpose registers are 64 bits wide.
13671 Assume that floating-point registers are 32 bits wide.
13675 Assume that floating-point registers are 64 bits wide.
13678 @opindex mhard-float
13679 Use floating-point coprocessor instructions.
13682 @opindex msoft-float
13683 Do not use floating-point coprocessor instructions. Implement
13684 floating-point calculations using library calls instead.
13686 @item -msingle-float
13687 @opindex msingle-float
13688 Assume that the floating-point coprocessor only supports single-precision
13691 @item -mdouble-float
13692 @opindex mdouble-float
13693 Assume that the floating-point coprocessor supports double-precision
13694 operations. This is the default.
13700 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13701 implement atomic memory built-in functions. When neither option is
13702 specified, GCC will use the instructions if the target architecture
13705 @option{-mllsc} is useful if the runtime environment can emulate the
13706 instructions and @option{-mno-llsc} can be useful when compiling for
13707 nonstandard ISAs. You can make either option the default by
13708 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
13709 respectively. @option{--with-llsc} is the default for some
13710 configurations; see the installation documentation for details.
13716 Use (do not use) revision 1 of the MIPS DSP ASE@.
13717 @xref{MIPS DSP Built-in Functions}. This option defines the
13718 preprocessor macro @samp{__mips_dsp}. It also defines
13719 @samp{__mips_dsp_rev} to 1.
13725 Use (do not use) revision 2 of the MIPS DSP ASE@.
13726 @xref{MIPS DSP Built-in Functions}. This option defines the
13727 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
13728 It also defines @samp{__mips_dsp_rev} to 2.
13731 @itemx -mno-smartmips
13732 @opindex msmartmips
13733 @opindex mno-smartmips
13734 Use (do not use) the MIPS SmartMIPS ASE.
13736 @item -mpaired-single
13737 @itemx -mno-paired-single
13738 @opindex mpaired-single
13739 @opindex mno-paired-single
13740 Use (do not use) paired-single floating-point instructions.
13741 @xref{MIPS Paired-Single Support}. This option requires
13742 hardware floating-point support to be enabled.
13748 Use (do not use) MIPS Digital Media Extension instructions.
13749 This option can only be used when generating 64-bit code and requires
13750 hardware floating-point support to be enabled.
13755 @opindex mno-mips3d
13756 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
13757 The option @option{-mips3d} implies @option{-mpaired-single}.
13763 Use (do not use) MT Multithreading instructions.
13767 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
13768 an explanation of the default and the way that the pointer size is
13773 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
13775 The default size of @code{int}s, @code{long}s and pointers depends on
13776 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
13777 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
13778 32-bit @code{long}s. Pointers are the same size as @code{long}s,
13779 or the same size as integer registers, whichever is smaller.
13785 Assume (do not assume) that all symbols have 32-bit values, regardless
13786 of the selected ABI@. This option is useful in combination with
13787 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
13788 to generate shorter and faster references to symbolic addresses.
13792 Put definitions of externally-visible data in a small data section
13793 if that data is no bigger than @var{num} bytes. GCC can then access
13794 the data more efficiently; see @option{-mgpopt} for details.
13796 The default @option{-G} option depends on the configuration.
13798 @item -mlocal-sdata
13799 @itemx -mno-local-sdata
13800 @opindex mlocal-sdata
13801 @opindex mno-local-sdata
13802 Extend (do not extend) the @option{-G} behavior to local data too,
13803 such as to static variables in C@. @option{-mlocal-sdata} is the
13804 default for all configurations.
13806 If the linker complains that an application is using too much small data,
13807 you might want to try rebuilding the less performance-critical parts with
13808 @option{-mno-local-sdata}. You might also want to build large
13809 libraries with @option{-mno-local-sdata}, so that the libraries leave
13810 more room for the main program.
13812 @item -mextern-sdata
13813 @itemx -mno-extern-sdata
13814 @opindex mextern-sdata
13815 @opindex mno-extern-sdata
13816 Assume (do not assume) that externally-defined data will be in
13817 a small data section if that data is within the @option{-G} limit.
13818 @option{-mextern-sdata} is the default for all configurations.
13820 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
13821 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
13822 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
13823 is placed in a small data section. If @var{Var} is defined by another
13824 module, you must either compile that module with a high-enough
13825 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
13826 definition. If @var{Var} is common, you must link the application
13827 with a high-enough @option{-G} setting.
13829 The easiest way of satisfying these restrictions is to compile
13830 and link every module with the same @option{-G} option. However,
13831 you may wish to build a library that supports several different
13832 small data limits. You can do this by compiling the library with
13833 the highest supported @option{-G} setting and additionally using
13834 @option{-mno-extern-sdata} to stop the library from making assumptions
13835 about externally-defined data.
13841 Use (do not use) GP-relative accesses for symbols that are known to be
13842 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
13843 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
13846 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
13847 might not hold the value of @code{_gp}. For example, if the code is
13848 part of a library that might be used in a boot monitor, programs that
13849 call boot monitor routines will pass an unknown value in @code{$gp}.
13850 (In such situations, the boot monitor itself would usually be compiled
13851 with @option{-G0}.)
13853 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
13854 @option{-mno-extern-sdata}.
13856 @item -membedded-data
13857 @itemx -mno-embedded-data
13858 @opindex membedded-data
13859 @opindex mno-embedded-data
13860 Allocate variables to the read-only data section first if possible, then
13861 next in the small data section if possible, otherwise in data. This gives
13862 slightly slower code than the default, but reduces the amount of RAM required
13863 when executing, and thus may be preferred for some embedded systems.
13865 @item -muninit-const-in-rodata
13866 @itemx -mno-uninit-const-in-rodata
13867 @opindex muninit-const-in-rodata
13868 @opindex mno-uninit-const-in-rodata
13869 Put uninitialized @code{const} variables in the read-only data section.
13870 This option is only meaningful in conjunction with @option{-membedded-data}.
13872 @item -mcode-readable=@var{setting}
13873 @opindex mcode-readable
13874 Specify whether GCC may generate code that reads from executable sections.
13875 There are three possible settings:
13878 @item -mcode-readable=yes
13879 Instructions may freely access executable sections. This is the
13882 @item -mcode-readable=pcrel
13883 MIPS16 PC-relative load instructions can access executable sections,
13884 but other instructions must not do so. This option is useful on 4KSc
13885 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
13886 It is also useful on processors that can be configured to have a dual
13887 instruction/data SRAM interface and that, like the M4K, automatically
13888 redirect PC-relative loads to the instruction RAM.
13890 @item -mcode-readable=no
13891 Instructions must not access executable sections. This option can be
13892 useful on targets that are configured to have a dual instruction/data
13893 SRAM interface but that (unlike the M4K) do not automatically redirect
13894 PC-relative loads to the instruction RAM.
13897 @item -msplit-addresses
13898 @itemx -mno-split-addresses
13899 @opindex msplit-addresses
13900 @opindex mno-split-addresses
13901 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
13902 relocation operators. This option has been superseded by
13903 @option{-mexplicit-relocs} but is retained for backwards compatibility.
13905 @item -mexplicit-relocs
13906 @itemx -mno-explicit-relocs
13907 @opindex mexplicit-relocs
13908 @opindex mno-explicit-relocs
13909 Use (do not use) assembler relocation operators when dealing with symbolic
13910 addresses. The alternative, selected by @option{-mno-explicit-relocs},
13911 is to use assembler macros instead.
13913 @option{-mexplicit-relocs} is the default if GCC was configured
13914 to use an assembler that supports relocation operators.
13916 @item -mcheck-zero-division
13917 @itemx -mno-check-zero-division
13918 @opindex mcheck-zero-division
13919 @opindex mno-check-zero-division
13920 Trap (do not trap) on integer division by zero.
13922 The default is @option{-mcheck-zero-division}.
13924 @item -mdivide-traps
13925 @itemx -mdivide-breaks
13926 @opindex mdivide-traps
13927 @opindex mdivide-breaks
13928 MIPS systems check for division by zero by generating either a
13929 conditional trap or a break instruction. Using traps results in
13930 smaller code, but is only supported on MIPS II and later. Also, some
13931 versions of the Linux kernel have a bug that prevents trap from
13932 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
13933 allow conditional traps on architectures that support them and
13934 @option{-mdivide-breaks} to force the use of breaks.
13936 The default is usually @option{-mdivide-traps}, but this can be
13937 overridden at configure time using @option{--with-divide=breaks}.
13938 Divide-by-zero checks can be completely disabled using
13939 @option{-mno-check-zero-division}.
13944 @opindex mno-memcpy
13945 Force (do not force) the use of @code{memcpy()} for non-trivial block
13946 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
13947 most constant-sized copies.
13950 @itemx -mno-long-calls
13951 @opindex mlong-calls
13952 @opindex mno-long-calls
13953 Disable (do not disable) use of the @code{jal} instruction. Calling
13954 functions using @code{jal} is more efficient but requires the caller
13955 and callee to be in the same 256 megabyte segment.
13957 This option has no effect on abicalls code. The default is
13958 @option{-mno-long-calls}.
13964 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
13965 instructions, as provided by the R4650 ISA@.
13968 @itemx -mno-fused-madd
13969 @opindex mfused-madd
13970 @opindex mno-fused-madd
13971 Enable (disable) use of the floating point multiply-accumulate
13972 instructions, when they are available. The default is
13973 @option{-mfused-madd}.
13975 When multiply-accumulate instructions are used, the intermediate
13976 product is calculated to infinite precision and is not subject to
13977 the FCSR Flush to Zero bit. This may be undesirable in some
13982 Tell the MIPS assembler to not run its preprocessor over user
13983 assembler files (with a @samp{.s} suffix) when assembling them.
13986 @itemx -mno-fix-r4000
13987 @opindex mfix-r4000
13988 @opindex mno-fix-r4000
13989 Work around certain R4000 CPU errata:
13992 A double-word or a variable shift may give an incorrect result if executed
13993 immediately after starting an integer division.
13995 A double-word or a variable shift may give an incorrect result if executed
13996 while an integer multiplication is in progress.
13998 An integer division may give an incorrect result if started in a delay slot
13999 of a taken branch or a jump.
14003 @itemx -mno-fix-r4400
14004 @opindex mfix-r4400
14005 @opindex mno-fix-r4400
14006 Work around certain R4400 CPU errata:
14009 A double-word or a variable shift may give an incorrect result if executed
14010 immediately after starting an integer division.
14014 @itemx -mno-fix-r10000
14015 @opindex mfix-r10000
14016 @opindex mno-fix-r10000
14017 Work around certain R10000 errata:
14020 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14021 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14024 This option can only be used if the target architecture supports
14025 branch-likely instructions. @option{-mfix-r10000} is the default when
14026 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14030 @itemx -mno-fix-vr4120
14031 @opindex mfix-vr4120
14032 Work around certain VR4120 errata:
14035 @code{dmultu} does not always produce the correct result.
14037 @code{div} and @code{ddiv} do not always produce the correct result if one
14038 of the operands is negative.
14040 The workarounds for the division errata rely on special functions in
14041 @file{libgcc.a}. At present, these functions are only provided by
14042 the @code{mips64vr*-elf} configurations.
14044 Other VR4120 errata require a nop to be inserted between certain pairs of
14045 instructions. These errata are handled by the assembler, not by GCC itself.
14048 @opindex mfix-vr4130
14049 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14050 workarounds are implemented by the assembler rather than by GCC,
14051 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14052 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14053 instructions are available instead.
14056 @itemx -mno-fix-sb1
14058 Work around certain SB-1 CPU core errata.
14059 (This flag currently works around the SB-1 revision 2
14060 ``F1'' and ``F2'' floating point errata.)
14062 @item -mr10k-cache-barrier=@var{setting}
14063 @opindex mr10k-cache-barrier
14064 Specify whether GCC should insert cache barriers to avoid the
14065 side-effects of speculation on R10K processors.
14067 In common with many processors, the R10K tries to predict the outcome
14068 of a conditional branch and speculatively executes instructions from
14069 the ``taken'' branch. It later aborts these instructions if the
14070 predicted outcome was wrong. However, on the R10K, even aborted
14071 instructions can have side effects.
14073 This problem only affects kernel stores and, depending on the system,
14074 kernel loads. As an example, a speculatively-executed store may load
14075 the target memory into cache and mark the cache line as dirty, even if
14076 the store itself is later aborted. If a DMA operation writes to the
14077 same area of memory before the ``dirty'' line is flushed, the cached
14078 data will overwrite the DMA-ed data. See the R10K processor manual
14079 for a full description, including other potential problems.
14081 One workaround is to insert cache barrier instructions before every memory
14082 access that might be speculatively executed and that might have side
14083 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14084 controls GCC's implementation of this workaround. It assumes that
14085 aborted accesses to any byte in the following regions will not have
14090 the memory occupied by the current function's stack frame;
14093 the memory occupied by an incoming stack argument;
14096 the memory occupied by an object with a link-time-constant address.
14099 It is the kernel's responsibility to ensure that speculative
14100 accesses to these regions are indeed safe.
14102 If the input program contains a function declaration such as:
14108 then the implementation of @code{foo} must allow @code{j foo} and
14109 @code{jal foo} to be executed speculatively. GCC honors this
14110 restriction for functions it compiles itself. It expects non-GCC
14111 functions (such as hand-written assembly code) to do the same.
14113 The option has three forms:
14116 @item -mr10k-cache-barrier=load-store
14117 Insert a cache barrier before a load or store that might be
14118 speculatively executed and that might have side effects even
14121 @item -mr10k-cache-barrier=store
14122 Insert a cache barrier before a store that might be speculatively
14123 executed and that might have side effects even if aborted.
14125 @item -mr10k-cache-barrier=none
14126 Disable the insertion of cache barriers. This is the default setting.
14129 @item -mflush-func=@var{func}
14130 @itemx -mno-flush-func
14131 @opindex mflush-func
14132 Specifies the function to call to flush the I and D caches, or to not
14133 call any such function. If called, the function must take the same
14134 arguments as the common @code{_flush_func()}, that is, the address of the
14135 memory range for which the cache is being flushed, the size of the
14136 memory range, and the number 3 (to flush both caches). The default
14137 depends on the target GCC was configured for, but commonly is either
14138 @samp{_flush_func} or @samp{__cpu_flush}.
14140 @item mbranch-cost=@var{num}
14141 @opindex mbranch-cost
14142 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14143 This cost is only a heuristic and is not guaranteed to produce
14144 consistent results across releases. A zero cost redundantly selects
14145 the default, which is based on the @option{-mtune} setting.
14147 @item -mbranch-likely
14148 @itemx -mno-branch-likely
14149 @opindex mbranch-likely
14150 @opindex mno-branch-likely
14151 Enable or disable use of Branch Likely instructions, regardless of the
14152 default for the selected architecture. By default, Branch Likely
14153 instructions may be generated if they are supported by the selected
14154 architecture. An exception is for the MIPS32 and MIPS64 architectures
14155 and processors which implement those architectures; for those, Branch
14156 Likely instructions will not be generated by default because the MIPS32
14157 and MIPS64 architectures specifically deprecate their use.
14159 @item -mfp-exceptions
14160 @itemx -mno-fp-exceptions
14161 @opindex mfp-exceptions
14162 Specifies whether FP exceptions are enabled. This affects how we schedule
14163 FP instructions for some processors. The default is that FP exceptions are
14166 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14167 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14170 @item -mvr4130-align
14171 @itemx -mno-vr4130-align
14172 @opindex mvr4130-align
14173 The VR4130 pipeline is two-way superscalar, but can only issue two
14174 instructions together if the first one is 8-byte aligned. When this
14175 option is enabled, GCC will align pairs of instructions that it
14176 thinks should execute in parallel.
14178 This option only has an effect when optimizing for the VR4130.
14179 It normally makes code faster, but at the expense of making it bigger.
14180 It is enabled by default at optimization level @option{-O3}.
14185 Enable (disable) generation of @code{synci} instructions on
14186 architectures that support it. The @code{synci} instructions (if
14187 enabled) will be generated when @code{__builtin___clear_cache()} is
14190 This option defaults to @code{-mno-synci}, but the default can be
14191 overridden by configuring with @code{--with-synci}.
14193 When compiling code for single processor systems, it is generally safe
14194 to use @code{synci}. However, on many multi-core (SMP) systems, it
14195 will not invalidate the instruction caches on all cores and may lead
14196 to undefined behavior.
14198 @item -mrelax-pic-calls
14199 @itemx -mno-relax-pic-calls
14200 @opindex mrelax-pic-calls
14201 Try to turn PIC calls that are normally dispatched via register
14202 @code{$25} into direct calls. This is only possible if the linker can
14203 resolve the destination at link-time and if the destination is within
14204 range for a direct call.
14206 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14207 an assembler and a linker that supports the @code{.reloc} assembly
14208 directive and @code{-mexplicit-relocs} is in effect. With
14209 @code{-mno-explicit-relocs}, this optimization can be performed by the
14210 assembler and the linker alone without help from the compiler.
14212 @item -mmcount-ra-address
14213 @itemx -mno-mcount-ra-address
14214 @opindex mmcount-ra-address
14215 @opindex mno-mcount-ra-address
14216 Emit (do not emit) code that allows @code{_mcount} to modify the
14217 colling function's return address. When enabled, this option extends
14218 the usual @code{_mcount} interface with a new @var{ra-address}
14219 parameter, which has type @code{intptr_t *} and is passed in register
14220 @code{$12}. @code{_mcount} can then modify the return address by
14221 doing both of the following:
14224 Returning the new address in register @code{$31}.
14226 Storing the new address in @code{*@var{ra-address}},
14227 if @var{ra-address} is nonnull.
14230 The default is @option{-mno-mcount-ra-address}.
14235 @subsection MMIX Options
14236 @cindex MMIX Options
14238 These options are defined for the MMIX:
14242 @itemx -mno-libfuncs
14244 @opindex mno-libfuncs
14245 Specify that intrinsic library functions are being compiled, passing all
14246 values in registers, no matter the size.
14249 @itemx -mno-epsilon
14251 @opindex mno-epsilon
14252 Generate floating-point comparison instructions that compare with respect
14253 to the @code{rE} epsilon register.
14255 @item -mabi=mmixware
14257 @opindex mabi=mmixware
14259 Generate code that passes function parameters and return values that (in
14260 the called function) are seen as registers @code{$0} and up, as opposed to
14261 the GNU ABI which uses global registers @code{$231} and up.
14263 @item -mzero-extend
14264 @itemx -mno-zero-extend
14265 @opindex mzero-extend
14266 @opindex mno-zero-extend
14267 When reading data from memory in sizes shorter than 64 bits, use (do not
14268 use) zero-extending load instructions by default, rather than
14269 sign-extending ones.
14272 @itemx -mno-knuthdiv
14274 @opindex mno-knuthdiv
14275 Make the result of a division yielding a remainder have the same sign as
14276 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14277 remainder follows the sign of the dividend. Both methods are
14278 arithmetically valid, the latter being almost exclusively used.
14280 @item -mtoplevel-symbols
14281 @itemx -mno-toplevel-symbols
14282 @opindex mtoplevel-symbols
14283 @opindex mno-toplevel-symbols
14284 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14285 code can be used with the @code{PREFIX} assembly directive.
14289 Generate an executable in the ELF format, rather than the default
14290 @samp{mmo} format used by the @command{mmix} simulator.
14292 @item -mbranch-predict
14293 @itemx -mno-branch-predict
14294 @opindex mbranch-predict
14295 @opindex mno-branch-predict
14296 Use (do not use) the probable-branch instructions, when static branch
14297 prediction indicates a probable branch.
14299 @item -mbase-addresses
14300 @itemx -mno-base-addresses
14301 @opindex mbase-addresses
14302 @opindex mno-base-addresses
14303 Generate (do not generate) code that uses @emph{base addresses}. Using a
14304 base address automatically generates a request (handled by the assembler
14305 and the linker) for a constant to be set up in a global register. The
14306 register is used for one or more base address requests within the range 0
14307 to 255 from the value held in the register. The generally leads to short
14308 and fast code, but the number of different data items that can be
14309 addressed is limited. This means that a program that uses lots of static
14310 data may require @option{-mno-base-addresses}.
14312 @item -msingle-exit
14313 @itemx -mno-single-exit
14314 @opindex msingle-exit
14315 @opindex mno-single-exit
14316 Force (do not force) generated code to have a single exit point in each
14320 @node MN10300 Options
14321 @subsection MN10300 Options
14322 @cindex MN10300 options
14324 These @option{-m} options are defined for Matsushita MN10300 architectures:
14329 Generate code to avoid bugs in the multiply instructions for the MN10300
14330 processors. This is the default.
14332 @item -mno-mult-bug
14333 @opindex mno-mult-bug
14334 Do not generate code to avoid bugs in the multiply instructions for the
14335 MN10300 processors.
14339 Generate code which uses features specific to the AM33 processor.
14343 Do not generate code which uses features specific to the AM33 processor. This
14346 @item -mreturn-pointer-on-d0
14347 @opindex mreturn-pointer-on-d0
14348 When generating a function which returns a pointer, return the pointer
14349 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14350 only in a0, and attempts to call such functions without a prototype
14351 would result in errors. Note that this option is on by default; use
14352 @option{-mno-return-pointer-on-d0} to disable it.
14356 Do not link in the C run-time initialization object file.
14360 Indicate to the linker that it should perform a relaxation optimization pass
14361 to shorten branches, calls and absolute memory addresses. This option only
14362 has an effect when used on the command line for the final link step.
14364 This option makes symbolic debugging impossible.
14367 @node PDP-11 Options
14368 @subsection PDP-11 Options
14369 @cindex PDP-11 Options
14371 These options are defined for the PDP-11:
14376 Use hardware FPP floating point. This is the default. (FIS floating
14377 point on the PDP-11/40 is not supported.)
14380 @opindex msoft-float
14381 Do not use hardware floating point.
14385 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14389 Return floating-point results in memory. This is the default.
14393 Generate code for a PDP-11/40.
14397 Generate code for a PDP-11/45. This is the default.
14401 Generate code for a PDP-11/10.
14403 @item -mbcopy-builtin
14404 @opindex mbcopy-builtin
14405 Use inline @code{movmemhi} patterns for copying memory. This is the
14410 Do not use inline @code{movmemhi} patterns for copying memory.
14416 Use 16-bit @code{int}. This is the default.
14422 Use 32-bit @code{int}.
14425 @itemx -mno-float32
14427 @opindex mno-float32
14428 Use 64-bit @code{float}. This is the default.
14431 @itemx -mno-float64
14433 @opindex mno-float64
14434 Use 32-bit @code{float}.
14438 Use @code{abshi2} pattern. This is the default.
14442 Do not use @code{abshi2} pattern.
14444 @item -mbranch-expensive
14445 @opindex mbranch-expensive
14446 Pretend that branches are expensive. This is for experimenting with
14447 code generation only.
14449 @item -mbranch-cheap
14450 @opindex mbranch-cheap
14451 Do not pretend that branches are expensive. This is the default.
14455 Generate code for a system with split I&D@.
14459 Generate code for a system without split I&D@. This is the default.
14463 Use Unix assembler syntax. This is the default when configured for
14464 @samp{pdp11-*-bsd}.
14468 Use DEC assembler syntax. This is the default when configured for any
14469 PDP-11 target other than @samp{pdp11-*-bsd}.
14472 @node picoChip Options
14473 @subsection picoChip Options
14474 @cindex picoChip options
14476 These @samp{-m} options are defined for picoChip implementations:
14480 @item -mae=@var{ae_type}
14482 Set the instruction set, register set, and instruction scheduling
14483 parameters for array element type @var{ae_type}. Supported values
14484 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14486 @option{-mae=ANY} selects a completely generic AE type. Code
14487 generated with this option will run on any of the other AE types. The
14488 code will not be as efficient as it would be if compiled for a specific
14489 AE type, and some types of operation (e.g., multiplication) will not
14490 work properly on all types of AE.
14492 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14493 for compiled code, and is the default.
14495 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14496 option may suffer from poor performance of byte (char) manipulation,
14497 since the DSP AE does not provide hardware support for byte load/stores.
14499 @item -msymbol-as-address
14500 Enable the compiler to directly use a symbol name as an address in a
14501 load/store instruction, without first loading it into a
14502 register. Typically, the use of this option will generate larger
14503 programs, which run faster than when the option isn't used. However, the
14504 results vary from program to program, so it is left as a user option,
14505 rather than being permanently enabled.
14507 @item -mno-inefficient-warnings
14508 Disables warnings about the generation of inefficient code. These
14509 warnings can be generated, for example, when compiling code which
14510 performs byte-level memory operations on the MAC AE type. The MAC AE has
14511 no hardware support for byte-level memory operations, so all byte
14512 load/stores must be synthesized from word load/store operations. This is
14513 inefficient and a warning will be generated indicating to the programmer
14514 that they should rewrite the code to avoid byte operations, or to target
14515 an AE type which has the necessary hardware support. This option enables
14516 the warning to be turned off.
14520 @node PowerPC Options
14521 @subsection PowerPC Options
14522 @cindex PowerPC options
14524 These are listed under @xref{RS/6000 and PowerPC Options}.
14526 @node RS/6000 and PowerPC Options
14527 @subsection IBM RS/6000 and PowerPC Options
14528 @cindex RS/6000 and PowerPC Options
14529 @cindex IBM RS/6000 and PowerPC Options
14531 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14538 @itemx -mno-powerpc
14539 @itemx -mpowerpc-gpopt
14540 @itemx -mno-powerpc-gpopt
14541 @itemx -mpowerpc-gfxopt
14542 @itemx -mno-powerpc-gfxopt
14544 @itemx -mno-powerpc64
14548 @itemx -mno-popcntb
14550 @itemx -mno-popcntd
14558 @itemx -mno-hard-dfp
14562 @opindex mno-power2
14564 @opindex mno-powerpc
14565 @opindex mpowerpc-gpopt
14566 @opindex mno-powerpc-gpopt
14567 @opindex mpowerpc-gfxopt
14568 @opindex mno-powerpc-gfxopt
14569 @opindex mpowerpc64
14570 @opindex mno-powerpc64
14574 @opindex mno-popcntb
14576 @opindex mno-popcntd
14582 @opindex mno-mfpgpr
14584 @opindex mno-hard-dfp
14585 GCC supports two related instruction set architectures for the
14586 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14587 instructions supported by the @samp{rios} chip set used in the original
14588 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14589 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14590 the IBM 4xx, 6xx, and follow-on microprocessors.
14592 Neither architecture is a subset of the other. However there is a
14593 large common subset of instructions supported by both. An MQ
14594 register is included in processors supporting the POWER architecture.
14596 You use these options to specify which instructions are available on the
14597 processor you are using. The default value of these options is
14598 determined when configuring GCC@. Specifying the
14599 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14600 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14601 rather than the options listed above.
14603 The @option{-mpower} option allows GCC to generate instructions that
14604 are found only in the POWER architecture and to use the MQ register.
14605 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14606 to generate instructions that are present in the POWER2 architecture but
14607 not the original POWER architecture.
14609 The @option{-mpowerpc} option allows GCC to generate instructions that
14610 are found only in the 32-bit subset of the PowerPC architecture.
14611 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14612 GCC to use the optional PowerPC architecture instructions in the
14613 General Purpose group, including floating-point square root. Specifying
14614 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14615 use the optional PowerPC architecture instructions in the Graphics
14616 group, including floating-point select.
14618 The @option{-mmfcrf} option allows GCC to generate the move from
14619 condition register field instruction implemented on the POWER4
14620 processor and other processors that support the PowerPC V2.01
14622 The @option{-mpopcntb} option allows GCC to generate the popcount and
14623 double precision FP reciprocal estimate instruction implemented on the
14624 POWER5 processor and other processors that support the PowerPC V2.02
14626 The @option{-mpopcntd} option allows GCC to generate the popcount
14627 instruction implemented on the POWER7 processor and other processors
14628 that support the PowerPC V2.06 architecture.
14629 The @option{-mfprnd} option allows GCC to generate the FP round to
14630 integer instructions implemented on the POWER5+ processor and other
14631 processors that support the PowerPC V2.03 architecture.
14632 The @option{-mcmpb} option allows GCC to generate the compare bytes
14633 instruction implemented on the POWER6 processor and other processors
14634 that support the PowerPC V2.05 architecture.
14635 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14636 general purpose register instructions implemented on the POWER6X
14637 processor and other processors that support the extended PowerPC V2.05
14639 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14640 point instructions implemented on some POWER processors.
14642 The @option{-mpowerpc64} option allows GCC to generate the additional
14643 64-bit instructions that are found in the full PowerPC64 architecture
14644 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14645 @option{-mno-powerpc64}.
14647 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14648 will use only the instructions in the common subset of both
14649 architectures plus some special AIX common-mode calls, and will not use
14650 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14651 permits GCC to use any instruction from either architecture and to
14652 allow use of the MQ register; specify this for the Motorola MPC601.
14654 @item -mnew-mnemonics
14655 @itemx -mold-mnemonics
14656 @opindex mnew-mnemonics
14657 @opindex mold-mnemonics
14658 Select which mnemonics to use in the generated assembler code. With
14659 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14660 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14661 assembler mnemonics defined for the POWER architecture. Instructions
14662 defined in only one architecture have only one mnemonic; GCC uses that
14663 mnemonic irrespective of which of these options is specified.
14665 GCC defaults to the mnemonics appropriate for the architecture in
14666 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14667 value of these option. Unless you are building a cross-compiler, you
14668 should normally not specify either @option{-mnew-mnemonics} or
14669 @option{-mold-mnemonics}, but should instead accept the default.
14671 @item -mcpu=@var{cpu_type}
14673 Set architecture type, register usage, choice of mnemonics, and
14674 instruction scheduling parameters for machine type @var{cpu_type}.
14675 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14676 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14677 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
14678 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
14679 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14680 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
14681 @samp{e300c3}, @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
14682 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
14683 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7},
14684 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14685 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14687 @option{-mcpu=common} selects a completely generic processor. Code
14688 generated under this option will run on any POWER or PowerPC processor.
14689 GCC will use only the instructions in the common subset of both
14690 architectures, and will not use the MQ register. GCC assumes a generic
14691 processor model for scheduling purposes.
14693 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14694 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14695 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14696 types, with an appropriate, generic processor model assumed for
14697 scheduling purposes.
14699 The other options specify a specific processor. Code generated under
14700 those options will run best on that processor, and may not run at all on
14703 The @option{-mcpu} options automatically enable or disable the
14706 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
14707 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
14708 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
14709 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
14711 The particular options set for any particular CPU will vary between
14712 compiler versions, depending on what setting seems to produce optimal
14713 code for that CPU; it doesn't necessarily reflect the actual hardware's
14714 capabilities. If you wish to set an individual option to a particular
14715 value, you may specify it after the @option{-mcpu} option, like
14716 @samp{-mcpu=970 -mno-altivec}.
14718 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
14719 not enabled or disabled by the @option{-mcpu} option at present because
14720 AIX does not have full support for these options. You may still
14721 enable or disable them individually if you're sure it'll work in your
14724 @item -mtune=@var{cpu_type}
14726 Set the instruction scheduling parameters for machine type
14727 @var{cpu_type}, but do not set the architecture type, register usage, or
14728 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
14729 values for @var{cpu_type} are used for @option{-mtune} as for
14730 @option{-mcpu}. If both are specified, the code generated will use the
14731 architecture, registers, and mnemonics set by @option{-mcpu}, but the
14732 scheduling parameters set by @option{-mtune}.
14738 Generate code to compute division as reciprocal estimate and iterative
14739 refinement, creating opportunities for increased throughput. This
14740 feature requires: optional PowerPC Graphics instruction set for single
14741 precision and FRE instruction for double precision, assuming divides
14742 cannot generate user-visible traps, and the domain values not include
14743 Infinities, denormals or zero denominator.
14746 @itemx -mno-altivec
14748 @opindex mno-altivec
14749 Generate code that uses (does not use) AltiVec instructions, and also
14750 enable the use of built-in functions that allow more direct access to
14751 the AltiVec instruction set. You may also need to set
14752 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
14758 @opindex mno-vrsave
14759 Generate VRSAVE instructions when generating AltiVec code.
14761 @item -mgen-cell-microcode
14762 @opindex mgen-cell-microcode
14763 Generate Cell microcode instructions
14765 @item -mwarn-cell-microcode
14766 @opindex mwarn-cell-microcode
14767 Warning when a Cell microcode instruction is going to emitted. An example
14768 of a Cell microcode instruction is a variable shift.
14771 @opindex msecure-plt
14772 Generate code that allows ld and ld.so to build executables and shared
14773 libraries with non-exec .plt and .got sections. This is a PowerPC
14774 32-bit SYSV ABI option.
14778 Generate code that uses a BSS .plt section that ld.so fills in, and
14779 requires .plt and .got sections that are both writable and executable.
14780 This is a PowerPC 32-bit SYSV ABI option.
14786 This switch enables or disables the generation of ISEL instructions.
14788 @item -misel=@var{yes/no}
14789 This switch has been deprecated. Use @option{-misel} and
14790 @option{-mno-isel} instead.
14796 This switch enables or disables the generation of SPE simd
14802 @opindex mno-paired
14803 This switch enables or disables the generation of PAIRED simd
14806 @item -mspe=@var{yes/no}
14807 This option has been deprecated. Use @option{-mspe} and
14808 @option{-mno-spe} instead.
14814 Generate code that uses (does not use) vector/scalar (VSX)
14815 instructions, and also enable the use of built-in functions that allow
14816 more direct access to the VSX instruction set.
14818 @item -mfloat-gprs=@var{yes/single/double/no}
14819 @itemx -mfloat-gprs
14820 @opindex mfloat-gprs
14821 This switch enables or disables the generation of floating point
14822 operations on the general purpose registers for architectures that
14825 The argument @var{yes} or @var{single} enables the use of
14826 single-precision floating point operations.
14828 The argument @var{double} enables the use of single and
14829 double-precision floating point operations.
14831 The argument @var{no} disables floating point operations on the
14832 general purpose registers.
14834 This option is currently only available on the MPC854x.
14840 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
14841 targets (including GNU/Linux). The 32-bit environment sets int, long
14842 and pointer to 32 bits and generates code that runs on any PowerPC
14843 variant. The 64-bit environment sets int to 32 bits and long and
14844 pointer to 64 bits, and generates code for PowerPC64, as for
14845 @option{-mpowerpc64}.
14848 @itemx -mno-fp-in-toc
14849 @itemx -mno-sum-in-toc
14850 @itemx -mminimal-toc
14852 @opindex mno-fp-in-toc
14853 @opindex mno-sum-in-toc
14854 @opindex mminimal-toc
14855 Modify generation of the TOC (Table Of Contents), which is created for
14856 every executable file. The @option{-mfull-toc} option is selected by
14857 default. In that case, GCC will allocate at least one TOC entry for
14858 each unique non-automatic variable reference in your program. GCC
14859 will also place floating-point constants in the TOC@. However, only
14860 16,384 entries are available in the TOC@.
14862 If you receive a linker error message that saying you have overflowed
14863 the available TOC space, you can reduce the amount of TOC space used
14864 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
14865 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
14866 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
14867 generate code to calculate the sum of an address and a constant at
14868 run-time instead of putting that sum into the TOC@. You may specify one
14869 or both of these options. Each causes GCC to produce very slightly
14870 slower and larger code at the expense of conserving TOC space.
14872 If you still run out of space in the TOC even when you specify both of
14873 these options, specify @option{-mminimal-toc} instead. This option causes
14874 GCC to make only one TOC entry for every file. When you specify this
14875 option, GCC will produce code that is slower and larger but which
14876 uses extremely little TOC space. You may wish to use this option
14877 only on files that contain less frequently executed code.
14883 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
14884 @code{long} type, and the infrastructure needed to support them.
14885 Specifying @option{-maix64} implies @option{-mpowerpc64} and
14886 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
14887 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
14890 @itemx -mno-xl-compat
14891 @opindex mxl-compat
14892 @opindex mno-xl-compat
14893 Produce code that conforms more closely to IBM XL compiler semantics
14894 when using AIX-compatible ABI@. Pass floating-point arguments to
14895 prototyped functions beyond the register save area (RSA) on the stack
14896 in addition to argument FPRs. Do not assume that most significant
14897 double in 128-bit long double value is properly rounded when comparing
14898 values and converting to double. Use XL symbol names for long double
14901 The AIX calling convention was extended but not initially documented to
14902 handle an obscure K&R C case of calling a function that takes the
14903 address of its arguments with fewer arguments than declared. IBM XL
14904 compilers access floating point arguments which do not fit in the
14905 RSA from the stack when a subroutine is compiled without
14906 optimization. Because always storing floating-point arguments on the
14907 stack is inefficient and rarely needed, this option is not enabled by
14908 default and only is necessary when calling subroutines compiled by IBM
14909 XL compilers without optimization.
14913 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
14914 application written to use message passing with special startup code to
14915 enable the application to run. The system must have PE installed in the
14916 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
14917 must be overridden with the @option{-specs=} option to specify the
14918 appropriate directory location. The Parallel Environment does not
14919 support threads, so the @option{-mpe} option and the @option{-pthread}
14920 option are incompatible.
14922 @item -malign-natural
14923 @itemx -malign-power
14924 @opindex malign-natural
14925 @opindex malign-power
14926 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
14927 @option{-malign-natural} overrides the ABI-defined alignment of larger
14928 types, such as floating-point doubles, on their natural size-based boundary.
14929 The option @option{-malign-power} instructs GCC to follow the ABI-specified
14930 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
14932 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
14936 @itemx -mhard-float
14937 @opindex msoft-float
14938 @opindex mhard-float
14939 Generate code that does not use (uses) the floating-point register set.
14940 Software floating point emulation is provided if you use the
14941 @option{-msoft-float} option, and pass the option to GCC when linking.
14943 @item -msingle-float
14944 @itemx -mdouble-float
14945 @opindex msingle-float
14946 @opindex mdouble-float
14947 Generate code for single or double-precision floating point operations.
14948 @option{-mdouble-float} implies @option{-msingle-float}.
14951 @opindex msimple-fpu
14952 Do not generate sqrt and div instructions for hardware floating point unit.
14956 Specify type of floating point unit. Valid values are @var{sp_lite}
14957 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
14958 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
14959 and @var{dp_full} (equivalent to -mdouble-float).
14962 @opindex mxilinx-fpu
14963 Perform optimizations for floating point unit on Xilinx PPC 405/440.
14966 @itemx -mno-multiple
14968 @opindex mno-multiple
14969 Generate code that uses (does not use) the load multiple word
14970 instructions and the store multiple word instructions. These
14971 instructions are generated by default on POWER systems, and not
14972 generated on PowerPC systems. Do not use @option{-mmultiple} on little
14973 endian PowerPC systems, since those instructions do not work when the
14974 processor is in little endian mode. The exceptions are PPC740 and
14975 PPC750 which permit the instructions usage in little endian mode.
14980 @opindex mno-string
14981 Generate code that uses (does not use) the load string instructions
14982 and the store string word instructions to save multiple registers and
14983 do small block moves. These instructions are generated by default on
14984 POWER systems, and not generated on PowerPC systems. Do not use
14985 @option{-mstring} on little endian PowerPC systems, since those
14986 instructions do not work when the processor is in little endian mode.
14987 The exceptions are PPC740 and PPC750 which permit the instructions
14988 usage in little endian mode.
14993 @opindex mno-update
14994 Generate code that uses (does not use) the load or store instructions
14995 that update the base register to the address of the calculated memory
14996 location. These instructions are generated by default. If you use
14997 @option{-mno-update}, there is a small window between the time that the
14998 stack pointer is updated and the address of the previous frame is
14999 stored, which means code that walks the stack frame across interrupts or
15000 signals may get corrupted data.
15002 @item -mavoid-indexed-addresses
15003 @item -mno-avoid-indexed-addresses
15004 @opindex mavoid-indexed-addresses
15005 @opindex mno-avoid-indexed-addresses
15006 Generate code that tries to avoid (not avoid) the use of indexed load
15007 or store instructions. These instructions can incur a performance
15008 penalty on Power6 processors in certain situations, such as when
15009 stepping through large arrays that cross a 16M boundary. This option
15010 is enabled by default when targetting Power6 and disabled otherwise.
15013 @itemx -mno-fused-madd
15014 @opindex mfused-madd
15015 @opindex mno-fused-madd
15016 Generate code that uses (does not use) the floating point multiply and
15017 accumulate instructions. These instructions are generated by default if
15018 hardware floating is used.
15024 Generate code that uses (does not use) the half-word multiply and
15025 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15026 These instructions are generated by default when targetting those
15033 Generate code that uses (does not use) the string-search @samp{dlmzb}
15034 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15035 generated by default when targetting those processors.
15037 @item -mno-bit-align
15039 @opindex mno-bit-align
15040 @opindex mbit-align
15041 On System V.4 and embedded PowerPC systems do not (do) force structures
15042 and unions that contain bit-fields to be aligned to the base type of the
15045 For example, by default a structure containing nothing but 8
15046 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15047 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15048 the structure would be aligned to a 1 byte boundary and be one byte in
15051 @item -mno-strict-align
15052 @itemx -mstrict-align
15053 @opindex mno-strict-align
15054 @opindex mstrict-align
15055 On System V.4 and embedded PowerPC systems do not (do) assume that
15056 unaligned memory references will be handled by the system.
15058 @item -mrelocatable
15059 @itemx -mno-relocatable
15060 @opindex mrelocatable
15061 @opindex mno-relocatable
15062 On embedded PowerPC systems generate code that allows (does not allow)
15063 the program to be relocated to a different address at runtime. If you
15064 use @option{-mrelocatable} on any module, all objects linked together must
15065 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
15067 @item -mrelocatable-lib
15068 @itemx -mno-relocatable-lib
15069 @opindex mrelocatable-lib
15070 @opindex mno-relocatable-lib
15071 On embedded PowerPC systems generate code that allows (does not allow)
15072 the program to be relocated to a different address at runtime. Modules
15073 compiled with @option{-mrelocatable-lib} can be linked with either modules
15074 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
15075 with modules compiled with the @option{-mrelocatable} options.
15081 On System V.4 and embedded PowerPC systems do not (do) assume that
15082 register 2 contains a pointer to a global area pointing to the addresses
15083 used in the program.
15086 @itemx -mlittle-endian
15088 @opindex mlittle-endian
15089 On System V.4 and embedded PowerPC systems compile code for the
15090 processor in little endian mode. The @option{-mlittle-endian} option is
15091 the same as @option{-mlittle}.
15094 @itemx -mbig-endian
15096 @opindex mbig-endian
15097 On System V.4 and embedded PowerPC systems compile code for the
15098 processor in big endian mode. The @option{-mbig-endian} option is
15099 the same as @option{-mbig}.
15101 @item -mdynamic-no-pic
15102 @opindex mdynamic-no-pic
15103 On Darwin and Mac OS X systems, compile code so that it is not
15104 relocatable, but that its external references are relocatable. The
15105 resulting code is suitable for applications, but not shared
15108 @item -mprioritize-restricted-insns=@var{priority}
15109 @opindex mprioritize-restricted-insns
15110 This option controls the priority that is assigned to
15111 dispatch-slot restricted instructions during the second scheduling
15112 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15113 @var{no/highest/second-highest} priority to dispatch slot restricted
15116 @item -msched-costly-dep=@var{dependence_type}
15117 @opindex msched-costly-dep
15118 This option controls which dependences are considered costly
15119 by the target during instruction scheduling. The argument
15120 @var{dependence_type} takes one of the following values:
15121 @var{no}: no dependence is costly,
15122 @var{all}: all dependences are costly,
15123 @var{true_store_to_load}: a true dependence from store to load is costly,
15124 @var{store_to_load}: any dependence from store to load is costly,
15125 @var{number}: any dependence which latency >= @var{number} is costly.
15127 @item -minsert-sched-nops=@var{scheme}
15128 @opindex minsert-sched-nops
15129 This option controls which nop insertion scheme will be used during
15130 the second scheduling pass. The argument @var{scheme} takes one of the
15132 @var{no}: Don't insert nops.
15133 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15134 according to the scheduler's grouping.
15135 @var{regroup_exact}: Insert nops to force costly dependent insns into
15136 separate groups. Insert exactly as many nops as needed to force an insn
15137 to a new group, according to the estimated processor grouping.
15138 @var{number}: Insert nops to force costly dependent insns into
15139 separate groups. Insert @var{number} nops to force an insn to a new group.
15142 @opindex mcall-sysv
15143 On System V.4 and embedded PowerPC systems compile code using calling
15144 conventions that adheres to the March 1995 draft of the System V
15145 Application Binary Interface, PowerPC processor supplement. This is the
15146 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15148 @item -mcall-sysv-eabi
15150 @opindex mcall-sysv-eabi
15151 @opindex mcall-eabi
15152 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15154 @item -mcall-sysv-noeabi
15155 @opindex mcall-sysv-noeabi
15156 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15158 @item -mcall-aixdesc
15160 On System V.4 and embedded PowerPC systems compile code for the AIX
15164 @opindex mcall-linux
15165 On System V.4 and embedded PowerPC systems compile code for the
15166 Linux-based GNU system.
15170 On System V.4 and embedded PowerPC systems compile code for the
15171 Hurd-based GNU system.
15173 @item -mcall-freebsd
15174 @opindex mcall-freebsd
15175 On System V.4 and embedded PowerPC systems compile code for the
15176 FreeBSD operating system.
15178 @item -mcall-netbsd
15179 @opindex mcall-netbsd
15180 On System V.4 and embedded PowerPC systems compile code for the
15181 NetBSD operating system.
15183 @item -mcall-openbsd
15184 @opindex mcall-netbsd
15185 On System V.4 and embedded PowerPC systems compile code for the
15186 OpenBSD operating system.
15188 @item -maix-struct-return
15189 @opindex maix-struct-return
15190 Return all structures in memory (as specified by the AIX ABI)@.
15192 @item -msvr4-struct-return
15193 @opindex msvr4-struct-return
15194 Return structures smaller than 8 bytes in registers (as specified by the
15197 @item -mabi=@var{abi-type}
15199 Extend the current ABI with a particular extension, or remove such extension.
15200 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15201 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15205 Extend the current ABI with SPE ABI extensions. This does not change
15206 the default ABI, instead it adds the SPE ABI extensions to the current
15210 @opindex mabi=no-spe
15211 Disable Booke SPE ABI extensions for the current ABI@.
15213 @item -mabi=ibmlongdouble
15214 @opindex mabi=ibmlongdouble
15215 Change the current ABI to use IBM extended precision long double.
15216 This is a PowerPC 32-bit SYSV ABI option.
15218 @item -mabi=ieeelongdouble
15219 @opindex mabi=ieeelongdouble
15220 Change the current ABI to use IEEE extended precision long double.
15221 This is a PowerPC 32-bit Linux ABI option.
15224 @itemx -mno-prototype
15225 @opindex mprototype
15226 @opindex mno-prototype
15227 On System V.4 and embedded PowerPC systems assume that all calls to
15228 variable argument functions are properly prototyped. Otherwise, the
15229 compiler must insert an instruction before every non prototyped call to
15230 set or clear bit 6 of the condition code register (@var{CR}) to
15231 indicate whether floating point values were passed in the floating point
15232 registers in case the function takes a variable arguments. With
15233 @option{-mprototype}, only calls to prototyped variable argument functions
15234 will set or clear the bit.
15238 On embedded PowerPC systems, assume that the startup module is called
15239 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15240 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15245 On embedded PowerPC systems, assume that the startup module is called
15246 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15251 On embedded PowerPC systems, assume that the startup module is called
15252 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15255 @item -myellowknife
15256 @opindex myellowknife
15257 On embedded PowerPC systems, assume that the startup module is called
15258 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15263 On System V.4 and embedded PowerPC systems, specify that you are
15264 compiling for a VxWorks system.
15268 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15269 header to indicate that @samp{eabi} extended relocations are used.
15275 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15276 Embedded Applications Binary Interface (eabi) which is a set of
15277 modifications to the System V.4 specifications. Selecting @option{-meabi}
15278 means that the stack is aligned to an 8 byte boundary, a function
15279 @code{__eabi} is called to from @code{main} to set up the eabi
15280 environment, and the @option{-msdata} option can use both @code{r2} and
15281 @code{r13} to point to two separate small data areas. Selecting
15282 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15283 do not call an initialization function from @code{main}, and the
15284 @option{-msdata} option will only use @code{r13} to point to a single
15285 small data area. The @option{-meabi} option is on by default if you
15286 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15289 @opindex msdata=eabi
15290 On System V.4 and embedded PowerPC systems, put small initialized
15291 @code{const} global and static data in the @samp{.sdata2} section, which
15292 is pointed to by register @code{r2}. Put small initialized
15293 non-@code{const} global and static data in the @samp{.sdata} section,
15294 which is pointed to by register @code{r13}. Put small uninitialized
15295 global and static data in the @samp{.sbss} section, which is adjacent to
15296 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15297 incompatible with the @option{-mrelocatable} option. The
15298 @option{-msdata=eabi} option also sets the @option{-memb} option.
15301 @opindex msdata=sysv
15302 On System V.4 and embedded PowerPC systems, put small global and static
15303 data in the @samp{.sdata} section, which is pointed to by register
15304 @code{r13}. Put small uninitialized global and static data in the
15305 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15306 The @option{-msdata=sysv} option is incompatible with the
15307 @option{-mrelocatable} option.
15309 @item -msdata=default
15311 @opindex msdata=default
15313 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15314 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15315 same as @option{-msdata=sysv}.
15318 @opindex msdata=data
15319 On System V.4 and embedded PowerPC systems, put small global
15320 data in the @samp{.sdata} section. Put small uninitialized global
15321 data in the @samp{.sbss} section. Do not use register @code{r13}
15322 to address small data however. This is the default behavior unless
15323 other @option{-msdata} options are used.
15327 @opindex msdata=none
15329 On embedded PowerPC systems, put all initialized global and static data
15330 in the @samp{.data} section, and all uninitialized data in the
15331 @samp{.bss} section.
15335 @cindex smaller data references (PowerPC)
15336 @cindex .sdata/.sdata2 references (PowerPC)
15337 On embedded PowerPC systems, put global and static items less than or
15338 equal to @var{num} bytes into the small data or bss sections instead of
15339 the normal data or bss section. By default, @var{num} is 8. The
15340 @option{-G @var{num}} switch is also passed to the linker.
15341 All modules should be compiled with the same @option{-G @var{num}} value.
15344 @itemx -mno-regnames
15346 @opindex mno-regnames
15347 On System V.4 and embedded PowerPC systems do (do not) emit register
15348 names in the assembly language output using symbolic forms.
15351 @itemx -mno-longcall
15353 @opindex mno-longcall
15354 By default assume that all calls are far away so that a longer more
15355 expensive calling sequence is required. This is required for calls
15356 further than 32 megabytes (33,554,432 bytes) from the current location.
15357 A short call will be generated if the compiler knows
15358 the call cannot be that far away. This setting can be overridden by
15359 the @code{shortcall} function attribute, or by @code{#pragma
15362 Some linkers are capable of detecting out-of-range calls and generating
15363 glue code on the fly. On these systems, long calls are unnecessary and
15364 generate slower code. As of this writing, the AIX linker can do this,
15365 as can the GNU linker for PowerPC/64. It is planned to add this feature
15366 to the GNU linker for 32-bit PowerPC systems as well.
15368 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15369 callee, L42'', plus a ``branch island'' (glue code). The two target
15370 addresses represent the callee and the ``branch island''. The
15371 Darwin/PPC linker will prefer the first address and generate a ``bl
15372 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15373 otherwise, the linker will generate ``bl L42'' to call the ``branch
15374 island''. The ``branch island'' is appended to the body of the
15375 calling function; it computes the full 32-bit address of the callee
15378 On Mach-O (Darwin) systems, this option directs the compiler emit to
15379 the glue for every direct call, and the Darwin linker decides whether
15380 to use or discard it.
15382 In the future, we may cause GCC to ignore all longcall specifications
15383 when the linker is known to generate glue.
15385 @item -mtls-markers
15386 @itemx -mno-tls-markers
15387 @opindex mtls-markers
15388 @opindex mno-tls-markers
15389 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15390 specifying the function argument. The relocation allows ld to
15391 reliably associate function call with argument setup instructions for
15392 TLS optimization, which in turn allows gcc to better schedule the
15397 Adds support for multithreading with the @dfn{pthreads} library.
15398 This option sets flags for both the preprocessor and linker.
15403 @subsection RX Options
15406 These @option{-m} options are defined for RX implementations:
15409 @item -m64bit-doubles
15410 @itemx -m32bit-doubles
15411 @opindex m64bit-doubles
15412 @opindex m32bit-doubles
15413 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
15414 or 32-bits (@option{-m32bit-doubles}) in size. The default is
15415 @option{-m32bit-doubles}. @emph{Note} the RX's hardware floating
15416 point instructions are only used for 32-bit floating point values, and
15417 then only if @option{-ffast-math} has been specified on the command
15418 line. This is because the RX FPU instructions do not properly support
15419 denormal (or sub-normal) values.
15421 @item -mbig-endian-data
15422 @itemx -mlittle-endian-data
15423 @opindex mbig-endian-data
15424 @opindex mlittle-endian-data
15425 Store data (but not code) in the big-endian format. The default is
15426 @option{-mlittle-endian-data}, ie to store data in the little endian
15429 @item -msmall-data-limit=@var{N}
15430 @opindex msmall-data-limit
15431 Specifies the maximum size in bytes of global and static variables
15432 which can be placed into the small data area. Using the small data
15433 area can lead to smaller and faster code, but the size of area is
15434 limited and it is up to the programmer to ensure that the area does
15435 not overflow. Also when the small data area is used one of the RX's
15436 registers (@code{r13}) is reserved for use pointing to this area, so
15437 it is no longer available for use by the compiler. This could result
15438 in slower and/or larger code if variables which once could have been
15439 held in @code{r13} are now pushed onto the stack.
15441 Note, common variables (variables which have not been initialised) and
15442 constants are not placed into the small data area as they are assigned
15443 to other sections in the output executeable.
15445 The default value is zero, which disables this feature. Note, this
15446 feature is not enabled by default with higher optimization levels
15447 (@option{-O2} etc) because of the potentially deterimental effects of
15448 reserving register @code{r13}. It is up to the programmer to
15449 experiment and discover whether this feature is of benefit to their
15456 Use the simulator runtime. The default is to use the libgloss board
15459 @item -mas100-syntax
15460 @item -mno-as100-syntax
15461 @opindex mas100-syntax
15462 @opindex mno-as100-syntax
15463 When generating assembler output use a syntax that is compatible with
15464 Renesas's AS100 assembler. This syntax can also be handled by the GAS
15465 assembler but it has some restrictions so generating it is not the
15468 @item -mmax-constant-size=@var{N}
15469 @opindex mmax-constant-size
15470 Specifies the maxium size, in bytes, of a constant that can be used as
15471 an operand in a RX instruction. Although the RX instruction set does
15472 allow consants of up to 4 bytes in length to be used in instructions,
15473 a longer value equates to a longer instruction. Thus in some
15474 circumstances it can be beneficial to restrict the size of constants
15475 that are used in instructions. Constants that are too big are instead
15476 placed into a constant pool and referenced via register indirection.
15478 The value @var{N} can be between 0 and 3. A value of 0, the default,
15479 means that constants of any size are allowed.
15483 Enable linker relaxation. Linker relaxation is a process whereby the
15484 linker will attempt to reduce the size of a program by finding shorter
15485 versions of various instructions. Disabled by default.
15487 @item -mint-register=@var{N}
15488 @opindex mint-register
15489 Specify the number of registers to reserve for fast interrupt handler
15490 functions. The value @var{N} can be between 0 and 4. A value of 1
15491 means that register @code{r13} will be reserved for ther exclusive use
15492 of fast interrupt handlers. A value of 2 reserves @code{r13} and
15493 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
15494 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
15495 A value of 0, the default, does not reserve any registers.
15498 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
15499 has special significance to the RX port when used with the
15500 @code{interrupt} function attribute. This attribute indicates a
15501 function intended to process fast interrupts. GCC will will ensure
15502 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
15503 and/or @code{r13} and only provided that the normal use of the
15504 corresponding registers have been restricted via the
15505 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
15508 @node S/390 and zSeries Options
15509 @subsection S/390 and zSeries Options
15510 @cindex S/390 and zSeries Options
15512 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15516 @itemx -msoft-float
15517 @opindex mhard-float
15518 @opindex msoft-float
15519 Use (do not use) the hardware floating-point instructions and registers
15520 for floating-point operations. When @option{-msoft-float} is specified,
15521 functions in @file{libgcc.a} will be used to perform floating-point
15522 operations. When @option{-mhard-float} is specified, the compiler
15523 generates IEEE floating-point instructions. This is the default.
15526 @itemx -mno-hard-dfp
15528 @opindex mno-hard-dfp
15529 Use (do not use) the hardware decimal-floating-point instructions for
15530 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15531 specified, functions in @file{libgcc.a} will be used to perform
15532 decimal-floating-point operations. When @option{-mhard-dfp} is
15533 specified, the compiler generates decimal-floating-point hardware
15534 instructions. This is the default for @option{-march=z9-ec} or higher.
15536 @item -mlong-double-64
15537 @itemx -mlong-double-128
15538 @opindex mlong-double-64
15539 @opindex mlong-double-128
15540 These switches control the size of @code{long double} type. A size
15541 of 64bit makes the @code{long double} type equivalent to the @code{double}
15542 type. This is the default.
15545 @itemx -mno-backchain
15546 @opindex mbackchain
15547 @opindex mno-backchain
15548 Store (do not store) the address of the caller's frame as backchain pointer
15549 into the callee's stack frame.
15550 A backchain may be needed to allow debugging using tools that do not understand
15551 DWARF-2 call frame information.
15552 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15553 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15554 the backchain is placed into the topmost word of the 96/160 byte register
15557 In general, code compiled with @option{-mbackchain} is call-compatible with
15558 code compiled with @option{-mmo-backchain}; however, use of the backchain
15559 for debugging purposes usually requires that the whole binary is built with
15560 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15561 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15562 to build a linux kernel use @option{-msoft-float}.
15564 The default is to not maintain the backchain.
15566 @item -mpacked-stack
15567 @itemx -mno-packed-stack
15568 @opindex mpacked-stack
15569 @opindex mno-packed-stack
15570 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15571 specified, the compiler uses the all fields of the 96/160 byte register save
15572 area only for their default purpose; unused fields still take up stack space.
15573 When @option{-mpacked-stack} is specified, register save slots are densely
15574 packed at the top of the register save area; unused space is reused for other
15575 purposes, allowing for more efficient use of the available stack space.
15576 However, when @option{-mbackchain} is also in effect, the topmost word of
15577 the save area is always used to store the backchain, and the return address
15578 register is always saved two words below the backchain.
15580 As long as the stack frame backchain is not used, code generated with
15581 @option{-mpacked-stack} is call-compatible with code generated with
15582 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15583 S/390 or zSeries generated code that uses the stack frame backchain at run
15584 time, not just for debugging purposes. Such code is not call-compatible
15585 with code compiled with @option{-mpacked-stack}. Also, note that the
15586 combination of @option{-mbackchain},
15587 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15588 to build a linux kernel use @option{-msoft-float}.
15590 The default is to not use the packed stack layout.
15593 @itemx -mno-small-exec
15594 @opindex msmall-exec
15595 @opindex mno-small-exec
15596 Generate (or do not generate) code using the @code{bras} instruction
15597 to do subroutine calls.
15598 This only works reliably if the total executable size does not
15599 exceed 64k. The default is to use the @code{basr} instruction instead,
15600 which does not have this limitation.
15606 When @option{-m31} is specified, generate code compliant to the
15607 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15608 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15609 particular to generate 64-bit instructions. For the @samp{s390}
15610 targets, the default is @option{-m31}, while the @samp{s390x}
15611 targets default to @option{-m64}.
15617 When @option{-mzarch} is specified, generate code using the
15618 instructions available on z/Architecture.
15619 When @option{-mesa} is specified, generate code using the
15620 instructions available on ESA/390. Note that @option{-mesa} is
15621 not possible with @option{-m64}.
15622 When generating code compliant to the GNU/Linux for S/390 ABI,
15623 the default is @option{-mesa}. When generating code compliant
15624 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
15630 Generate (or do not generate) code using the @code{mvcle} instruction
15631 to perform block moves. When @option{-mno-mvcle} is specified,
15632 use a @code{mvc} loop instead. This is the default unless optimizing for
15639 Print (or do not print) additional debug information when compiling.
15640 The default is to not print debug information.
15642 @item -march=@var{cpu-type}
15644 Generate code that will run on @var{cpu-type}, which is the name of a system
15645 representing a certain processor type. Possible values for
15646 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
15647 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
15648 When generating code using the instructions available on z/Architecture,
15649 the default is @option{-march=z900}. Otherwise, the default is
15650 @option{-march=g5}.
15652 @item -mtune=@var{cpu-type}
15654 Tune to @var{cpu-type} everything applicable about the generated code,
15655 except for the ABI and the set of available instructions.
15656 The list of @var{cpu-type} values is the same as for @option{-march}.
15657 The default is the value used for @option{-march}.
15660 @itemx -mno-tpf-trace
15661 @opindex mtpf-trace
15662 @opindex mno-tpf-trace
15663 Generate code that adds (does not add) in TPF OS specific branches to trace
15664 routines in the operating system. This option is off by default, even
15665 when compiling for the TPF OS@.
15668 @itemx -mno-fused-madd
15669 @opindex mfused-madd
15670 @opindex mno-fused-madd
15671 Generate code that uses (does not use) the floating point multiply and
15672 accumulate instructions. These instructions are generated by default if
15673 hardware floating point is used.
15675 @item -mwarn-framesize=@var{framesize}
15676 @opindex mwarn-framesize
15677 Emit a warning if the current function exceeds the given frame size. Because
15678 this is a compile time check it doesn't need to be a real problem when the program
15679 runs. It is intended to identify functions which most probably cause
15680 a stack overflow. It is useful to be used in an environment with limited stack
15681 size e.g.@: the linux kernel.
15683 @item -mwarn-dynamicstack
15684 @opindex mwarn-dynamicstack
15685 Emit a warning if the function calls alloca or uses dynamically
15686 sized arrays. This is generally a bad idea with a limited stack size.
15688 @item -mstack-guard=@var{stack-guard}
15689 @itemx -mstack-size=@var{stack-size}
15690 @opindex mstack-guard
15691 @opindex mstack-size
15692 If these options are provided the s390 back end emits additional instructions in
15693 the function prologue which trigger a trap if the stack size is @var{stack-guard}
15694 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
15695 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
15696 the frame size of the compiled function is chosen.
15697 These options are intended to be used to help debugging stack overflow problems.
15698 The additionally emitted code causes only little overhead and hence can also be
15699 used in production like systems without greater performance degradation. The given
15700 values have to be exact powers of 2 and @var{stack-size} has to be greater than
15701 @var{stack-guard} without exceeding 64k.
15702 In order to be efficient the extra code makes the assumption that the stack starts
15703 at an address aligned to the value given by @var{stack-size}.
15704 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
15707 @node Score Options
15708 @subsection Score Options
15709 @cindex Score Options
15711 These options are defined for Score implementations:
15716 Compile code for big endian mode. This is the default.
15720 Compile code for little endian mode.
15724 Disable generate bcnz instruction.
15728 Enable generate unaligned load and store instruction.
15732 Enable the use of multiply-accumulate instructions. Disabled by default.
15736 Specify the SCORE5 as the target architecture.
15740 Specify the SCORE5U of the target architecture.
15744 Specify the SCORE7 as the target architecture. This is the default.
15748 Specify the SCORE7D as the target architecture.
15752 @subsection SH Options
15754 These @samp{-m} options are defined for the SH implementations:
15759 Generate code for the SH1.
15763 Generate code for the SH2.
15766 Generate code for the SH2e.
15770 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
15771 that the floating-point unit is not used.
15773 @item -m2a-single-only
15774 @opindex m2a-single-only
15775 Generate code for the SH2a-FPU, in such a way that no double-precision
15776 floating point operations are used.
15779 @opindex m2a-single
15780 Generate code for the SH2a-FPU assuming the floating-point unit is in
15781 single-precision mode by default.
15785 Generate code for the SH2a-FPU assuming the floating-point unit is in
15786 double-precision mode by default.
15790 Generate code for the SH3.
15794 Generate code for the SH3e.
15798 Generate code for the SH4 without a floating-point unit.
15800 @item -m4-single-only
15801 @opindex m4-single-only
15802 Generate code for the SH4 with a floating-point unit that only
15803 supports single-precision arithmetic.
15807 Generate code for the SH4 assuming the floating-point unit is in
15808 single-precision mode by default.
15812 Generate code for the SH4.
15816 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
15817 floating-point unit is not used.
15819 @item -m4a-single-only
15820 @opindex m4a-single-only
15821 Generate code for the SH4a, in such a way that no double-precision
15822 floating point operations are used.
15825 @opindex m4a-single
15826 Generate code for the SH4a assuming the floating-point unit is in
15827 single-precision mode by default.
15831 Generate code for the SH4a.
15835 Same as @option{-m4a-nofpu}, except that it implicitly passes
15836 @option{-dsp} to the assembler. GCC doesn't generate any DSP
15837 instructions at the moment.
15841 Compile code for the processor in big endian mode.
15845 Compile code for the processor in little endian mode.
15849 Align doubles at 64-bit boundaries. Note that this changes the calling
15850 conventions, and thus some functions from the standard C library will
15851 not work unless you recompile it first with @option{-mdalign}.
15855 Shorten some address references at link time, when possible; uses the
15856 linker option @option{-relax}.
15860 Use 32-bit offsets in @code{switch} tables. The default is to use
15865 Enable the use of bit manipulation instructions on SH2A.
15869 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
15870 alignment constraints.
15874 Comply with the calling conventions defined by Renesas.
15878 Comply with the calling conventions defined by Renesas.
15882 Comply with the calling conventions defined for GCC before the Renesas
15883 conventions were available. This option is the default for all
15884 targets of the SH toolchain except for @samp{sh-symbianelf}.
15887 @opindex mnomacsave
15888 Mark the @code{MAC} register as call-clobbered, even if
15889 @option{-mhitachi} is given.
15893 Increase IEEE-compliance of floating-point code.
15894 At the moment, this is equivalent to @option{-fno-finite-math-only}.
15895 When generating 16 bit SH opcodes, getting IEEE-conforming results for
15896 comparisons of NANs / infinities incurs extra overhead in every
15897 floating point comparison, therefore the default is set to
15898 @option{-ffinite-math-only}.
15900 @item -minline-ic_invalidate
15901 @opindex minline-ic_invalidate
15902 Inline code to invalidate instruction cache entries after setting up
15903 nested function trampolines.
15904 This option has no effect if -musermode is in effect and the selected
15905 code generation option (e.g. -m4) does not allow the use of the icbi
15907 If the selected code generation option does not allow the use of the icbi
15908 instruction, and -musermode is not in effect, the inlined code will
15909 manipulate the instruction cache address array directly with an associative
15910 write. This not only requires privileged mode, but it will also
15911 fail if the cache line had been mapped via the TLB and has become unmapped.
15915 Dump instruction size and location in the assembly code.
15918 @opindex mpadstruct
15919 This option is deprecated. It pads structures to multiple of 4 bytes,
15920 which is incompatible with the SH ABI@.
15924 Optimize for space instead of speed. Implied by @option{-Os}.
15927 @opindex mprefergot
15928 When generating position-independent code, emit function calls using
15929 the Global Offset Table instead of the Procedure Linkage Table.
15933 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
15934 if the inlined code would not work in user mode.
15935 This is the default when the target is @code{sh-*-linux*}.
15937 @item -multcost=@var{number}
15938 @opindex multcost=@var{number}
15939 Set the cost to assume for a multiply insn.
15941 @item -mdiv=@var{strategy}
15942 @opindex mdiv=@var{strategy}
15943 Set the division strategy to use for SHmedia code. @var{strategy} must be
15944 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
15945 inv:call2, inv:fp .
15946 "fp" performs the operation in floating point. This has a very high latency,
15947 but needs only a few instructions, so it might be a good choice if
15948 your code has enough easily exploitable ILP to allow the compiler to
15949 schedule the floating point instructions together with other instructions.
15950 Division by zero causes a floating point exception.
15951 "inv" uses integer operations to calculate the inverse of the divisor,
15952 and then multiplies the dividend with the inverse. This strategy allows
15953 cse and hoisting of the inverse calculation. Division by zero calculates
15954 an unspecified result, but does not trap.
15955 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
15956 have been found, or if the entire operation has been hoisted to the same
15957 place, the last stages of the inverse calculation are intertwined with the
15958 final multiply to reduce the overall latency, at the expense of using a few
15959 more instructions, and thus offering fewer scheduling opportunities with
15961 "call" calls a library function that usually implements the inv:minlat
15963 This gives high code density for m5-*media-nofpu compilations.
15964 "call2" uses a different entry point of the same library function, where it
15965 assumes that a pointer to a lookup table has already been set up, which
15966 exposes the pointer load to cse / code hoisting optimizations.
15967 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
15968 code generation, but if the code stays unoptimized, revert to the "call",
15969 "call2", or "fp" strategies, respectively. Note that the
15970 potentially-trapping side effect of division by zero is carried by a
15971 separate instruction, so it is possible that all the integer instructions
15972 are hoisted out, but the marker for the side effect stays where it is.
15973 A recombination to fp operations or a call is not possible in that case.
15974 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
15975 that the inverse calculation was nor separated from the multiply, they speed
15976 up division where the dividend fits into 20 bits (plus sign where applicable),
15977 by inserting a test to skip a number of operations in this case; this test
15978 slows down the case of larger dividends. inv20u assumes the case of a such
15979 a small dividend to be unlikely, and inv20l assumes it to be likely.
15981 @item -mdivsi3_libfunc=@var{name}
15982 @opindex mdivsi3_libfunc=@var{name}
15983 Set the name of the library function used for 32 bit signed division to
15984 @var{name}. This only affect the name used in the call and inv:call
15985 division strategies, and the compiler will still expect the same
15986 sets of input/output/clobbered registers as if this option was not present.
15988 @item -mfixed-range=@var{register-range}
15989 @opindex mfixed-range
15990 Generate code treating the given register range as fixed registers.
15991 A fixed register is one that the register allocator can not use. This is
15992 useful when compiling kernel code. A register range is specified as
15993 two registers separated by a dash. Multiple register ranges can be
15994 specified separated by a comma.
15996 @item -madjust-unroll
15997 @opindex madjust-unroll
15998 Throttle unrolling to avoid thrashing target registers.
15999 This option only has an effect if the gcc code base supports the
16000 TARGET_ADJUST_UNROLL_MAX target hook.
16002 @item -mindexed-addressing
16003 @opindex mindexed-addressing
16004 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16005 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16006 semantics for the indexed addressing mode. The architecture allows the
16007 implementation of processors with 64 bit MMU, which the OS could use to
16008 get 32 bit addressing, but since no current hardware implementation supports
16009 this or any other way to make the indexed addressing mode safe to use in
16010 the 32 bit ABI, the default is -mno-indexed-addressing.
16012 @item -mgettrcost=@var{number}
16013 @opindex mgettrcost=@var{number}
16014 Set the cost assumed for the gettr instruction to @var{number}.
16015 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16019 Assume pt* instructions won't trap. This will generally generate better
16020 scheduled code, but is unsafe on current hardware. The current architecture
16021 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16022 This has the unintentional effect of making it unsafe to schedule ptabs /
16023 ptrel before a branch, or hoist it out of a loop. For example,
16024 __do_global_ctors, a part of libgcc that runs constructors at program
16025 startup, calls functions in a list which is delimited by @minus{}1. With the
16026 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16027 That means that all the constructors will be run a bit quicker, but when
16028 the loop comes to the end of the list, the program crashes because ptabs
16029 loads @minus{}1 into a target register. Since this option is unsafe for any
16030 hardware implementing the current architecture specification, the default
16031 is -mno-pt-fixed. Unless the user specifies a specific cost with
16032 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16033 this deters register allocation using target registers for storing
16036 @item -minvalid-symbols
16037 @opindex minvalid-symbols
16038 Assume symbols might be invalid. Ordinary function symbols generated by
16039 the compiler will always be valid to load with movi/shori/ptabs or
16040 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16041 to generate symbols that will cause ptabs / ptrel to trap.
16042 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16043 It will then prevent cross-basic-block cse, hoisting and most scheduling
16044 of symbol loads. The default is @option{-mno-invalid-symbols}.
16047 @node SPARC Options
16048 @subsection SPARC Options
16049 @cindex SPARC options
16051 These @samp{-m} options are supported on the SPARC:
16054 @item -mno-app-regs
16056 @opindex mno-app-regs
16058 Specify @option{-mapp-regs} to generate output using the global registers
16059 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16062 To be fully SVR4 ABI compliant at the cost of some performance loss,
16063 specify @option{-mno-app-regs}. You should compile libraries and system
16064 software with this option.
16067 @itemx -mhard-float
16069 @opindex mhard-float
16070 Generate output containing floating point instructions. This is the
16074 @itemx -msoft-float
16076 @opindex msoft-float
16077 Generate output containing library calls for floating point.
16078 @strong{Warning:} the requisite libraries are not available for all SPARC
16079 targets. Normally the facilities of the machine's usual C compiler are
16080 used, but this cannot be done directly in cross-compilation. You must make
16081 your own arrangements to provide suitable library functions for
16082 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16083 @samp{sparclite-*-*} do provide software floating point support.
16085 @option{-msoft-float} changes the calling convention in the output file;
16086 therefore, it is only useful if you compile @emph{all} of a program with
16087 this option. In particular, you need to compile @file{libgcc.a}, the
16088 library that comes with GCC, with @option{-msoft-float} in order for
16091 @item -mhard-quad-float
16092 @opindex mhard-quad-float
16093 Generate output containing quad-word (long double) floating point
16096 @item -msoft-quad-float
16097 @opindex msoft-quad-float
16098 Generate output containing library calls for quad-word (long double)
16099 floating point instructions. The functions called are those specified
16100 in the SPARC ABI@. This is the default.
16102 As of this writing, there are no SPARC implementations that have hardware
16103 support for the quad-word floating point instructions. They all invoke
16104 a trap handler for one of these instructions, and then the trap handler
16105 emulates the effect of the instruction. Because of the trap handler overhead,
16106 this is much slower than calling the ABI library routines. Thus the
16107 @option{-msoft-quad-float} option is the default.
16109 @item -mno-unaligned-doubles
16110 @itemx -munaligned-doubles
16111 @opindex mno-unaligned-doubles
16112 @opindex munaligned-doubles
16113 Assume that doubles have 8 byte alignment. This is the default.
16115 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16116 alignment only if they are contained in another type, or if they have an
16117 absolute address. Otherwise, it assumes they have 4 byte alignment.
16118 Specifying this option avoids some rare compatibility problems with code
16119 generated by other compilers. It is not the default because it results
16120 in a performance loss, especially for floating point code.
16122 @item -mno-faster-structs
16123 @itemx -mfaster-structs
16124 @opindex mno-faster-structs
16125 @opindex mfaster-structs
16126 With @option{-mfaster-structs}, the compiler assumes that structures
16127 should have 8 byte alignment. This enables the use of pairs of
16128 @code{ldd} and @code{std} instructions for copies in structure
16129 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16130 However, the use of this changed alignment directly violates the SPARC
16131 ABI@. Thus, it's intended only for use on targets where the developer
16132 acknowledges that their resulting code will not be directly in line with
16133 the rules of the ABI@.
16135 @item -mimpure-text
16136 @opindex mimpure-text
16137 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16138 the compiler to not pass @option{-z text} to the linker when linking a
16139 shared object. Using this option, you can link position-dependent
16140 code into a shared object.
16142 @option{-mimpure-text} suppresses the ``relocations remain against
16143 allocatable but non-writable sections'' linker error message.
16144 However, the necessary relocations will trigger copy-on-write, and the
16145 shared object is not actually shared across processes. Instead of
16146 using @option{-mimpure-text}, you should compile all source code with
16147 @option{-fpic} or @option{-fPIC}.
16149 This option is only available on SunOS and Solaris.
16151 @item -mcpu=@var{cpu_type}
16153 Set the instruction set, register set, and instruction scheduling parameters
16154 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16155 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16156 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16157 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16158 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16160 Default instruction scheduling parameters are used for values that select
16161 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16162 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16164 Here is a list of each supported architecture and their supported
16169 v8: supersparc, hypersparc
16170 sparclite: f930, f934, sparclite86x
16172 v9: ultrasparc, ultrasparc3, niagara, niagara2
16175 By default (unless configured otherwise), GCC generates code for the V7
16176 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16177 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16178 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16179 SPARCStation 1, 2, IPX etc.
16181 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16182 architecture. The only difference from V7 code is that the compiler emits
16183 the integer multiply and integer divide instructions which exist in SPARC-V8
16184 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16185 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16188 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16189 the SPARC architecture. This adds the integer multiply, integer divide step
16190 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16191 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16192 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16193 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16194 MB86934 chip, which is the more recent SPARClite with FPU@.
16196 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16197 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16198 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16199 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16200 optimizes it for the TEMIC SPARClet chip.
16202 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16203 architecture. This adds 64-bit integer and floating-point move instructions,
16204 3 additional floating-point condition code registers and conditional move
16205 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16206 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16207 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
16208 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
16209 @option{-mcpu=niagara}, the compiler additionally optimizes it for
16210 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
16211 additionally optimizes it for Sun UltraSPARC T2 chips.
16213 @item -mtune=@var{cpu_type}
16215 Set the instruction scheduling parameters for machine type
16216 @var{cpu_type}, but do not set the instruction set or register set that the
16217 option @option{-mcpu=@var{cpu_type}} would.
16219 The same values for @option{-mcpu=@var{cpu_type}} can be used for
16220 @option{-mtune=@var{cpu_type}}, but the only useful values are those
16221 that select a particular cpu implementation. Those are @samp{cypress},
16222 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
16223 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
16224 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
16229 @opindex mno-v8plus
16230 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
16231 difference from the V8 ABI is that the global and out registers are
16232 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
16233 mode for all SPARC-V9 processors.
16239 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
16240 Visual Instruction Set extensions. The default is @option{-mno-vis}.
16243 These @samp{-m} options are supported in addition to the above
16244 on SPARC-V9 processors in 64-bit environments:
16247 @item -mlittle-endian
16248 @opindex mlittle-endian
16249 Generate code for a processor running in little-endian mode. It is only
16250 available for a few configurations and most notably not on Solaris and Linux.
16256 Generate code for a 32-bit or 64-bit environment.
16257 The 32-bit environment sets int, long and pointer to 32 bits.
16258 The 64-bit environment sets int to 32 bits and long and pointer
16261 @item -mcmodel=medlow
16262 @opindex mcmodel=medlow
16263 Generate code for the Medium/Low code model: 64-bit addresses, programs
16264 must be linked in the low 32 bits of memory. Programs can be statically
16265 or dynamically linked.
16267 @item -mcmodel=medmid
16268 @opindex mcmodel=medmid
16269 Generate code for the Medium/Middle code model: 64-bit addresses, programs
16270 must be linked in the low 44 bits of memory, the text and data segments must
16271 be less than 2GB in size and the data segment must be located within 2GB of
16274 @item -mcmodel=medany
16275 @opindex mcmodel=medany
16276 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
16277 may be linked anywhere in memory, the text and data segments must be less
16278 than 2GB in size and the data segment must be located within 2GB of the
16281 @item -mcmodel=embmedany
16282 @opindex mcmodel=embmedany
16283 Generate code for the Medium/Anywhere code model for embedded systems:
16284 64-bit addresses, the text and data segments must be less than 2GB in
16285 size, both starting anywhere in memory (determined at link time). The
16286 global register %g4 points to the base of the data segment. Programs
16287 are statically linked and PIC is not supported.
16290 @itemx -mno-stack-bias
16291 @opindex mstack-bias
16292 @opindex mno-stack-bias
16293 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
16294 frame pointer if present, are offset by @minus{}2047 which must be added back
16295 when making stack frame references. This is the default in 64-bit mode.
16296 Otherwise, assume no such offset is present.
16299 These switches are supported in addition to the above on Solaris:
16304 Add support for multithreading using the Solaris threads library. This
16305 option sets flags for both the preprocessor and linker. This option does
16306 not affect the thread safety of object code produced by the compiler or
16307 that of libraries supplied with it.
16311 Add support for multithreading using the POSIX threads library. This
16312 option sets flags for both the preprocessor and linker. This option does
16313 not affect the thread safety of object code produced by the compiler or
16314 that of libraries supplied with it.
16318 This is a synonym for @option{-pthreads}.
16322 @subsection SPU Options
16323 @cindex SPU options
16325 These @samp{-m} options are supported on the SPU:
16329 @itemx -merror-reloc
16330 @opindex mwarn-reloc
16331 @opindex merror-reloc
16333 The loader for SPU does not handle dynamic relocations. By default, GCC
16334 will give an error when it generates code that requires a dynamic
16335 relocation. @option{-mno-error-reloc} disables the error,
16336 @option{-mwarn-reloc} will generate a warning instead.
16339 @itemx -munsafe-dma
16341 @opindex munsafe-dma
16343 Instructions which initiate or test completion of DMA must not be
16344 reordered with respect to loads and stores of the memory which is being
16345 accessed. Users typically address this problem using the volatile
16346 keyword, but that can lead to inefficient code in places where the
16347 memory is known to not change. Rather than mark the memory as volatile
16348 we treat the DMA instructions as potentially effecting all memory. With
16349 @option{-munsafe-dma} users must use the volatile keyword to protect
16352 @item -mbranch-hints
16353 @opindex mbranch-hints
16355 By default, GCC will generate a branch hint instruction to avoid
16356 pipeline stalls for always taken or probably taken branches. A hint
16357 will not be generated closer than 8 instructions away from its branch.
16358 There is little reason to disable them, except for debugging purposes,
16359 or to make an object a little bit smaller.
16363 @opindex msmall-mem
16364 @opindex mlarge-mem
16366 By default, GCC generates code assuming that addresses are never larger
16367 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
16368 a full 32 bit address.
16373 By default, GCC links against startup code that assumes the SPU-style
16374 main function interface (which has an unconventional parameter list).
16375 With @option{-mstdmain}, GCC will link your program against startup
16376 code that assumes a C99-style interface to @code{main}, including a
16377 local copy of @code{argv} strings.
16379 @item -mfixed-range=@var{register-range}
16380 @opindex mfixed-range
16381 Generate code treating the given register range as fixed registers.
16382 A fixed register is one that the register allocator can not use. This is
16383 useful when compiling kernel code. A register range is specified as
16384 two registers separated by a dash. Multiple register ranges can be
16385 specified separated by a comma.
16391 Compile code assuming that pointers to the PPU address space accessed
16392 via the @code{__ea} named address space qualifier are either 32 or 64
16393 bits wide. The default is 32 bits. As this is an ABI changing option,
16394 all object code in an executable must be compiled with the same setting.
16396 @item -maddress-space-conversion
16397 @itemx -mno-address-space-conversion
16398 @opindex maddress-space-conversion
16399 @opindex mno-address-space-conversion
16400 Allow/disallow treating the @code{__ea} address space as superset
16401 of the generic address space. This enables explicit type casts
16402 between @code{__ea} and generic pointer as well as implicit
16403 conversions of generic pointers to @code{__ea} pointers. The
16404 default is to allow address space pointer conversions.
16406 @item -mcache-size=@var{cache-size}
16407 @opindex mcache-size
16408 This option controls the version of libgcc that the compiler links to an
16409 executable and selects a software-managed cache for accessing variables
16410 in the @code{__ea} address space with a particular cache size. Possible
16411 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
16412 and @samp{128}. The default cache size is 64KB.
16414 @item -matomic-updates
16415 @itemx -mno-atomic-updates
16416 @opindex matomic-updates
16417 @opindex mno-atomic-updates
16418 This option controls the version of libgcc that the compiler links to an
16419 executable and selects whether atomic updates to the software-managed
16420 cache of PPU-side variables are used. If you use atomic updates, changes
16421 to a PPU variable from SPU code using the @code{__ea} named address space
16422 qualifier will not interfere with changes to other PPU variables residing
16423 in the same cache line from PPU code. If you do not use atomic updates,
16424 such interference may occur; however, writing back cache lines will be
16425 more efficient. The default behavior is to use atomic updates.
16428 @itemx -mdual-nops=@var{n}
16429 @opindex mdual-nops
16430 By default, GCC will insert nops to increase dual issue when it expects
16431 it to increase performance. @var{n} can be a value from 0 to 10. A
16432 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
16433 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
16435 @item -mhint-max-nops=@var{n}
16436 @opindex mhint-max-nops
16437 Maximum number of nops to insert for a branch hint. A branch hint must
16438 be at least 8 instructions away from the branch it is effecting. GCC
16439 will insert up to @var{n} nops to enforce this, otherwise it will not
16440 generate the branch hint.
16442 @item -mhint-max-distance=@var{n}
16443 @opindex mhint-max-distance
16444 The encoding of the branch hint instruction limits the hint to be within
16445 256 instructions of the branch it is effecting. By default, GCC makes
16446 sure it is within 125.
16449 @opindex msafe-hints
16450 Work around a hardware bug which causes the SPU to stall indefinitely.
16451 By default, GCC will insert the @code{hbrp} instruction to make sure
16452 this stall won't happen.
16456 @node System V Options
16457 @subsection Options for System V
16459 These additional options are available on System V Release 4 for
16460 compatibility with other compilers on those systems:
16465 Create a shared object.
16466 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16470 Identify the versions of each tool used by the compiler, in a
16471 @code{.ident} assembler directive in the output.
16475 Refrain from adding @code{.ident} directives to the output file (this is
16478 @item -YP,@var{dirs}
16480 Search the directories @var{dirs}, and no others, for libraries
16481 specified with @option{-l}.
16483 @item -Ym,@var{dir}
16485 Look in the directory @var{dir} to find the M4 preprocessor.
16486 The assembler uses this option.
16487 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16488 @c the generic assembler that comes with Solaris takes just -Ym.
16492 @subsection V850 Options
16493 @cindex V850 Options
16495 These @samp{-m} options are defined for V850 implementations:
16499 @itemx -mno-long-calls
16500 @opindex mlong-calls
16501 @opindex mno-long-calls
16502 Treat all calls as being far away (near). If calls are assumed to be
16503 far away, the compiler will always load the functions address up into a
16504 register, and call indirect through the pointer.
16510 Do not optimize (do optimize) basic blocks that use the same index
16511 pointer 4 or more times to copy pointer into the @code{ep} register, and
16512 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16513 option is on by default if you optimize.
16515 @item -mno-prolog-function
16516 @itemx -mprolog-function
16517 @opindex mno-prolog-function
16518 @opindex mprolog-function
16519 Do not use (do use) external functions to save and restore registers
16520 at the prologue and epilogue of a function. The external functions
16521 are slower, but use less code space if more than one function saves
16522 the same number of registers. The @option{-mprolog-function} option
16523 is on by default if you optimize.
16527 Try to make the code as small as possible. At present, this just turns
16528 on the @option{-mep} and @option{-mprolog-function} options.
16530 @item -mtda=@var{n}
16532 Put static or global variables whose size is @var{n} bytes or less into
16533 the tiny data area that register @code{ep} points to. The tiny data
16534 area can hold up to 256 bytes in total (128 bytes for byte references).
16536 @item -msda=@var{n}
16538 Put static or global variables whose size is @var{n} bytes or less into
16539 the small data area that register @code{gp} points to. The small data
16540 area can hold up to 64 kilobytes.
16542 @item -mzda=@var{n}
16544 Put static or global variables whose size is @var{n} bytes or less into
16545 the first 32 kilobytes of memory.
16549 Specify that the target processor is the V850.
16552 @opindex mbig-switch
16553 Generate code suitable for big switch tables. Use this option only if
16554 the assembler/linker complain about out of range branches within a switch
16559 This option will cause r2 and r5 to be used in the code generated by
16560 the compiler. This setting is the default.
16562 @item -mno-app-regs
16563 @opindex mno-app-regs
16564 This option will cause r2 and r5 to be treated as fixed registers.
16568 Specify that the target processor is the V850E1. The preprocessor
16569 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16570 this option is used.
16574 Specify that the target processor is the V850E@. The preprocessor
16575 constant @samp{__v850e__} will be defined if this option is used.
16577 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16578 are defined then a default target processor will be chosen and the
16579 relevant @samp{__v850*__} preprocessor constant will be defined.
16581 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16582 defined, regardless of which processor variant is the target.
16584 @item -mdisable-callt
16585 @opindex mdisable-callt
16586 This option will suppress generation of the CALLT instruction for the
16587 v850e and v850e1 flavors of the v850 architecture. The default is
16588 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16593 @subsection VAX Options
16594 @cindex VAX options
16596 These @samp{-m} options are defined for the VAX:
16601 Do not output certain jump instructions (@code{aobleq} and so on)
16602 that the Unix assembler for the VAX cannot handle across long
16607 Do output those jump instructions, on the assumption that you
16608 will assemble with the GNU assembler.
16612 Output code for g-format floating point numbers instead of d-format.
16615 @node VxWorks Options
16616 @subsection VxWorks Options
16617 @cindex VxWorks Options
16619 The options in this section are defined for all VxWorks targets.
16620 Options specific to the target hardware are listed with the other
16621 options for that target.
16626 GCC can generate code for both VxWorks kernels and real time processes
16627 (RTPs). This option switches from the former to the latter. It also
16628 defines the preprocessor macro @code{__RTP__}.
16631 @opindex non-static
16632 Link an RTP executable against shared libraries rather than static
16633 libraries. The options @option{-static} and @option{-shared} can
16634 also be used for RTPs (@pxref{Link Options}); @option{-static}
16641 These options are passed down to the linker. They are defined for
16642 compatibility with Diab.
16645 @opindex Xbind-lazy
16646 Enable lazy binding of function calls. This option is equivalent to
16647 @option{-Wl,-z,now} and is defined for compatibility with Diab.
16651 Disable lazy binding of function calls. This option is the default and
16652 is defined for compatibility with Diab.
16655 @node x86-64 Options
16656 @subsection x86-64 Options
16657 @cindex x86-64 options
16659 These are listed under @xref{i386 and x86-64 Options}.
16661 @node i386 and x86-64 Windows Options
16662 @subsection i386 and x86-64 Windows Options
16663 @cindex i386 and x86-64 Windows Options
16665 These additional options are available for Windows targets:
16670 This option is available for Cygwin and MinGW targets. It
16671 specifies that a console application is to be generated, by
16672 instructing the linker to set the PE header subsystem type
16673 required for console applications.
16674 This is the default behavior for Cygwin and MinGW targets.
16678 This option is available for Cygwin targets. It specifies that
16679 the Cygwin internal interface is to be used for predefined
16680 preprocessor macros, C runtime libraries and related linker
16681 paths and options. For Cygwin targets this is the default behavior.
16682 This option is deprecated and will be removed in a future release.
16685 @opindex mno-cygwin
16686 This option is available for Cygwin targets. It specifies that
16687 the MinGW internal interface is to be used instead of Cygwin's, by
16688 setting MinGW-related predefined macros and linker paths and default
16690 This option is deprecated and will be removed in a future release.
16694 This option is available for Cygwin and MinGW targets. It
16695 specifies that a DLL - a dynamic link library - is to be
16696 generated, enabling the selection of the required runtime
16697 startup object and entry point.
16699 @item -mnop-fun-dllimport
16700 @opindex mnop-fun-dllimport
16701 This option is available for Cygwin and MinGW targets. It
16702 specifies that the dllimport attribute should be ignored.
16706 This option is available for MinGW targets. It specifies
16707 that MinGW-specific thread support is to be used.
16711 This option is available for mingw-w64 targets. It specifies
16712 that the UNICODE macro is getting pre-defined and that the
16713 unicode capable runtime startup code is choosen.
16717 This option is available for Cygwin and MinGW targets. It
16718 specifies that the typical Windows pre-defined macros are to
16719 be set in the pre-processor, but does not influence the choice
16720 of runtime library/startup code.
16724 This option is available for Cygwin and MinGW targets. It
16725 specifies that a GUI application is to be generated by
16726 instructing the linker to set the PE header subsystem type
16729 @item -fno-set-stack-executable
16730 @opindex fno-set-stack-executable
16731 This option is available for MinGW targets. It specifies that
16732 the executable flag for stack used by nested functions isn't
16733 set. This is necessary for binaries running in kernel mode of
16734 Windows, as there the user32 API, which is used to set executable
16735 privileges, isn't available.
16737 @item -mpe-aligned-commons
16738 @opindex mpe-aligned-commons
16739 This option is available for Cygwin and MinGW targets. It
16740 specifies that the GNU extension to the PE file format that
16741 permits the correct alignment of COMMON variables should be
16742 used when generating code. It will be enabled by default if
16743 GCC detects that the target assembler found during configuration
16744 supports the feature.
16747 See also under @ref{i386 and x86-64 Options} for standard options.
16749 @node Xstormy16 Options
16750 @subsection Xstormy16 Options
16751 @cindex Xstormy16 Options
16753 These options are defined for Xstormy16:
16758 Choose startup files and linker script suitable for the simulator.
16761 @node Xtensa Options
16762 @subsection Xtensa Options
16763 @cindex Xtensa Options
16765 These options are supported for Xtensa targets:
16769 @itemx -mno-const16
16771 @opindex mno-const16
16772 Enable or disable use of @code{CONST16} instructions for loading
16773 constant values. The @code{CONST16} instruction is currently not a
16774 standard option from Tensilica. When enabled, @code{CONST16}
16775 instructions are always used in place of the standard @code{L32R}
16776 instructions. The use of @code{CONST16} is enabled by default only if
16777 the @code{L32R} instruction is not available.
16780 @itemx -mno-fused-madd
16781 @opindex mfused-madd
16782 @opindex mno-fused-madd
16783 Enable or disable use of fused multiply/add and multiply/subtract
16784 instructions in the floating-point option. This has no effect if the
16785 floating-point option is not also enabled. Disabling fused multiply/add
16786 and multiply/subtract instructions forces the compiler to use separate
16787 instructions for the multiply and add/subtract operations. This may be
16788 desirable in some cases where strict IEEE 754-compliant results are
16789 required: the fused multiply add/subtract instructions do not round the
16790 intermediate result, thereby producing results with @emph{more} bits of
16791 precision than specified by the IEEE standard. Disabling fused multiply
16792 add/subtract instructions also ensures that the program output is not
16793 sensitive to the compiler's ability to combine multiply and add/subtract
16796 @item -mserialize-volatile
16797 @itemx -mno-serialize-volatile
16798 @opindex mserialize-volatile
16799 @opindex mno-serialize-volatile
16800 When this option is enabled, GCC inserts @code{MEMW} instructions before
16801 @code{volatile} memory references to guarantee sequential consistency.
16802 The default is @option{-mserialize-volatile}. Use
16803 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
16805 @item -mtext-section-literals
16806 @itemx -mno-text-section-literals
16807 @opindex mtext-section-literals
16808 @opindex mno-text-section-literals
16809 Control the treatment of literal pools. The default is
16810 @option{-mno-text-section-literals}, which places literals in a separate
16811 section in the output file. This allows the literal pool to be placed
16812 in a data RAM/ROM, and it also allows the linker to combine literal
16813 pools from separate object files to remove redundant literals and
16814 improve code size. With @option{-mtext-section-literals}, the literals
16815 are interspersed in the text section in order to keep them as close as
16816 possible to their references. This may be necessary for large assembly
16819 @item -mtarget-align
16820 @itemx -mno-target-align
16821 @opindex mtarget-align
16822 @opindex mno-target-align
16823 When this option is enabled, GCC instructs the assembler to
16824 automatically align instructions to reduce branch penalties at the
16825 expense of some code density. The assembler attempts to widen density
16826 instructions to align branch targets and the instructions following call
16827 instructions. If there are not enough preceding safe density
16828 instructions to align a target, no widening will be performed. The
16829 default is @option{-mtarget-align}. These options do not affect the
16830 treatment of auto-aligned instructions like @code{LOOP}, which the
16831 assembler will always align, either by widening density instructions or
16832 by inserting no-op instructions.
16835 @itemx -mno-longcalls
16836 @opindex mlongcalls
16837 @opindex mno-longcalls
16838 When this option is enabled, GCC instructs the assembler to translate
16839 direct calls to indirect calls unless it can determine that the target
16840 of a direct call is in the range allowed by the call instruction. This
16841 translation typically occurs for calls to functions in other source
16842 files. Specifically, the assembler translates a direct @code{CALL}
16843 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
16844 The default is @option{-mno-longcalls}. This option should be used in
16845 programs where the call target can potentially be out of range. This
16846 option is implemented in the assembler, not the compiler, so the
16847 assembly code generated by GCC will still show direct call
16848 instructions---look at the disassembled object code to see the actual
16849 instructions. Note that the assembler will use an indirect call for
16850 every cross-file call, not just those that really will be out of range.
16853 @node zSeries Options
16854 @subsection zSeries Options
16855 @cindex zSeries options
16857 These are listed under @xref{S/390 and zSeries Options}.
16859 @node Code Gen Options
16860 @section Options for Code Generation Conventions
16861 @cindex code generation conventions
16862 @cindex options, code generation
16863 @cindex run-time options
16865 These machine-independent options control the interface conventions
16866 used in code generation.
16868 Most of them have both positive and negative forms; the negative form
16869 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
16870 one of the forms is listed---the one which is not the default. You
16871 can figure out the other form by either removing @samp{no-} or adding
16875 @item -fbounds-check
16876 @opindex fbounds-check
16877 For front-ends that support it, generate additional code to check that
16878 indices used to access arrays are within the declared range. This is
16879 currently only supported by the Java and Fortran front-ends, where
16880 this option defaults to true and false respectively.
16884 This option generates traps for signed overflow on addition, subtraction,
16885 multiplication operations.
16889 This option instructs the compiler to assume that signed arithmetic
16890 overflow of addition, subtraction and multiplication wraps around
16891 using twos-complement representation. This flag enables some optimizations
16892 and disables others. This option is enabled by default for the Java
16893 front-end, as required by the Java language specification.
16896 @opindex fexceptions
16897 Enable exception handling. Generates extra code needed to propagate
16898 exceptions. For some targets, this implies GCC will generate frame
16899 unwind information for all functions, which can produce significant data
16900 size overhead, although it does not affect execution. If you do not
16901 specify this option, GCC will enable it by default for languages like
16902 C++ which normally require exception handling, and disable it for
16903 languages like C that do not normally require it. However, you may need
16904 to enable this option when compiling C code that needs to interoperate
16905 properly with exception handlers written in C++. You may also wish to
16906 disable this option if you are compiling older C++ programs that don't
16907 use exception handling.
16909 @item -fnon-call-exceptions
16910 @opindex fnon-call-exceptions
16911 Generate code that allows trapping instructions to throw exceptions.
16912 Note that this requires platform-specific runtime support that does
16913 not exist everywhere. Moreover, it only allows @emph{trapping}
16914 instructions to throw exceptions, i.e.@: memory references or floating
16915 point instructions. It does not allow exceptions to be thrown from
16916 arbitrary signal handlers such as @code{SIGALRM}.
16918 @item -funwind-tables
16919 @opindex funwind-tables
16920 Similar to @option{-fexceptions}, except that it will just generate any needed
16921 static data, but will not affect the generated code in any other way.
16922 You will normally not enable this option; instead, a language processor
16923 that needs this handling would enable it on your behalf.
16925 @item -fasynchronous-unwind-tables
16926 @opindex fasynchronous-unwind-tables
16927 Generate unwind table in dwarf2 format, if supported by target machine. The
16928 table is exact at each instruction boundary, so it can be used for stack
16929 unwinding from asynchronous events (such as debugger or garbage collector).
16931 @item -fpcc-struct-return
16932 @opindex fpcc-struct-return
16933 Return ``short'' @code{struct} and @code{union} values in memory like
16934 longer ones, rather than in registers. This convention is less
16935 efficient, but it has the advantage of allowing intercallability between
16936 GCC-compiled files and files compiled with other compilers, particularly
16937 the Portable C Compiler (pcc).
16939 The precise convention for returning structures in memory depends
16940 on the target configuration macros.
16942 Short structures and unions are those whose size and alignment match
16943 that of some integer type.
16945 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
16946 switch is not binary compatible with code compiled with the
16947 @option{-freg-struct-return} switch.
16948 Use it to conform to a non-default application binary interface.
16950 @item -freg-struct-return
16951 @opindex freg-struct-return
16952 Return @code{struct} and @code{union} values in registers when possible.
16953 This is more efficient for small structures than
16954 @option{-fpcc-struct-return}.
16956 If you specify neither @option{-fpcc-struct-return} nor
16957 @option{-freg-struct-return}, GCC defaults to whichever convention is
16958 standard for the target. If there is no standard convention, GCC
16959 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
16960 the principal compiler. In those cases, we can choose the standard, and
16961 we chose the more efficient register return alternative.
16963 @strong{Warning:} code compiled with the @option{-freg-struct-return}
16964 switch is not binary compatible with code compiled with the
16965 @option{-fpcc-struct-return} switch.
16966 Use it to conform to a non-default application binary interface.
16968 @item -fshort-enums
16969 @opindex fshort-enums
16970 Allocate to an @code{enum} type only as many bytes as it needs for the
16971 declared range of possible values. Specifically, the @code{enum} type
16972 will be equivalent to the smallest integer type which has enough room.
16974 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
16975 code that is not binary compatible with code generated without that switch.
16976 Use it to conform to a non-default application binary interface.
16978 @item -fshort-double
16979 @opindex fshort-double
16980 Use the same size for @code{double} as for @code{float}.
16982 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
16983 code that is not binary compatible with code generated without that switch.
16984 Use it to conform to a non-default application binary interface.
16986 @item -fshort-wchar
16987 @opindex fshort-wchar
16988 Override the underlying type for @samp{wchar_t} to be @samp{short
16989 unsigned int} instead of the default for the target. This option is
16990 useful for building programs to run under WINE@.
16992 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
16993 code that is not binary compatible with code generated without that switch.
16994 Use it to conform to a non-default application binary interface.
16997 @opindex fno-common
16998 In C code, controls the placement of uninitialized global variables.
16999 Unix C compilers have traditionally permitted multiple definitions of
17000 such variables in different compilation units by placing the variables
17002 This is the behavior specified by @option{-fcommon}, and is the default
17003 for GCC on most targets.
17004 On the other hand, this behavior is not required by ISO C, and on some
17005 targets may carry a speed or code size penalty on variable references.
17006 The @option{-fno-common} option specifies that the compiler should place
17007 uninitialized global variables in the data section of the object file,
17008 rather than generating them as common blocks.
17009 This has the effect that if the same variable is declared
17010 (without @code{extern}) in two different compilations,
17011 you will get a multiple-definition error when you link them.
17012 In this case, you must compile with @option{-fcommon} instead.
17013 Compiling with @option{-fno-common} is useful on targets for which
17014 it provides better performance, or if you wish to verify that the
17015 program will work on other systems which always treat uninitialized
17016 variable declarations this way.
17020 Ignore the @samp{#ident} directive.
17022 @item -finhibit-size-directive
17023 @opindex finhibit-size-directive
17024 Don't output a @code{.size} assembler directive, or anything else that
17025 would cause trouble if the function is split in the middle, and the
17026 two halves are placed at locations far apart in memory. This option is
17027 used when compiling @file{crtstuff.c}; you should not need to use it
17030 @item -fverbose-asm
17031 @opindex fverbose-asm
17032 Put extra commentary information in the generated assembly code to
17033 make it more readable. This option is generally only of use to those
17034 who actually need to read the generated assembly code (perhaps while
17035 debugging the compiler itself).
17037 @option{-fno-verbose-asm}, the default, causes the
17038 extra information to be omitted and is useful when comparing two assembler
17041 @item -frecord-gcc-switches
17042 @opindex frecord-gcc-switches
17043 This switch causes the command line that was used to invoke the
17044 compiler to be recorded into the object file that is being created.
17045 This switch is only implemented on some targets and the exact format
17046 of the recording is target and binary file format dependent, but it
17047 usually takes the form of a section containing ASCII text. This
17048 switch is related to the @option{-fverbose-asm} switch, but that
17049 switch only records information in the assembler output file as
17050 comments, so it never reaches the object file.
17054 @cindex global offset table
17056 Generate position-independent code (PIC) suitable for use in a shared
17057 library, if supported for the target machine. Such code accesses all
17058 constant addresses through a global offset table (GOT)@. The dynamic
17059 loader resolves the GOT entries when the program starts (the dynamic
17060 loader is not part of GCC; it is part of the operating system). If
17061 the GOT size for the linked executable exceeds a machine-specific
17062 maximum size, you get an error message from the linker indicating that
17063 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17064 instead. (These maximums are 8k on the SPARC and 32k
17065 on the m68k and RS/6000. The 386 has no such limit.)
17067 Position-independent code requires special support, and therefore works
17068 only on certain machines. For the 386, GCC supports PIC for System V
17069 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17070 position-independent.
17072 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17077 If supported for the target machine, emit position-independent code,
17078 suitable for dynamic linking and avoiding any limit on the size of the
17079 global offset table. This option makes a difference on the m68k,
17080 PowerPC and SPARC@.
17082 Position-independent code requires special support, and therefore works
17083 only on certain machines.
17085 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17092 These options are similar to @option{-fpic} and @option{-fPIC}, but
17093 generated position independent code can be only linked into executables.
17094 Usually these options are used when @option{-pie} GCC option will be
17095 used during linking.
17097 @option{-fpie} and @option{-fPIE} both define the macros
17098 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17099 for @option{-fpie} and 2 for @option{-fPIE}.
17101 @item -fno-jump-tables
17102 @opindex fno-jump-tables
17103 Do not use jump tables for switch statements even where it would be
17104 more efficient than other code generation strategies. This option is
17105 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17106 building code which forms part of a dynamic linker and cannot
17107 reference the address of a jump table. On some targets, jump tables
17108 do not require a GOT and this option is not needed.
17110 @item -ffixed-@var{reg}
17112 Treat the register named @var{reg} as a fixed register; generated code
17113 should never refer to it (except perhaps as a stack pointer, frame
17114 pointer or in some other fixed role).
17116 @var{reg} must be the name of a register. The register names accepted
17117 are machine-specific and are defined in the @code{REGISTER_NAMES}
17118 macro in the machine description macro file.
17120 This flag does not have a negative form, because it specifies a
17123 @item -fcall-used-@var{reg}
17124 @opindex fcall-used
17125 Treat the register named @var{reg} as an allocable register that is
17126 clobbered by function calls. It may be allocated for temporaries or
17127 variables that do not live across a call. Functions compiled this way
17128 will not save and restore the register @var{reg}.
17130 It is an error to used this flag with the frame pointer or stack pointer.
17131 Use of this flag for other registers that have fixed pervasive roles in
17132 the machine's execution model will produce disastrous results.
17134 This flag does not have a negative form, because it specifies a
17137 @item -fcall-saved-@var{reg}
17138 @opindex fcall-saved
17139 Treat the register named @var{reg} as an allocable register saved by
17140 functions. It may be allocated even for temporaries or variables that
17141 live across a call. Functions compiled this way will save and restore
17142 the register @var{reg} if they use it.
17144 It is an error to used this flag with the frame pointer or stack pointer.
17145 Use of this flag for other registers that have fixed pervasive roles in
17146 the machine's execution model will produce disastrous results.
17148 A different sort of disaster will result from the use of this flag for
17149 a register in which function values may be returned.
17151 This flag does not have a negative form, because it specifies a
17154 @item -fpack-struct[=@var{n}]
17155 @opindex fpack-struct
17156 Without a value specified, pack all structure members together without
17157 holes. When a value is specified (which must be a small power of two), pack
17158 structure members according to this value, representing the maximum
17159 alignment (that is, objects with default alignment requirements larger than
17160 this will be output potentially unaligned at the next fitting location.
17162 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17163 code that is not binary compatible with code generated without that switch.
17164 Additionally, it makes the code suboptimal.
17165 Use it to conform to a non-default application binary interface.
17167 @item -finstrument-functions
17168 @opindex finstrument-functions
17169 Generate instrumentation calls for entry and exit to functions. Just
17170 after function entry and just before function exit, the following
17171 profiling functions will be called with the address of the current
17172 function and its call site. (On some platforms,
17173 @code{__builtin_return_address} does not work beyond the current
17174 function, so the call site information may not be available to the
17175 profiling functions otherwise.)
17178 void __cyg_profile_func_enter (void *this_fn,
17180 void __cyg_profile_func_exit (void *this_fn,
17184 The first argument is the address of the start of the current function,
17185 which may be looked up exactly in the symbol table.
17187 This instrumentation is also done for functions expanded inline in other
17188 functions. The profiling calls will indicate where, conceptually, the
17189 inline function is entered and exited. This means that addressable
17190 versions of such functions must be available. If all your uses of a
17191 function are expanded inline, this may mean an additional expansion of
17192 code size. If you use @samp{extern inline} in your C code, an
17193 addressable version of such functions must be provided. (This is
17194 normally the case anyways, but if you get lucky and the optimizer always
17195 expands the functions inline, you might have gotten away without
17196 providing static copies.)
17198 A function may be given the attribute @code{no_instrument_function}, in
17199 which case this instrumentation will not be done. This can be used, for
17200 example, for the profiling functions listed above, high-priority
17201 interrupt routines, and any functions from which the profiling functions
17202 cannot safely be called (perhaps signal handlers, if the profiling
17203 routines generate output or allocate memory).
17205 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17206 @opindex finstrument-functions-exclude-file-list
17208 Set the list of functions that are excluded from instrumentation (see
17209 the description of @code{-finstrument-functions}). If the file that
17210 contains a function definition matches with one of @var{file}, then
17211 that function is not instrumented. The match is done on substrings:
17212 if the @var{file} parameter is a substring of the file name, it is
17213 considered to be a match.
17216 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
17217 will exclude any inline function defined in files whose pathnames
17218 contain @code{/bits/stl} or @code{include/sys}.
17220 If, for some reason, you want to include letter @code{','} in one of
17221 @var{sym}, write @code{'\,'}. For example,
17222 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17223 (note the single quote surrounding the option).
17225 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
17226 @opindex finstrument-functions-exclude-function-list
17228 This is similar to @code{-finstrument-functions-exclude-file-list},
17229 but this option sets the list of function names to be excluded from
17230 instrumentation. The function name to be matched is its user-visible
17231 name, such as @code{vector<int> blah(const vector<int> &)}, not the
17232 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
17233 match is done on substrings: if the @var{sym} parameter is a substring
17234 of the function name, it is considered to be a match. For C99 and C++
17235 extended identifiers, the function name must be given in UTF-8, not
17236 using universal character names.
17238 @item -fstack-check
17239 @opindex fstack-check
17240 Generate code to verify that you do not go beyond the boundary of the
17241 stack. You should specify this flag if you are running in an
17242 environment with multiple threads, but only rarely need to specify it in
17243 a single-threaded environment since stack overflow is automatically
17244 detected on nearly all systems if there is only one stack.
17246 Note that this switch does not actually cause checking to be done; the
17247 operating system or the language runtime must do that. The switch causes
17248 generation of code to ensure that they see the stack being extended.
17250 You can additionally specify a string parameter: @code{no} means no
17251 checking, @code{generic} means force the use of old-style checking,
17252 @code{specific} means use the best checking method and is equivalent
17253 to bare @option{-fstack-check}.
17255 Old-style checking is a generic mechanism that requires no specific
17256 target support in the compiler but comes with the following drawbacks:
17260 Modified allocation strategy for large objects: they will always be
17261 allocated dynamically if their size exceeds a fixed threshold.
17264 Fixed limit on the size of the static frame of functions: when it is
17265 topped by a particular function, stack checking is not reliable and
17266 a warning is issued by the compiler.
17269 Inefficiency: because of both the modified allocation strategy and the
17270 generic implementation, the performances of the code are hampered.
17273 Note that old-style stack checking is also the fallback method for
17274 @code{specific} if no target support has been added in the compiler.
17276 @item -fstack-limit-register=@var{reg}
17277 @itemx -fstack-limit-symbol=@var{sym}
17278 @itemx -fno-stack-limit
17279 @opindex fstack-limit-register
17280 @opindex fstack-limit-symbol
17281 @opindex fno-stack-limit
17282 Generate code to ensure that the stack does not grow beyond a certain value,
17283 either the value of a register or the address of a symbol. If the stack
17284 would grow beyond the value, a signal is raised. For most targets,
17285 the signal is raised before the stack overruns the boundary, so
17286 it is possible to catch the signal without taking special precautions.
17288 For instance, if the stack starts at absolute address @samp{0x80000000}
17289 and grows downwards, you can use the flags
17290 @option{-fstack-limit-symbol=__stack_limit} and
17291 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
17292 of 128KB@. Note that this may only work with the GNU linker.
17294 @cindex aliasing of parameters
17295 @cindex parameters, aliased
17296 @item -fargument-alias
17297 @itemx -fargument-noalias
17298 @itemx -fargument-noalias-global
17299 @itemx -fargument-noalias-anything
17300 @opindex fargument-alias
17301 @opindex fargument-noalias
17302 @opindex fargument-noalias-global
17303 @opindex fargument-noalias-anything
17304 Specify the possible relationships among parameters and between
17305 parameters and global data.
17307 @option{-fargument-alias} specifies that arguments (parameters) may
17308 alias each other and may alias global storage.@*
17309 @option{-fargument-noalias} specifies that arguments do not alias
17310 each other, but may alias global storage.@*
17311 @option{-fargument-noalias-global} specifies that arguments do not
17312 alias each other and do not alias global storage.
17313 @option{-fargument-noalias-anything} specifies that arguments do not
17314 alias any other storage.
17316 Each language will automatically use whatever option is required by
17317 the language standard. You should not need to use these options yourself.
17319 @item -fleading-underscore
17320 @opindex fleading-underscore
17321 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
17322 change the way C symbols are represented in the object file. One use
17323 is to help link with legacy assembly code.
17325 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
17326 generate code that is not binary compatible with code generated without that
17327 switch. Use it to conform to a non-default application binary interface.
17328 Not all targets provide complete support for this switch.
17330 @item -ftls-model=@var{model}
17331 @opindex ftls-model
17332 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
17333 The @var{model} argument should be one of @code{global-dynamic},
17334 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
17336 The default without @option{-fpic} is @code{initial-exec}; with
17337 @option{-fpic} the default is @code{global-dynamic}.
17339 @item -fvisibility=@var{default|internal|hidden|protected}
17340 @opindex fvisibility
17341 Set the default ELF image symbol visibility to the specified option---all
17342 symbols will be marked with this unless overridden within the code.
17343 Using this feature can very substantially improve linking and
17344 load times of shared object libraries, produce more optimized
17345 code, provide near-perfect API export and prevent symbol clashes.
17346 It is @strong{strongly} recommended that you use this in any shared objects
17349 Despite the nomenclature, @code{default} always means public ie;
17350 available to be linked against from outside the shared object.
17351 @code{protected} and @code{internal} are pretty useless in real-world
17352 usage so the only other commonly used option will be @code{hidden}.
17353 The default if @option{-fvisibility} isn't specified is
17354 @code{default}, i.e., make every
17355 symbol public---this causes the same behavior as previous versions of
17358 A good explanation of the benefits offered by ensuring ELF
17359 symbols have the correct visibility is given by ``How To Write
17360 Shared Libraries'' by Ulrich Drepper (which can be found at
17361 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
17362 solution made possible by this option to marking things hidden when
17363 the default is public is to make the default hidden and mark things
17364 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
17365 and @code{__attribute__ ((visibility("default")))} instead of
17366 @code{__declspec(dllexport)} you get almost identical semantics with
17367 identical syntax. This is a great boon to those working with
17368 cross-platform projects.
17370 For those adding visibility support to existing code, you may find
17371 @samp{#pragma GCC visibility} of use. This works by you enclosing
17372 the declarations you wish to set visibility for with (for example)
17373 @samp{#pragma GCC visibility push(hidden)} and
17374 @samp{#pragma GCC visibility pop}.
17375 Bear in mind that symbol visibility should be viewed @strong{as
17376 part of the API interface contract} and thus all new code should
17377 always specify visibility when it is not the default ie; declarations
17378 only for use within the local DSO should @strong{always} be marked explicitly
17379 as hidden as so to avoid PLT indirection overheads---making this
17380 abundantly clear also aids readability and self-documentation of the code.
17381 Note that due to ISO C++ specification requirements, operator new and
17382 operator delete must always be of default visibility.
17384 Be aware that headers from outside your project, in particular system
17385 headers and headers from any other library you use, may not be
17386 expecting to be compiled with visibility other than the default. You
17387 may need to explicitly say @samp{#pragma GCC visibility push(default)}
17388 before including any such headers.
17390 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
17391 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
17392 no modifications. However, this means that calls to @samp{extern}
17393 functions with no explicit visibility will use the PLT, so it is more
17394 effective to use @samp{__attribute ((visibility))} and/or
17395 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
17396 declarations should be treated as hidden.
17398 Note that @samp{-fvisibility} does affect C++ vague linkage
17399 entities. This means that, for instance, an exception class that will
17400 be thrown between DSOs must be explicitly marked with default
17401 visibility so that the @samp{type_info} nodes will be unified between
17404 An overview of these techniques, their benefits and how to use them
17405 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
17411 @node Environment Variables
17412 @section Environment Variables Affecting GCC
17413 @cindex environment variables
17415 @c man begin ENVIRONMENT
17416 This section describes several environment variables that affect how GCC
17417 operates. Some of them work by specifying directories or prefixes to use
17418 when searching for various kinds of files. Some are used to specify other
17419 aspects of the compilation environment.
17421 Note that you can also specify places to search using options such as
17422 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
17423 take precedence over places specified using environment variables, which
17424 in turn take precedence over those specified by the configuration of GCC@.
17425 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
17426 GNU Compiler Collection (GCC) Internals}.
17431 @c @itemx LC_COLLATE
17433 @c @itemx LC_MONETARY
17434 @c @itemx LC_NUMERIC
17439 @c @findex LC_COLLATE
17440 @findex LC_MESSAGES
17441 @c @findex LC_MONETARY
17442 @c @findex LC_NUMERIC
17446 These environment variables control the way that GCC uses
17447 localization information that allow GCC to work with different
17448 national conventions. GCC inspects the locale categories
17449 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
17450 so. These locale categories can be set to any value supported by your
17451 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
17452 Kingdom encoded in UTF-8.
17454 The @env{LC_CTYPE} environment variable specifies character
17455 classification. GCC uses it to determine the character boundaries in
17456 a string; this is needed for some multibyte encodings that contain quote
17457 and escape characters that would otherwise be interpreted as a string
17460 The @env{LC_MESSAGES} environment variable specifies the language to
17461 use in diagnostic messages.
17463 If the @env{LC_ALL} environment variable is set, it overrides the value
17464 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17465 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17466 environment variable. If none of these variables are set, GCC
17467 defaults to traditional C English behavior.
17471 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17472 files. GCC uses temporary files to hold the output of one stage of
17473 compilation which is to be used as input to the next stage: for example,
17474 the output of the preprocessor, which is the input to the compiler
17477 @item GCC_EXEC_PREFIX
17478 @findex GCC_EXEC_PREFIX
17479 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17480 names of the subprograms executed by the compiler. No slash is added
17481 when this prefix is combined with the name of a subprogram, but you can
17482 specify a prefix that ends with a slash if you wish.
17484 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17485 an appropriate prefix to use based on the pathname it was invoked with.
17487 If GCC cannot find the subprogram using the specified prefix, it
17488 tries looking in the usual places for the subprogram.
17490 The default value of @env{GCC_EXEC_PREFIX} is
17491 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17492 the installed compiler. In many cases @var{prefix} is the value
17493 of @code{prefix} when you ran the @file{configure} script.
17495 Other prefixes specified with @option{-B} take precedence over this prefix.
17497 This prefix is also used for finding files such as @file{crt0.o} that are
17500 In addition, the prefix is used in an unusual way in finding the
17501 directories to search for header files. For each of the standard
17502 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17503 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17504 replacing that beginning with the specified prefix to produce an
17505 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17506 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17507 These alternate directories are searched first; the standard directories
17508 come next. If a standard directory begins with the configured
17509 @var{prefix} then the value of @var{prefix} is replaced by
17510 @env{GCC_EXEC_PREFIX} when looking for header files.
17512 @item COMPILER_PATH
17513 @findex COMPILER_PATH
17514 The value of @env{COMPILER_PATH} is a colon-separated list of
17515 directories, much like @env{PATH}. GCC tries the directories thus
17516 specified when searching for subprograms, if it can't find the
17517 subprograms using @env{GCC_EXEC_PREFIX}.
17520 @findex LIBRARY_PATH
17521 The value of @env{LIBRARY_PATH} is a colon-separated list of
17522 directories, much like @env{PATH}. When configured as a native compiler,
17523 GCC tries the directories thus specified when searching for special
17524 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17525 using GCC also uses these directories when searching for ordinary
17526 libraries for the @option{-l} option (but directories specified with
17527 @option{-L} come first).
17531 @cindex locale definition
17532 This variable is used to pass locale information to the compiler. One way in
17533 which this information is used is to determine the character set to be used
17534 when character literals, string literals and comments are parsed in C and C++.
17535 When the compiler is configured to allow multibyte characters,
17536 the following values for @env{LANG} are recognized:
17540 Recognize JIS characters.
17542 Recognize SJIS characters.
17544 Recognize EUCJP characters.
17547 If @env{LANG} is not defined, or if it has some other value, then the
17548 compiler will use mblen and mbtowc as defined by the default locale to
17549 recognize and translate multibyte characters.
17553 Some additional environments variables affect the behavior of the
17556 @include cppenv.texi
17560 @node Precompiled Headers
17561 @section Using Precompiled Headers
17562 @cindex precompiled headers
17563 @cindex speed of compilation
17565 Often large projects have many header files that are included in every
17566 source file. The time the compiler takes to process these header files
17567 over and over again can account for nearly all of the time required to
17568 build the project. To make builds faster, GCC allows users to
17569 `precompile' a header file; then, if builds can use the precompiled
17570 header file they will be much faster.
17572 To create a precompiled header file, simply compile it as you would any
17573 other file, if necessary using the @option{-x} option to make the driver
17574 treat it as a C or C++ header file. You will probably want to use a
17575 tool like @command{make} to keep the precompiled header up-to-date when
17576 the headers it contains change.
17578 A precompiled header file will be searched for when @code{#include} is
17579 seen in the compilation. As it searches for the included file
17580 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17581 compiler looks for a precompiled header in each directory just before it
17582 looks for the include file in that directory. The name searched for is
17583 the name specified in the @code{#include} with @samp{.gch} appended. If
17584 the precompiled header file can't be used, it is ignored.
17586 For instance, if you have @code{#include "all.h"}, and you have
17587 @file{all.h.gch} in the same directory as @file{all.h}, then the
17588 precompiled header file will be used if possible, and the original
17589 header will be used otherwise.
17591 Alternatively, you might decide to put the precompiled header file in a
17592 directory and use @option{-I} to ensure that directory is searched
17593 before (or instead of) the directory containing the original header.
17594 Then, if you want to check that the precompiled header file is always
17595 used, you can put a file of the same name as the original header in this
17596 directory containing an @code{#error} command.
17598 This also works with @option{-include}. So yet another way to use
17599 precompiled headers, good for projects not designed with precompiled
17600 header files in mind, is to simply take most of the header files used by
17601 a project, include them from another header file, precompile that header
17602 file, and @option{-include} the precompiled header. If the header files
17603 have guards against multiple inclusion, they will be skipped because
17604 they've already been included (in the precompiled header).
17606 If you need to precompile the same header file for different
17607 languages, targets, or compiler options, you can instead make a
17608 @emph{directory} named like @file{all.h.gch}, and put each precompiled
17609 header in the directory, perhaps using @option{-o}. It doesn't matter
17610 what you call the files in the directory, every precompiled header in
17611 the directory will be considered. The first precompiled header
17612 encountered in the directory that is valid for this compilation will
17613 be used; they're searched in no particular order.
17615 There are many other possibilities, limited only by your imagination,
17616 good sense, and the constraints of your build system.
17618 A precompiled header file can be used only when these conditions apply:
17622 Only one precompiled header can be used in a particular compilation.
17625 A precompiled header can't be used once the first C token is seen. You
17626 can have preprocessor directives before a precompiled header; you can
17627 even include a precompiled header from inside another header, so long as
17628 there are no C tokens before the @code{#include}.
17631 The precompiled header file must be produced for the same language as
17632 the current compilation. You can't use a C precompiled header for a C++
17636 The precompiled header file must have been produced by the same compiler
17637 binary as the current compilation is using.
17640 Any macros defined before the precompiled header is included must
17641 either be defined in the same way as when the precompiled header was
17642 generated, or must not affect the precompiled header, which usually
17643 means that they don't appear in the precompiled header at all.
17645 The @option{-D} option is one way to define a macro before a
17646 precompiled header is included; using a @code{#define} can also do it.
17647 There are also some options that define macros implicitly, like
17648 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
17651 @item If debugging information is output when using the precompiled
17652 header, using @option{-g} or similar, the same kind of debugging information
17653 must have been output when building the precompiled header. However,
17654 a precompiled header built using @option{-g} can be used in a compilation
17655 when no debugging information is being output.
17657 @item The same @option{-m} options must generally be used when building
17658 and using the precompiled header. @xref{Submodel Options},
17659 for any cases where this rule is relaxed.
17661 @item Each of the following options must be the same when building and using
17662 the precompiled header:
17664 @gccoptlist{-fexceptions}
17667 Some other command-line options starting with @option{-f},
17668 @option{-p}, or @option{-O} must be defined in the same way as when
17669 the precompiled header was generated. At present, it's not clear
17670 which options are safe to change and which are not; the safest choice
17671 is to use exactly the same options when generating and using the
17672 precompiled header. The following are known to be safe:
17674 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
17675 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
17676 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
17681 For all of these except the last, the compiler will automatically
17682 ignore the precompiled header if the conditions aren't met. If you
17683 find an option combination that doesn't work and doesn't cause the
17684 precompiled header to be ignored, please consider filing a bug report,
17687 If you do use differing options when generating and using the
17688 precompiled header, the actual behavior will be a mixture of the
17689 behavior for the options. For instance, if you use @option{-g} to
17690 generate the precompiled header but not when using it, you may or may
17691 not get debugging information for routines in the precompiled header.