1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
156 @section Option Summary
158 Here is a summary of all the options, grouped by type. Explanations are
159 in the following sections.
162 @item Overall Options
163 @xref{Overall Options,,Options Controlling the Kind of Output}.
164 @gccoptlist{-c -S -E -o @var{file} -combine -no-canonical-prefixes @gol
165 -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
167 --version -wrapper@@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -fno-pretty-templates @gol
193 -frepo -fno-rtti -fstats -ftemplate-depth=@var{n} @gol
194 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
195 -fno-default-inline -fvisibility-inlines-hidden @gol
196 -fvisibility-ms-compat @gol
197 -Wabi -Wconversion-null -Wctor-dtor-privacy @gol
198 -Wnon-virtual-dtor -Wreorder @gol
199 -Weffc++ -Wstrict-null-sentinel @gol
200 -Wno-non-template-friend -Wold-style-cast @gol
201 -Woverloaded-virtual -Wno-pmf-conversions @gol
204 @item Objective-C and Objective-C++ Language Options
205 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
206 Objective-C and Objective-C++ Dialects}.
207 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
208 -fgnu-runtime -fnext-runtime @gol
209 -fno-nil-receivers @gol
210 -fobjc-call-cxx-cdtors @gol
211 -fobjc-direct-dispatch @gol
212 -fobjc-exceptions @gol
214 -freplace-objc-classes @gol
217 -Wassign-intercept @gol
218 -Wno-protocol -Wselector @gol
219 -Wstrict-selector-match @gol
220 -Wundeclared-selector}
222 @item Language Independent Options
223 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
224 @gccoptlist{-fmessage-length=@var{n} @gol
225 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
226 -fdiagnostics-show-option}
228 @item Warning Options
229 @xref{Warning Options,,Options to Request or Suppress Warnings}.
230 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
231 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
232 -Wno-attributes -Wno-builtin-macro-redefined @gol
233 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
234 -Wchar-subscripts -Wclobbered -Wcomment @gol
235 -Wconversion -Wcoverage-mismatch -Wcpp -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 -Wsuggest-attribute=@r{[}const@r{|}pure@r{]} @gol
261 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
262 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
263 -Wunknown-pragmas -Wno-pragmas @gol
264 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
265 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value -Wunused-variable @gol
266 -Wunused-but-set-parameter -Wunused-but-set-variable -Wvariadic-macros -Wvla @gol
267 -Wvolatile-register-var -Wwrite-strings}
269 @item C and Objective-C-only Warning Options
270 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
271 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
272 -Wold-style-declaration -Wold-style-definition @gol
273 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
274 -Wdeclaration-after-statement -Wpointer-sign}
276 @item Debugging Options
277 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
278 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
279 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
280 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
281 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
282 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
283 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
284 -fdump-statistics @gol
286 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
290 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
299 -fdump-tree-nrv -fdump-tree-vect @gol
300 -fdump-tree-sink @gol
301 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
302 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
303 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
304 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
305 -ftree-vectorizer-verbose=@var{n} @gol
306 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
307 -fdump-final-insns=@var{file} @gol
308 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
309 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
310 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
311 -fenable-icf-debug @gol
312 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
313 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
314 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
315 -ftest-coverage -ftime-report -fvar-tracking @gol
316 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
317 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
318 -ggdb -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
319 -gvms -gxcoff -gxcoff+ @gol
320 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
321 -fdebug-prefix-map=@var{old}=@var{new} @gol
322 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
323 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
324 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
325 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
326 -print-prog-name=@var{program} -print-search-dirs -Q @gol
327 -print-sysroot -print-sysroot-headers-suffix @gol
328 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
330 @item Optimization Options
331 @xref{Optimize Options,,Options that Control Optimization}.
333 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
334 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
335 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
336 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
337 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
338 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
339 -fdata-sections -fdce -fdce @gol
340 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
341 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffast-math @gol
342 -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
343 -fforward-propagate -ffunction-sections @gol
344 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
345 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
346 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
347 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
348 -fipa-profile -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
349 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
350 -fira-region=@var{region} -fira-coalesce @gol
351 -fira-loop-pressure -fno-ira-share-save-slots @gol
352 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
353 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
354 -floop-block -floop-interchange -floop-strip-mine -fgraphite-identity @gol
355 -floop-parallelize-all -flto -flto-compression-level -flto-report -fltrans @gol
356 -fltrans-output-list -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
357 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
358 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
359 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
360 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
361 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
362 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
363 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
364 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
365 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
366 -fprofile-generate=@var{path} @gol
367 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
368 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
369 -freorder-blocks-and-partition -freorder-functions @gol
370 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
371 -frounding-math -fsched2-use-superblocks -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} -iplugindir=@var{dir}}
430 -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-
433 @item Machine Dependent Options
434 @xref{Submodel Options,,Hardware Models and Configurations}.
435 @c This list is ordered alphanumerically by subsection name.
436 @c Try and put the significant identifier (CPU or system) first,
437 @c so users have a clue at guessing where the ones they want will be.
440 @gccoptlist{-EB -EL @gol
441 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
442 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
445 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
446 -mabi=@var{name} @gol
447 -mapcs-stack-check -mno-apcs-stack-check @gol
448 -mapcs-float -mno-apcs-float @gol
449 -mapcs-reentrant -mno-apcs-reentrant @gol
450 -msched-prolog -mno-sched-prolog @gol
451 -mlittle-endian -mbig-endian -mwords-little-endian @gol
452 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
453 -mfp16-format=@var{name}
454 -mthumb-interwork -mno-thumb-interwork @gol
455 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
456 -mstructure-size-boundary=@var{n} @gol
457 -mabort-on-noreturn @gol
458 -mlong-calls -mno-long-calls @gol
459 -msingle-pic-base -mno-single-pic-base @gol
460 -mpic-register=@var{reg} @gol
461 -mnop-fun-dllimport @gol
462 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
463 -mpoke-function-name @gol
465 -mtpcs-frame -mtpcs-leaf-frame @gol
466 -mcaller-super-interworking -mcallee-super-interworking @gol
468 -mword-relocations @gol
469 -mfix-cortex-m3-ldrd}
472 @gccoptlist{-mmcu=@var{mcu} -mno-interrupts @gol
473 -mcall-prologues -mtiny-stack -mint8}
475 @emph{Blackfin Options}
476 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
477 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
478 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
479 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
480 -mno-id-shared-library -mshared-library-id=@var{n} @gol
481 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
482 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
483 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
487 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
488 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
489 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
490 -mstack-align -mdata-align -mconst-align @gol
491 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
492 -melf -maout -melinux -mlinux -sim -sim2 @gol
493 -mmul-bug-workaround -mno-mul-bug-workaround}
496 @gccoptlist{-mmac -mpush-args}
498 @emph{Darwin Options}
499 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
500 -arch_only -bind_at_load -bundle -bundle_loader @gol
501 -client_name -compatibility_version -current_version @gol
503 -dependency-file -dylib_file -dylinker_install_name @gol
504 -dynamic -dynamiclib -exported_symbols_list @gol
505 -filelist -flat_namespace -force_cpusubtype_ALL @gol
506 -force_flat_namespace -headerpad_max_install_names @gol
508 -image_base -init -install_name -keep_private_externs @gol
509 -multi_module -multiply_defined -multiply_defined_unused @gol
510 -noall_load -no_dead_strip_inits_and_terms @gol
511 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
512 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
513 -private_bundle -read_only_relocs -sectalign @gol
514 -sectobjectsymbols -whyload -seg1addr @gol
515 -sectcreate -sectobjectsymbols -sectorder @gol
516 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
517 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
518 -segprot -segs_read_only_addr -segs_read_write_addr @gol
519 -single_module -static -sub_library -sub_umbrella @gol
520 -twolevel_namespace -umbrella -undefined @gol
521 -unexported_symbols_list -weak_reference_mismatches @gol
522 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
523 -mkernel -mone-byte-bool}
525 @emph{DEC Alpha Options}
526 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
527 -mieee -mieee-with-inexact -mieee-conformant @gol
528 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
529 -mtrap-precision=@var{mode} -mbuild-constants @gol
530 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
531 -mbwx -mmax -mfix -mcix @gol
532 -mfloat-vax -mfloat-ieee @gol
533 -mexplicit-relocs -msmall-data -mlarge-data @gol
534 -msmall-text -mlarge-text @gol
535 -mmemory-latency=@var{time}}
537 @emph{DEC Alpha/VMS Options}
538 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
541 @gccoptlist{-msmall-model -mno-lsim}
544 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
545 -mhard-float -msoft-float @gol
546 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
547 -mdouble -mno-double @gol
548 -mmedia -mno-media -mmuladd -mno-muladd @gol
549 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
550 -mlinked-fp -mlong-calls -malign-labels @gol
551 -mlibrary-pic -macc-4 -macc-8 @gol
552 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
553 -moptimize-membar -mno-optimize-membar @gol
554 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
555 -mvliw-branch -mno-vliw-branch @gol
556 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
557 -mno-nested-cond-exec -mtomcat-stats @gol
561 @emph{GNU/Linux Options}
562 @gccoptlist{-muclibc}
564 @emph{H8/300 Options}
565 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
568 @gccoptlist{-march=@var{architecture-type} @gol
569 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
570 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
571 -mfixed-range=@var{register-range} @gol
572 -mjump-in-delay -mlinker-opt -mlong-calls @gol
573 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
574 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
575 -mno-jump-in-delay -mno-long-load-store @gol
576 -mno-portable-runtime -mno-soft-float @gol
577 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
578 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
579 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
580 -munix=@var{unix-std} -nolibdld -static -threads}
582 @emph{i386 and x86-64 Options}
583 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
584 -mfpmath=@var{unit} @gol
585 -masm=@var{dialect} -mno-fancy-math-387 @gol
586 -mno-fp-ret-in-387 -msoft-float @gol
587 -mno-wide-multiply -mrtd -malign-double @gol
588 -mpreferred-stack-boundary=@var{num}
589 -mincoming-stack-boundary=@var{num}
590 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip @gol
591 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
592 -maes -mpclmul -mfused-madd @gol
593 -msse4a -m3dnow -mpopcnt -mabm -mfma4 -mxop -mlwp @gol
594 -mthreads -mno-align-stringops -minline-all-stringops @gol
595 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
596 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
597 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
598 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
599 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
600 -mcmodel=@var{code-model} -mabi=@var{name} @gol
601 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
605 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
606 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
607 -mconstant-gp -mauto-pic -mfused-madd @gol
608 -minline-float-divide-min-latency @gol
609 -minline-float-divide-max-throughput @gol
610 -mno-inline-float-divide @gol
611 -minline-int-divide-min-latency @gol
612 -minline-int-divide-max-throughput @gol
613 -mno-inline-int-divide @gol
614 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
615 -mno-inline-sqrt @gol
616 -mdwarf2-asm -mearly-stop-bits @gol
617 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
618 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
619 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
620 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
621 -msched-spec-ldc -msched-spec-control-ldc @gol
622 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
623 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
624 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
625 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
627 @emph{IA-64/VMS Options}
628 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
631 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
632 -msign-extend-enabled -muser-enabled}
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 -fpu -nofpu@gol
789 -mbig-endian-data -mlittle-endian-data @gol
792 -mas100-syntax -mno-as100-syntax@gol
794 -mmax-constant-size=@gol
796 -msave-acc-in-interrupts}
798 @emph{S/390 and zSeries Options}
799 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
800 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
801 -mlong-double-64 -mlong-double-128 @gol
802 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
803 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
804 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
805 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
806 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
809 @gccoptlist{-meb -mel @gol
813 -mscore5 -mscore5u -mscore7 -mscore7d}
816 @gccoptlist{-m1 -m2 -m2e @gol
817 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
819 -m4-nofpu -m4-single-only -m4-single -m4 @gol
820 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
821 -m5-64media -m5-64media-nofpu @gol
822 -m5-32media -m5-32media-nofpu @gol
823 -m5-compact -m5-compact-nofpu @gol
824 -mb -ml -mdalign -mrelax @gol
825 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
826 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
827 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
828 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
829 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
830 -maccumulate-outgoing-args -minvalid-symbols}
833 @gccoptlist{-mcpu=@var{cpu-type} @gol
834 -mtune=@var{cpu-type} @gol
835 -mcmodel=@var{code-model} @gol
836 -m32 -m64 -mapp-regs -mno-app-regs @gol
837 -mfaster-structs -mno-faster-structs @gol
838 -mfpu -mno-fpu -mhard-float -msoft-float @gol
839 -mhard-quad-float -msoft-quad-float @gol
840 -mimpure-text -mno-impure-text -mlittle-endian @gol
841 -mstack-bias -mno-stack-bias @gol
842 -munaligned-doubles -mno-unaligned-doubles @gol
843 -mv8plus -mno-v8plus -mvis -mno-vis
844 -threads -pthreads -pthread}
847 @gccoptlist{-mwarn-reloc -merror-reloc @gol
848 -msafe-dma -munsafe-dma @gol
850 -msmall-mem -mlarge-mem -mstdmain @gol
851 -mfixed-range=@var{register-range} @gol
853 -maddress-space-conversion -mno-address-space-conversion @gol
854 -mcache-size=@var{cache-size} @gol
855 -matomic-updates -mno-atomic-updates}
857 @emph{System V Options}
858 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
861 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
862 -mprolog-function -mno-prolog-function -mspace @gol
863 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
864 -mapp-regs -mno-app-regs @gol
865 -mdisable-callt -mno-disable-callt @gol
871 @gccoptlist{-mg -mgnu -munix}
873 @emph{VxWorks Options}
874 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
875 -Xbind-lazy -Xbind-now}
877 @emph{x86-64 Options}
878 See i386 and x86-64 Options.
880 @emph{i386 and x86-64 Windows Options}
881 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
882 -mnop-fun-dllimport -mthread -municode -mwin32 -mwindows
883 -fno-set-stack-executable}
885 @emph{Xstormy16 Options}
888 @emph{Xtensa Options}
889 @gccoptlist{-mconst16 -mno-const16 @gol
890 -mfused-madd -mno-fused-madd @gol
891 -mserialize-volatile -mno-serialize-volatile @gol
892 -mtext-section-literals -mno-text-section-literals @gol
893 -mtarget-align -mno-target-align @gol
894 -mlongcalls -mno-longcalls}
896 @emph{zSeries Options}
897 See S/390 and zSeries Options.
899 @item Code Generation Options
900 @xref{Code Gen Options,,Options for Code Generation Conventions}.
901 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
902 -ffixed-@var{reg} -fexceptions @gol
903 -fnon-call-exceptions -funwind-tables @gol
904 -fasynchronous-unwind-tables @gol
905 -finhibit-size-directive -finstrument-functions @gol
906 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
907 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
908 -fno-common -fno-ident @gol
909 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
910 -fno-jump-tables @gol
911 -frecord-gcc-switches @gol
912 -freg-struct-return -fshort-enums @gol
913 -fshort-double -fshort-wchar @gol
914 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
915 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
916 -fno-stack-limit @gol
917 -fleading-underscore -ftls-model=@var{model} @gol
918 -ftrapv -fwrapv -fbounds-check @gol
923 * Overall Options:: Controlling the kind of output:
924 an executable, object files, assembler files,
925 or preprocessed source.
926 * C Dialect Options:: Controlling the variant of C language compiled.
927 * C++ Dialect Options:: Variations on C++.
928 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
930 * Language Independent Options:: Controlling how diagnostics should be
932 * Warning Options:: How picky should the compiler be?
933 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
934 * Optimize Options:: How much optimization?
935 * Preprocessor Options:: Controlling header files and macro definitions.
936 Also, getting dependency information for Make.
937 * Assembler Options:: Passing options to the assembler.
938 * Link Options:: Specifying libraries and so on.
939 * Directory Options:: Where to find header files and libraries.
940 Where to find the compiler executable files.
941 * Spec Files:: How to pass switches to sub-processes.
942 * Target Options:: Running a cross-compiler, or an old version of GCC.
945 @node Overall Options
946 @section Options Controlling the Kind of Output
948 Compilation can involve up to four stages: preprocessing, compilation
949 proper, assembly and linking, always in that order. GCC is capable of
950 preprocessing and compiling several files either into several
951 assembler input files, or into one assembler input file; then each
952 assembler input file produces an object file, and linking combines all
953 the object files (those newly compiled, and those specified as input)
954 into an executable file.
956 @cindex file name suffix
957 For any given input file, the file name suffix determines what kind of
962 C source code which must be preprocessed.
965 C source code which should not be preprocessed.
968 C++ source code which should not be preprocessed.
971 Objective-C source code. Note that you must link with the @file{libobjc}
972 library to make an Objective-C program work.
975 Objective-C source code which should not be preprocessed.
979 Objective-C++ source code. Note that you must link with the @file{libobjc}
980 library to make an Objective-C++ program work. Note that @samp{.M} refers
981 to a literal capital M@.
984 Objective-C++ source code which should not be preprocessed.
987 C, C++, Objective-C or Objective-C++ header file to be turned into a
992 @itemx @var{file}.cxx
993 @itemx @var{file}.cpp
994 @itemx @var{file}.CPP
995 @itemx @var{file}.c++
997 C++ source code which must be preprocessed. Note that in @samp{.cxx},
998 the last two letters must both be literally @samp{x}. Likewise,
999 @samp{.C} refers to a literal capital C@.
1003 Objective-C++ source code which must be preprocessed.
1005 @item @var{file}.mii
1006 Objective-C++ source code which should not be preprocessed.
1010 @itemx @var{file}.hp
1011 @itemx @var{file}.hxx
1012 @itemx @var{file}.hpp
1013 @itemx @var{file}.HPP
1014 @itemx @var{file}.h++
1015 @itemx @var{file}.tcc
1016 C++ header file to be turned into a precompiled header.
1019 @itemx @var{file}.for
1020 @itemx @var{file}.ftn
1021 Fixed form Fortran source code which should not be preprocessed.
1024 @itemx @var{file}.FOR
1025 @itemx @var{file}.fpp
1026 @itemx @var{file}.FPP
1027 @itemx @var{file}.FTN
1028 Fixed form Fortran source code which must be preprocessed (with the traditional
1031 @item @var{file}.f90
1032 @itemx @var{file}.f95
1033 @itemx @var{file}.f03
1034 @itemx @var{file}.f08
1035 Free form Fortran source code which should not be preprocessed.
1037 @item @var{file}.F90
1038 @itemx @var{file}.F95
1039 @itemx @var{file}.F03
1040 @itemx @var{file}.F08
1041 Free form Fortran source code which must be preprocessed (with the
1042 traditional preprocessor).
1044 @c FIXME: Descriptions of Java file types.
1050 @item @var{file}.ads
1051 Ada source code file which contains a library unit declaration (a
1052 declaration of a package, subprogram, or generic, or a generic
1053 instantiation), or a library unit renaming declaration (a package,
1054 generic, or subprogram renaming declaration). Such files are also
1057 @item @var{file}.adb
1058 Ada source code file containing a library unit body (a subprogram or
1059 package body). Such files are also called @dfn{bodies}.
1061 @c GCC also knows about some suffixes for languages not yet included:
1072 @itemx @var{file}.sx
1073 Assembler code which must be preprocessed.
1076 An object file to be fed straight into linking.
1077 Any file name with no recognized suffix is treated this way.
1081 You can specify the input language explicitly with the @option{-x} option:
1084 @item -x @var{language}
1085 Specify explicitly the @var{language} for the following input files
1086 (rather than letting the compiler choose a default based on the file
1087 name suffix). This option applies to all following input files until
1088 the next @option{-x} option. Possible values for @var{language} are:
1090 c c-header c-cpp-output
1091 c++ c++-header c++-cpp-output
1092 objective-c objective-c-header objective-c-cpp-output
1093 objective-c++ objective-c++-header objective-c++-cpp-output
1094 assembler assembler-with-cpp
1096 f77 f77-cpp-input f95 f95-cpp-input
1101 Turn off any specification of a language, so that subsequent files are
1102 handled according to their file name suffixes (as they are if @option{-x}
1103 has not been used at all).
1105 @item -pass-exit-codes
1106 @opindex pass-exit-codes
1107 Normally the @command{gcc} program will exit with the code of 1 if any
1108 phase of the compiler returns a non-success return code. If you specify
1109 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1110 numerically highest error produced by any phase that returned an error
1111 indication. The C, C++, and Fortran frontends return 4, if an internal
1112 compiler error is encountered.
1115 If you only want some of the stages of compilation, you can use
1116 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1117 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1118 @command{gcc} is to stop. Note that some combinations (for example,
1119 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1124 Compile or assemble the source files, but do not link. The linking
1125 stage simply is not done. The ultimate output is in the form of an
1126 object file for each source file.
1128 By default, the object file name for a source file is made by replacing
1129 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1131 Unrecognized input files, not requiring compilation or assembly, are
1136 Stop after the stage of compilation proper; do not assemble. The output
1137 is in the form of an assembler code file for each non-assembler input
1140 By default, the assembler file name for a source file is made by
1141 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1143 Input files that don't require compilation are ignored.
1147 Stop after the preprocessing stage; do not run the compiler proper. The
1148 output is in the form of preprocessed source code, which is sent to the
1151 Input files which don't require preprocessing are ignored.
1153 @cindex output file option
1156 Place output in file @var{file}. This applies regardless to whatever
1157 sort of output is being produced, whether it be an executable file,
1158 an object file, an assembler file or preprocessed C code.
1160 If @option{-o} is not specified, the default is to put an executable
1161 file in @file{a.out}, the object file for
1162 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1163 assembler file in @file{@var{source}.s}, a precompiled header file in
1164 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1169 Print (on standard error output) the commands executed to run the stages
1170 of compilation. Also print the version number of the compiler driver
1171 program and of the preprocessor and the compiler proper.
1175 Like @option{-v} except the commands are not executed and all command
1176 arguments are quoted. This is useful for shell scripts to capture the
1177 driver-generated command lines.
1181 Use pipes rather than temporary files for communication between the
1182 various stages of compilation. This fails to work on some systems where
1183 the assembler is unable to read from a pipe; but the GNU assembler has
1188 If you are compiling multiple source files, this option tells the driver
1189 to pass all the source files to the compiler at once (for those
1190 languages for which the compiler can handle this). This will allow
1191 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1192 language for which this is supported is C@. If you pass source files for
1193 multiple languages to the driver, using this option, the driver will invoke
1194 the compiler(s) that support IMA once each, passing each compiler all the
1195 source files appropriate for it. For those languages that do not support
1196 IMA this option will be ignored, and the compiler will be invoked once for
1197 each source file in that language. If you use this option in conjunction
1198 with @option{-save-temps}, the compiler will generate multiple
1200 (one for each source file), but only one (combined) @file{.o} or
1205 Print (on the standard output) a description of the command line options
1206 understood by @command{gcc}. If the @option{-v} option is also specified
1207 then @option{--help} will also be passed on to the various processes
1208 invoked by @command{gcc}, so that they can display the command line options
1209 they accept. If the @option{-Wextra} option has also been specified
1210 (prior to the @option{--help} option), then command line options which
1211 have no documentation associated with them will also be displayed.
1214 @opindex target-help
1215 Print (on the standard output) a description of target-specific command
1216 line options for each tool. For some targets extra target-specific
1217 information may also be printed.
1219 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1220 Print (on the standard output) a description of the command line
1221 options understood by the compiler that fit into all specified classes
1222 and qualifiers. These are the supported classes:
1225 @item @samp{optimizers}
1226 This will display all of the optimization options supported by the
1229 @item @samp{warnings}
1230 This will display all of the options controlling warning messages
1231 produced by the compiler.
1234 This will display target-specific options. Unlike the
1235 @option{--target-help} option however, target-specific options of the
1236 linker and assembler will not be displayed. This is because those
1237 tools do not currently support the extended @option{--help=} syntax.
1240 This will display the values recognized by the @option{--param}
1243 @item @var{language}
1244 This will display the options supported for @var{language}, where
1245 @var{language} is the name of one of the languages supported in this
1249 This will display the options that are common to all languages.
1252 These are the supported qualifiers:
1255 @item @samp{undocumented}
1256 Display only those options which are undocumented.
1259 Display options which take an argument that appears after an equal
1260 sign in the same continuous piece of text, such as:
1261 @samp{--help=target}.
1263 @item @samp{separate}
1264 Display options which take an argument that appears as a separate word
1265 following the original option, such as: @samp{-o output-file}.
1268 Thus for example to display all the undocumented target-specific
1269 switches supported by the compiler the following can be used:
1272 --help=target,undocumented
1275 The sense of a qualifier can be inverted by prefixing it with the
1276 @samp{^} character, so for example to display all binary warning
1277 options (i.e., ones that are either on or off and that do not take an
1278 argument), which have a description the following can be used:
1281 --help=warnings,^joined,^undocumented
1284 The argument to @option{--help=} should not consist solely of inverted
1287 Combining several classes is possible, although this usually
1288 restricts the output by so much that there is nothing to display. One
1289 case where it does work however is when one of the classes is
1290 @var{target}. So for example to display all the target-specific
1291 optimization options the following can be used:
1294 --help=target,optimizers
1297 The @option{--help=} option can be repeated on the command line. Each
1298 successive use will display its requested class of options, skipping
1299 those that have already been displayed.
1301 If the @option{-Q} option appears on the command line before the
1302 @option{--help=} option, then the descriptive text displayed by
1303 @option{--help=} is changed. Instead of describing the displayed
1304 options, an indication is given as to whether the option is enabled,
1305 disabled or set to a specific value (assuming that the compiler
1306 knows this at the point where the @option{--help=} option is used).
1308 Here is a truncated example from the ARM port of @command{gcc}:
1311 % gcc -Q -mabi=2 --help=target -c
1312 The following options are target specific:
1314 -mabort-on-noreturn [disabled]
1318 The output is sensitive to the effects of previous command line
1319 options, so for example it is possible to find out which optimizations
1320 are enabled at @option{-O2} by using:
1323 -Q -O2 --help=optimizers
1326 Alternatively you can discover which binary optimizations are enabled
1327 by @option{-O3} by using:
1330 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1331 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1332 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1335 @item -no-canonical-prefixes
1336 @opindex no-canonical-prefixes
1337 Do not expand any symbolic links, resolve references to @samp{/../}
1338 or @samp{/./}, or make the path absolute when generating a relative
1343 Display the version number and copyrights of the invoked GCC@.
1347 Invoke all subcommands under a wrapper program. It takes a single
1348 comma separated list as an argument, which will be used to invoke
1352 gcc -c t.c -wrapper gdb,--args
1355 This will invoke all subprograms of gcc under "gdb --args",
1356 thus cc1 invocation will be "gdb --args cc1 ...".
1358 @item -fplugin=@var{name}.so
1359 Load the plugin code in file @var{name}.so, assumed to be a
1360 shared object to be dlopen'd by the compiler. The base name of
1361 the shared object file is used to identify the plugin for the
1362 purposes of argument parsing (See
1363 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1364 Each plugin should define the callback functions specified in the
1367 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1368 Define an argument called @var{key} with a value of @var{value}
1369 for the plugin called @var{name}.
1371 @include @value{srcdir}/../libiberty/at-file.texi
1375 @section Compiling C++ Programs
1377 @cindex suffixes for C++ source
1378 @cindex C++ source file suffixes
1379 C++ source files conventionally use one of the suffixes @samp{.C},
1380 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1381 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1382 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1383 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1384 files with these names and compiles them as C++ programs even if you
1385 call the compiler the same way as for compiling C programs (usually
1386 with the name @command{gcc}).
1390 However, the use of @command{gcc} does not add the C++ library.
1391 @command{g++} is a program that calls GCC and treats @samp{.c},
1392 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1393 files unless @option{-x} is used, and automatically specifies linking
1394 against the C++ library. This program is also useful when
1395 precompiling a C header file with a @samp{.h} extension for use in C++
1396 compilations. On many systems, @command{g++} is also installed with
1397 the name @command{c++}.
1399 @cindex invoking @command{g++}
1400 When you compile C++ programs, you may specify many of the same
1401 command-line options that you use for compiling programs in any
1402 language; or command-line options meaningful for C and related
1403 languages; or options that are meaningful only for C++ programs.
1404 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1405 explanations of options for languages related to C@.
1406 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1407 explanations of options that are meaningful only for C++ programs.
1409 @node C Dialect Options
1410 @section Options Controlling C Dialect
1411 @cindex dialect options
1412 @cindex language dialect options
1413 @cindex options, dialect
1415 The following options control the dialect of C (or languages derived
1416 from C, such as C++, Objective-C and Objective-C++) that the compiler
1420 @cindex ANSI support
1424 In C mode, this is equivalent to @samp{-std=c90}. In C++ mode, it is
1425 equivalent to @samp{-std=c++98}.
1427 This turns off certain features of GCC that are incompatible with ISO
1428 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1429 such as the @code{asm} and @code{typeof} keywords, and
1430 predefined macros such as @code{unix} and @code{vax} that identify the
1431 type of system you are using. It also enables the undesirable and
1432 rarely used ISO trigraph feature. For the C compiler,
1433 it disables recognition of C++ style @samp{//} comments as well as
1434 the @code{inline} keyword.
1436 The alternate keywords @code{__asm__}, @code{__extension__},
1437 @code{__inline__} and @code{__typeof__} continue to work despite
1438 @option{-ansi}. You would not want to use them in an ISO C program, of
1439 course, but it is useful to put them in header files that might be included
1440 in compilations done with @option{-ansi}. Alternate predefined macros
1441 such as @code{__unix__} and @code{__vax__} are also available, with or
1442 without @option{-ansi}.
1444 The @option{-ansi} option does not cause non-ISO programs to be
1445 rejected gratuitously. For that, @option{-pedantic} is required in
1446 addition to @option{-ansi}. @xref{Warning Options}.
1448 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1449 option is used. Some header files may notice this macro and refrain
1450 from declaring certain functions or defining certain macros that the
1451 ISO standard doesn't call for; this is to avoid interfering with any
1452 programs that might use these names for other things.
1454 Functions that would normally be built in but do not have semantics
1455 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1456 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1457 built-in functions provided by GCC}, for details of the functions
1462 Determine the language standard. @xref{Standards,,Language Standards
1463 Supported by GCC}, for details of these standard versions. This option
1464 is currently only supported when compiling C or C++.
1466 The compiler can accept several base standards, such as @samp{c90} or
1467 @samp{c++98}, and GNU dialects of those standards, such as
1468 @samp{gnu90} or @samp{gnu++98}. By specifying a base standard, the
1469 compiler will accept all programs following that standard and those
1470 using GNU extensions that do not contradict it. For example,
1471 @samp{-std=c90} turns off certain features of GCC that are
1472 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1473 keywords, but not other GNU extensions that do not have a meaning in
1474 ISO C90, such as omitting the middle term of a @code{?:}
1475 expression. On the other hand, by specifying a GNU dialect of a
1476 standard, all features the compiler support are enabled, even when
1477 those features change the meaning of the base standard and some
1478 strict-conforming programs may be rejected. The particular standard
1479 is used by @option{-pedantic} to identify which features are GNU
1480 extensions given that version of the standard. For example
1481 @samp{-std=gnu90 -pedantic} would warn about C++ style @samp{//}
1482 comments, while @samp{-std=gnu99 -pedantic} would not.
1484 A value for this option must be provided; possible values are
1490 Support all ISO C90 programs (certain GNU extensions that conflict
1491 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1493 @item iso9899:199409
1494 ISO C90 as modified in amendment 1.
1500 ISO C99. Note that this standard is not yet fully supported; see
1501 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1502 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1505 ISO C1X, the draft of the next revision of the ISO C standard.
1506 Support is limited and experimental and features enabled by this
1507 option may be changed or removed if changed in or removed from the
1512 GNU dialect of ISO C90 (including some C99 features). This
1513 is the default for C code.
1517 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1518 this will become the default. The name @samp{gnu9x} is deprecated.
1521 GNU dialect of ISO C1X. Support is limited and experimental and
1522 features enabled by this option may be changed or removed if changed
1523 in or removed from the standard draft.
1526 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1530 GNU dialect of @option{-std=c++98}. This is the default for
1534 The working draft of the upcoming ISO C++0x standard. This option
1535 enables experimental features that are likely to be included in
1536 C++0x. The working draft is constantly changing, and any feature that is
1537 enabled by this flag may be removed from future versions of GCC if it is
1538 not part of the C++0x standard.
1541 GNU dialect of @option{-std=c++0x}. This option enables
1542 experimental features that may be removed in future versions of GCC.
1545 @item -fgnu89-inline
1546 @opindex fgnu89-inline
1547 The option @option{-fgnu89-inline} tells GCC to use the traditional
1548 GNU semantics for @code{inline} functions when in C99 mode.
1549 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1550 is accepted and ignored by GCC versions 4.1.3 up to but not including
1551 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1552 C99 mode. Using this option is roughly equivalent to adding the
1553 @code{gnu_inline} function attribute to all inline functions
1554 (@pxref{Function Attributes}).
1556 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1557 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1558 specifies the default behavior). This option was first supported in
1559 GCC 4.3. This option is not supported in @option{-std=c90} or
1560 @option{-std=gnu90} mode.
1562 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1563 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1564 in effect for @code{inline} functions. @xref{Common Predefined
1565 Macros,,,cpp,The C Preprocessor}.
1567 @item -aux-info @var{filename}
1569 Output to the given filename prototyped declarations for all functions
1570 declared and/or defined in a translation unit, including those in header
1571 files. This option is silently ignored in any language other than C@.
1573 Besides declarations, the file indicates, in comments, the origin of
1574 each declaration (source file and line), whether the declaration was
1575 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1576 @samp{O} for old, respectively, in the first character after the line
1577 number and the colon), and whether it came from a declaration or a
1578 definition (@samp{C} or @samp{F}, respectively, in the following
1579 character). In the case of function definitions, a K&R-style list of
1580 arguments followed by their declarations is also provided, inside
1581 comments, after the declaration.
1585 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1586 keyword, so that code can use these words as identifiers. You can use
1587 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1588 instead. @option{-ansi} implies @option{-fno-asm}.
1590 In C++, this switch only affects the @code{typeof} keyword, since
1591 @code{asm} and @code{inline} are standard keywords. You may want to
1592 use the @option{-fno-gnu-keywords} flag instead, which has the same
1593 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1594 switch only affects the @code{asm} and @code{typeof} keywords, since
1595 @code{inline} is a standard keyword in ISO C99.
1598 @itemx -fno-builtin-@var{function}
1599 @opindex fno-builtin
1600 @cindex built-in functions
1601 Don't recognize built-in functions that do not begin with
1602 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1603 functions provided by GCC}, for details of the functions affected,
1604 including those which are not built-in functions when @option{-ansi} or
1605 @option{-std} options for strict ISO C conformance are used because they
1606 do not have an ISO standard meaning.
1608 GCC normally generates special code to handle certain built-in functions
1609 more efficiently; for instance, calls to @code{alloca} may become single
1610 instructions that adjust the stack directly, and calls to @code{memcpy}
1611 may become inline copy loops. The resulting code is often both smaller
1612 and faster, but since the function calls no longer appear as such, you
1613 cannot set a breakpoint on those calls, nor can you change the behavior
1614 of the functions by linking with a different library. In addition,
1615 when a function is recognized as a built-in function, GCC may use
1616 information about that function to warn about problems with calls to
1617 that function, or to generate more efficient code, even if the
1618 resulting code still contains calls to that function. For example,
1619 warnings are given with @option{-Wformat} for bad calls to
1620 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1621 known not to modify global memory.
1623 With the @option{-fno-builtin-@var{function}} option
1624 only the built-in function @var{function} is
1625 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1626 function is named that is not built-in in this version of GCC, this
1627 option is ignored. There is no corresponding
1628 @option{-fbuiltin-@var{function}} option; if you wish to enable
1629 built-in functions selectively when using @option{-fno-builtin} or
1630 @option{-ffreestanding}, you may define macros such as:
1633 #define abs(n) __builtin_abs ((n))
1634 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1639 @cindex hosted environment
1641 Assert that compilation takes place in a hosted environment. This implies
1642 @option{-fbuiltin}. A hosted environment is one in which the
1643 entire standard library is available, and in which @code{main} has a return
1644 type of @code{int}. Examples are nearly everything except a kernel.
1645 This is equivalent to @option{-fno-freestanding}.
1647 @item -ffreestanding
1648 @opindex ffreestanding
1649 @cindex hosted environment
1651 Assert that compilation takes place in a freestanding environment. This
1652 implies @option{-fno-builtin}. A freestanding environment
1653 is one in which the standard library may not exist, and program startup may
1654 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1655 This is equivalent to @option{-fno-hosted}.
1657 @xref{Standards,,Language Standards Supported by GCC}, for details of
1658 freestanding and hosted environments.
1662 @cindex openmp parallel
1663 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1664 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1665 compiler generates parallel code according to the OpenMP Application
1666 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1667 implies @option{-pthread}, and thus is only supported on targets that
1668 have support for @option{-pthread}.
1670 @item -fms-extensions
1671 @opindex fms-extensions
1672 Accept some non-standard constructs used in Microsoft header files.
1674 Some cases of unnamed fields in structures and unions are only
1675 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1676 fields within structs/unions}, for details.
1680 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1681 options for strict ISO C conformance) implies @option{-trigraphs}.
1683 @item -no-integrated-cpp
1684 @opindex no-integrated-cpp
1685 Performs a compilation in two passes: preprocessing and compiling. This
1686 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1687 @option{-B} option. The user supplied compilation step can then add in
1688 an additional preprocessing step after normal preprocessing but before
1689 compiling. The default is to use the integrated cpp (internal cpp)
1691 The semantics of this option will change if "cc1", "cc1plus", and
1692 "cc1obj" are merged.
1694 @cindex traditional C language
1695 @cindex C language, traditional
1697 @itemx -traditional-cpp
1698 @opindex traditional-cpp
1699 @opindex traditional
1700 Formerly, these options caused GCC to attempt to emulate a pre-standard
1701 C compiler. They are now only supported with the @option{-E} switch.
1702 The preprocessor continues to support a pre-standard mode. See the GNU
1703 CPP manual for details.
1705 @item -fcond-mismatch
1706 @opindex fcond-mismatch
1707 Allow conditional expressions with mismatched types in the second and
1708 third arguments. The value of such an expression is void. This option
1709 is not supported for C++.
1711 @item -flax-vector-conversions
1712 @opindex flax-vector-conversions
1713 Allow implicit conversions between vectors with differing numbers of
1714 elements and/or incompatible element types. This option should not be
1717 @item -funsigned-char
1718 @opindex funsigned-char
1719 Let the type @code{char} be unsigned, like @code{unsigned char}.
1721 Each kind of machine has a default for what @code{char} should
1722 be. It is either like @code{unsigned char} by default or like
1723 @code{signed char} by default.
1725 Ideally, a portable program should always use @code{signed char} or
1726 @code{unsigned char} when it depends on the signedness of an object.
1727 But many programs have been written to use plain @code{char} and
1728 expect it to be signed, or expect it to be unsigned, depending on the
1729 machines they were written for. This option, and its inverse, let you
1730 make such a program work with the opposite default.
1732 The type @code{char} is always a distinct type from each of
1733 @code{signed char} or @code{unsigned char}, even though its behavior
1734 is always just like one of those two.
1737 @opindex fsigned-char
1738 Let the type @code{char} be signed, like @code{signed char}.
1740 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1741 the negative form of @option{-funsigned-char}. Likewise, the option
1742 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1744 @item -fsigned-bitfields
1745 @itemx -funsigned-bitfields
1746 @itemx -fno-signed-bitfields
1747 @itemx -fno-unsigned-bitfields
1748 @opindex fsigned-bitfields
1749 @opindex funsigned-bitfields
1750 @opindex fno-signed-bitfields
1751 @opindex fno-unsigned-bitfields
1752 These options control whether a bit-field is signed or unsigned, when the
1753 declaration does not use either @code{signed} or @code{unsigned}. By
1754 default, such a bit-field is signed, because this is consistent: the
1755 basic integer types such as @code{int} are signed types.
1758 @node C++ Dialect Options
1759 @section Options Controlling C++ Dialect
1761 @cindex compiler options, C++
1762 @cindex C++ options, command line
1763 @cindex options, C++
1764 This section describes the command-line options that are only meaningful
1765 for C++ programs; but you can also use most of the GNU compiler options
1766 regardless of what language your program is in. For example, you
1767 might compile a file @code{firstClass.C} like this:
1770 g++ -g -frepo -O -c firstClass.C
1774 In this example, only @option{-frepo} is an option meant
1775 only for C++ programs; you can use the other options with any
1776 language supported by GCC@.
1778 Here is a list of options that are @emph{only} for compiling C++ programs:
1782 @item -fabi-version=@var{n}
1783 @opindex fabi-version
1784 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1785 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1786 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1787 the version that conforms most closely to the C++ ABI specification.
1788 Therefore, the ABI obtained using version 0 will change as ABI bugs
1791 The default is version 2.
1793 Version 3 corrects an error in mangling a constant address as a
1796 Version 4 implements a standard mangling for vector types.
1798 See also @option{-Wabi}.
1800 @item -fno-access-control
1801 @opindex fno-access-control
1802 Turn off all access checking. This switch is mainly useful for working
1803 around bugs in the access control code.
1807 Check that the pointer returned by @code{operator new} is non-null
1808 before attempting to modify the storage allocated. This check is
1809 normally unnecessary because the C++ standard specifies that
1810 @code{operator new} will only return @code{0} if it is declared
1811 @samp{throw()}, in which case the compiler will always check the
1812 return value even without this option. In all other cases, when
1813 @code{operator new} has a non-empty exception specification, memory
1814 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1815 @samp{new (nothrow)}.
1817 @item -fconserve-space
1818 @opindex fconserve-space
1819 Put uninitialized or runtime-initialized global variables into the
1820 common segment, as C does. This saves space in the executable at the
1821 cost of not diagnosing duplicate definitions. If you compile with this
1822 flag and your program mysteriously crashes after @code{main()} has
1823 completed, you may have an object that is being destroyed twice because
1824 two definitions were merged.
1826 This option is no longer useful on most targets, now that support has
1827 been added for putting variables into BSS without making them common.
1829 @item -fno-deduce-init-list
1830 @opindex fno-deduce-init-list
1831 Disable deduction of a template type parameter as
1832 std::initializer_list from a brace-enclosed initializer list, i.e.
1835 template <class T> auto forward(T t) -> decltype (realfn (t))
1842 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1846 This option is present because this deduction is an extension to the
1847 current specification in the C++0x working draft, and there was
1848 some concern about potential overload resolution problems.
1850 @item -ffriend-injection
1851 @opindex ffriend-injection
1852 Inject friend functions into the enclosing namespace, so that they are
1853 visible outside the scope of the class in which they are declared.
1854 Friend functions were documented to work this way in the old Annotated
1855 C++ Reference Manual, and versions of G++ before 4.1 always worked
1856 that way. However, in ISO C++ a friend function which is not declared
1857 in an enclosing scope can only be found using argument dependent
1858 lookup. This option causes friends to be injected as they were in
1861 This option is for compatibility, and may be removed in a future
1864 @item -fno-elide-constructors
1865 @opindex fno-elide-constructors
1866 The C++ standard allows an implementation to omit creating a temporary
1867 which is only used to initialize another object of the same type.
1868 Specifying this option disables that optimization, and forces G++ to
1869 call the copy constructor in all cases.
1871 @item -fno-enforce-eh-specs
1872 @opindex fno-enforce-eh-specs
1873 Don't generate code to check for violation of exception specifications
1874 at runtime. This option violates the C++ standard, but may be useful
1875 for reducing code size in production builds, much like defining
1876 @samp{NDEBUG}. This does not give user code permission to throw
1877 exceptions in violation of the exception specifications; the compiler
1878 will still optimize based on the specifications, so throwing an
1879 unexpected exception will result in undefined behavior.
1882 @itemx -fno-for-scope
1884 @opindex fno-for-scope
1885 If @option{-ffor-scope} is specified, the scope of variables declared in
1886 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1887 as specified by the C++ standard.
1888 If @option{-fno-for-scope} is specified, the scope of variables declared in
1889 a @i{for-init-statement} extends to the end of the enclosing scope,
1890 as was the case in old versions of G++, and other (traditional)
1891 implementations of C++.
1893 The default if neither flag is given to follow the standard,
1894 but to allow and give a warning for old-style code that would
1895 otherwise be invalid, or have different behavior.
1897 @item -fno-gnu-keywords
1898 @opindex fno-gnu-keywords
1899 Do not recognize @code{typeof} as a keyword, so that code can use this
1900 word as an identifier. You can use the keyword @code{__typeof__} instead.
1901 @option{-ansi} implies @option{-fno-gnu-keywords}.
1903 @item -fno-implicit-templates
1904 @opindex fno-implicit-templates
1905 Never emit code for non-inline templates which are instantiated
1906 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1907 @xref{Template Instantiation}, for more information.
1909 @item -fno-implicit-inline-templates
1910 @opindex fno-implicit-inline-templates
1911 Don't emit code for implicit instantiations of inline templates, either.
1912 The default is to handle inlines differently so that compiles with and
1913 without optimization will need the same set of explicit instantiations.
1915 @item -fno-implement-inlines
1916 @opindex fno-implement-inlines
1917 To save space, do not emit out-of-line copies of inline functions
1918 controlled by @samp{#pragma implementation}. This will cause linker
1919 errors if these functions are not inlined everywhere they are called.
1921 @item -fms-extensions
1922 @opindex fms-extensions
1923 Disable pedantic warnings about constructs used in MFC, such as implicit
1924 int and getting a pointer to member function via non-standard syntax.
1926 @item -fno-nonansi-builtins
1927 @opindex fno-nonansi-builtins
1928 Disable built-in declarations of functions that are not mandated by
1929 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1930 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1932 @item -fno-operator-names
1933 @opindex fno-operator-names
1934 Do not treat the operator name keywords @code{and}, @code{bitand},
1935 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1936 synonyms as keywords.
1938 @item -fno-optional-diags
1939 @opindex fno-optional-diags
1940 Disable diagnostics that the standard says a compiler does not need to
1941 issue. Currently, the only such diagnostic issued by G++ is the one for
1942 a name having multiple meanings within a class.
1945 @opindex fpermissive
1946 Downgrade some diagnostics about nonconformant code from errors to
1947 warnings. Thus, using @option{-fpermissive} will allow some
1948 nonconforming code to compile.
1950 @item -fno-pretty-templates
1951 @opindex fno-pretty-templates
1952 When an error message refers to a specialization of a function
1953 template, the compiler will normally print the signature of the
1954 template followed by the template arguments and any typedefs or
1955 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1956 rather than @code{void f(int)}) so that it's clear which template is
1957 involved. When an error message refers to a specialization of a class
1958 template, the compiler will omit any template arguments which match
1959 the default template arguments for that template. If either of these
1960 behaviors make it harder to understand the error message rather than
1961 easier, using @option{-fno-pretty-templates} will disable them.
1965 Enable automatic template instantiation at link time. This option also
1966 implies @option{-fno-implicit-templates}. @xref{Template
1967 Instantiation}, for more information.
1971 Disable generation of information about every class with virtual
1972 functions for use by the C++ runtime type identification features
1973 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1974 of the language, you can save some space by using this flag. Note that
1975 exception handling uses the same information, but it will generate it as
1976 needed. The @samp{dynamic_cast} operator can still be used for casts that
1977 do not require runtime type information, i.e.@: casts to @code{void *} or to
1978 unambiguous base classes.
1982 Emit statistics about front-end processing at the end of the compilation.
1983 This information is generally only useful to the G++ development team.
1985 @item -fstrict-enums
1986 @opindex fstrict-enums
1987 Allow the compiler to optimize using the assumption that a value of
1988 enumeration type can only be one of the values of the enumeration (as
1989 defined in the C++ standard; basically, a value which can be
1990 represented in the minimum number of bits needed to represent all the
1991 enumerators). This assumption may not be valid if the program uses a
1992 cast to convert an arbitrary integer value to the enumeration type.
1994 @item -ftemplate-depth=@var{n}
1995 @opindex ftemplate-depth
1996 Set the maximum instantiation depth for template classes to @var{n}.
1997 A limit on the template instantiation depth is needed to detect
1998 endless recursions during template class instantiation. ANSI/ISO C++
1999 conforming programs must not rely on a maximum depth greater than 17
2000 (changed to 1024 in C++0x).
2002 @item -fno-threadsafe-statics
2003 @opindex fno-threadsafe-statics
2004 Do not emit the extra code to use the routines specified in the C++
2005 ABI for thread-safe initialization of local statics. You can use this
2006 option to reduce code size slightly in code that doesn't need to be
2009 @item -fuse-cxa-atexit
2010 @opindex fuse-cxa-atexit
2011 Register destructors for objects with static storage duration with the
2012 @code{__cxa_atexit} function rather than the @code{atexit} function.
2013 This option is required for fully standards-compliant handling of static
2014 destructors, but will only work if your C library supports
2015 @code{__cxa_atexit}.
2017 @item -fno-use-cxa-get-exception-ptr
2018 @opindex fno-use-cxa-get-exception-ptr
2019 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2020 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2021 if the runtime routine is not available.
2023 @item -fvisibility-inlines-hidden
2024 @opindex fvisibility-inlines-hidden
2025 This switch declares that the user does not attempt to compare
2026 pointers to inline methods where the addresses of the two functions
2027 were taken in different shared objects.
2029 The effect of this is that GCC may, effectively, mark inline methods with
2030 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2031 appear in the export table of a DSO and do not require a PLT indirection
2032 when used within the DSO@. Enabling this option can have a dramatic effect
2033 on load and link times of a DSO as it massively reduces the size of the
2034 dynamic export table when the library makes heavy use of templates.
2036 The behavior of this switch is not quite the same as marking the
2037 methods as hidden directly, because it does not affect static variables
2038 local to the function or cause the compiler to deduce that
2039 the function is defined in only one shared object.
2041 You may mark a method as having a visibility explicitly to negate the
2042 effect of the switch for that method. For example, if you do want to
2043 compare pointers to a particular inline method, you might mark it as
2044 having default visibility. Marking the enclosing class with explicit
2045 visibility will have no effect.
2047 Explicitly instantiated inline methods are unaffected by this option
2048 as their linkage might otherwise cross a shared library boundary.
2049 @xref{Template Instantiation}.
2051 @item -fvisibility-ms-compat
2052 @opindex fvisibility-ms-compat
2053 This flag attempts to use visibility settings to make GCC's C++
2054 linkage model compatible with that of Microsoft Visual Studio.
2056 The flag makes these changes to GCC's linkage model:
2060 It sets the default visibility to @code{hidden}, like
2061 @option{-fvisibility=hidden}.
2064 Types, but not their members, are not hidden by default.
2067 The One Definition Rule is relaxed for types without explicit
2068 visibility specifications which are defined in more than one different
2069 shared object: those declarations are permitted if they would have
2070 been permitted when this option was not used.
2073 In new code it is better to use @option{-fvisibility=hidden} and
2074 export those classes which are intended to be externally visible.
2075 Unfortunately it is possible for code to rely, perhaps accidentally,
2076 on the Visual Studio behavior.
2078 Among the consequences of these changes are that static data members
2079 of the same type with the same name but defined in different shared
2080 objects will be different, so changing one will not change the other;
2081 and that pointers to function members defined in different shared
2082 objects may not compare equal. When this flag is given, it is a
2083 violation of the ODR to define types with the same name differently.
2087 Do not use weak symbol support, even if it is provided by the linker.
2088 By default, G++ will use weak symbols if they are available. This
2089 option exists only for testing, and should not be used by end-users;
2090 it will result in inferior code and has no benefits. This option may
2091 be removed in a future release of G++.
2095 Do not search for header files in the standard directories specific to
2096 C++, but do still search the other standard directories. (This option
2097 is used when building the C++ library.)
2100 In addition, these optimization, warning, and code generation options
2101 have meanings only for C++ programs:
2104 @item -fno-default-inline
2105 @opindex fno-default-inline
2106 Do not assume @samp{inline} for functions defined inside a class scope.
2107 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2108 functions will have linkage like inline functions; they just won't be
2111 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2114 Warn when G++ generates code that is probably not compatible with the
2115 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2116 all such cases, there are probably some cases that are not warned about,
2117 even though G++ is generating incompatible code. There may also be
2118 cases where warnings are emitted even though the code that is generated
2121 You should rewrite your code to avoid these warnings if you are
2122 concerned about the fact that code generated by G++ may not be binary
2123 compatible with code generated by other compilers.
2125 The known incompatibilities in @option{-fabi-version=2} (the default) include:
2130 A template with a non-type template parameter of reference type is
2131 mangled incorrectly:
2134 template <int &> struct S @{@};
2138 This is fixed in @option{-fabi-version=3}.
2141 SIMD vector types declared using @code{__attribute ((vector_size))} are
2142 mangled in a non-standard way that does not allow for overloading of
2143 functions taking vectors of different sizes.
2145 The mangling is changed in @option{-fabi-version=4}.
2148 The known incompatibilities in @option{-fabi-version=1} include:
2153 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2154 pack data into the same byte as a base class. For example:
2157 struct A @{ virtual void f(); int f1 : 1; @};
2158 struct B : public A @{ int f2 : 1; @};
2162 In this case, G++ will place @code{B::f2} into the same byte
2163 as@code{A::f1}; other compilers will not. You can avoid this problem
2164 by explicitly padding @code{A} so that its size is a multiple of the
2165 byte size on your platform; that will cause G++ and other compilers to
2166 layout @code{B} identically.
2169 Incorrect handling of tail-padding for virtual bases. G++ does not use
2170 tail padding when laying out virtual bases. For example:
2173 struct A @{ virtual void f(); char c1; @};
2174 struct B @{ B(); char c2; @};
2175 struct C : public A, public virtual B @{@};
2179 In this case, G++ will not place @code{B} into the tail-padding for
2180 @code{A}; other compilers will. You can avoid this problem by
2181 explicitly padding @code{A} so that its size is a multiple of its
2182 alignment (ignoring virtual base classes); that will cause G++ and other
2183 compilers to layout @code{C} identically.
2186 Incorrect handling of bit-fields with declared widths greater than that
2187 of their underlying types, when the bit-fields appear in a union. For
2191 union U @{ int i : 4096; @};
2195 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2196 union too small by the number of bits in an @code{int}.
2199 Empty classes can be placed at incorrect offsets. For example:
2209 struct C : public B, public A @{@};
2213 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2214 it should be placed at offset zero. G++ mistakenly believes that the
2215 @code{A} data member of @code{B} is already at offset zero.
2218 Names of template functions whose types involve @code{typename} or
2219 template template parameters can be mangled incorrectly.
2222 template <typename Q>
2223 void f(typename Q::X) @{@}
2225 template <template <typename> class Q>
2226 void f(typename Q<int>::X) @{@}
2230 Instantiations of these templates may be mangled incorrectly.
2234 It also warns psABI related changes. The known psABI changes at this
2240 For SYSV/x86-64, when passing union with long double, it is changed to
2241 pass in memory as specified in psABI. For example:
2251 @code{union U} will always be passed in memory.
2255 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2256 @opindex Wctor-dtor-privacy
2257 @opindex Wno-ctor-dtor-privacy
2258 Warn when a class seems unusable because all the constructors or
2259 destructors in that class are private, and it has neither friends nor
2260 public static member functions.
2262 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2263 @opindex Wnon-virtual-dtor
2264 @opindex Wno-non-virtual-dtor
2265 Warn when a class has virtual functions and accessible non-virtual
2266 destructor, in which case it would be possible but unsafe to delete
2267 an instance of a derived class through a pointer to the base class.
2268 This warning is also enabled if -Weffc++ is specified.
2270 @item -Wreorder @r{(C++ and Objective-C++ only)}
2272 @opindex Wno-reorder
2273 @cindex reordering, warning
2274 @cindex warning for reordering of member initializers
2275 Warn when the order of member initializers given in the code does not
2276 match the order in which they must be executed. For instance:
2282 A(): j (0), i (1) @{ @}
2286 The compiler will rearrange the member initializers for @samp{i}
2287 and @samp{j} to match the declaration order of the members, emitting
2288 a warning to that effect. This warning is enabled by @option{-Wall}.
2291 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2294 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2297 Warn about violations of the following style guidelines from Scott Meyers'
2298 @cite{Effective C++} book:
2302 Item 11: Define a copy constructor and an assignment operator for classes
2303 with dynamically allocated memory.
2306 Item 12: Prefer initialization to assignment in constructors.
2309 Item 14: Make destructors virtual in base classes.
2312 Item 15: Have @code{operator=} return a reference to @code{*this}.
2315 Item 23: Don't try to return a reference when you must return an object.
2319 Also warn about violations of the following style guidelines from
2320 Scott Meyers' @cite{More Effective C++} book:
2324 Item 6: Distinguish between prefix and postfix forms of increment and
2325 decrement operators.
2328 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2332 When selecting this option, be aware that the standard library
2333 headers do not obey all of these guidelines; use @samp{grep -v}
2334 to filter out those warnings.
2336 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2337 @opindex Wstrict-null-sentinel
2338 @opindex Wno-strict-null-sentinel
2339 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2340 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2341 to @code{__null}. Although it is a null pointer constant not a null pointer,
2342 it is guaranteed to be of the same size as a pointer. But this use is
2343 not portable across different compilers.
2345 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2346 @opindex Wno-non-template-friend
2347 @opindex Wnon-template-friend
2348 Disable warnings when non-templatized friend functions are declared
2349 within a template. Since the advent of explicit template specification
2350 support in G++, if the name of the friend is an unqualified-id (i.e.,
2351 @samp{friend foo(int)}), the C++ language specification demands that the
2352 friend declare or define an ordinary, nontemplate function. (Section
2353 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2354 could be interpreted as a particular specialization of a templatized
2355 function. Because this non-conforming behavior is no longer the default
2356 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2357 check existing code for potential trouble spots and is on by default.
2358 This new compiler behavior can be turned off with
2359 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2360 but disables the helpful warning.
2362 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2363 @opindex Wold-style-cast
2364 @opindex Wno-old-style-cast
2365 Warn if an old-style (C-style) cast to a non-void type is used within
2366 a C++ program. The new-style casts (@samp{dynamic_cast},
2367 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2368 less vulnerable to unintended effects and much easier to search for.
2370 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2371 @opindex Woverloaded-virtual
2372 @opindex Wno-overloaded-virtual
2373 @cindex overloaded virtual fn, warning
2374 @cindex warning for overloaded virtual fn
2375 Warn when a function declaration hides virtual functions from a
2376 base class. For example, in:
2383 struct B: public A @{
2388 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2396 will fail to compile.
2398 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2399 @opindex Wno-pmf-conversions
2400 @opindex Wpmf-conversions
2401 Disable the diagnostic for converting a bound pointer to member function
2404 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2405 @opindex Wsign-promo
2406 @opindex Wno-sign-promo
2407 Warn when overload resolution chooses a promotion from unsigned or
2408 enumerated type to a signed type, over a conversion to an unsigned type of
2409 the same size. Previous versions of G++ would try to preserve
2410 unsignedness, but the standard mandates the current behavior.
2415 A& operator = (int);
2425 In this example, G++ will synthesize a default @samp{A& operator =
2426 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2429 @node Objective-C and Objective-C++ Dialect Options
2430 @section Options Controlling Objective-C and Objective-C++ Dialects
2432 @cindex compiler options, Objective-C and Objective-C++
2433 @cindex Objective-C and Objective-C++ options, command line
2434 @cindex options, Objective-C and Objective-C++
2435 (NOTE: This manual does not describe the Objective-C and Objective-C++
2436 languages themselves. See @xref{Standards,,Language Standards
2437 Supported by GCC}, for references.)
2439 This section describes the command-line options that are only meaningful
2440 for Objective-C and Objective-C++ programs, but you can also use most of
2441 the language-independent GNU compiler options.
2442 For example, you might compile a file @code{some_class.m} like this:
2445 gcc -g -fgnu-runtime -O -c some_class.m
2449 In this example, @option{-fgnu-runtime} is an option meant only for
2450 Objective-C and Objective-C++ programs; you can use the other options with
2451 any language supported by GCC@.
2453 Note that since Objective-C is an extension of the C language, Objective-C
2454 compilations may also use options specific to the C front-end (e.g.,
2455 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2456 C++-specific options (e.g., @option{-Wabi}).
2458 Here is a list of options that are @emph{only} for compiling Objective-C
2459 and Objective-C++ programs:
2462 @item -fconstant-string-class=@var{class-name}
2463 @opindex fconstant-string-class
2464 Use @var{class-name} as the name of the class to instantiate for each
2465 literal string specified with the syntax @code{@@"@dots{}"}. The default
2466 class name is @code{NXConstantString} if the GNU runtime is being used, and
2467 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2468 @option{-fconstant-cfstrings} option, if also present, will override the
2469 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2470 to be laid out as constant CoreFoundation strings.
2473 @opindex fgnu-runtime
2474 Generate object code compatible with the standard GNU Objective-C
2475 runtime. This is the default for most types of systems.
2477 @item -fnext-runtime
2478 @opindex fnext-runtime
2479 Generate output compatible with the NeXT runtime. This is the default
2480 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2481 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2484 @item -fno-nil-receivers
2485 @opindex fno-nil-receivers
2486 Assume that all Objective-C message dispatches (e.g.,
2487 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2488 is not @code{nil}. This allows for more efficient entry points in the runtime
2489 to be used. Currently, this option is only available in conjunction with
2490 the NeXT runtime on Mac OS X 10.3 and later.
2492 @item -fobjc-call-cxx-cdtors
2493 @opindex fobjc-call-cxx-cdtors
2494 For each Objective-C class, check if any of its instance variables is a
2495 C++ object with a non-trivial default constructor. If so, synthesize a
2496 special @code{- (id) .cxx_construct} instance method that will run
2497 non-trivial default constructors on any such instance variables, in order,
2498 and then return @code{self}. Similarly, check if any instance variable
2499 is a C++ object with a non-trivial destructor, and if so, synthesize a
2500 special @code{- (void) .cxx_destruct} method that will run
2501 all such default destructors, in reverse order.
2503 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2504 thusly generated will only operate on instance variables declared in the
2505 current Objective-C class, and not those inherited from superclasses. It
2506 is the responsibility of the Objective-C runtime to invoke all such methods
2507 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2508 will be invoked by the runtime immediately after a new object
2509 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2510 be invoked immediately before the runtime deallocates an object instance.
2512 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2513 support for invoking the @code{- (id) .cxx_construct} and
2514 @code{- (void) .cxx_destruct} methods.
2516 @item -fobjc-direct-dispatch
2517 @opindex fobjc-direct-dispatch
2518 Allow fast jumps to the message dispatcher. On Darwin this is
2519 accomplished via the comm page.
2521 @item -fobjc-exceptions
2522 @opindex fobjc-exceptions
2523 Enable syntactic support for structured exception handling in Objective-C,
2524 similar to what is offered by C++ and Java. This option is
2525 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2534 @@catch (AnObjCClass *exc) @{
2541 @@catch (AnotherClass *exc) @{
2544 @@catch (id allOthers) @{
2554 The @code{@@throw} statement may appear anywhere in an Objective-C or
2555 Objective-C++ program; when used inside of a @code{@@catch} block, the
2556 @code{@@throw} may appear without an argument (as shown above), in which case
2557 the object caught by the @code{@@catch} will be rethrown.
2559 Note that only (pointers to) Objective-C objects may be thrown and
2560 caught using this scheme. When an object is thrown, it will be caught
2561 by the nearest @code{@@catch} clause capable of handling objects of that type,
2562 analogously to how @code{catch} blocks work in C++ and Java. A
2563 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2564 any and all Objective-C exceptions not caught by previous @code{@@catch}
2567 The @code{@@finally} clause, if present, will be executed upon exit from the
2568 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2569 regardless of whether any exceptions are thrown, caught or rethrown
2570 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2571 of the @code{finally} clause in Java.
2573 There are several caveats to using the new exception mechanism:
2577 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2578 idioms provided by the @code{NSException} class, the new
2579 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2580 systems, due to additional functionality needed in the (NeXT) Objective-C
2584 As mentioned above, the new exceptions do not support handling
2585 types other than Objective-C objects. Furthermore, when used from
2586 Objective-C++, the Objective-C exception model does not interoperate with C++
2587 exceptions at this time. This means you cannot @code{@@throw} an exception
2588 from Objective-C and @code{catch} it in C++, or vice versa
2589 (i.e., @code{throw @dots{} @@catch}).
2592 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2593 blocks for thread-safe execution:
2596 @@synchronized (ObjCClass *guard) @{
2601 Upon entering the @code{@@synchronized} block, a thread of execution shall
2602 first check whether a lock has been placed on the corresponding @code{guard}
2603 object by another thread. If it has, the current thread shall wait until
2604 the other thread relinquishes its lock. Once @code{guard} becomes available,
2605 the current thread will place its own lock on it, execute the code contained in
2606 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2607 making @code{guard} available to other threads).
2609 Unlike Java, Objective-C does not allow for entire methods to be marked
2610 @code{@@synchronized}. Note that throwing exceptions out of
2611 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2612 to be unlocked properly.
2616 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2618 @item -freplace-objc-classes
2619 @opindex freplace-objc-classes
2620 Emit a special marker instructing @command{ld(1)} not to statically link in
2621 the resulting object file, and allow @command{dyld(1)} to load it in at
2622 run time instead. This is used in conjunction with the Fix-and-Continue
2623 debugging mode, where the object file in question may be recompiled and
2624 dynamically reloaded in the course of program execution, without the need
2625 to restart the program itself. Currently, Fix-and-Continue functionality
2626 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2631 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2632 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2633 compile time) with static class references that get initialized at load time,
2634 which improves run-time performance. Specifying the @option{-fzero-link} flag
2635 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2636 to be retained. This is useful in Zero-Link debugging mode, since it allows
2637 for individual class implementations to be modified during program execution.
2641 Dump interface declarations for all classes seen in the source file to a
2642 file named @file{@var{sourcename}.decl}.
2644 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2645 @opindex Wassign-intercept
2646 @opindex Wno-assign-intercept
2647 Warn whenever an Objective-C assignment is being intercepted by the
2650 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2651 @opindex Wno-protocol
2653 If a class is declared to implement a protocol, a warning is issued for
2654 every method in the protocol that is not implemented by the class. The
2655 default behavior is to issue a warning for every method not explicitly
2656 implemented in the class, even if a method implementation is inherited
2657 from the superclass. If you use the @option{-Wno-protocol} option, then
2658 methods inherited from the superclass are considered to be implemented,
2659 and no warning is issued for them.
2661 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2663 @opindex Wno-selector
2664 Warn if multiple methods of different types for the same selector are
2665 found during compilation. The check is performed on the list of methods
2666 in the final stage of compilation. Additionally, a check is performed
2667 for each selector appearing in a @code{@@selector(@dots{})}
2668 expression, and a corresponding method for that selector has been found
2669 during compilation. Because these checks scan the method table only at
2670 the end of compilation, these warnings are not produced if the final
2671 stage of compilation is not reached, for example because an error is
2672 found during compilation, or because the @option{-fsyntax-only} option is
2675 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2676 @opindex Wstrict-selector-match
2677 @opindex Wno-strict-selector-match
2678 Warn if multiple methods with differing argument and/or return types are
2679 found for a given selector when attempting to send a message using this
2680 selector to a receiver of type @code{id} or @code{Class}. When this flag
2681 is off (which is the default behavior), the compiler will omit such warnings
2682 if any differences found are confined to types which share the same size
2685 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2686 @opindex Wundeclared-selector
2687 @opindex Wno-undeclared-selector
2688 Warn if a @code{@@selector(@dots{})} expression referring to an
2689 undeclared selector is found. A selector is considered undeclared if no
2690 method with that name has been declared before the
2691 @code{@@selector(@dots{})} expression, either explicitly in an
2692 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2693 an @code{@@implementation} section. This option always performs its
2694 checks as soon as a @code{@@selector(@dots{})} expression is found,
2695 while @option{-Wselector} only performs its checks in the final stage of
2696 compilation. This also enforces the coding style convention
2697 that methods and selectors must be declared before being used.
2699 @item -print-objc-runtime-info
2700 @opindex print-objc-runtime-info
2701 Generate C header describing the largest structure that is passed by
2706 @node Language Independent Options
2707 @section Options to Control Diagnostic Messages Formatting
2708 @cindex options to control diagnostics formatting
2709 @cindex diagnostic messages
2710 @cindex message formatting
2712 Traditionally, diagnostic messages have been formatted irrespective of
2713 the output device's aspect (e.g.@: its width, @dots{}). The options described
2714 below can be used to control the diagnostic messages formatting
2715 algorithm, e.g.@: how many characters per line, how often source location
2716 information should be reported. Right now, only the C++ front end can
2717 honor these options. However it is expected, in the near future, that
2718 the remaining front ends would be able to digest them correctly.
2721 @item -fmessage-length=@var{n}
2722 @opindex fmessage-length
2723 Try to format error messages so that they fit on lines of about @var{n}
2724 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2725 the front ends supported by GCC@. If @var{n} is zero, then no
2726 line-wrapping will be done; each error message will appear on a single
2729 @opindex fdiagnostics-show-location
2730 @item -fdiagnostics-show-location=once
2731 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2732 reporter to emit @emph{once} source location information; that is, in
2733 case the message is too long to fit on a single physical line and has to
2734 be wrapped, the source location won't be emitted (as prefix) again,
2735 over and over, in subsequent continuation lines. This is the default
2738 @item -fdiagnostics-show-location=every-line
2739 Only meaningful in line-wrapping mode. Instructs the diagnostic
2740 messages reporter to emit the same source location information (as
2741 prefix) for physical lines that result from the process of breaking
2742 a message which is too long to fit on a single line.
2744 @item -fdiagnostics-show-option
2745 @opindex fdiagnostics-show-option
2746 This option instructs the diagnostic machinery to add text to each
2747 diagnostic emitted, which indicates which command line option directly
2748 controls that diagnostic, when such an option is known to the
2749 diagnostic machinery.
2751 @item -Wcoverage-mismatch
2752 @opindex Wcoverage-mismatch
2753 Warn if feedback profiles do not match when using the
2754 @option{-fprofile-use} option.
2755 If a source file was changed between @option{-fprofile-gen} and
2756 @option{-fprofile-use}, the files with the profile feedback can fail
2757 to match the source file and GCC can not use the profile feedback
2758 information. By default, GCC emits an error message in this case.
2759 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2760 error. GCC does not use appropriate feedback profiles, so using this
2761 option can result in poorly optimized code. This option is useful
2762 only in the case of very minor changes such as bug fixes to an
2767 @node Warning Options
2768 @section Options to Request or Suppress Warnings
2769 @cindex options to control warnings
2770 @cindex warning messages
2771 @cindex messages, warning
2772 @cindex suppressing warnings
2774 Warnings are diagnostic messages that report constructions which
2775 are not inherently erroneous but which are risky or suggest there
2776 may have been an error.
2778 The following language-independent options do not enable specific
2779 warnings but control the kinds of diagnostics produced by GCC.
2782 @cindex syntax checking
2784 @opindex fsyntax-only
2785 Check the code for syntax errors, but don't do anything beyond that.
2789 Inhibit all warning messages.
2794 Make all warnings into errors.
2799 Make the specified warning into an error. The specifier for a warning
2800 is appended, for example @option{-Werror=switch} turns the warnings
2801 controlled by @option{-Wswitch} into errors. This switch takes a
2802 negative form, to be used to negate @option{-Werror} for specific
2803 warnings, for example @option{-Wno-error=switch} makes
2804 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2805 is in effect. You can use the @option{-fdiagnostics-show-option}
2806 option to have each controllable warning amended with the option which
2807 controls it, to determine what to use with this option.
2809 Note that specifying @option{-Werror=}@var{foo} automatically implies
2810 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2813 @item -Wfatal-errors
2814 @opindex Wfatal-errors
2815 @opindex Wno-fatal-errors
2816 This option causes the compiler to abort compilation on the first error
2817 occurred rather than trying to keep going and printing further error
2822 You can request many specific warnings with options beginning
2823 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2824 implicit declarations. Each of these specific warning options also
2825 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2826 example, @option{-Wno-implicit}. This manual lists only one of the
2827 two forms, whichever is not the default. For further,
2828 language-specific options also refer to @ref{C++ Dialect Options} and
2829 @ref{Objective-C and Objective-C++ Dialect Options}.
2831 When an unrecognized warning label is requested (e.g.,
2832 @option{-Wunknown-warning}), GCC will emit a diagnostic stating
2833 that the option is not recognized. However, if the @samp{-Wno-} form
2834 is used, the behavior is slightly different: No diagnostic will be
2835 produced for @option{-Wno-unknown-warning} unless other diagnostics
2836 are being produced. This allows the use of new @option{-Wno-} options
2837 with old compilers, but if something goes wrong, the compiler will
2838 warn that an unrecognized option was used.
2843 Issue all the warnings demanded by strict ISO C and ISO C++;
2844 reject all programs that use forbidden extensions, and some other
2845 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2846 version of the ISO C standard specified by any @option{-std} option used.
2848 Valid ISO C and ISO C++ programs should compile properly with or without
2849 this option (though a rare few will require @option{-ansi} or a
2850 @option{-std} option specifying the required version of ISO C)@. However,
2851 without this option, certain GNU extensions and traditional C and C++
2852 features are supported as well. With this option, they are rejected.
2854 @option{-pedantic} does not cause warning messages for use of the
2855 alternate keywords whose names begin and end with @samp{__}. Pedantic
2856 warnings are also disabled in the expression that follows
2857 @code{__extension__}. However, only system header files should use
2858 these escape routes; application programs should avoid them.
2859 @xref{Alternate Keywords}.
2861 Some users try to use @option{-pedantic} to check programs for strict ISO
2862 C conformance. They soon find that it does not do quite what they want:
2863 it finds some non-ISO practices, but not all---only those for which
2864 ISO C @emph{requires} a diagnostic, and some others for which
2865 diagnostics have been added.
2867 A feature to report any failure to conform to ISO C might be useful in
2868 some instances, but would require considerable additional work and would
2869 be quite different from @option{-pedantic}. We don't have plans to
2870 support such a feature in the near future.
2872 Where the standard specified with @option{-std} represents a GNU
2873 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
2874 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2875 extended dialect is based. Warnings from @option{-pedantic} are given
2876 where they are required by the base standard. (It would not make sense
2877 for such warnings to be given only for features not in the specified GNU
2878 C dialect, since by definition the GNU dialects of C include all
2879 features the compiler supports with the given option, and there would be
2880 nothing to warn about.)
2882 @item -pedantic-errors
2883 @opindex pedantic-errors
2884 Like @option{-pedantic}, except that errors are produced rather than
2890 This enables all the warnings about constructions that some users
2891 consider questionable, and that are easy to avoid (or modify to
2892 prevent the warning), even in conjunction with macros. This also
2893 enables some language-specific warnings described in @ref{C++ Dialect
2894 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2896 @option{-Wall} turns on the following warning flags:
2898 @gccoptlist{-Waddress @gol
2899 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2901 -Wchar-subscripts @gol
2902 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2904 -Wimplicit-function-declaration @gol
2907 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2908 -Wmissing-braces @gol
2914 -Wsequence-point @gol
2915 -Wsign-compare @r{(only in C++)} @gol
2916 -Wstrict-aliasing @gol
2917 -Wstrict-overflow=1 @gol
2920 -Wuninitialized @gol
2921 -Wunknown-pragmas @gol
2922 -Wunused-function @gol
2925 -Wunused-variable @gol
2926 -Wvolatile-register-var @gol
2929 Note that some warning flags are not implied by @option{-Wall}. Some of
2930 them warn about constructions that users generally do not consider
2931 questionable, but which occasionally you might wish to check for;
2932 others warn about constructions that are necessary or hard to avoid in
2933 some cases, and there is no simple way to modify the code to suppress
2934 the warning. Some of them are enabled by @option{-Wextra} but many of
2935 them must be enabled individually.
2941 This enables some extra warning flags that are not enabled by
2942 @option{-Wall}. (This option used to be called @option{-W}. The older
2943 name is still supported, but the newer name is more descriptive.)
2945 @gccoptlist{-Wclobbered @gol
2947 -Wignored-qualifiers @gol
2948 -Wmissing-field-initializers @gol
2949 -Wmissing-parameter-type @r{(C only)} @gol
2950 -Wold-style-declaration @r{(C only)} @gol
2951 -Woverride-init @gol
2954 -Wuninitialized @gol
2955 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2956 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2959 The option @option{-Wextra} also prints warning messages for the
2965 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2966 @samp{>}, or @samp{>=}.
2969 (C++ only) An enumerator and a non-enumerator both appear in a
2970 conditional expression.
2973 (C++ only) Ambiguous virtual bases.
2976 (C++ only) Subscripting an array which has been declared @samp{register}.
2979 (C++ only) Taking the address of a variable which has been declared
2983 (C++ only) A base class is not initialized in a derived class' copy
2988 @item -Wchar-subscripts
2989 @opindex Wchar-subscripts
2990 @opindex Wno-char-subscripts
2991 Warn if an array subscript has type @code{char}. This is a common cause
2992 of error, as programmers often forget that this type is signed on some
2994 This warning is enabled by @option{-Wall}.
2998 @opindex Wno-comment
2999 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3000 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3001 This warning is enabled by @option{-Wall}.
3004 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3006 Suppress warning messages emitted by @code{#warning} directives.
3011 @opindex ffreestanding
3012 @opindex fno-builtin
3013 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3014 the arguments supplied have types appropriate to the format string
3015 specified, and that the conversions specified in the format string make
3016 sense. This includes standard functions, and others specified by format
3017 attributes (@pxref{Function Attributes}), in the @code{printf},
3018 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3019 not in the C standard) families (or other target-specific families).
3020 Which functions are checked without format attributes having been
3021 specified depends on the standard version selected, and such checks of
3022 functions without the attribute specified are disabled by
3023 @option{-ffreestanding} or @option{-fno-builtin}.
3025 The formats are checked against the format features supported by GNU
3026 libc version 2.2. These include all ISO C90 and C99 features, as well
3027 as features from the Single Unix Specification and some BSD and GNU
3028 extensions. Other library implementations may not support all these
3029 features; GCC does not support warning about features that go beyond a
3030 particular library's limitations. However, if @option{-pedantic} is used
3031 with @option{-Wformat}, warnings will be given about format features not
3032 in the selected standard version (but not for @code{strfmon} formats,
3033 since those are not in any version of the C standard). @xref{C Dialect
3034 Options,,Options Controlling C Dialect}.
3036 Since @option{-Wformat} also checks for null format arguments for
3037 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3039 @option{-Wformat} is included in @option{-Wall}. For more control over some
3040 aspects of format checking, the options @option{-Wformat-y2k},
3041 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3042 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3043 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
3046 @opindex Wformat-y2k
3047 @opindex Wno-format-y2k
3048 If @option{-Wformat} is specified, also warn about @code{strftime}
3049 formats which may yield only a two-digit year.
3051 @item -Wno-format-contains-nul
3052 @opindex Wno-format-contains-nul
3053 @opindex Wformat-contains-nul
3054 If @option{-Wformat} is specified, do not warn about format strings that
3057 @item -Wno-format-extra-args
3058 @opindex Wno-format-extra-args
3059 @opindex Wformat-extra-args
3060 If @option{-Wformat} is specified, do not warn about excess arguments to a
3061 @code{printf} or @code{scanf} format function. The C standard specifies
3062 that such arguments are ignored.
3064 Where the unused arguments lie between used arguments that are
3065 specified with @samp{$} operand number specifications, normally
3066 warnings are still given, since the implementation could not know what
3067 type to pass to @code{va_arg} to skip the unused arguments. However,
3068 in the case of @code{scanf} formats, this option will suppress the
3069 warning if the unused arguments are all pointers, since the Single
3070 Unix Specification says that such unused arguments are allowed.
3072 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3073 @opindex Wno-format-zero-length
3074 @opindex Wformat-zero-length
3075 If @option{-Wformat} is specified, do not warn about zero-length formats.
3076 The C standard specifies that zero-length formats are allowed.
3078 @item -Wformat-nonliteral
3079 @opindex Wformat-nonliteral
3080 @opindex Wno-format-nonliteral
3081 If @option{-Wformat} is specified, also warn if the format string is not a
3082 string literal and so cannot be checked, unless the format function
3083 takes its format arguments as a @code{va_list}.
3085 @item -Wformat-security
3086 @opindex Wformat-security
3087 @opindex Wno-format-security
3088 If @option{-Wformat} is specified, also warn about uses of format
3089 functions that represent possible security problems. At present, this
3090 warns about calls to @code{printf} and @code{scanf} functions where the
3091 format string is not a string literal and there are no format arguments,
3092 as in @code{printf (foo);}. This may be a security hole if the format
3093 string came from untrusted input and contains @samp{%n}. (This is
3094 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3095 in future warnings may be added to @option{-Wformat-security} that are not
3096 included in @option{-Wformat-nonliteral}.)
3100 @opindex Wno-format=2
3101 Enable @option{-Wformat} plus format checks not included in
3102 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3103 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3105 @item -Wnonnull @r{(C and Objective-C only)}
3107 @opindex Wno-nonnull
3108 Warn about passing a null pointer for arguments marked as
3109 requiring a non-null value by the @code{nonnull} function attribute.
3111 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3112 can be disabled with the @option{-Wno-nonnull} option.
3114 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3116 @opindex Wno-init-self
3117 Warn about uninitialized variables which are initialized with themselves.
3118 Note this option can only be used with the @option{-Wuninitialized} option.
3120 For example, GCC will warn about @code{i} being uninitialized in the
3121 following snippet only when @option{-Winit-self} has been specified:
3132 @item -Wimplicit-int @r{(C and Objective-C only)}
3133 @opindex Wimplicit-int
3134 @opindex Wno-implicit-int
3135 Warn when a declaration does not specify a type.
3136 This warning is enabled by @option{-Wall}.
3138 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3139 @opindex Wimplicit-function-declaration
3140 @opindex Wno-implicit-function-declaration
3141 Give a warning whenever a function is used before being declared. In
3142 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3143 enabled by default and it is made into an error by
3144 @option{-pedantic-errors}. This warning is also enabled by
3149 @opindex Wno-implicit
3150 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3151 This warning is enabled by @option{-Wall}.
3153 @item -Wignored-qualifiers @r{(C and C++ only)}
3154 @opindex Wignored-qualifiers
3155 @opindex Wno-ignored-qualifiers
3156 Warn if the return type of a function has a type qualifier
3157 such as @code{const}. For ISO C such a type qualifier has no effect,
3158 since the value returned by a function is not an lvalue.
3159 For C++, the warning is only emitted for scalar types or @code{void}.
3160 ISO C prohibits qualified @code{void} return types on function
3161 definitions, so such return types always receive a warning
3162 even without this option.
3164 This warning is also enabled by @option{-Wextra}.
3169 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3170 a function with external linkage, returning int, taking either zero
3171 arguments, two, or three arguments of appropriate types. This warning
3172 is enabled by default in C++ and is enabled by either @option{-Wall}
3173 or @option{-pedantic}.
3175 @item -Wmissing-braces
3176 @opindex Wmissing-braces
3177 @opindex Wno-missing-braces
3178 Warn if an aggregate or union initializer is not fully bracketed. In
3179 the following example, the initializer for @samp{a} is not fully
3180 bracketed, but that for @samp{b} is fully bracketed.
3183 int a[2][2] = @{ 0, 1, 2, 3 @};
3184 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3187 This warning is enabled by @option{-Wall}.
3189 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3190 @opindex Wmissing-include-dirs
3191 @opindex Wno-missing-include-dirs
3192 Warn if a user-supplied include directory does not exist.
3195 @opindex Wparentheses
3196 @opindex Wno-parentheses
3197 Warn if parentheses are omitted in certain contexts, such
3198 as when there is an assignment in a context where a truth value
3199 is expected, or when operators are nested whose precedence people
3200 often get confused about.
3202 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3203 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3204 interpretation from that of ordinary mathematical notation.
3206 Also warn about constructions where there may be confusion to which
3207 @code{if} statement an @code{else} branch belongs. Here is an example of
3222 In C/C++, every @code{else} branch belongs to the innermost possible
3223 @code{if} statement, which in this example is @code{if (b)}. This is
3224 often not what the programmer expected, as illustrated in the above
3225 example by indentation the programmer chose. When there is the
3226 potential for this confusion, GCC will issue a warning when this flag
3227 is specified. To eliminate the warning, add explicit braces around
3228 the innermost @code{if} statement so there is no way the @code{else}
3229 could belong to the enclosing @code{if}. The resulting code would
3246 This warning is enabled by @option{-Wall}.
3248 @item -Wsequence-point
3249 @opindex Wsequence-point
3250 @opindex Wno-sequence-point
3251 Warn about code that may have undefined semantics because of violations
3252 of sequence point rules in the C and C++ standards.
3254 The C and C++ standards defines the order in which expressions in a C/C++
3255 program are evaluated in terms of @dfn{sequence points}, which represent
3256 a partial ordering between the execution of parts of the program: those
3257 executed before the sequence point, and those executed after it. These
3258 occur after the evaluation of a full expression (one which is not part
3259 of a larger expression), after the evaluation of the first operand of a
3260 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3261 function is called (but after the evaluation of its arguments and the
3262 expression denoting the called function), and in certain other places.
3263 Other than as expressed by the sequence point rules, the order of
3264 evaluation of subexpressions of an expression is not specified. All
3265 these rules describe only a partial order rather than a total order,
3266 since, for example, if two functions are called within one expression
3267 with no sequence point between them, the order in which the functions
3268 are called is not specified. However, the standards committee have
3269 ruled that function calls do not overlap.
3271 It is not specified when between sequence points modifications to the
3272 values of objects take effect. Programs whose behavior depends on this
3273 have undefined behavior; the C and C++ standards specify that ``Between
3274 the previous and next sequence point an object shall have its stored
3275 value modified at most once by the evaluation of an expression.
3276 Furthermore, the prior value shall be read only to determine the value
3277 to be stored.''. If a program breaks these rules, the results on any
3278 particular implementation are entirely unpredictable.
3280 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3281 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3282 diagnosed by this option, and it may give an occasional false positive
3283 result, but in general it has been found fairly effective at detecting
3284 this sort of problem in programs.
3286 The standard is worded confusingly, therefore there is some debate
3287 over the precise meaning of the sequence point rules in subtle cases.
3288 Links to discussions of the problem, including proposed formal
3289 definitions, may be found on the GCC readings page, at
3290 @w{@uref{http://gcc.gnu.org/readings.html}}.
3292 This warning is enabled by @option{-Wall} for C and C++.
3295 @opindex Wreturn-type
3296 @opindex Wno-return-type
3297 Warn whenever a function is defined with a return-type that defaults
3298 to @code{int}. Also warn about any @code{return} statement with no
3299 return-value in a function whose return-type is not @code{void}
3300 (falling off the end of the function body is considered returning
3301 without a value), and about a @code{return} statement with an
3302 expression in a function whose return-type is @code{void}.
3304 For C++, a function without return type always produces a diagnostic
3305 message, even when @option{-Wno-return-type} is specified. The only
3306 exceptions are @samp{main} and functions defined in system headers.
3308 This warning is enabled by @option{-Wall}.
3313 Warn whenever a @code{switch} statement has an index of enumerated type
3314 and lacks a @code{case} for one or more of the named codes of that
3315 enumeration. (The presence of a @code{default} label prevents this
3316 warning.) @code{case} labels outside the enumeration range also
3317 provoke warnings when this option is used (even if there is a
3318 @code{default} label).
3319 This warning is enabled by @option{-Wall}.
3321 @item -Wswitch-default
3322 @opindex Wswitch-default
3323 @opindex Wno-switch-default
3324 Warn whenever a @code{switch} statement does not have a @code{default}
3328 @opindex Wswitch-enum
3329 @opindex Wno-switch-enum
3330 Warn whenever a @code{switch} statement has an index of enumerated type
3331 and lacks a @code{case} for one or more of the named codes of that
3332 enumeration. @code{case} labels outside the enumeration range also
3333 provoke warnings when this option is used. The only difference
3334 between @option{-Wswitch} and this option is that this option gives a
3335 warning about an omitted enumeration code even if there is a
3336 @code{default} label.
3338 @item -Wsync-nand @r{(C and C++ only)}
3340 @opindex Wno-sync-nand
3341 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3342 built-in functions are used. These functions changed semantics in GCC 4.4.
3346 @opindex Wno-trigraphs
3347 Warn if any trigraphs are encountered that might change the meaning of
3348 the program (trigraphs within comments are not warned about).
3349 This warning is enabled by @option{-Wall}.
3351 @item -Wunused-but-set-parameter
3352 @opindex Wunused-but-set-parameter
3353 @opindex Wno-unused-but-set-parameter
3354 Warn whenever a function parameter is assigned to, but otherwise unused
3355 (aside from its declaration).
3357 To suppress this warning use the @samp{unused} attribute
3358 (@pxref{Variable Attributes}).
3360 This warning is also enabled by @option{-Wunused} together with
3363 @item -Wunused-but-set-variable
3364 @opindex Wunused-but-set-variable
3365 @opindex Wno-unused-but-set-variable
3366 Warn whenever a local variable is assigned to, but otherwise unused
3367 (aside from its declaration).
3368 This warning is enabled by @option{-Wall}.
3370 To suppress this warning use the @samp{unused} attribute
3371 (@pxref{Variable Attributes}).
3373 This warning is also enabled by @option{-Wunused}, which is enabled
3376 @item -Wunused-function
3377 @opindex Wunused-function
3378 @opindex Wno-unused-function
3379 Warn whenever a static function is declared but not defined or a
3380 non-inline static function is unused.
3381 This warning is enabled by @option{-Wall}.
3383 @item -Wunused-label
3384 @opindex Wunused-label
3385 @opindex Wno-unused-label
3386 Warn whenever a label is declared but not used.
3387 This warning is enabled by @option{-Wall}.
3389 To suppress this warning use the @samp{unused} attribute
3390 (@pxref{Variable Attributes}).
3392 @item -Wunused-parameter
3393 @opindex Wunused-parameter
3394 @opindex Wno-unused-parameter
3395 Warn whenever a function parameter is unused aside from its declaration.
3397 To suppress this warning use the @samp{unused} attribute
3398 (@pxref{Variable Attributes}).
3400 @item -Wno-unused-result
3401 @opindex Wunused-result
3402 @opindex Wno-unused-result
3403 Do not warn if a caller of a function marked with attribute
3404 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3405 its return value. The default is @option{-Wunused-result}.
3407 @item -Wunused-variable
3408 @opindex Wunused-variable
3409 @opindex Wno-unused-variable
3410 Warn whenever a local variable or non-constant static variable is unused
3411 aside from its declaration.
3412 This warning is enabled by @option{-Wall}.
3414 To suppress this warning use the @samp{unused} attribute
3415 (@pxref{Variable Attributes}).
3417 @item -Wunused-value
3418 @opindex Wunused-value
3419 @opindex Wno-unused-value
3420 Warn whenever a statement computes a result that is explicitly not
3421 used. To suppress this warning cast the unused expression to
3422 @samp{void}. This includes an expression-statement or the left-hand
3423 side of a comma expression that contains no side effects. For example,
3424 an expression such as @samp{x[i,j]} will cause a warning, while
3425 @samp{x[(void)i,j]} will not.
3427 This warning is enabled by @option{-Wall}.
3432 All the above @option{-Wunused} options combined.
3434 In order to get a warning about an unused function parameter, you must
3435 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3436 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3438 @item -Wuninitialized
3439 @opindex Wuninitialized
3440 @opindex Wno-uninitialized
3441 Warn if an automatic variable is used without first being initialized
3442 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3443 warn if a non-static reference or non-static @samp{const} member
3444 appears in a class without constructors.
3446 If you want to warn about code which uses the uninitialized value of the
3447 variable in its own initializer, use the @option{-Winit-self} option.
3449 These warnings occur for individual uninitialized or clobbered
3450 elements of structure, union or array variables as well as for
3451 variables which are uninitialized or clobbered as a whole. They do
3452 not occur for variables or elements declared @code{volatile}. Because
3453 these warnings depend on optimization, the exact variables or elements
3454 for which there are warnings will depend on the precise optimization
3455 options and version of GCC used.
3457 Note that there may be no warning about a variable that is used only
3458 to compute a value that itself is never used, because such
3459 computations may be deleted by data flow analysis before the warnings
3462 These warnings are made optional because GCC is not smart
3463 enough to see all the reasons why the code might be correct
3464 despite appearing to have an error. Here is one example of how
3485 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3486 always initialized, but GCC doesn't know this. Here is
3487 another common case:
3492 if (change_y) save_y = y, y = new_y;
3494 if (change_y) y = save_y;
3499 This has no bug because @code{save_y} is used only if it is set.
3501 @cindex @code{longjmp} warnings
3502 This option also warns when a non-volatile automatic variable might be
3503 changed by a call to @code{longjmp}. These warnings as well are possible
3504 only in optimizing compilation.
3506 The compiler sees only the calls to @code{setjmp}. It cannot know
3507 where @code{longjmp} will be called; in fact, a signal handler could
3508 call it at any point in the code. As a result, you may get a warning
3509 even when there is in fact no problem because @code{longjmp} cannot
3510 in fact be called at the place which would cause a problem.
3512 Some spurious warnings can be avoided if you declare all the functions
3513 you use that never return as @code{noreturn}. @xref{Function
3516 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3518 @item -Wunknown-pragmas
3519 @opindex Wunknown-pragmas
3520 @opindex Wno-unknown-pragmas
3521 @cindex warning for unknown pragmas
3522 @cindex unknown pragmas, warning
3523 @cindex pragmas, warning of unknown
3524 Warn when a #pragma directive is encountered which is not understood by
3525 GCC@. If this command line option is used, warnings will even be issued
3526 for unknown pragmas in system header files. This is not the case if
3527 the warnings were only enabled by the @option{-Wall} command line option.
3530 @opindex Wno-pragmas
3532 Do not warn about misuses of pragmas, such as incorrect parameters,
3533 invalid syntax, or conflicts between pragmas. See also
3534 @samp{-Wunknown-pragmas}.
3536 @item -Wstrict-aliasing
3537 @opindex Wstrict-aliasing
3538 @opindex Wno-strict-aliasing
3539 This option is only active when @option{-fstrict-aliasing} is active.
3540 It warns about code which might break the strict aliasing rules that the
3541 compiler is using for optimization. The warning does not catch all
3542 cases, but does attempt to catch the more common pitfalls. It is
3543 included in @option{-Wall}.
3544 It is equivalent to @option{-Wstrict-aliasing=3}
3546 @item -Wstrict-aliasing=n
3547 @opindex Wstrict-aliasing=n
3548 @opindex Wno-strict-aliasing=n
3549 This option is only active when @option{-fstrict-aliasing} is active.
3550 It warns about code which might break the strict aliasing rules that the
3551 compiler is using for optimization.
3552 Higher levels correspond to higher accuracy (fewer false positives).
3553 Higher levels also correspond to more effort, similar to the way -O works.
3554 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3557 Level 1: Most aggressive, quick, least accurate.
3558 Possibly useful when higher levels
3559 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3560 false negatives. However, it has many false positives.
3561 Warns for all pointer conversions between possibly incompatible types,
3562 even if never dereferenced. Runs in the frontend only.
3564 Level 2: Aggressive, quick, not too precise.
3565 May still have many false positives (not as many as level 1 though),
3566 and few false negatives (but possibly more than level 1).
3567 Unlike level 1, it only warns when an address is taken. Warns about
3568 incomplete types. Runs in the frontend only.
3570 Level 3 (default for @option{-Wstrict-aliasing}):
3571 Should have very few false positives and few false
3572 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3573 Takes care of the common pun+dereference pattern in the frontend:
3574 @code{*(int*)&some_float}.
3575 If optimization is enabled, it also runs in the backend, where it deals
3576 with multiple statement cases using flow-sensitive points-to information.
3577 Only warns when the converted pointer is dereferenced.
3578 Does not warn about incomplete types.
3580 @item -Wstrict-overflow
3581 @itemx -Wstrict-overflow=@var{n}
3582 @opindex Wstrict-overflow
3583 @opindex Wno-strict-overflow
3584 This option is only active when @option{-fstrict-overflow} is active.
3585 It warns about cases where the compiler optimizes based on the
3586 assumption that signed overflow does not occur. Note that it does not
3587 warn about all cases where the code might overflow: it only warns
3588 about cases where the compiler implements some optimization. Thus
3589 this warning depends on the optimization level.
3591 An optimization which assumes that signed overflow does not occur is
3592 perfectly safe if the values of the variables involved are such that
3593 overflow never does, in fact, occur. Therefore this warning can
3594 easily give a false positive: a warning about code which is not
3595 actually a problem. To help focus on important issues, several
3596 warning levels are defined. No warnings are issued for the use of
3597 undefined signed overflow when estimating how many iterations a loop
3598 will require, in particular when determining whether a loop will be
3602 @item -Wstrict-overflow=1
3603 Warn about cases which are both questionable and easy to avoid. For
3604 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3605 compiler will simplify this to @code{1}. This level of
3606 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3607 are not, and must be explicitly requested.
3609 @item -Wstrict-overflow=2
3610 Also warn about other cases where a comparison is simplified to a
3611 constant. For example: @code{abs (x) >= 0}. This can only be
3612 simplified when @option{-fstrict-overflow} is in effect, because
3613 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3614 zero. @option{-Wstrict-overflow} (with no level) is the same as
3615 @option{-Wstrict-overflow=2}.
3617 @item -Wstrict-overflow=3
3618 Also warn about other cases where a comparison is simplified. For
3619 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3621 @item -Wstrict-overflow=4
3622 Also warn about other simplifications not covered by the above cases.
3623 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3625 @item -Wstrict-overflow=5
3626 Also warn about cases where the compiler reduces the magnitude of a
3627 constant involved in a comparison. For example: @code{x + 2 > y} will
3628 be simplified to @code{x + 1 >= y}. This is reported only at the
3629 highest warning level because this simplification applies to many
3630 comparisons, so this warning level will give a very large number of
3634 @item -Wsuggest-attribute=@r{[}const@r{|}pure@r{]}
3635 @opindex Wsuggest-attribute=
3636 @opindex Wno-suggest-attribute=
3637 Warn for cases where adding an attribute may be beneficial. The
3638 attributes currently supported are listed below.
3641 @item -Wsuggest-attribute=pure
3642 @itemx -Wsuggest-attribute=const
3643 @opindex Wsuggest-attribute=pure
3644 @opindex Wno-suggest-attribute=pure
3645 @opindex Wsuggest-attribute=const
3646 @opindex Wno-suggest-attribute=const
3648 Warn about functions which might be candidates for attributes
3649 @code{pure} or @code{const}. The compiler only warns for functions
3650 visible in other compilation units or if it cannot prove that the
3651 function returns normally. A function returns normally if it doesn't
3652 contain an infinite loop nor returns abnormally by throwing, calling
3653 @code{abort()} or trapping. This analysis requires option
3654 @option{-fipa-pure-const}, which is enabled by default at @option{-O}
3655 and higher. Higher optimization levels improve the accuracy of the
3659 @item -Warray-bounds
3660 @opindex Wno-array-bounds
3661 @opindex Warray-bounds
3662 This option is only active when @option{-ftree-vrp} is active
3663 (default for @option{-O2} and above). It warns about subscripts to arrays
3664 that are always out of bounds. This warning is enabled by @option{-Wall}.
3666 @item -Wno-div-by-zero
3667 @opindex Wno-div-by-zero
3668 @opindex Wdiv-by-zero
3669 Do not warn about compile-time integer division by zero. Floating point
3670 division by zero is not warned about, as it can be a legitimate way of
3671 obtaining infinities and NaNs.
3673 @item -Wsystem-headers
3674 @opindex Wsystem-headers
3675 @opindex Wno-system-headers
3676 @cindex warnings from system headers
3677 @cindex system headers, warnings from
3678 Print warning messages for constructs found in system header files.
3679 Warnings from system headers are normally suppressed, on the assumption
3680 that they usually do not indicate real problems and would only make the
3681 compiler output harder to read. Using this command line option tells
3682 GCC to emit warnings from system headers as if they occurred in user
3683 code. However, note that using @option{-Wall} in conjunction with this
3684 option will @emph{not} warn about unknown pragmas in system
3685 headers---for that, @option{-Wunknown-pragmas} must also be used.
3688 @opindex Wfloat-equal
3689 @opindex Wno-float-equal
3690 Warn if floating point values are used in equality comparisons.
3692 The idea behind this is that sometimes it is convenient (for the
3693 programmer) to consider floating-point values as approximations to
3694 infinitely precise real numbers. If you are doing this, then you need
3695 to compute (by analyzing the code, or in some other way) the maximum or
3696 likely maximum error that the computation introduces, and allow for it
3697 when performing comparisons (and when producing output, but that's a
3698 different problem). In particular, instead of testing for equality, you
3699 would check to see whether the two values have ranges that overlap; and
3700 this is done with the relational operators, so equality comparisons are
3703 @item -Wtraditional @r{(C and Objective-C only)}
3704 @opindex Wtraditional
3705 @opindex Wno-traditional
3706 Warn about certain constructs that behave differently in traditional and
3707 ISO C@. Also warn about ISO C constructs that have no traditional C
3708 equivalent, and/or problematic constructs which should be avoided.
3712 Macro parameters that appear within string literals in the macro body.
3713 In traditional C macro replacement takes place within string literals,
3714 but does not in ISO C@.
3717 In traditional C, some preprocessor directives did not exist.
3718 Traditional preprocessors would only consider a line to be a directive
3719 if the @samp{#} appeared in column 1 on the line. Therefore
3720 @option{-Wtraditional} warns about directives that traditional C
3721 understands but would ignore because the @samp{#} does not appear as the
3722 first character on the line. It also suggests you hide directives like
3723 @samp{#pragma} not understood by traditional C by indenting them. Some
3724 traditional implementations would not recognize @samp{#elif}, so it
3725 suggests avoiding it altogether.
3728 A function-like macro that appears without arguments.
3731 The unary plus operator.
3734 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3735 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3736 constants.) Note, these suffixes appear in macros defined in the system
3737 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3738 Use of these macros in user code might normally lead to spurious
3739 warnings, however GCC's integrated preprocessor has enough context to
3740 avoid warning in these cases.
3743 A function declared external in one block and then used after the end of
3747 A @code{switch} statement has an operand of type @code{long}.
3750 A non-@code{static} function declaration follows a @code{static} one.
3751 This construct is not accepted by some traditional C compilers.
3754 The ISO type of an integer constant has a different width or
3755 signedness from its traditional type. This warning is only issued if
3756 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3757 typically represent bit patterns, are not warned about.
3760 Usage of ISO string concatenation is detected.
3763 Initialization of automatic aggregates.
3766 Identifier conflicts with labels. Traditional C lacks a separate
3767 namespace for labels.
3770 Initialization of unions. If the initializer is zero, the warning is
3771 omitted. This is done under the assumption that the zero initializer in
3772 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3773 initializer warnings and relies on default initialization to zero in the
3777 Conversions by prototypes between fixed/floating point values and vice
3778 versa. The absence of these prototypes when compiling with traditional
3779 C would cause serious problems. This is a subset of the possible
3780 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3783 Use of ISO C style function definitions. This warning intentionally is
3784 @emph{not} issued for prototype declarations or variadic functions
3785 because these ISO C features will appear in your code when using
3786 libiberty's traditional C compatibility macros, @code{PARAMS} and
3787 @code{VPARAMS}. This warning is also bypassed for nested functions
3788 because that feature is already a GCC extension and thus not relevant to
3789 traditional C compatibility.
3792 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3793 @opindex Wtraditional-conversion
3794 @opindex Wno-traditional-conversion
3795 Warn if a prototype causes a type conversion that is different from what
3796 would happen to the same argument in the absence of a prototype. This
3797 includes conversions of fixed point to floating and vice versa, and
3798 conversions changing the width or signedness of a fixed point argument
3799 except when the same as the default promotion.
3801 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3802 @opindex Wdeclaration-after-statement
3803 @opindex Wno-declaration-after-statement
3804 Warn when a declaration is found after a statement in a block. This
3805 construct, known from C++, was introduced with ISO C99 and is by default
3806 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3807 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3812 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3814 @item -Wno-endif-labels
3815 @opindex Wno-endif-labels
3816 @opindex Wendif-labels
3817 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3822 Warn whenever a local variable shadows another local variable, parameter or
3823 global variable or whenever a built-in function is shadowed.
3825 @item -Wlarger-than=@var{len}
3826 @opindex Wlarger-than=@var{len}
3827 @opindex Wlarger-than-@var{len}
3828 Warn whenever an object of larger than @var{len} bytes is defined.
3830 @item -Wframe-larger-than=@var{len}
3831 @opindex Wframe-larger-than
3832 Warn if the size of a function frame is larger than @var{len} bytes.
3833 The computation done to determine the stack frame size is approximate
3834 and not conservative.
3835 The actual requirements may be somewhat greater than @var{len}
3836 even if you do not get a warning. In addition, any space allocated
3837 via @code{alloca}, variable-length arrays, or related constructs
3838 is not included by the compiler when determining
3839 whether or not to issue a warning.
3841 @item -Wunsafe-loop-optimizations
3842 @opindex Wunsafe-loop-optimizations
3843 @opindex Wno-unsafe-loop-optimizations
3844 Warn if the loop cannot be optimized because the compiler could not
3845 assume anything on the bounds of the loop indices. With
3846 @option{-funsafe-loop-optimizations} warn if the compiler made
3849 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3850 @opindex Wno-pedantic-ms-format
3851 @opindex Wpedantic-ms-format
3852 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3853 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3854 depending on the MS runtime, when you are using the options @option{-Wformat}
3855 and @option{-pedantic} without gnu-extensions.
3857 @item -Wpointer-arith
3858 @opindex Wpointer-arith
3859 @opindex Wno-pointer-arith
3860 Warn about anything that depends on the ``size of'' a function type or
3861 of @code{void}. GNU C assigns these types a size of 1, for
3862 convenience in calculations with @code{void *} pointers and pointers
3863 to functions. In C++, warn also when an arithmetic operation involves
3864 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3867 @opindex Wtype-limits
3868 @opindex Wno-type-limits
3869 Warn if a comparison is always true or always false due to the limited
3870 range of the data type, but do not warn for constant expressions. For
3871 example, warn if an unsigned variable is compared against zero with
3872 @samp{<} or @samp{>=}. This warning is also enabled by
3875 @item -Wbad-function-cast @r{(C and Objective-C only)}
3876 @opindex Wbad-function-cast
3877 @opindex Wno-bad-function-cast
3878 Warn whenever a function call is cast to a non-matching type.
3879 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3881 @item -Wc++-compat @r{(C and Objective-C only)}
3882 Warn about ISO C constructs that are outside of the common subset of
3883 ISO C and ISO C++, e.g.@: request for implicit conversion from
3884 @code{void *} to a pointer to non-@code{void} type.
3886 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3887 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3888 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3889 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3893 @opindex Wno-cast-qual
3894 Warn whenever a pointer is cast so as to remove a type qualifier from
3895 the target type. For example, warn if a @code{const char *} is cast
3896 to an ordinary @code{char *}.
3898 Also warn when making a cast which introduces a type qualifier in an
3899 unsafe way. For example, casting @code{char **} to @code{const char **}
3900 is unsafe, as in this example:
3903 /* p is char ** value. */
3904 const char **q = (const char **) p;
3905 /* Assignment of readonly string to const char * is OK. */
3907 /* Now char** pointer points to read-only memory. */
3912 @opindex Wcast-align
3913 @opindex Wno-cast-align
3914 Warn whenever a pointer is cast such that the required alignment of the
3915 target is increased. For example, warn if a @code{char *} is cast to
3916 an @code{int *} on machines where integers can only be accessed at
3917 two- or four-byte boundaries.
3919 @item -Wwrite-strings
3920 @opindex Wwrite-strings
3921 @opindex Wno-write-strings
3922 When compiling C, give string constants the type @code{const
3923 char[@var{length}]} so that copying the address of one into a
3924 non-@code{const} @code{char *} pointer will get a warning. These
3925 warnings will help you find at compile time code that can try to write
3926 into a string constant, but only if you have been very careful about
3927 using @code{const} in declarations and prototypes. Otherwise, it will
3928 just be a nuisance. This is why we did not make @option{-Wall} request
3931 When compiling C++, warn about the deprecated conversion from string
3932 literals to @code{char *}. This warning is enabled by default for C++
3937 @opindex Wno-clobbered
3938 Warn for variables that might be changed by @samp{longjmp} or
3939 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3942 @opindex Wconversion
3943 @opindex Wno-conversion
3944 Warn for implicit conversions that may alter a value. This includes
3945 conversions between real and integer, like @code{abs (x)} when
3946 @code{x} is @code{double}; conversions between signed and unsigned,
3947 like @code{unsigned ui = -1}; and conversions to smaller types, like
3948 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3949 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3950 changed by the conversion like in @code{abs (2.0)}. Warnings about
3951 conversions between signed and unsigned integers can be disabled by
3952 using @option{-Wno-sign-conversion}.
3954 For C++, also warn for confusing overload resolution for user-defined
3955 conversions; and conversions that will never use a type conversion
3956 operator: conversions to @code{void}, the same type, a base class or a
3957 reference to them. Warnings about conversions between signed and
3958 unsigned integers are disabled by default in C++ unless
3959 @option{-Wsign-conversion} is explicitly enabled.
3961 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
3962 @opindex Wconversion-null
3963 @opindex Wno-conversion-null
3964 Do not warn for conversions between @code{NULL} and non-pointer
3965 types. @option{-Wconversion-null} is enabled by default.
3968 @opindex Wempty-body
3969 @opindex Wno-empty-body
3970 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3971 while} statement. This warning is also enabled by @option{-Wextra}.
3973 @item -Wenum-compare
3974 @opindex Wenum-compare
3975 @opindex Wno-enum-compare
3976 Warn about a comparison between values of different enum types. In C++
3977 this warning is enabled by default. In C this warning is enabled by
3980 @item -Wjump-misses-init @r{(C, Objective-C only)}
3981 @opindex Wjump-misses-init
3982 @opindex Wno-jump-misses-init
3983 Warn if a @code{goto} statement or a @code{switch} statement jumps
3984 forward across the initialization of a variable, or jumps backward to a
3985 label after the variable has been initialized. This only warns about
3986 variables which are initialized when they are declared. This warning is
3987 only supported for C and Objective C; in C++ this sort of branch is an
3990 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
3991 can be disabled with the @option{-Wno-jump-misses-init} option.
3993 @item -Wsign-compare
3994 @opindex Wsign-compare
3995 @opindex Wno-sign-compare
3996 @cindex warning for comparison of signed and unsigned values
3997 @cindex comparison of signed and unsigned values, warning
3998 @cindex signed and unsigned values, comparison warning
3999 Warn when a comparison between signed and unsigned values could produce
4000 an incorrect result when the signed value is converted to unsigned.
4001 This warning is also enabled by @option{-Wextra}; to get the other warnings
4002 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
4004 @item -Wsign-conversion
4005 @opindex Wsign-conversion
4006 @opindex Wno-sign-conversion
4007 Warn for implicit conversions that may change the sign of an integer
4008 value, like assigning a signed integer expression to an unsigned
4009 integer variable. An explicit cast silences the warning. In C, this
4010 option is enabled also by @option{-Wconversion}.
4014 @opindex Wno-address
4015 Warn about suspicious uses of memory addresses. These include using
4016 the address of a function in a conditional expression, such as
4017 @code{void func(void); if (func)}, and comparisons against the memory
4018 address of a string literal, such as @code{if (x == "abc")}. Such
4019 uses typically indicate a programmer error: the address of a function
4020 always evaluates to true, so their use in a conditional usually
4021 indicate that the programmer forgot the parentheses in a function
4022 call; and comparisons against string literals result in unspecified
4023 behavior and are not portable in C, so they usually indicate that the
4024 programmer intended to use @code{strcmp}. This warning is enabled by
4028 @opindex Wlogical-op
4029 @opindex Wno-logical-op
4030 Warn about suspicious uses of logical operators in expressions.
4031 This includes using logical operators in contexts where a
4032 bit-wise operator is likely to be expected.
4034 @item -Waggregate-return
4035 @opindex Waggregate-return
4036 @opindex Wno-aggregate-return
4037 Warn if any functions that return structures or unions are defined or
4038 called. (In languages where you can return an array, this also elicits
4041 @item -Wno-attributes
4042 @opindex Wno-attributes
4043 @opindex Wattributes
4044 Do not warn if an unexpected @code{__attribute__} is used, such as
4045 unrecognized attributes, function attributes applied to variables,
4046 etc. This will not stop errors for incorrect use of supported
4049 @item -Wno-builtin-macro-redefined
4050 @opindex Wno-builtin-macro-redefined
4051 @opindex Wbuiltin-macro-redefined
4052 Do not warn if certain built-in macros are redefined. This suppresses
4053 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4054 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4056 @item -Wstrict-prototypes @r{(C and Objective-C only)}
4057 @opindex Wstrict-prototypes
4058 @opindex Wno-strict-prototypes
4059 Warn if a function is declared or defined without specifying the
4060 argument types. (An old-style function definition is permitted without
4061 a warning if preceded by a declaration which specifies the argument
4064 @item -Wold-style-declaration @r{(C and Objective-C only)}
4065 @opindex Wold-style-declaration
4066 @opindex Wno-old-style-declaration
4067 Warn for obsolescent usages, according to the C Standard, in a
4068 declaration. For example, warn if storage-class specifiers like
4069 @code{static} are not the first things in a declaration. This warning
4070 is also enabled by @option{-Wextra}.
4072 @item -Wold-style-definition @r{(C and Objective-C only)}
4073 @opindex Wold-style-definition
4074 @opindex Wno-old-style-definition
4075 Warn if an old-style function definition is used. A warning is given
4076 even if there is a previous prototype.
4078 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
4079 @opindex Wmissing-parameter-type
4080 @opindex Wno-missing-parameter-type
4081 A function parameter is declared without a type specifier in K&R-style
4088 This warning is also enabled by @option{-Wextra}.
4090 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4091 @opindex Wmissing-prototypes
4092 @opindex Wno-missing-prototypes
4093 Warn if a global function is defined without a previous prototype
4094 declaration. This warning is issued even if the definition itself
4095 provides a prototype. The aim is to detect global functions that fail
4096 to be declared in header files.
4098 @item -Wmissing-declarations
4099 @opindex Wmissing-declarations
4100 @opindex Wno-missing-declarations
4101 Warn if a global function is defined without a previous declaration.
4102 Do so even if the definition itself provides a prototype.
4103 Use this option to detect global functions that are not declared in
4104 header files. In C++, no warnings are issued for function templates,
4105 or for inline functions, or for functions in anonymous namespaces.
4107 @item -Wmissing-field-initializers
4108 @opindex Wmissing-field-initializers
4109 @opindex Wno-missing-field-initializers
4113 Warn if a structure's initializer has some fields missing. For
4114 example, the following code would cause such a warning, because
4115 @code{x.h} is implicitly zero:
4118 struct s @{ int f, g, h; @};
4119 struct s x = @{ 3, 4 @};
4122 This option does not warn about designated initializers, so the following
4123 modification would not trigger a warning:
4126 struct s @{ int f, g, h; @};
4127 struct s x = @{ .f = 3, .g = 4 @};
4130 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4131 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4133 @item -Wmissing-noreturn
4134 @opindex Wmissing-noreturn
4135 @opindex Wno-missing-noreturn
4136 Warn about functions which might be candidates for attribute @code{noreturn}.
4137 Note these are only possible candidates, not absolute ones. Care should
4138 be taken to manually verify functions actually do not ever return before
4139 adding the @code{noreturn} attribute, otherwise subtle code generation
4140 bugs could be introduced. You will not get a warning for @code{main} in
4141 hosted C environments.
4143 @item -Wmissing-format-attribute
4144 @opindex Wmissing-format-attribute
4145 @opindex Wno-missing-format-attribute
4148 Warn about function pointers which might be candidates for @code{format}
4149 attributes. Note these are only possible candidates, not absolute ones.
4150 GCC will guess that function pointers with @code{format} attributes that
4151 are used in assignment, initialization, parameter passing or return
4152 statements should have a corresponding @code{format} attribute in the
4153 resulting type. I.e.@: the left-hand side of the assignment or
4154 initialization, the type of the parameter variable, or the return type
4155 of the containing function respectively should also have a @code{format}
4156 attribute to avoid the warning.
4158 GCC will also warn about function definitions which might be
4159 candidates for @code{format} attributes. Again, these are only
4160 possible candidates. GCC will guess that @code{format} attributes
4161 might be appropriate for any function that calls a function like
4162 @code{vprintf} or @code{vscanf}, but this might not always be the
4163 case, and some functions for which @code{format} attributes are
4164 appropriate may not be detected.
4166 @item -Wno-multichar
4167 @opindex Wno-multichar
4169 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4170 Usually they indicate a typo in the user's code, as they have
4171 implementation-defined values, and should not be used in portable code.
4173 @item -Wnormalized=<none|id|nfc|nfkc>
4174 @opindex Wnormalized=
4177 @cindex character set, input normalization
4178 In ISO C and ISO C++, two identifiers are different if they are
4179 different sequences of characters. However, sometimes when characters
4180 outside the basic ASCII character set are used, you can have two
4181 different character sequences that look the same. To avoid confusion,
4182 the ISO 10646 standard sets out some @dfn{normalization rules} which
4183 when applied ensure that two sequences that look the same are turned into
4184 the same sequence. GCC can warn you if you are using identifiers which
4185 have not been normalized; this option controls that warning.
4187 There are four levels of warning that GCC supports. The default is
4188 @option{-Wnormalized=nfc}, which warns about any identifier which is
4189 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4190 recommended form for most uses.
4192 Unfortunately, there are some characters which ISO C and ISO C++ allow
4193 in identifiers that when turned into NFC aren't allowable as
4194 identifiers. That is, there's no way to use these symbols in portable
4195 ISO C or C++ and have all your identifiers in NFC@.
4196 @option{-Wnormalized=id} suppresses the warning for these characters.
4197 It is hoped that future versions of the standards involved will correct
4198 this, which is why this option is not the default.
4200 You can switch the warning off for all characters by writing
4201 @option{-Wnormalized=none}. You would only want to do this if you
4202 were using some other normalization scheme (like ``D''), because
4203 otherwise you can easily create bugs that are literally impossible to see.
4205 Some characters in ISO 10646 have distinct meanings but look identical
4206 in some fonts or display methodologies, especially once formatting has
4207 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4208 LETTER N'', will display just like a regular @code{n} which has been
4209 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4210 normalization scheme to convert all these into a standard form as
4211 well, and GCC will warn if your code is not in NFKC if you use
4212 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4213 about every identifier that contains the letter O because it might be
4214 confused with the digit 0, and so is not the default, but may be
4215 useful as a local coding convention if the programming environment is
4216 unable to be fixed to display these characters distinctly.
4218 @item -Wno-deprecated
4219 @opindex Wno-deprecated
4220 @opindex Wdeprecated
4221 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4223 @item -Wno-deprecated-declarations
4224 @opindex Wno-deprecated-declarations
4225 @opindex Wdeprecated-declarations
4226 Do not warn about uses of functions (@pxref{Function Attributes}),
4227 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4228 Attributes}) marked as deprecated by using the @code{deprecated}
4232 @opindex Wno-overflow
4234 Do not warn about compile-time overflow in constant expressions.
4236 @item -Woverride-init @r{(C and Objective-C only)}
4237 @opindex Woverride-init
4238 @opindex Wno-override-init
4242 Warn if an initialized field without side effects is overridden when
4243 using designated initializers (@pxref{Designated Inits, , Designated
4246 This warning is included in @option{-Wextra}. To get other
4247 @option{-Wextra} warnings without this one, use @samp{-Wextra
4248 -Wno-override-init}.
4253 Warn if a structure is given the packed attribute, but the packed
4254 attribute has no effect on the layout or size of the structure.
4255 Such structures may be mis-aligned for little benefit. For
4256 instance, in this code, the variable @code{f.x} in @code{struct bar}
4257 will be misaligned even though @code{struct bar} does not itself
4258 have the packed attribute:
4265 @} __attribute__((packed));
4273 @item -Wpacked-bitfield-compat
4274 @opindex Wpacked-bitfield-compat
4275 @opindex Wno-packed-bitfield-compat
4276 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4277 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4278 the change can lead to differences in the structure layout. GCC
4279 informs you when the offset of such a field has changed in GCC 4.4.
4280 For example there is no longer a 4-bit padding between field @code{a}
4281 and @code{b} in this structure:
4288 @} __attribute__ ((packed));
4291 This warning is enabled by default. Use
4292 @option{-Wno-packed-bitfield-compat} to disable this warning.
4297 Warn if padding is included in a structure, either to align an element
4298 of the structure or to align the whole structure. Sometimes when this
4299 happens it is possible to rearrange the fields of the structure to
4300 reduce the padding and so make the structure smaller.
4302 @item -Wredundant-decls
4303 @opindex Wredundant-decls
4304 @opindex Wno-redundant-decls
4305 Warn if anything is declared more than once in the same scope, even in
4306 cases where multiple declaration is valid and changes nothing.
4308 @item -Wnested-externs @r{(C and Objective-C only)}
4309 @opindex Wnested-externs
4310 @opindex Wno-nested-externs
4311 Warn if an @code{extern} declaration is encountered within a function.
4316 Warn if a function can not be inlined and it was declared as inline.
4317 Even with this option, the compiler will not warn about failures to
4318 inline functions declared in system headers.
4320 The compiler uses a variety of heuristics to determine whether or not
4321 to inline a function. For example, the compiler takes into account
4322 the size of the function being inlined and the amount of inlining
4323 that has already been done in the current function. Therefore,
4324 seemingly insignificant changes in the source program can cause the
4325 warnings produced by @option{-Winline} to appear or disappear.
4327 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4328 @opindex Wno-invalid-offsetof
4329 @opindex Winvalid-offsetof
4330 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4331 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4332 to a non-POD type is undefined. In existing C++ implementations,
4333 however, @samp{offsetof} typically gives meaningful results even when
4334 applied to certain kinds of non-POD types. (Such as a simple
4335 @samp{struct} that fails to be a POD type only by virtue of having a
4336 constructor.) This flag is for users who are aware that they are
4337 writing nonportable code and who have deliberately chosen to ignore the
4340 The restrictions on @samp{offsetof} may be relaxed in a future version
4341 of the C++ standard.
4343 @item -Wno-int-to-pointer-cast
4344 @opindex Wno-int-to-pointer-cast
4345 @opindex Wint-to-pointer-cast
4346 Suppress warnings from casts to pointer type of an integer of a
4347 different size. In C++, casting to a pointer type of smaller size is
4348 an error. @option{Wint-to-pointer-cast} is enabled by default.
4351 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4352 @opindex Wno-pointer-to-int-cast
4353 @opindex Wpointer-to-int-cast
4354 Suppress warnings from casts from a pointer to an integer type of a
4358 @opindex Winvalid-pch
4359 @opindex Wno-invalid-pch
4360 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4361 the search path but can't be used.
4365 @opindex Wno-long-long
4366 Warn if @samp{long long} type is used. This is enabled by either
4367 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4368 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4370 @item -Wvariadic-macros
4371 @opindex Wvariadic-macros
4372 @opindex Wno-variadic-macros
4373 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4374 alternate syntax when in pedantic ISO C99 mode. This is default.
4375 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4380 Warn if variable length array is used in the code.
4381 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4382 the variable length array.
4384 @item -Wvolatile-register-var
4385 @opindex Wvolatile-register-var
4386 @opindex Wno-volatile-register-var
4387 Warn if a register variable is declared volatile. The volatile
4388 modifier does not inhibit all optimizations that may eliminate reads
4389 and/or writes to register variables. This warning is enabled by
4392 @item -Wdisabled-optimization
4393 @opindex Wdisabled-optimization
4394 @opindex Wno-disabled-optimization
4395 Warn if a requested optimization pass is disabled. This warning does
4396 not generally indicate that there is anything wrong with your code; it
4397 merely indicates that GCC's optimizers were unable to handle the code
4398 effectively. Often, the problem is that your code is too big or too
4399 complex; GCC will refuse to optimize programs when the optimization
4400 itself is likely to take inordinate amounts of time.
4402 @item -Wpointer-sign @r{(C and Objective-C only)}
4403 @opindex Wpointer-sign
4404 @opindex Wno-pointer-sign
4405 Warn for pointer argument passing or assignment with different signedness.
4406 This option is only supported for C and Objective-C@. It is implied by
4407 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4408 @option{-Wno-pointer-sign}.
4410 @item -Wstack-protector
4411 @opindex Wstack-protector
4412 @opindex Wno-stack-protector
4413 This option is only active when @option{-fstack-protector} is active. It
4414 warns about functions that will not be protected against stack smashing.
4417 @opindex Wno-mudflap
4418 Suppress warnings about constructs that cannot be instrumented by
4421 @item -Woverlength-strings
4422 @opindex Woverlength-strings
4423 @opindex Wno-overlength-strings
4424 Warn about string constants which are longer than the ``minimum
4425 maximum'' length specified in the C standard. Modern compilers
4426 generally allow string constants which are much longer than the
4427 standard's minimum limit, but very portable programs should avoid
4428 using longer strings.
4430 The limit applies @emph{after} string constant concatenation, and does
4431 not count the trailing NUL@. In C90, the limit was 509 characters; in
4432 C99, it was raised to 4095. C++98 does not specify a normative
4433 minimum maximum, so we do not diagnose overlength strings in C++@.
4435 This option is implied by @option{-pedantic}, and can be disabled with
4436 @option{-Wno-overlength-strings}.
4438 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4439 @opindex Wunsuffixed-float-constants
4441 GCC will issue a warning for any floating constant that does not have
4442 a suffix. When used together with @option{-Wsystem-headers} it will
4443 warn about such constants in system header files. This can be useful
4444 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4445 from the decimal floating-point extension to C99.
4448 @node Debugging Options
4449 @section Options for Debugging Your Program or GCC
4450 @cindex options, debugging
4451 @cindex debugging information options
4453 GCC has various special options that are used for debugging
4454 either your program or GCC:
4459 Produce debugging information in the operating system's native format
4460 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4463 On most systems that use stabs format, @option{-g} enables use of extra
4464 debugging information that only GDB can use; this extra information
4465 makes debugging work better in GDB but will probably make other debuggers
4467 refuse to read the program. If you want to control for certain whether
4468 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4469 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4471 GCC allows you to use @option{-g} with
4472 @option{-O}. The shortcuts taken by optimized code may occasionally
4473 produce surprising results: some variables you declared may not exist
4474 at all; flow of control may briefly move where you did not expect it;
4475 some statements may not be executed because they compute constant
4476 results or their values were already at hand; some statements may
4477 execute in different places because they were moved out of loops.
4479 Nevertheless it proves possible to debug optimized output. This makes
4480 it reasonable to use the optimizer for programs that might have bugs.
4482 The following options are useful when GCC is generated with the
4483 capability for more than one debugging format.
4487 Produce debugging information for use by GDB@. This means to use the
4488 most expressive format available (DWARF 2, stabs, or the native format
4489 if neither of those are supported), including GDB extensions if at all
4494 Produce debugging information in stabs format (if that is supported),
4495 without GDB extensions. This is the format used by DBX on most BSD
4496 systems. On MIPS, Alpha and System V Release 4 systems this option
4497 produces stabs debugging output which is not understood by DBX or SDB@.
4498 On System V Release 4 systems this option requires the GNU assembler.
4500 @item -feliminate-unused-debug-symbols
4501 @opindex feliminate-unused-debug-symbols
4502 Produce debugging information in stabs format (if that is supported),
4503 for only symbols that are actually used.
4505 @item -femit-class-debug-always
4506 Instead of emitting debugging information for a C++ class in only one
4507 object file, emit it in all object files using the class. This option
4508 should be used only with debuggers that are unable to handle the way GCC
4509 normally emits debugging information for classes because using this
4510 option will increase the size of debugging information by as much as a
4515 Produce debugging information in stabs format (if that is supported),
4516 using GNU extensions understood only by the GNU debugger (GDB)@. The
4517 use of these extensions is likely to make other debuggers crash or
4518 refuse to read the program.
4522 Produce debugging information in COFF format (if that is supported).
4523 This is the format used by SDB on most System V systems prior to
4528 Produce debugging information in XCOFF format (if that is supported).
4529 This is the format used by the DBX debugger on IBM RS/6000 systems.
4533 Produce debugging information in XCOFF format (if that is supported),
4534 using GNU extensions understood only by the GNU debugger (GDB)@. The
4535 use of these extensions is likely to make other debuggers crash or
4536 refuse to read the program, and may cause assemblers other than the GNU
4537 assembler (GAS) to fail with an error.
4539 @item -gdwarf-@var{version}
4540 @opindex gdwarf-@var{version}
4541 Produce debugging information in DWARF format (if that is
4542 supported). This is the format used by DBX on IRIX 6. The value
4543 of @var{version} may be either 2, 3 or 4; the default version is 2.
4545 Note that with DWARF version 2 some ports require, and will always
4546 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4548 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4549 for maximum benefit.
4551 @item -gstrict-dwarf
4552 @opindex gstrict-dwarf
4553 Disallow using extensions of later DWARF standard version than selected
4554 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4555 DWARF extensions from later standard versions is allowed.
4557 @item -gno-strict-dwarf
4558 @opindex gno-strict-dwarf
4559 Allow using extensions of later DWARF standard version than selected with
4560 @option{-gdwarf-@var{version}}.
4564 Produce debugging information in VMS debug format (if that is
4565 supported). This is the format used by DEBUG on VMS systems.
4568 @itemx -ggdb@var{level}
4569 @itemx -gstabs@var{level}
4570 @itemx -gcoff@var{level}
4571 @itemx -gxcoff@var{level}
4572 @itemx -gvms@var{level}
4573 Request debugging information and also use @var{level} to specify how
4574 much information. The default level is 2.
4576 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4579 Level 1 produces minimal information, enough for making backtraces in
4580 parts of the program that you don't plan to debug. This includes
4581 descriptions of functions and external variables, but no information
4582 about local variables and no line numbers.
4584 Level 3 includes extra information, such as all the macro definitions
4585 present in the program. Some debuggers support macro expansion when
4586 you use @option{-g3}.
4588 @option{-gdwarf-2} does not accept a concatenated debug level, because
4589 GCC used to support an option @option{-gdwarf} that meant to generate
4590 debug information in version 1 of the DWARF format (which is very
4591 different from version 2), and it would have been too confusing. That
4592 debug format is long obsolete, but the option cannot be changed now.
4593 Instead use an additional @option{-g@var{level}} option to change the
4594 debug level for DWARF.
4598 Turn off generation of debug info, if leaving out this option would have
4599 generated it, or turn it on at level 2 otherwise. The position of this
4600 argument in the command line does not matter, it takes effect after all
4601 other options are processed, and it does so only once, no matter how
4602 many times it is given. This is mainly intended to be used with
4603 @option{-fcompare-debug}.
4605 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4606 @opindex fdump-final-insns
4607 Dump the final internal representation (RTL) to @var{file}. If the
4608 optional argument is omitted (or if @var{file} is @code{.}), the name
4609 of the dump file will be determined by appending @code{.gkd} to the
4610 compilation output file name.
4612 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4613 @opindex fcompare-debug
4614 @opindex fno-compare-debug
4615 If no error occurs during compilation, run the compiler a second time,
4616 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4617 passed to the second compilation. Dump the final internal
4618 representation in both compilations, and print an error if they differ.
4620 If the equal sign is omitted, the default @option{-gtoggle} is used.
4622 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4623 and nonzero, implicitly enables @option{-fcompare-debug}. If
4624 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4625 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4628 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4629 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4630 of the final representation and the second compilation, preventing even
4631 @env{GCC_COMPARE_DEBUG} from taking effect.
4633 To verify full coverage during @option{-fcompare-debug} testing, set
4634 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4635 which GCC will reject as an invalid option in any actual compilation
4636 (rather than preprocessing, assembly or linking). To get just a
4637 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4638 not overridden} will do.
4640 @item -fcompare-debug-second
4641 @opindex fcompare-debug-second
4642 This option is implicitly passed to the compiler for the second
4643 compilation requested by @option{-fcompare-debug}, along with options to
4644 silence warnings, and omitting other options that would cause
4645 side-effect compiler outputs to files or to the standard output. Dump
4646 files and preserved temporary files are renamed so as to contain the
4647 @code{.gk} additional extension during the second compilation, to avoid
4648 overwriting those generated by the first.
4650 When this option is passed to the compiler driver, it causes the
4651 @emph{first} compilation to be skipped, which makes it useful for little
4652 other than debugging the compiler proper.
4654 @item -feliminate-dwarf2-dups
4655 @opindex feliminate-dwarf2-dups
4656 Compress DWARF2 debugging information by eliminating duplicated
4657 information about each symbol. This option only makes sense when
4658 generating DWARF2 debugging information with @option{-gdwarf-2}.
4660 @item -femit-struct-debug-baseonly
4661 Emit debug information for struct-like types
4662 only when the base name of the compilation source file
4663 matches the base name of file in which the struct was defined.
4665 This option substantially reduces the size of debugging information,
4666 but at significant potential loss in type information to the debugger.
4667 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4668 See @option{-femit-struct-debug-detailed} for more detailed control.
4670 This option works only with DWARF 2.
4672 @item -femit-struct-debug-reduced
4673 Emit debug information for struct-like types
4674 only when the base name of the compilation source file
4675 matches the base name of file in which the type was defined,
4676 unless the struct is a template or defined in a system header.
4678 This option significantly reduces the size of debugging information,
4679 with some potential loss in type information to the debugger.
4680 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4681 See @option{-femit-struct-debug-detailed} for more detailed control.
4683 This option works only with DWARF 2.
4685 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4686 Specify the struct-like types
4687 for which the compiler will generate debug information.
4688 The intent is to reduce duplicate struct debug information
4689 between different object files within the same program.
4691 This option is a detailed version of
4692 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4693 which will serve for most needs.
4695 A specification has the syntax
4696 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4698 The optional first word limits the specification to
4699 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4700 A struct type is used directly when it is the type of a variable, member.
4701 Indirect uses arise through pointers to structs.
4702 That is, when use of an incomplete struct would be legal, the use is indirect.
4704 @samp{struct one direct; struct two * indirect;}.
4706 The optional second word limits the specification to
4707 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4708 Generic structs are a bit complicated to explain.
4709 For C++, these are non-explicit specializations of template classes,
4710 or non-template classes within the above.
4711 Other programming languages have generics,
4712 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4714 The third word specifies the source files for those
4715 structs for which the compiler will emit debug information.
4716 The values @samp{none} and @samp{any} have the normal meaning.
4717 The value @samp{base} means that
4718 the base of name of the file in which the type declaration appears
4719 must match the base of the name of the main compilation file.
4720 In practice, this means that
4721 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4722 but types declared in other header will not.
4723 The value @samp{sys} means those types satisfying @samp{base}
4724 or declared in system or compiler headers.
4726 You may need to experiment to determine the best settings for your application.
4728 The default is @samp{-femit-struct-debug-detailed=all}.
4730 This option works only with DWARF 2.
4732 @item -fenable-icf-debug
4733 @opindex fenable-icf-debug
4734 Generate additional debug information to support identical code folding (ICF).
4735 This option only works with DWARF version 2 or higher.
4737 @item -fno-merge-debug-strings
4738 @opindex fmerge-debug-strings
4739 @opindex fno-merge-debug-strings
4740 Direct the linker to not merge together strings in the debugging
4741 information which are identical in different object files. Merging is
4742 not supported by all assemblers or linkers. Merging decreases the size
4743 of the debug information in the output file at the cost of increasing
4744 link processing time. Merging is enabled by default.
4746 @item -fdebug-prefix-map=@var{old}=@var{new}
4747 @opindex fdebug-prefix-map
4748 When compiling files in directory @file{@var{old}}, record debugging
4749 information describing them as in @file{@var{new}} instead.
4751 @item -fno-dwarf2-cfi-asm
4752 @opindex fdwarf2-cfi-asm
4753 @opindex fno-dwarf2-cfi-asm
4754 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4755 instead of using GAS @code{.cfi_*} directives.
4757 @cindex @command{prof}
4760 Generate extra code to write profile information suitable for the
4761 analysis program @command{prof}. You must use this option when compiling
4762 the source files you want data about, and you must also use it when
4765 @cindex @command{gprof}
4768 Generate extra code to write profile information suitable for the
4769 analysis program @command{gprof}. You must use this option when compiling
4770 the source files you want data about, and you must also use it when
4775 Makes the compiler print out each function name as it is compiled, and
4776 print some statistics about each pass when it finishes.
4779 @opindex ftime-report
4780 Makes the compiler print some statistics about the time consumed by each
4781 pass when it finishes.
4784 @opindex fmem-report
4785 Makes the compiler print some statistics about permanent memory
4786 allocation when it finishes.
4788 @item -fpre-ipa-mem-report
4789 @opindex fpre-ipa-mem-report
4790 @item -fpost-ipa-mem-report
4791 @opindex fpost-ipa-mem-report
4792 Makes the compiler print some statistics about permanent memory
4793 allocation before or after interprocedural optimization.
4795 @item -fprofile-arcs
4796 @opindex fprofile-arcs
4797 Add code so that program flow @dfn{arcs} are instrumented. During
4798 execution the program records how many times each branch and call is
4799 executed and how many times it is taken or returns. When the compiled
4800 program exits it saves this data to a file called
4801 @file{@var{auxname}.gcda} for each source file. The data may be used for
4802 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4803 test coverage analysis (@option{-ftest-coverage}). Each object file's
4804 @var{auxname} is generated from the name of the output file, if
4805 explicitly specified and it is not the final executable, otherwise it is
4806 the basename of the source file. In both cases any suffix is removed
4807 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4808 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4809 @xref{Cross-profiling}.
4811 @cindex @command{gcov}
4815 This option is used to compile and link code instrumented for coverage
4816 analysis. The option is a synonym for @option{-fprofile-arcs}
4817 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4818 linking). See the documentation for those options for more details.
4823 Compile the source files with @option{-fprofile-arcs} plus optimization
4824 and code generation options. For test coverage analysis, use the
4825 additional @option{-ftest-coverage} option. You do not need to profile
4826 every source file in a program.
4829 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4830 (the latter implies the former).
4833 Run the program on a representative workload to generate the arc profile
4834 information. This may be repeated any number of times. You can run
4835 concurrent instances of your program, and provided that the file system
4836 supports locking, the data files will be correctly updated. Also
4837 @code{fork} calls are detected and correctly handled (double counting
4841 For profile-directed optimizations, compile the source files again with
4842 the same optimization and code generation options plus
4843 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4844 Control Optimization}).
4847 For test coverage analysis, use @command{gcov} to produce human readable
4848 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4849 @command{gcov} documentation for further information.
4853 With @option{-fprofile-arcs}, for each function of your program GCC
4854 creates a program flow graph, then finds a spanning tree for the graph.
4855 Only arcs that are not on the spanning tree have to be instrumented: the
4856 compiler adds code to count the number of times that these arcs are
4857 executed. When an arc is the only exit or only entrance to a block, the
4858 instrumentation code can be added to the block; otherwise, a new basic
4859 block must be created to hold the instrumentation code.
4862 @item -ftest-coverage
4863 @opindex ftest-coverage
4864 Produce a notes file that the @command{gcov} code-coverage utility
4865 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4866 show program coverage. Each source file's note file is called
4867 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4868 above for a description of @var{auxname} and instructions on how to
4869 generate test coverage data. Coverage data will match the source files
4870 more closely, if you do not optimize.
4872 @item -fdbg-cnt-list
4873 @opindex fdbg-cnt-list
4874 Print the name and the counter upperbound for all debug counters.
4876 @item -fdbg-cnt=@var{counter-value-list}
4878 Set the internal debug counter upperbound. @var{counter-value-list}
4879 is a comma-separated list of @var{name}:@var{value} pairs
4880 which sets the upperbound of each debug counter @var{name} to @var{value}.
4881 All debug counters have the initial upperbound of @var{UINT_MAX},
4882 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4883 e.g. With -fdbg-cnt=dce:10,tail_call:0
4884 dbg_cnt(dce) will return true only for first 10 invocations
4885 and dbg_cnt(tail_call) will return false always.
4887 @item -d@var{letters}
4888 @itemx -fdump-rtl-@var{pass}
4890 Says to make debugging dumps during compilation at times specified by
4891 @var{letters}. This is used for debugging the RTL-based passes of the
4892 compiler. The file names for most of the dumps are made by appending
4893 a pass number and a word to the @var{dumpname}, and the files are
4894 created in the directory of the output file. @var{dumpname} is
4895 generated from the name of the output file, if explicitly specified
4896 and it is not an executable, otherwise it is the basename of the
4897 source file. These switches may have different effects when
4898 @option{-E} is used for preprocessing.
4900 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4901 @option{-d} option @var{letters}. Here are the possible
4902 letters for use in @var{pass} and @var{letters}, and their meanings:
4906 @item -fdump-rtl-alignments
4907 @opindex fdump-rtl-alignments
4908 Dump after branch alignments have been computed.
4910 @item -fdump-rtl-asmcons
4911 @opindex fdump-rtl-asmcons
4912 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4914 @item -fdump-rtl-auto_inc_dec
4915 @opindex fdump-rtl-auto_inc_dec
4916 Dump after auto-inc-dec discovery. This pass is only run on
4917 architectures that have auto inc or auto dec instructions.
4919 @item -fdump-rtl-barriers
4920 @opindex fdump-rtl-barriers
4921 Dump after cleaning up the barrier instructions.
4923 @item -fdump-rtl-bbpart
4924 @opindex fdump-rtl-bbpart
4925 Dump after partitioning hot and cold basic blocks.
4927 @item -fdump-rtl-bbro
4928 @opindex fdump-rtl-bbro
4929 Dump after block reordering.
4931 @item -fdump-rtl-btl1
4932 @itemx -fdump-rtl-btl2
4933 @opindex fdump-rtl-btl2
4934 @opindex fdump-rtl-btl2
4935 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4936 after the two branch
4937 target load optimization passes.
4939 @item -fdump-rtl-bypass
4940 @opindex fdump-rtl-bypass
4941 Dump after jump bypassing and control flow optimizations.
4943 @item -fdump-rtl-combine
4944 @opindex fdump-rtl-combine
4945 Dump after the RTL instruction combination pass.
4947 @item -fdump-rtl-compgotos
4948 @opindex fdump-rtl-compgotos
4949 Dump after duplicating the computed gotos.
4951 @item -fdump-rtl-ce1
4952 @itemx -fdump-rtl-ce2
4953 @itemx -fdump-rtl-ce3
4954 @opindex fdump-rtl-ce1
4955 @opindex fdump-rtl-ce2
4956 @opindex fdump-rtl-ce3
4957 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4958 @option{-fdump-rtl-ce3} enable dumping after the three
4959 if conversion passes.
4961 @itemx -fdump-rtl-cprop_hardreg
4962 @opindex fdump-rtl-cprop_hardreg
4963 Dump after hard register copy propagation.
4965 @itemx -fdump-rtl-csa
4966 @opindex fdump-rtl-csa
4967 Dump after combining stack adjustments.
4969 @item -fdump-rtl-cse1
4970 @itemx -fdump-rtl-cse2
4971 @opindex fdump-rtl-cse1
4972 @opindex fdump-rtl-cse2
4973 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4974 the two common sub-expression elimination passes.
4976 @itemx -fdump-rtl-dce
4977 @opindex fdump-rtl-dce
4978 Dump after the standalone dead code elimination passes.
4980 @itemx -fdump-rtl-dbr
4981 @opindex fdump-rtl-dbr
4982 Dump after delayed branch scheduling.
4984 @item -fdump-rtl-dce1
4985 @itemx -fdump-rtl-dce2
4986 @opindex fdump-rtl-dce1
4987 @opindex fdump-rtl-dce2
4988 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4989 the two dead store elimination passes.
4992 @opindex fdump-rtl-eh
4993 Dump after finalization of EH handling code.
4995 @item -fdump-rtl-eh_ranges
4996 @opindex fdump-rtl-eh_ranges
4997 Dump after conversion of EH handling range regions.
4999 @item -fdump-rtl-expand
5000 @opindex fdump-rtl-expand
5001 Dump after RTL generation.
5003 @item -fdump-rtl-fwprop1
5004 @itemx -fdump-rtl-fwprop2
5005 @opindex fdump-rtl-fwprop1
5006 @opindex fdump-rtl-fwprop2
5007 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
5008 dumping after the two forward propagation passes.
5010 @item -fdump-rtl-gcse1
5011 @itemx -fdump-rtl-gcse2
5012 @opindex fdump-rtl-gcse1
5013 @opindex fdump-rtl-gcse2
5014 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
5015 after global common subexpression elimination.
5017 @item -fdump-rtl-init-regs
5018 @opindex fdump-rtl-init-regs
5019 Dump after the initialization of the registers.
5021 @item -fdump-rtl-initvals
5022 @opindex fdump-rtl-initvals
5023 Dump after the computation of the initial value sets.
5025 @itemx -fdump-rtl-into_cfglayout
5026 @opindex fdump-rtl-into_cfglayout
5027 Dump after converting to cfglayout mode.
5029 @item -fdump-rtl-ira
5030 @opindex fdump-rtl-ira
5031 Dump after iterated register allocation.
5033 @item -fdump-rtl-jump
5034 @opindex fdump-rtl-jump
5035 Dump after the second jump optimization.
5037 @item -fdump-rtl-loop2
5038 @opindex fdump-rtl-loop2
5039 @option{-fdump-rtl-loop2} enables dumping after the rtl
5040 loop optimization passes.
5042 @item -fdump-rtl-mach
5043 @opindex fdump-rtl-mach
5044 Dump after performing the machine dependent reorganization pass, if that
5047 @item -fdump-rtl-mode_sw
5048 @opindex fdump-rtl-mode_sw
5049 Dump after removing redundant mode switches.
5051 @item -fdump-rtl-rnreg
5052 @opindex fdump-rtl-rnreg
5053 Dump after register renumbering.
5055 @itemx -fdump-rtl-outof_cfglayout
5056 @opindex fdump-rtl-outof_cfglayout
5057 Dump after converting from cfglayout mode.
5059 @item -fdump-rtl-peephole2
5060 @opindex fdump-rtl-peephole2
5061 Dump after the peephole pass.
5063 @item -fdump-rtl-postreload
5064 @opindex fdump-rtl-postreload
5065 Dump after post-reload optimizations.
5067 @itemx -fdump-rtl-pro_and_epilogue
5068 @opindex fdump-rtl-pro_and_epilogue
5069 Dump after generating the function pro and epilogues.
5071 @item -fdump-rtl-regmove
5072 @opindex fdump-rtl-regmove
5073 Dump after the register move pass.
5075 @item -fdump-rtl-sched1
5076 @itemx -fdump-rtl-sched2
5077 @opindex fdump-rtl-sched1
5078 @opindex fdump-rtl-sched2
5079 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
5080 after the basic block scheduling passes.
5082 @item -fdump-rtl-see
5083 @opindex fdump-rtl-see
5084 Dump after sign extension elimination.
5086 @item -fdump-rtl-seqabstr
5087 @opindex fdump-rtl-seqabstr
5088 Dump after common sequence discovery.
5090 @item -fdump-rtl-shorten
5091 @opindex fdump-rtl-shorten
5092 Dump after shortening branches.
5094 @item -fdump-rtl-sibling
5095 @opindex fdump-rtl-sibling
5096 Dump after sibling call optimizations.
5098 @item -fdump-rtl-split1
5099 @itemx -fdump-rtl-split2
5100 @itemx -fdump-rtl-split3
5101 @itemx -fdump-rtl-split4
5102 @itemx -fdump-rtl-split5
5103 @opindex fdump-rtl-split1
5104 @opindex fdump-rtl-split2
5105 @opindex fdump-rtl-split3
5106 @opindex fdump-rtl-split4
5107 @opindex fdump-rtl-split5
5108 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5109 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5110 @option{-fdump-rtl-split5} enable dumping after five rounds of
5111 instruction splitting.
5113 @item -fdump-rtl-sms
5114 @opindex fdump-rtl-sms
5115 Dump after modulo scheduling. This pass is only run on some
5118 @item -fdump-rtl-stack
5119 @opindex fdump-rtl-stack
5120 Dump after conversion from GCC's "flat register file" registers to the
5121 x87's stack-like registers. This pass is only run on x86 variants.
5123 @item -fdump-rtl-subreg1
5124 @itemx -fdump-rtl-subreg2
5125 @opindex fdump-rtl-subreg1
5126 @opindex fdump-rtl-subreg2
5127 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5128 the two subreg expansion passes.
5130 @item -fdump-rtl-unshare
5131 @opindex fdump-rtl-unshare
5132 Dump after all rtl has been unshared.
5134 @item -fdump-rtl-vartrack
5135 @opindex fdump-rtl-vartrack
5136 Dump after variable tracking.
5138 @item -fdump-rtl-vregs
5139 @opindex fdump-rtl-vregs
5140 Dump after converting virtual registers to hard registers.
5142 @item -fdump-rtl-web
5143 @opindex fdump-rtl-web
5144 Dump after live range splitting.
5146 @item -fdump-rtl-regclass
5147 @itemx -fdump-rtl-subregs_of_mode_init
5148 @itemx -fdump-rtl-subregs_of_mode_finish
5149 @itemx -fdump-rtl-dfinit
5150 @itemx -fdump-rtl-dfinish
5151 @opindex fdump-rtl-regclass
5152 @opindex fdump-rtl-subregs_of_mode_init
5153 @opindex fdump-rtl-subregs_of_mode_finish
5154 @opindex fdump-rtl-dfinit
5155 @opindex fdump-rtl-dfinish
5156 These dumps are defined but always produce empty files.
5158 @item -fdump-rtl-all
5159 @opindex fdump-rtl-all
5160 Produce all the dumps listed above.
5164 Annotate the assembler output with miscellaneous debugging information.
5168 Dump all macro definitions, at the end of preprocessing, in addition to
5173 Produce a core dump whenever an error occurs.
5177 Print statistics on memory usage, at the end of the run, to
5182 Annotate the assembler output with a comment indicating which
5183 pattern and alternative was used. The length of each instruction is
5188 Dump the RTL in the assembler output as a comment before each instruction.
5189 Also turns on @option{-dp} annotation.
5193 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5194 dump a representation of the control flow graph suitable for viewing with VCG
5195 to @file{@var{file}.@var{pass}.vcg}.
5199 Just generate RTL for a function instead of compiling it. Usually used
5200 with @option{-fdump-rtl-expand}.
5204 Dump debugging information during parsing, to standard error.
5208 @opindex fdump-noaddr
5209 When doing debugging dumps, suppress address output. This makes it more
5210 feasible to use diff on debugging dumps for compiler invocations with
5211 different compiler binaries and/or different
5212 text / bss / data / heap / stack / dso start locations.
5214 @item -fdump-unnumbered
5215 @opindex fdump-unnumbered
5216 When doing debugging dumps, suppress instruction numbers and address output.
5217 This makes it more feasible to use diff on debugging dumps for compiler
5218 invocations with different options, in particular with and without
5221 @item -fdump-unnumbered-links
5222 @opindex fdump-unnumbered-links
5223 When doing debugging dumps (see @option{-d} option above), suppress
5224 instruction numbers for the links to the previous and next instructions
5227 @item -fdump-translation-unit @r{(C++ only)}
5228 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5229 @opindex fdump-translation-unit
5230 Dump a representation of the tree structure for the entire translation
5231 unit to a file. The file name is made by appending @file{.tu} to the
5232 source file name, and the file is created in the same directory as the
5233 output file. If the @samp{-@var{options}} form is used, @var{options}
5234 controls the details of the dump as described for the
5235 @option{-fdump-tree} options.
5237 @item -fdump-class-hierarchy @r{(C++ only)}
5238 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5239 @opindex fdump-class-hierarchy
5240 Dump a representation of each class's hierarchy and virtual function
5241 table layout to a file. The file name is made by appending
5242 @file{.class} to the source file name, and the file is created in the
5243 same directory as the output file. If the @samp{-@var{options}} form
5244 is used, @var{options} controls the details of the dump as described
5245 for the @option{-fdump-tree} options.
5247 @item -fdump-ipa-@var{switch}
5249 Control the dumping at various stages of inter-procedural analysis
5250 language tree to a file. The file name is generated by appending a
5251 switch specific suffix to the source file name, and the file is created
5252 in the same directory as the output file. The following dumps are
5257 Enables all inter-procedural analysis dumps.
5260 Dumps information about call-graph optimization, unused function removal,
5261 and inlining decisions.
5264 Dump after function inlining.
5268 @item -fdump-statistics-@var{option}
5269 @opindex fdump-statistics
5270 Enable and control dumping of pass statistics in a separate file. The
5271 file name is generated by appending a suffix ending in
5272 @samp{.statistics} to the source file name, and the file is created in
5273 the same directory as the output file. If the @samp{-@var{option}}
5274 form is used, @samp{-stats} will cause counters to be summed over the
5275 whole compilation unit while @samp{-details} will dump every event as
5276 the passes generate them. The default with no option is to sum
5277 counters for each function compiled.
5279 @item -fdump-tree-@var{switch}
5280 @itemx -fdump-tree-@var{switch}-@var{options}
5282 Control the dumping at various stages of processing the intermediate
5283 language tree to a file. The file name is generated by appending a
5284 switch specific suffix to the source file name, and the file is
5285 created in the same directory as the output file. If the
5286 @samp{-@var{options}} form is used, @var{options} is a list of
5287 @samp{-} separated options that control the details of the dump. Not
5288 all options are applicable to all dumps, those which are not
5289 meaningful will be ignored. The following options are available
5293 Print the address of each node. Usually this is not meaningful as it
5294 changes according to the environment and source file. Its primary use
5295 is for tying up a dump file with a debug environment.
5297 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5298 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5299 use working backward from mangled names in the assembly file.
5301 Inhibit dumping of members of a scope or body of a function merely
5302 because that scope has been reached. Only dump such items when they
5303 are directly reachable by some other path. When dumping pretty-printed
5304 trees, this option inhibits dumping the bodies of control structures.
5306 Print a raw representation of the tree. By default, trees are
5307 pretty-printed into a C-like representation.
5309 Enable more detailed dumps (not honored by every dump option).
5311 Enable dumping various statistics about the pass (not honored by every dump
5314 Enable showing basic block boundaries (disabled in raw dumps).
5316 Enable showing virtual operands for every statement.
5318 Enable showing line numbers for statements.
5320 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5322 Enable showing the tree dump for each statement.
5324 Enable showing the EH region number holding each statement.
5326 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5327 and @option{lineno}.
5330 The following tree dumps are possible:
5334 @opindex fdump-tree-original
5335 Dump before any tree based optimization, to @file{@var{file}.original}.
5338 @opindex fdump-tree-optimized
5339 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5342 @opindex fdump-tree-gimple
5343 Dump each function before and after the gimplification pass to a file. The
5344 file name is made by appending @file{.gimple} to the source file name.
5347 @opindex fdump-tree-cfg
5348 Dump the control flow graph of each function to a file. The file name is
5349 made by appending @file{.cfg} to the source file name.
5352 @opindex fdump-tree-vcg
5353 Dump the control flow graph of each function to a file in VCG format. The
5354 file name is made by appending @file{.vcg} to the source file name. Note
5355 that if the file contains more than one function, the generated file cannot
5356 be used directly by VCG@. You will need to cut and paste each function's
5357 graph into its own separate file first.
5360 @opindex fdump-tree-ch
5361 Dump each function after copying loop headers. The file name is made by
5362 appending @file{.ch} to the source file name.
5365 @opindex fdump-tree-ssa
5366 Dump SSA related information to a file. The file name is made by appending
5367 @file{.ssa} to the source file name.
5370 @opindex fdump-tree-alias
5371 Dump aliasing information for each function. The file name is made by
5372 appending @file{.alias} to the source file name.
5375 @opindex fdump-tree-ccp
5376 Dump each function after CCP@. The file name is made by appending
5377 @file{.ccp} to the source file name.
5380 @opindex fdump-tree-storeccp
5381 Dump each function after STORE-CCP@. The file name is made by appending
5382 @file{.storeccp} to the source file name.
5385 @opindex fdump-tree-pre
5386 Dump trees after partial redundancy elimination. The file name is made
5387 by appending @file{.pre} to the source file name.
5390 @opindex fdump-tree-fre
5391 Dump trees after full redundancy elimination. The file name is made
5392 by appending @file{.fre} to the source file name.
5395 @opindex fdump-tree-copyprop
5396 Dump trees after copy propagation. The file name is made
5397 by appending @file{.copyprop} to the source file name.
5399 @item store_copyprop
5400 @opindex fdump-tree-store_copyprop
5401 Dump trees after store copy-propagation. The file name is made
5402 by appending @file{.store_copyprop} to the source file name.
5405 @opindex fdump-tree-dce
5406 Dump each function after dead code elimination. The file name is made by
5407 appending @file{.dce} to the source file name.
5410 @opindex fdump-tree-mudflap
5411 Dump each function after adding mudflap instrumentation. The file name is
5412 made by appending @file{.mudflap} to the source file name.
5415 @opindex fdump-tree-sra
5416 Dump each function after performing scalar replacement of aggregates. The
5417 file name is made by appending @file{.sra} to the source file name.
5420 @opindex fdump-tree-sink
5421 Dump each function after performing code sinking. The file name is made
5422 by appending @file{.sink} to the source file name.
5425 @opindex fdump-tree-dom
5426 Dump each function after applying dominator tree optimizations. The file
5427 name is made by appending @file{.dom} to the source file name.
5430 @opindex fdump-tree-dse
5431 Dump each function after applying dead store elimination. The file
5432 name is made by appending @file{.dse} to the source file name.
5435 @opindex fdump-tree-phiopt
5436 Dump each function after optimizing PHI nodes into straightline code. The file
5437 name is made by appending @file{.phiopt} to the source file name.
5440 @opindex fdump-tree-forwprop
5441 Dump each function after forward propagating single use variables. The file
5442 name is made by appending @file{.forwprop} to the source file name.
5445 @opindex fdump-tree-copyrename
5446 Dump each function after applying the copy rename optimization. The file
5447 name is made by appending @file{.copyrename} to the source file name.
5450 @opindex fdump-tree-nrv
5451 Dump each function after applying the named return value optimization on
5452 generic trees. The file name is made by appending @file{.nrv} to the source
5456 @opindex fdump-tree-vect
5457 Dump each function after applying vectorization of loops. The file name is
5458 made by appending @file{.vect} to the source file name.
5461 @opindex fdump-tree-slp
5462 Dump each function after applying vectorization of basic blocks. The file name
5463 is made by appending @file{.slp} to the source file name.
5466 @opindex fdump-tree-vrp
5467 Dump each function after Value Range Propagation (VRP). The file name
5468 is made by appending @file{.vrp} to the source file name.
5471 @opindex fdump-tree-all
5472 Enable all the available tree dumps with the flags provided in this option.
5475 @item -ftree-vectorizer-verbose=@var{n}
5476 @opindex ftree-vectorizer-verbose
5477 This option controls the amount of debugging output the vectorizer prints.
5478 This information is written to standard error, unless
5479 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5480 in which case it is output to the usual dump listing file, @file{.vect}.
5481 For @var{n}=0 no diagnostic information is reported.
5482 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5483 and the total number of loops that got vectorized.
5484 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5485 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5486 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5487 level that @option{-fdump-tree-vect-stats} uses.
5488 Higher verbosity levels mean either more information dumped for each
5489 reported loop, or same amount of information reported for more loops:
5490 if @var{n}=3, vectorizer cost model information is reported.
5491 If @var{n}=4, alignment related information is added to the reports.
5492 If @var{n}=5, data-references related information (e.g.@: memory dependences,
5493 memory access-patterns) is added to the reports.
5494 If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5495 that did not pass the first analysis phase (i.e., may not be countable, or
5496 may have complicated control-flow).
5497 If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5498 If @var{n}=8, SLP related information is added to the reports.
5499 For @var{n}=9, all the information the vectorizer generates during its
5500 analysis and transformation is reported. This is the same verbosity level
5501 that @option{-fdump-tree-vect-details} uses.
5503 @item -frandom-seed=@var{string}
5504 @opindex frandom-seed
5505 This option provides a seed that GCC uses when it would otherwise use
5506 random numbers. It is used to generate certain symbol names
5507 that have to be different in every compiled file. It is also used to
5508 place unique stamps in coverage data files and the object files that
5509 produce them. You can use the @option{-frandom-seed} option to produce
5510 reproducibly identical object files.
5512 The @var{string} should be different for every file you compile.
5514 @item -fsched-verbose=@var{n}
5515 @opindex fsched-verbose
5516 On targets that use instruction scheduling, this option controls the
5517 amount of debugging output the scheduler prints. This information is
5518 written to standard error, unless @option{-fdump-rtl-sched1} or
5519 @option{-fdump-rtl-sched2} is specified, in which case it is output
5520 to the usual dump listing file, @file{.sched1} or @file{.sched2}
5521 respectively. However for @var{n} greater than nine, the output is
5522 always printed to standard error.
5524 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5525 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5526 For @var{n} greater than one, it also output basic block probabilities,
5527 detailed ready list information and unit/insn info. For @var{n} greater
5528 than two, it includes RTL at abort point, control-flow and regions info.
5529 And for @var{n} over four, @option{-fsched-verbose} also includes
5533 @itemx -save-temps=cwd
5535 Store the usual ``temporary'' intermediate files permanently; place them
5536 in the current directory and name them based on the source file. Thus,
5537 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5538 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5539 preprocessed @file{foo.i} output file even though the compiler now
5540 normally uses an integrated preprocessor.
5542 When used in combination with the @option{-x} command line option,
5543 @option{-save-temps} is sensible enough to avoid over writing an
5544 input source file with the same extension as an intermediate file.
5545 The corresponding intermediate file may be obtained by renaming the
5546 source file before using @option{-save-temps}.
5548 If you invoke GCC in parallel, compiling several different source
5549 files that share a common base name in different subdirectories or the
5550 same source file compiled for multiple output destinations, it is
5551 likely that the different parallel compilers will interfere with each
5552 other, and overwrite the temporary files. For instance:
5555 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5556 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5559 may result in @file{foo.i} and @file{foo.o} being written to
5560 simultaneously by both compilers.
5562 @item -save-temps=obj
5563 @opindex save-temps=obj
5564 Store the usual ``temporary'' intermediate files permanently. If the
5565 @option{-o} option is used, the temporary files are based on the
5566 object file. If the @option{-o} option is not used, the
5567 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5572 gcc -save-temps=obj -c foo.c
5573 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5574 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5577 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5578 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5579 @file{dir2/yfoobar.o}.
5581 @item -time@r{[}=@var{file}@r{]}
5583 Report the CPU time taken by each subprocess in the compilation
5584 sequence. For C source files, this is the compiler proper and assembler
5585 (plus the linker if linking is done).
5587 Without the specification of an output file, the output looks like this:
5594 The first number on each line is the ``user time'', that is time spent
5595 executing the program itself. The second number is ``system time'',
5596 time spent executing operating system routines on behalf of the program.
5597 Both numbers are in seconds.
5599 With the specification of an output file, the output is appended to the
5600 named file, and it looks like this:
5603 0.12 0.01 cc1 @var{options}
5604 0.00 0.01 as @var{options}
5607 The ``user time'' and the ``system time'' are moved before the program
5608 name, and the options passed to the program are displayed, so that one
5609 can later tell what file was being compiled, and with which options.
5611 @item -fvar-tracking
5612 @opindex fvar-tracking
5613 Run variable tracking pass. It computes where variables are stored at each
5614 position in code. Better debugging information is then generated
5615 (if the debugging information format supports this information).
5617 It is enabled by default when compiling with optimization (@option{-Os},
5618 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5619 the debug info format supports it.
5621 @item -fvar-tracking-assignments
5622 @opindex fvar-tracking-assignments
5623 @opindex fno-var-tracking-assignments
5624 Annotate assignments to user variables early in the compilation and
5625 attempt to carry the annotations over throughout the compilation all the
5626 way to the end, in an attempt to improve debug information while
5627 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5629 It can be enabled even if var-tracking is disabled, in which case
5630 annotations will be created and maintained, but discarded at the end.
5632 @item -fvar-tracking-assignments-toggle
5633 @opindex fvar-tracking-assignments-toggle
5634 @opindex fno-var-tracking-assignments-toggle
5635 Toggle @option{-fvar-tracking-assignments}, in the same way that
5636 @option{-gtoggle} toggles @option{-g}.
5638 @item -print-file-name=@var{library}
5639 @opindex print-file-name
5640 Print the full absolute name of the library file @var{library} that
5641 would be used when linking---and don't do anything else. With this
5642 option, GCC does not compile or link anything; it just prints the
5645 @item -print-multi-directory
5646 @opindex print-multi-directory
5647 Print the directory name corresponding to the multilib selected by any
5648 other switches present in the command line. This directory is supposed
5649 to exist in @env{GCC_EXEC_PREFIX}.
5651 @item -print-multi-lib
5652 @opindex print-multi-lib
5653 Print the mapping from multilib directory names to compiler switches
5654 that enable them. The directory name is separated from the switches by
5655 @samp{;}, and each switch starts with an @samp{@@} instead of the
5656 @samp{-}, without spaces between multiple switches. This is supposed to
5657 ease shell-processing.
5659 @item -print-multi-os-directory
5660 @opindex print-multi-os-directory
5661 Print the path to OS libraries for the selected
5662 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5663 present in the @file{lib} subdirectory and no multilibs are used, this is
5664 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5665 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5666 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5667 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5669 @item -print-prog-name=@var{program}
5670 @opindex print-prog-name
5671 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5673 @item -print-libgcc-file-name
5674 @opindex print-libgcc-file-name
5675 Same as @option{-print-file-name=libgcc.a}.
5677 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5678 but you do want to link with @file{libgcc.a}. You can do
5681 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5684 @item -print-search-dirs
5685 @opindex print-search-dirs
5686 Print the name of the configured installation directory and a list of
5687 program and library directories @command{gcc} will search---and don't do anything else.
5689 This is useful when @command{gcc} prints the error message
5690 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5691 To resolve this you either need to put @file{cpp0} and the other compiler
5692 components where @command{gcc} expects to find them, or you can set the environment
5693 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5694 Don't forget the trailing @samp{/}.
5695 @xref{Environment Variables}.
5697 @item -print-sysroot
5698 @opindex print-sysroot
5699 Print the target sysroot directory that will be used during
5700 compilation. This is the target sysroot specified either at configure
5701 time or using the @option{--sysroot} option, possibly with an extra
5702 suffix that depends on compilation options. If no target sysroot is
5703 specified, the option prints nothing.
5705 @item -print-sysroot-headers-suffix
5706 @opindex print-sysroot-headers-suffix
5707 Print the suffix added to the target sysroot when searching for
5708 headers, or give an error if the compiler is not configured with such
5709 a suffix---and don't do anything else.
5712 @opindex dumpmachine
5713 Print the compiler's target machine (for example,
5714 @samp{i686-pc-linux-gnu})---and don't do anything else.
5717 @opindex dumpversion
5718 Print the compiler version (for example, @samp{3.0})---and don't do
5723 Print the compiler's built-in specs---and don't do anything else. (This
5724 is used when GCC itself is being built.) @xref{Spec Files}.
5726 @item -feliminate-unused-debug-types
5727 @opindex feliminate-unused-debug-types
5728 Normally, when producing DWARF2 output, GCC will emit debugging
5729 information for all types declared in a compilation
5730 unit, regardless of whether or not they are actually used
5731 in that compilation unit. Sometimes this is useful, such as
5732 if, in the debugger, you want to cast a value to a type that is
5733 not actually used in your program (but is declared). More often,
5734 however, this results in a significant amount of wasted space.
5735 With this option, GCC will avoid producing debug symbol output
5736 for types that are nowhere used in the source file being compiled.
5739 @node Optimize Options
5740 @section Options That Control Optimization
5741 @cindex optimize options
5742 @cindex options, optimization
5744 These options control various sorts of optimizations.
5746 Without any optimization option, the compiler's goal is to reduce the
5747 cost of compilation and to make debugging produce the expected
5748 results. Statements are independent: if you stop the program with a
5749 breakpoint between statements, you can then assign a new value to any
5750 variable or change the program counter to any other statement in the
5751 function and get exactly the results you would expect from the source
5754 Turning on optimization flags makes the compiler attempt to improve
5755 the performance and/or code size at the expense of compilation time
5756 and possibly the ability to debug the program.
5758 The compiler performs optimization based on the knowledge it has of the
5759 program. Compiling multiple files at once to a single output file mode allows
5760 the compiler to use information gained from all of the files when compiling
5763 Not all optimizations are controlled directly by a flag. Only
5764 optimizations that have a flag are listed in this section.
5766 Most optimizations are only enabled if an @option{-O} level is set on
5767 the command line. Otherwise they are disabled, even if individual
5768 optimization flags are specified.
5770 Depending on the target and how GCC was configured, a slightly different
5771 set of optimizations may be enabled at each @option{-O} level than
5772 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5773 to find out the exact set of optimizations that are enabled at each level.
5774 @xref{Overall Options}, for examples.
5781 Optimize. Optimizing compilation takes somewhat more time, and a lot
5782 more memory for a large function.
5784 With @option{-O}, the compiler tries to reduce code size and execution
5785 time, without performing any optimizations that take a great deal of
5788 @option{-O} turns on the following optimization flags:
5791 -fcprop-registers @gol
5794 -fdelayed-branch @gol
5796 -fguess-branch-probability @gol
5797 -fif-conversion2 @gol
5798 -fif-conversion @gol
5799 -fipa-pure-const @gol
5801 -fipa-reference @gol
5803 -fsplit-wide-types @gol
5804 -ftree-builtin-call-dce @gol
5807 -ftree-copyrename @gol
5809 -ftree-dominator-opts @gol
5811 -ftree-forwprop @gol
5819 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5820 where doing so does not interfere with debugging.
5824 Optimize even more. GCC performs nearly all supported optimizations
5825 that do not involve a space-speed tradeoff.
5826 As compared to @option{-O}, this option increases both compilation time
5827 and the performance of the generated code.
5829 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5830 also turns on the following optimization flags:
5831 @gccoptlist{-fthread-jumps @gol
5832 -falign-functions -falign-jumps @gol
5833 -falign-loops -falign-labels @gol
5836 -fcse-follow-jumps -fcse-skip-blocks @gol
5837 -fdelete-null-pointer-checks @gol
5838 -fexpensive-optimizations @gol
5839 -fgcse -fgcse-lm @gol
5840 -finline-small-functions @gol
5841 -findirect-inlining @gol
5843 -foptimize-sibling-calls @gol
5846 -freorder-blocks -freorder-functions @gol
5847 -frerun-cse-after-loop @gol
5848 -fsched-interblock -fsched-spec @gol
5849 -fschedule-insns -fschedule-insns2 @gol
5850 -fstrict-aliasing -fstrict-overflow @gol
5851 -ftree-switch-conversion @gol
5855 Please note the warning under @option{-fgcse} about
5856 invoking @option{-O2} on programs that use computed gotos.
5860 Optimize yet more. @option{-O3} turns on all optimizations specified
5861 by @option{-O2} and also turns on the @option{-finline-functions},
5862 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5863 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5867 Reduce compilation time and make debugging produce the expected
5868 results. This is the default.
5872 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5873 do not typically increase code size. It also performs further
5874 optimizations designed to reduce code size.
5876 @option{-Os} disables the following optimization flags:
5877 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5878 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5879 -fprefetch-loop-arrays -ftree-vect-loop-version}
5881 If you use multiple @option{-O} options, with or without level numbers,
5882 the last such option is the one that is effective.
5885 Options of the form @option{-f@var{flag}} specify machine-independent
5886 flags. Most flags have both positive and negative forms; the negative
5887 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5888 below, only one of the forms is listed---the one you typically will
5889 use. You can figure out the other form by either removing @samp{no-}
5892 The following options control specific optimizations. They are either
5893 activated by @option{-O} options or are related to ones that are. You
5894 can use the following flags in the rare cases when ``fine-tuning'' of
5895 optimizations to be performed is desired.
5898 @item -fno-default-inline
5899 @opindex fno-default-inline
5900 Do not make member functions inline by default merely because they are
5901 defined inside the class scope (C++ only). Otherwise, when you specify
5902 @w{@option{-O}}, member functions defined inside class scope are compiled
5903 inline by default; i.e., you don't need to add @samp{inline} in front of
5904 the member function name.
5906 @item -fno-defer-pop
5907 @opindex fno-defer-pop
5908 Always pop the arguments to each function call as soon as that function
5909 returns. For machines which must pop arguments after a function call,
5910 the compiler normally lets arguments accumulate on the stack for several
5911 function calls and pops them all at once.
5913 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5915 @item -fforward-propagate
5916 @opindex fforward-propagate
5917 Perform a forward propagation pass on RTL@. The pass tries to combine two
5918 instructions and checks if the result can be simplified. If loop unrolling
5919 is active, two passes are performed and the second is scheduled after
5922 This option is enabled by default at optimization levels @option{-O},
5923 @option{-O2}, @option{-O3}, @option{-Os}.
5925 @item -fomit-frame-pointer
5926 @opindex fomit-frame-pointer
5927 Don't keep the frame pointer in a register for functions that
5928 don't need one. This avoids the instructions to save, set up and
5929 restore frame pointers; it also makes an extra register available
5930 in many functions. @strong{It also makes debugging impossible on
5933 On some machines, such as the VAX, this flag has no effect, because
5934 the standard calling sequence automatically handles the frame pointer
5935 and nothing is saved by pretending it doesn't exist. The
5936 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5937 whether a target machine supports this flag. @xref{Registers,,Register
5938 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5940 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5942 @item -foptimize-sibling-calls
5943 @opindex foptimize-sibling-calls
5944 Optimize sibling and tail recursive calls.
5946 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5950 Don't pay attention to the @code{inline} keyword. Normally this option
5951 is used to keep the compiler from expanding any functions inline.
5952 Note that if you are not optimizing, no functions can be expanded inline.
5954 @item -finline-small-functions
5955 @opindex finline-small-functions
5956 Integrate functions into their callers when their body is smaller than expected
5957 function call code (so overall size of program gets smaller). The compiler
5958 heuristically decides which functions are simple enough to be worth integrating
5961 Enabled at level @option{-O2}.
5963 @item -findirect-inlining
5964 @opindex findirect-inlining
5965 Inline also indirect calls that are discovered to be known at compile
5966 time thanks to previous inlining. This option has any effect only
5967 when inlining itself is turned on by the @option{-finline-functions}
5968 or @option{-finline-small-functions} options.
5970 Enabled at level @option{-O2}.
5972 @item -finline-functions
5973 @opindex finline-functions
5974 Integrate all simple functions into their callers. The compiler
5975 heuristically decides which functions are simple enough to be worth
5976 integrating in this way.
5978 If all calls to a given function are integrated, and the function is
5979 declared @code{static}, then the function is normally not output as
5980 assembler code in its own right.
5982 Enabled at level @option{-O3}.
5984 @item -finline-functions-called-once
5985 @opindex finline-functions-called-once
5986 Consider all @code{static} functions called once for inlining into their
5987 caller even if they are not marked @code{inline}. If a call to a given
5988 function is integrated, then the function is not output as assembler code
5991 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5993 @item -fearly-inlining
5994 @opindex fearly-inlining
5995 Inline functions marked by @code{always_inline} and functions whose body seems
5996 smaller than the function call overhead early before doing
5997 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5998 makes profiling significantly cheaper and usually inlining faster on programs
5999 having large chains of nested wrapper functions.
6005 Perform interprocedural scalar replacement of aggregates, removal of
6006 unused parameters and replacement of parameters passed by reference
6007 by parameters passed by value.
6009 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6011 @item -finline-limit=@var{n}
6012 @opindex finline-limit
6013 By default, GCC limits the size of functions that can be inlined. This flag
6014 allows coarse control of this limit. @var{n} is the size of functions that
6015 can be inlined in number of pseudo instructions.
6017 Inlining is actually controlled by a number of parameters, which may be
6018 specified individually by using @option{--param @var{name}=@var{value}}.
6019 The @option{-finline-limit=@var{n}} option sets some of these parameters
6023 @item max-inline-insns-single
6024 is set to @var{n}/2.
6025 @item max-inline-insns-auto
6026 is set to @var{n}/2.
6029 See below for a documentation of the individual
6030 parameters controlling inlining and for the defaults of these parameters.
6032 @emph{Note:} there may be no value to @option{-finline-limit} that results
6033 in default behavior.
6035 @emph{Note:} pseudo instruction represents, in this particular context, an
6036 abstract measurement of function's size. In no way does it represent a count
6037 of assembly instructions and as such its exact meaning might change from one
6038 release to an another.
6040 @item -fkeep-inline-functions
6041 @opindex fkeep-inline-functions
6042 In C, emit @code{static} functions that are declared @code{inline}
6043 into the object file, even if the function has been inlined into all
6044 of its callers. This switch does not affect functions using the
6045 @code{extern inline} extension in GNU C90@. In C++, emit any and all
6046 inline functions into the object file.
6048 @item -fkeep-static-consts
6049 @opindex fkeep-static-consts
6050 Emit variables declared @code{static const} when optimization isn't turned
6051 on, even if the variables aren't referenced.
6053 GCC enables this option by default. If you want to force the compiler to
6054 check if the variable was referenced, regardless of whether or not
6055 optimization is turned on, use the @option{-fno-keep-static-consts} option.
6057 @item -fmerge-constants
6058 @opindex fmerge-constants
6059 Attempt to merge identical constants (string constants and floating point
6060 constants) across compilation units.
6062 This option is the default for optimized compilation if the assembler and
6063 linker support it. Use @option{-fno-merge-constants} to inhibit this
6066 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6068 @item -fmerge-all-constants
6069 @opindex fmerge-all-constants
6070 Attempt to merge identical constants and identical variables.
6072 This option implies @option{-fmerge-constants}. In addition to
6073 @option{-fmerge-constants} this considers e.g.@: even constant initialized
6074 arrays or initialized constant variables with integral or floating point
6075 types. Languages like C or C++ require each variable, including multiple
6076 instances of the same variable in recursive calls, to have distinct locations,
6077 so using this option will result in non-conforming
6080 @item -fmodulo-sched
6081 @opindex fmodulo-sched
6082 Perform swing modulo scheduling immediately before the first scheduling
6083 pass. This pass looks at innermost loops and reorders their
6084 instructions by overlapping different iterations.
6086 @item -fmodulo-sched-allow-regmoves
6087 @opindex fmodulo-sched-allow-regmoves
6088 Perform more aggressive SMS based modulo scheduling with register moves
6089 allowed. By setting this flag certain anti-dependences edges will be
6090 deleted which will trigger the generation of reg-moves based on the
6091 life-range analysis. This option is effective only with
6092 @option{-fmodulo-sched} enabled.
6094 @item -fno-branch-count-reg
6095 @opindex fno-branch-count-reg
6096 Do not use ``decrement and branch'' instructions on a count register,
6097 but instead generate a sequence of instructions that decrement a
6098 register, compare it against zero, then branch based upon the result.
6099 This option is only meaningful on architectures that support such
6100 instructions, which include x86, PowerPC, IA-64 and S/390.
6102 The default is @option{-fbranch-count-reg}.
6104 @item -fno-function-cse
6105 @opindex fno-function-cse
6106 Do not put function addresses in registers; make each instruction that
6107 calls a constant function contain the function's address explicitly.
6109 This option results in less efficient code, but some strange hacks
6110 that alter the assembler output may be confused by the optimizations
6111 performed when this option is not used.
6113 The default is @option{-ffunction-cse}
6115 @item -fno-zero-initialized-in-bss
6116 @opindex fno-zero-initialized-in-bss
6117 If the target supports a BSS section, GCC by default puts variables that
6118 are initialized to zero into BSS@. This can save space in the resulting
6121 This option turns off this behavior because some programs explicitly
6122 rely on variables going to the data section. E.g., so that the
6123 resulting executable can find the beginning of that section and/or make
6124 assumptions based on that.
6126 The default is @option{-fzero-initialized-in-bss}.
6128 @item -fmudflap -fmudflapth -fmudflapir
6132 @cindex bounds checking
6134 For front-ends that support it (C and C++), instrument all risky
6135 pointer/array dereferencing operations, some standard library
6136 string/heap functions, and some other associated constructs with
6137 range/validity tests. Modules so instrumented should be immune to
6138 buffer overflows, invalid heap use, and some other classes of C/C++
6139 programming errors. The instrumentation relies on a separate runtime
6140 library (@file{libmudflap}), which will be linked into a program if
6141 @option{-fmudflap} is given at link time. Run-time behavior of the
6142 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6143 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6146 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6147 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6148 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6149 instrumentation should ignore pointer reads. This produces less
6150 instrumentation (and therefore faster execution) and still provides
6151 some protection against outright memory corrupting writes, but allows
6152 erroneously read data to propagate within a program.
6154 @item -fthread-jumps
6155 @opindex fthread-jumps
6156 Perform optimizations where we check to see if a jump branches to a
6157 location where another comparison subsumed by the first is found. If
6158 so, the first branch is redirected to either the destination of the
6159 second branch or a point immediately following it, depending on whether
6160 the condition is known to be true or false.
6162 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6164 @item -fsplit-wide-types
6165 @opindex fsplit-wide-types
6166 When using a type that occupies multiple registers, such as @code{long
6167 long} on a 32-bit system, split the registers apart and allocate them
6168 independently. This normally generates better code for those types,
6169 but may make debugging more difficult.
6171 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6174 @item -fcse-follow-jumps
6175 @opindex fcse-follow-jumps
6176 In common subexpression elimination (CSE), scan through jump instructions
6177 when the target of the jump is not reached by any other path. For
6178 example, when CSE encounters an @code{if} statement with an
6179 @code{else} clause, CSE will follow the jump when the condition
6182 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6184 @item -fcse-skip-blocks
6185 @opindex fcse-skip-blocks
6186 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6187 follow jumps which conditionally skip over blocks. When CSE
6188 encounters a simple @code{if} statement with no else clause,
6189 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6190 body of the @code{if}.
6192 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6194 @item -frerun-cse-after-loop
6195 @opindex frerun-cse-after-loop
6196 Re-run common subexpression elimination after loop optimizations has been
6199 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6203 Perform a global common subexpression elimination pass.
6204 This pass also performs global constant and copy propagation.
6206 @emph{Note:} When compiling a program using computed gotos, a GCC
6207 extension, you may get better runtime performance if you disable
6208 the global common subexpression elimination pass by adding
6209 @option{-fno-gcse} to the command line.
6211 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6215 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6216 attempt to move loads which are only killed by stores into themselves. This
6217 allows a loop containing a load/store sequence to be changed to a load outside
6218 the loop, and a copy/store within the loop.
6220 Enabled by default when gcse is enabled.
6224 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6225 global common subexpression elimination. This pass will attempt to move
6226 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6227 loops containing a load/store sequence can be changed to a load before
6228 the loop and a store after the loop.
6230 Not enabled at any optimization level.
6234 When @option{-fgcse-las} is enabled, the global common subexpression
6235 elimination pass eliminates redundant loads that come after stores to the
6236 same memory location (both partial and full redundancies).
6238 Not enabled at any optimization level.
6240 @item -fgcse-after-reload
6241 @opindex fgcse-after-reload
6242 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6243 pass is performed after reload. The purpose of this pass is to cleanup
6246 @item -funsafe-loop-optimizations
6247 @opindex funsafe-loop-optimizations
6248 If given, the loop optimizer will assume that loop indices do not
6249 overflow, and that the loops with nontrivial exit condition are not
6250 infinite. This enables a wider range of loop optimizations even if
6251 the loop optimizer itself cannot prove that these assumptions are valid.
6252 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6253 if it finds this kind of loop.
6255 @item -fcrossjumping
6256 @opindex fcrossjumping
6257 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6258 resulting code may or may not perform better than without cross-jumping.
6260 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6262 @item -fauto-inc-dec
6263 @opindex fauto-inc-dec
6264 Combine increments or decrements of addresses with memory accesses.
6265 This pass is always skipped on architectures that do not have
6266 instructions to support this. Enabled by default at @option{-O} and
6267 higher on architectures that support this.
6271 Perform dead code elimination (DCE) on RTL@.
6272 Enabled by default at @option{-O} and higher.
6276 Perform dead store elimination (DSE) on RTL@.
6277 Enabled by default at @option{-O} and higher.
6279 @item -fif-conversion
6280 @opindex fif-conversion
6281 Attempt to transform conditional jumps into branch-less equivalents. This
6282 include use of conditional moves, min, max, set flags and abs instructions, and
6283 some tricks doable by standard arithmetics. The use of conditional execution
6284 on chips where it is available is controlled by @code{if-conversion2}.
6286 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6288 @item -fif-conversion2
6289 @opindex fif-conversion2
6290 Use conditional execution (where available) to transform conditional jumps into
6291 branch-less equivalents.
6293 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6295 @item -fdelete-null-pointer-checks
6296 @opindex fdelete-null-pointer-checks
6297 Assume that programs cannot safely dereference null pointers, and that
6298 no code or data element resides there. This enables simple constant
6299 folding optimizations at all optimization levels. In addition, other
6300 optimization passes in GCC use this flag to control global dataflow
6301 analyses that eliminate useless checks for null pointers; these assume
6302 that if a pointer is checked after it has already been dereferenced,
6305 Note however that in some environments this assumption is not true.
6306 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6307 for programs which depend on that behavior.
6309 Some targets, especially embedded ones, disable this option at all levels.
6310 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6311 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6312 are enabled independently at different optimization levels.
6314 @item -fexpensive-optimizations
6315 @opindex fexpensive-optimizations
6316 Perform a number of minor optimizations that are relatively expensive.
6318 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6320 @item -foptimize-register-move
6322 @opindex foptimize-register-move
6324 Attempt to reassign register numbers in move instructions and as
6325 operands of other simple instructions in order to maximize the amount of
6326 register tying. This is especially helpful on machines with two-operand
6329 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6332 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6334 @item -fira-algorithm=@var{algorithm}
6335 Use specified coloring algorithm for the integrated register
6336 allocator. The @var{algorithm} argument should be @code{priority} or
6337 @code{CB}. The first algorithm specifies Chow's priority coloring,
6338 the second one specifies Chaitin-Briggs coloring. The second
6339 algorithm can be unimplemented for some architectures. If it is
6340 implemented, it is the default because Chaitin-Briggs coloring as a
6341 rule generates a better code.
6343 @item -fira-region=@var{region}
6344 Use specified regions for the integrated register allocator. The
6345 @var{region} argument should be one of @code{all}, @code{mixed}, or
6346 @code{one}. The first value means using all loops as register
6347 allocation regions, the second value which is the default means using
6348 all loops except for loops with small register pressure as the
6349 regions, and third one means using all function as a single region.
6350 The first value can give best result for machines with small size and
6351 irregular register set, the third one results in faster and generates
6352 decent code and the smallest size code, and the default value usually
6353 give the best results in most cases and for most architectures.
6355 @item -fira-coalesce
6356 @opindex fira-coalesce
6357 Do optimistic register coalescing. This option might be profitable for
6358 architectures with big regular register files.
6360 @item -fira-loop-pressure
6361 @opindex fira-loop-pressure
6362 Use IRA to evaluate register pressure in loops for decision to move
6363 loop invariants. Usage of this option usually results in generation
6364 of faster and smaller code on machines with big register files (>= 32
6365 registers) but it can slow compiler down.
6367 This option is enabled at level @option{-O3} for some targets.
6369 @item -fno-ira-share-save-slots
6370 @opindex fno-ira-share-save-slots
6371 Switch off sharing stack slots used for saving call used hard
6372 registers living through a call. Each hard register will get a
6373 separate stack slot and as a result function stack frame will be
6376 @item -fno-ira-share-spill-slots
6377 @opindex fno-ira-share-spill-slots
6378 Switch off sharing stack slots allocated for pseudo-registers. Each
6379 pseudo-register which did not get a hard register will get a separate
6380 stack slot and as a result function stack frame will be bigger.
6382 @item -fira-verbose=@var{n}
6383 @opindex fira-verbose
6384 Set up how verbose dump file for the integrated register allocator
6385 will be. Default value is 5. If the value is greater or equal to 10,
6386 the dump file will be stderr as if the value were @var{n} minus 10.
6388 @item -fdelayed-branch
6389 @opindex fdelayed-branch
6390 If supported for the target machine, attempt to reorder instructions
6391 to exploit instruction slots available after delayed branch
6394 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6396 @item -fschedule-insns
6397 @opindex fschedule-insns
6398 If supported for the target machine, attempt to reorder instructions to
6399 eliminate execution stalls due to required data being unavailable. This
6400 helps machines that have slow floating point or memory load instructions
6401 by allowing other instructions to be issued until the result of the load
6402 or floating point instruction is required.
6404 Enabled at levels @option{-O2}, @option{-O3}.
6406 @item -fschedule-insns2
6407 @opindex fschedule-insns2
6408 Similar to @option{-fschedule-insns}, but requests an additional pass of
6409 instruction scheduling after register allocation has been done. This is
6410 especially useful on machines with a relatively small number of
6411 registers and where memory load instructions take more than one cycle.
6413 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6415 @item -fno-sched-interblock
6416 @opindex fno-sched-interblock
6417 Don't schedule instructions across basic blocks. This is normally
6418 enabled by default when scheduling before register allocation, i.e.@:
6419 with @option{-fschedule-insns} or at @option{-O2} or higher.
6421 @item -fno-sched-spec
6422 @opindex fno-sched-spec
6423 Don't allow speculative motion of non-load instructions. This is normally
6424 enabled by default when scheduling before register allocation, i.e.@:
6425 with @option{-fschedule-insns} or at @option{-O2} or higher.
6427 @item -fsched-pressure
6428 @opindex fsched-pressure
6429 Enable register pressure sensitive insn scheduling before the register
6430 allocation. This only makes sense when scheduling before register
6431 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6432 @option{-O2} or higher. Usage of this option can improve the
6433 generated code and decrease its size by preventing register pressure
6434 increase above the number of available hard registers and as a
6435 consequence register spills in the register allocation.
6437 @item -fsched-spec-load
6438 @opindex fsched-spec-load
6439 Allow speculative motion of some load instructions. This only makes
6440 sense when scheduling before register allocation, i.e.@: with
6441 @option{-fschedule-insns} or at @option{-O2} or higher.
6443 @item -fsched-spec-load-dangerous
6444 @opindex fsched-spec-load-dangerous
6445 Allow speculative motion of more load instructions. This only makes
6446 sense when scheduling before register allocation, i.e.@: with
6447 @option{-fschedule-insns} or at @option{-O2} or higher.
6449 @item -fsched-stalled-insns
6450 @itemx -fsched-stalled-insns=@var{n}
6451 @opindex fsched-stalled-insns
6452 Define how many insns (if any) can be moved prematurely from the queue
6453 of stalled insns into the ready list, during the second scheduling pass.
6454 @option{-fno-sched-stalled-insns} means that no insns will be moved
6455 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6456 on how many queued insns can be moved prematurely.
6457 @option{-fsched-stalled-insns} without a value is equivalent to
6458 @option{-fsched-stalled-insns=1}.
6460 @item -fsched-stalled-insns-dep
6461 @itemx -fsched-stalled-insns-dep=@var{n}
6462 @opindex fsched-stalled-insns-dep
6463 Define how many insn groups (cycles) will be examined for a dependency
6464 on a stalled insn that is candidate for premature removal from the queue
6465 of stalled insns. This has an effect only during the second scheduling pass,
6466 and only if @option{-fsched-stalled-insns} is used.
6467 @option{-fno-sched-stalled-insns-dep} is equivalent to
6468 @option{-fsched-stalled-insns-dep=0}.
6469 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6470 @option{-fsched-stalled-insns-dep=1}.
6472 @item -fsched2-use-superblocks
6473 @opindex fsched2-use-superblocks
6474 When scheduling after register allocation, do use superblock scheduling
6475 algorithm. Superblock scheduling allows motion across basic block boundaries
6476 resulting on faster schedules. This option is experimental, as not all machine
6477 descriptions used by GCC model the CPU closely enough to avoid unreliable
6478 results from the algorithm.
6480 This only makes sense when scheduling after register allocation, i.e.@: with
6481 @option{-fschedule-insns2} or at @option{-O2} or higher.
6483 @item -fsched-group-heuristic
6484 @opindex fsched-group-heuristic
6485 Enable the group heuristic in the scheduler. This heuristic favors
6486 the instruction that belongs to a schedule group. This is enabled
6487 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6488 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6490 @item -fsched-critical-path-heuristic
6491 @opindex fsched-critical-path-heuristic
6492 Enable the critical-path heuristic in the scheduler. This heuristic favors
6493 instructions on the critical path. This is enabled by default when
6494 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6495 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6497 @item -fsched-spec-insn-heuristic
6498 @opindex fsched-spec-insn-heuristic
6499 Enable the speculative instruction heuristic in the scheduler. This
6500 heuristic favors speculative instructions with greater dependency weakness.
6501 This is enabled by default when scheduling is enabled, i.e.@:
6502 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6503 or at @option{-O2} or higher.
6505 @item -fsched-rank-heuristic
6506 @opindex fsched-rank-heuristic
6507 Enable the rank heuristic in the scheduler. This heuristic favors
6508 the instruction belonging to a basic block with greater size or frequency.
6509 This is enabled by default when scheduling is enabled, i.e.@:
6510 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6511 at @option{-O2} or higher.
6513 @item -fsched-last-insn-heuristic
6514 @opindex fsched-last-insn-heuristic
6515 Enable the last-instruction heuristic in the scheduler. This heuristic
6516 favors the instruction that is less dependent on the last instruction
6517 scheduled. This is enabled by default when scheduling is enabled,
6518 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6519 at @option{-O2} or higher.
6521 @item -fsched-dep-count-heuristic
6522 @opindex fsched-dep-count-heuristic
6523 Enable the dependent-count heuristic in the scheduler. This heuristic
6524 favors the instruction that has more instructions depending on it.
6525 This is enabled by default when scheduling is enabled, i.e.@:
6526 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6527 at @option{-O2} or higher.
6529 @item -freschedule-modulo-scheduled-loops
6530 @opindex freschedule-modulo-scheduled-loops
6531 The modulo scheduling comes before the traditional scheduling, if a loop
6532 was modulo scheduled we may want to prevent the later scheduling passes
6533 from changing its schedule, we use this option to control that.
6535 @item -fselective-scheduling
6536 @opindex fselective-scheduling
6537 Schedule instructions using selective scheduling algorithm. Selective
6538 scheduling runs instead of the first scheduler pass.
6540 @item -fselective-scheduling2
6541 @opindex fselective-scheduling2
6542 Schedule instructions using selective scheduling algorithm. Selective
6543 scheduling runs instead of the second scheduler pass.
6545 @item -fsel-sched-pipelining
6546 @opindex fsel-sched-pipelining
6547 Enable software pipelining of innermost loops during selective scheduling.
6548 This option has no effect until one of @option{-fselective-scheduling} or
6549 @option{-fselective-scheduling2} is turned on.
6551 @item -fsel-sched-pipelining-outer-loops
6552 @opindex fsel-sched-pipelining-outer-loops
6553 When pipelining loops during selective scheduling, also pipeline outer loops.
6554 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6556 @item -fcaller-saves
6557 @opindex fcaller-saves
6558 Enable values to be allocated in registers that will be clobbered by
6559 function calls, by emitting extra instructions to save and restore the
6560 registers around such calls. Such allocation is done only when it
6561 seems to result in better code than would otherwise be produced.
6563 This option is always enabled by default on certain machines, usually
6564 those which have no call-preserved registers to use instead.
6566 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6568 @item -fconserve-stack
6569 @opindex fconserve-stack
6570 Attempt to minimize stack usage. The compiler will attempt to use less
6571 stack space, even if that makes the program slower. This option
6572 implies setting the @option{large-stack-frame} parameter to 100
6573 and the @option{large-stack-frame-growth} parameter to 400.
6575 @item -ftree-reassoc
6576 @opindex ftree-reassoc
6577 Perform reassociation on trees. This flag is enabled by default
6578 at @option{-O} and higher.
6582 Perform partial redundancy elimination (PRE) on trees. This flag is
6583 enabled by default at @option{-O2} and @option{-O3}.
6585 @item -ftree-forwprop
6586 @opindex ftree-forwprop
6587 Perform forward propagation on trees. This flag is enabled by default
6588 at @option{-O} and higher.
6592 Perform full redundancy elimination (FRE) on trees. The difference
6593 between FRE and PRE is that FRE only considers expressions
6594 that are computed on all paths leading to the redundant computation.
6595 This analysis is faster than PRE, though it exposes fewer redundancies.
6596 This flag is enabled by default at @option{-O} and higher.
6598 @item -ftree-phiprop
6599 @opindex ftree-phiprop
6600 Perform hoisting of loads from conditional pointers on trees. This
6601 pass is enabled by default at @option{-O} and higher.
6603 @item -ftree-copy-prop
6604 @opindex ftree-copy-prop
6605 Perform copy propagation on trees. This pass eliminates unnecessary
6606 copy operations. This flag is enabled by default at @option{-O} and
6609 @item -fipa-pure-const
6610 @opindex fipa-pure-const
6611 Discover which functions are pure or constant.
6612 Enabled by default at @option{-O} and higher.
6614 @item -fipa-reference
6615 @opindex fipa-reference
6616 Discover which static variables do not escape cannot escape the
6618 Enabled by default at @option{-O} and higher.
6620 @item -fipa-struct-reorg
6621 @opindex fipa-struct-reorg
6622 Perform structure reorganization optimization, that change C-like structures
6623 layout in order to better utilize spatial locality. This transformation is
6624 affective for programs containing arrays of structures. Available in two
6625 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6626 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6627 to provide the safety of this transformation. It works only in whole program
6628 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6629 enabled. Structures considered @samp{cold} by this transformation are not
6630 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6632 With this flag, the program debug info reflects a new structure layout.
6636 Perform interprocedural pointer analysis and interprocedural modification
6637 and reference analysis. This option can cause excessive memory and
6638 compile-time usage on large compilation units. It is not enabled by
6639 default at any optimization level.
6642 @opindex fipa-profile
6643 Perform interprocedural profile propagation. The functions called only from
6644 cold functions are marked as cold. Also functions executed once (such as
6645 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
6646 functions and loop less parts of functions executed once are then optimized for
6648 Enabled by default at @option{-O} and higher.
6652 Perform interprocedural constant propagation.
6653 This optimization analyzes the program to determine when values passed
6654 to functions are constants and then optimizes accordingly.
6655 This optimization can substantially increase performance
6656 if the application has constants passed to functions.
6657 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6659 @item -fipa-cp-clone
6660 @opindex fipa-cp-clone
6661 Perform function cloning to make interprocedural constant propagation stronger.
6662 When enabled, interprocedural constant propagation will perform function cloning
6663 when externally visible function can be called with constant arguments.
6664 Because this optimization can create multiple copies of functions,
6665 it may significantly increase code size
6666 (see @option{--param ipcp-unit-growth=@var{value}}).
6667 This flag is enabled by default at @option{-O3}.
6669 @item -fipa-matrix-reorg
6670 @opindex fipa-matrix-reorg
6671 Perform matrix flattening and transposing.
6672 Matrix flattening tries to replace an @math{m}-dimensional matrix
6673 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6674 This reduces the level of indirection needed for accessing the elements
6675 of the matrix. The second optimization is matrix transposing that
6676 attempts to change the order of the matrix's dimensions in order to
6677 improve cache locality.
6678 Both optimizations need the @option{-fwhole-program} flag.
6679 Transposing is enabled only if profiling information is available.
6683 Perform forward store motion on trees. This flag is
6684 enabled by default at @option{-O} and higher.
6688 Perform sparse conditional constant propagation (CCP) on trees. This
6689 pass only operates on local scalar variables and is enabled by default
6690 at @option{-O} and higher.
6692 @item -ftree-switch-conversion
6693 Perform conversion of simple initializations in a switch to
6694 initializations from a scalar array. This flag is enabled by default
6695 at @option{-O2} and higher.
6699 Perform dead code elimination (DCE) on trees. This flag is enabled by
6700 default at @option{-O} and higher.
6702 @item -ftree-builtin-call-dce
6703 @opindex ftree-builtin-call-dce
6704 Perform conditional dead code elimination (DCE) for calls to builtin functions
6705 that may set @code{errno} but are otherwise side-effect free. This flag is
6706 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6709 @item -ftree-dominator-opts
6710 @opindex ftree-dominator-opts
6711 Perform a variety of simple scalar cleanups (constant/copy
6712 propagation, redundancy elimination, range propagation and expression
6713 simplification) based on a dominator tree traversal. This also
6714 performs jump threading (to reduce jumps to jumps). This flag is
6715 enabled by default at @option{-O} and higher.
6719 Perform dead store elimination (DSE) on trees. A dead store is a store into
6720 a memory location which will later be overwritten by another store without
6721 any intervening loads. In this case the earlier store can be deleted. This
6722 flag is enabled by default at @option{-O} and higher.
6726 Perform loop header copying on trees. This is beneficial since it increases
6727 effectiveness of code motion optimizations. It also saves one jump. This flag
6728 is enabled by default at @option{-O} and higher. It is not enabled
6729 for @option{-Os}, since it usually increases code size.
6731 @item -ftree-loop-optimize
6732 @opindex ftree-loop-optimize
6733 Perform loop optimizations on trees. This flag is enabled by default
6734 at @option{-O} and higher.
6736 @item -ftree-loop-linear
6737 @opindex ftree-loop-linear
6738 Perform linear loop transformations on tree. This flag can improve cache
6739 performance and allow further loop optimizations to take place.
6741 @item -floop-interchange
6742 Perform loop interchange transformations on loops. Interchanging two
6743 nested loops switches the inner and outer loops. For example, given a
6748 A(J, I) = A(J, I) * C
6752 loop interchange will transform the loop as if the user had written:
6756 A(J, I) = A(J, I) * C
6760 which can be beneficial when @code{N} is larger than the caches,
6761 because in Fortran, the elements of an array are stored in memory
6762 contiguously by column, and the original loop iterates over rows,
6763 potentially creating at each access a cache miss. This optimization
6764 applies to all the languages supported by GCC and is not limited to
6765 Fortran. To use this code transformation, GCC has to be configured
6766 with @option{--with-ppl} and @option{--with-cloog} to enable the
6767 Graphite loop transformation infrastructure.
6769 @item -floop-strip-mine
6770 Perform loop strip mining transformations on loops. Strip mining
6771 splits a loop into two nested loops. The outer loop has strides
6772 equal to the strip size and the inner loop has strides of the
6773 original loop within a strip. The strip length can be changed
6774 using the @option{loop-block-tile-size} parameter. For example,
6781 loop strip mining will transform the loop as if the user had written:
6784 DO I = II, min (II + 50, N)
6789 This optimization applies to all the languages supported by GCC and is
6790 not limited to Fortran. To use this code transformation, GCC has to
6791 be configured with @option{--with-ppl} and @option{--with-cloog} to
6792 enable the Graphite loop transformation infrastructure.
6795 Perform loop blocking transformations on loops. Blocking strip mines
6796 each loop in the loop nest such that the memory accesses of the
6797 element loops fit inside caches. The strip length can be changed
6798 using the @option{loop-block-tile-size} parameter. For example, given
6803 A(J, I) = B(I) + C(J)
6807 loop blocking will transform the loop as if the user had written:
6811 DO I = II, min (II + 50, N)
6812 DO J = JJ, min (JJ + 50, M)
6813 A(J, I) = B(I) + C(J)
6819 which can be beneficial when @code{M} is larger than the caches,
6820 because the innermost loop will iterate over a smaller amount of data
6821 that can be kept in the caches. This optimization applies to all the
6822 languages supported by GCC and is not limited to Fortran. To use this
6823 code transformation, GCC has to be configured with @option{--with-ppl}
6824 and @option{--with-cloog} to enable the Graphite loop transformation
6827 @item -fgraphite-identity
6828 @opindex fgraphite-identity
6829 Enable the identity transformation for graphite. For every SCoP we generate
6830 the polyhedral representation and transform it back to gimple. Using
6831 @option{-fgraphite-identity} we can check the costs or benefits of the
6832 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6833 are also performed by the code generator CLooG, like index splitting and
6834 dead code elimination in loops.
6836 @item -floop-parallelize-all
6837 Use the Graphite data dependence analysis to identify loops that can
6838 be parallelized. Parallelize all the loops that can be analyzed to
6839 not contain loop carried dependences without checking that it is
6840 profitable to parallelize the loops.
6842 @item -fcheck-data-deps
6843 @opindex fcheck-data-deps
6844 Compare the results of several data dependence analyzers. This option
6845 is used for debugging the data dependence analyzers.
6847 @item -ftree-loop-distribution
6848 Perform loop distribution. This flag can improve cache performance on
6849 big loop bodies and allow further loop optimizations, like
6850 parallelization or vectorization, to take place. For example, the loop
6867 @item -ftree-loop-im
6868 @opindex ftree-loop-im
6869 Perform loop invariant motion on trees. This pass moves only invariants that
6870 would be hard to handle at RTL level (function calls, operations that expand to
6871 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6872 operands of conditions that are invariant out of the loop, so that we can use
6873 just trivial invariantness analysis in loop unswitching. The pass also includes
6876 @item -ftree-loop-ivcanon
6877 @opindex ftree-loop-ivcanon
6878 Create a canonical counter for number of iterations in the loop for that
6879 determining number of iterations requires complicated analysis. Later
6880 optimizations then may determine the number easily. Useful especially
6881 in connection with unrolling.
6885 Perform induction variable optimizations (strength reduction, induction
6886 variable merging and induction variable elimination) on trees.
6888 @item -ftree-parallelize-loops=n
6889 @opindex ftree-parallelize-loops
6890 Parallelize loops, i.e., split their iteration space to run in n threads.
6891 This is only possible for loops whose iterations are independent
6892 and can be arbitrarily reordered. The optimization is only
6893 profitable on multiprocessor machines, for loops that are CPU-intensive,
6894 rather than constrained e.g.@: by memory bandwidth. This option
6895 implies @option{-pthread}, and thus is only supported on targets
6896 that have support for @option{-pthread}.
6900 Perform function-local points-to analysis on trees. This flag is
6901 enabled by default at @option{-O} and higher.
6905 Perform scalar replacement of aggregates. This pass replaces structure
6906 references with scalars to prevent committing structures to memory too
6907 early. This flag is enabled by default at @option{-O} and higher.
6909 @item -ftree-copyrename
6910 @opindex ftree-copyrename
6911 Perform copy renaming on trees. This pass attempts to rename compiler
6912 temporaries to other variables at copy locations, usually resulting in
6913 variable names which more closely resemble the original variables. This flag
6914 is enabled by default at @option{-O} and higher.
6918 Perform temporary expression replacement during the SSA->normal phase. Single
6919 use/single def temporaries are replaced at their use location with their
6920 defining expression. This results in non-GIMPLE code, but gives the expanders
6921 much more complex trees to work on resulting in better RTL generation. This is
6922 enabled by default at @option{-O} and higher.
6924 @item -ftree-vectorize
6925 @opindex ftree-vectorize
6926 Perform loop vectorization on trees. This flag is enabled by default at
6929 @item -ftree-slp-vectorize
6930 @opindex ftree-slp-vectorize
6931 Perform basic block vectorization on trees. This flag is enabled by default at
6932 @option{-O3} and when @option{-ftree-vectorize} is enabled.
6934 @item -ftree-vect-loop-version
6935 @opindex ftree-vect-loop-version
6936 Perform loop versioning when doing loop vectorization on trees. When a loop
6937 appears to be vectorizable except that data alignment or data dependence cannot
6938 be determined at compile time then vectorized and non-vectorized versions of
6939 the loop are generated along with runtime checks for alignment or dependence
6940 to control which version is executed. This option is enabled by default
6941 except at level @option{-Os} where it is disabled.
6943 @item -fvect-cost-model
6944 @opindex fvect-cost-model
6945 Enable cost model for vectorization.
6949 Perform Value Range Propagation on trees. This is similar to the
6950 constant propagation pass, but instead of values, ranges of values are
6951 propagated. This allows the optimizers to remove unnecessary range
6952 checks like array bound checks and null pointer checks. This is
6953 enabled by default at @option{-O2} and higher. Null pointer check
6954 elimination is only done if @option{-fdelete-null-pointer-checks} is
6959 Perform tail duplication to enlarge superblock size. This transformation
6960 simplifies the control flow of the function allowing other optimizations to do
6963 @item -funroll-loops
6964 @opindex funroll-loops
6965 Unroll loops whose number of iterations can be determined at compile
6966 time or upon entry to the loop. @option{-funroll-loops} implies
6967 @option{-frerun-cse-after-loop}. This option makes code larger,
6968 and may or may not make it run faster.
6970 @item -funroll-all-loops
6971 @opindex funroll-all-loops
6972 Unroll all loops, even if their number of iterations is uncertain when
6973 the loop is entered. This usually makes programs run more slowly.
6974 @option{-funroll-all-loops} implies the same options as
6975 @option{-funroll-loops},
6977 @item -fsplit-ivs-in-unroller
6978 @opindex fsplit-ivs-in-unroller
6979 Enables expressing of values of induction variables in later iterations
6980 of the unrolled loop using the value in the first iteration. This breaks
6981 long dependency chains, thus improving efficiency of the scheduling passes.
6983 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6984 same effect. However in cases the loop body is more complicated than
6985 a single basic block, this is not reliable. It also does not work at all
6986 on some of the architectures due to restrictions in the CSE pass.
6988 This optimization is enabled by default.
6990 @item -fvariable-expansion-in-unroller
6991 @opindex fvariable-expansion-in-unroller
6992 With this option, the compiler will create multiple copies of some
6993 local variables when unrolling a loop which can result in superior code.
6995 @item -fpredictive-commoning
6996 @opindex fpredictive-commoning
6997 Perform predictive commoning optimization, i.e., reusing computations
6998 (especially memory loads and stores) performed in previous
6999 iterations of loops.
7001 This option is enabled at level @option{-O3}.
7003 @item -fprefetch-loop-arrays
7004 @opindex fprefetch-loop-arrays
7005 If supported by the target machine, generate instructions to prefetch
7006 memory to improve the performance of loops that access large arrays.
7008 This option may generate better or worse code; results are highly
7009 dependent on the structure of loops within the source code.
7011 Disabled at level @option{-Os}.
7014 @itemx -fno-peephole2
7015 @opindex fno-peephole
7016 @opindex fno-peephole2
7017 Disable any machine-specific peephole optimizations. The difference
7018 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7019 are implemented in the compiler; some targets use one, some use the
7020 other, a few use both.
7022 @option{-fpeephole} is enabled by default.
7023 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7025 @item -fno-guess-branch-probability
7026 @opindex fno-guess-branch-probability
7027 Do not guess branch probabilities using heuristics.
7029 GCC will use heuristics to guess branch probabilities if they are
7030 not provided by profiling feedback (@option{-fprofile-arcs}). These
7031 heuristics are based on the control flow graph. If some branch probabilities
7032 are specified by @samp{__builtin_expect}, then the heuristics will be
7033 used to guess branch probabilities for the rest of the control flow graph,
7034 taking the @samp{__builtin_expect} info into account. The interactions
7035 between the heuristics and @samp{__builtin_expect} can be complex, and in
7036 some cases, it may be useful to disable the heuristics so that the effects
7037 of @samp{__builtin_expect} are easier to understand.
7039 The default is @option{-fguess-branch-probability} at levels
7040 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7042 @item -freorder-blocks
7043 @opindex freorder-blocks
7044 Reorder basic blocks in the compiled function in order to reduce number of
7045 taken branches and improve code locality.
7047 Enabled at levels @option{-O2}, @option{-O3}.
7049 @item -freorder-blocks-and-partition
7050 @opindex freorder-blocks-and-partition
7051 In addition to reordering basic blocks in the compiled function, in order
7052 to reduce number of taken branches, partitions hot and cold basic blocks
7053 into separate sections of the assembly and .o files, to improve
7054 paging and cache locality performance.
7056 This optimization is automatically turned off in the presence of
7057 exception handling, for linkonce sections, for functions with a user-defined
7058 section attribute and on any architecture that does not support named
7061 @item -freorder-functions
7062 @opindex freorder-functions
7063 Reorder functions in the object file in order to
7064 improve code locality. This is implemented by using special
7065 subsections @code{.text.hot} for most frequently executed functions and
7066 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
7067 the linker so object file format must support named sections and linker must
7068 place them in a reasonable way.
7070 Also profile feedback must be available in to make this option effective. See
7071 @option{-fprofile-arcs} for details.
7073 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7075 @item -fstrict-aliasing
7076 @opindex fstrict-aliasing
7077 Allow the compiler to assume the strictest aliasing rules applicable to
7078 the language being compiled. For C (and C++), this activates
7079 optimizations based on the type of expressions. In particular, an
7080 object of one type is assumed never to reside at the same address as an
7081 object of a different type, unless the types are almost the same. For
7082 example, an @code{unsigned int} can alias an @code{int}, but not a
7083 @code{void*} or a @code{double}. A character type may alias any other
7086 @anchor{Type-punning}Pay special attention to code like this:
7099 The practice of reading from a different union member than the one most
7100 recently written to (called ``type-punning'') is common. Even with
7101 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7102 is accessed through the union type. So, the code above will work as
7103 expected. @xref{Structures unions enumerations and bit-fields
7104 implementation}. However, this code might not:
7115 Similarly, access by taking the address, casting the resulting pointer
7116 and dereferencing the result has undefined behavior, even if the cast
7117 uses a union type, e.g.:
7121 return ((union a_union *) &d)->i;
7125 The @option{-fstrict-aliasing} option is enabled at levels
7126 @option{-O2}, @option{-O3}, @option{-Os}.
7128 @item -fstrict-overflow
7129 @opindex fstrict-overflow
7130 Allow the compiler to assume strict signed overflow rules, depending
7131 on the language being compiled. For C (and C++) this means that
7132 overflow when doing arithmetic with signed numbers is undefined, which
7133 means that the compiler may assume that it will not happen. This
7134 permits various optimizations. For example, the compiler will assume
7135 that an expression like @code{i + 10 > i} will always be true for
7136 signed @code{i}. This assumption is only valid if signed overflow is
7137 undefined, as the expression is false if @code{i + 10} overflows when
7138 using twos complement arithmetic. When this option is in effect any
7139 attempt to determine whether an operation on signed numbers will
7140 overflow must be written carefully to not actually involve overflow.
7142 This option also allows the compiler to assume strict pointer
7143 semantics: given a pointer to an object, if adding an offset to that
7144 pointer does not produce a pointer to the same object, the addition is
7145 undefined. This permits the compiler to conclude that @code{p + u >
7146 p} is always true for a pointer @code{p} and unsigned integer
7147 @code{u}. This assumption is only valid because pointer wraparound is
7148 undefined, as the expression is false if @code{p + u} overflows using
7149 twos complement arithmetic.
7151 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7152 that integer signed overflow is fully defined: it wraps. When
7153 @option{-fwrapv} is used, there is no difference between
7154 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7155 integers. With @option{-fwrapv} certain types of overflow are
7156 permitted. For example, if the compiler gets an overflow when doing
7157 arithmetic on constants, the overflowed value can still be used with
7158 @option{-fwrapv}, but not otherwise.
7160 The @option{-fstrict-overflow} option is enabled at levels
7161 @option{-O2}, @option{-O3}, @option{-Os}.
7163 @item -falign-functions
7164 @itemx -falign-functions=@var{n}
7165 @opindex falign-functions
7166 Align the start of functions to the next power-of-two greater than
7167 @var{n}, skipping up to @var{n} bytes. For instance,
7168 @option{-falign-functions=32} aligns functions to the next 32-byte
7169 boundary, but @option{-falign-functions=24} would align to the next
7170 32-byte boundary only if this can be done by skipping 23 bytes or less.
7172 @option{-fno-align-functions} and @option{-falign-functions=1} are
7173 equivalent and mean that functions will not be aligned.
7175 Some assemblers only support this flag when @var{n} is a power of two;
7176 in that case, it is rounded up.
7178 If @var{n} is not specified or is zero, use a machine-dependent default.
7180 Enabled at levels @option{-O2}, @option{-O3}.
7182 @item -falign-labels
7183 @itemx -falign-labels=@var{n}
7184 @opindex falign-labels
7185 Align all branch targets to a power-of-two boundary, skipping up to
7186 @var{n} bytes like @option{-falign-functions}. This option can easily
7187 make code slower, because it must insert dummy operations for when the
7188 branch target is reached in the usual flow of the code.
7190 @option{-fno-align-labels} and @option{-falign-labels=1} are
7191 equivalent and mean that labels will not be aligned.
7193 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7194 are greater than this value, then their values are used instead.
7196 If @var{n} is not specified or is zero, use a machine-dependent default
7197 which is very likely to be @samp{1}, meaning no alignment.
7199 Enabled at levels @option{-O2}, @option{-O3}.
7202 @itemx -falign-loops=@var{n}
7203 @opindex falign-loops
7204 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7205 like @option{-falign-functions}. The hope is that the loop will be
7206 executed many times, which will make up for any execution of the dummy
7209 @option{-fno-align-loops} and @option{-falign-loops=1} are
7210 equivalent and mean that loops will not be aligned.
7212 If @var{n} is not specified or is zero, use a machine-dependent default.
7214 Enabled at levels @option{-O2}, @option{-O3}.
7217 @itemx -falign-jumps=@var{n}
7218 @opindex falign-jumps
7219 Align branch targets to a power-of-two boundary, for branch targets
7220 where the targets can only be reached by jumping, skipping up to @var{n}
7221 bytes like @option{-falign-functions}. In this case, no dummy operations
7224 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7225 equivalent and mean that loops will not be aligned.
7227 If @var{n} is not specified or is zero, use a machine-dependent default.
7229 Enabled at levels @option{-O2}, @option{-O3}.
7231 @item -funit-at-a-time
7232 @opindex funit-at-a-time
7233 This option is left for compatibility reasons. @option{-funit-at-a-time}
7234 has no effect, while @option{-fno-unit-at-a-time} implies
7235 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7239 @item -fno-toplevel-reorder
7240 @opindex fno-toplevel-reorder
7241 Do not reorder top-level functions, variables, and @code{asm}
7242 statements. Output them in the same order that they appear in the
7243 input file. When this option is used, unreferenced static variables
7244 will not be removed. This option is intended to support existing code
7245 which relies on a particular ordering. For new code, it is better to
7248 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7249 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7254 Constructs webs as commonly used for register allocation purposes and assign
7255 each web individual pseudo register. This allows the register allocation pass
7256 to operate on pseudos directly, but also strengthens several other optimization
7257 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7258 however, make debugging impossible, since variables will no longer stay in a
7261 Enabled by default with @option{-funroll-loops}.
7263 @item -fwhole-program
7264 @opindex fwhole-program
7265 Assume that the current compilation unit represents the whole program being
7266 compiled. All public functions and variables with the exception of @code{main}
7267 and those merged by attribute @code{externally_visible} become static functions
7268 and in effect are optimized more aggressively by interprocedural optimizers.
7269 While this option is equivalent to proper use of the @code{static} keyword for
7270 programs consisting of a single file, in combination with option
7271 @option{-combine}, @option{-flto} or @option{-fwhopr} this flag can be used to
7272 compile many smaller scale programs since the functions and variables become
7273 local for the whole combined compilation unit, not for the single source file
7276 This option implies @option{-fwhole-file} for Fortran programs.
7280 This option runs the standard link-time optimizer. When invoked
7281 with source code, it generates GIMPLE (one of GCC's internal
7282 representations) and writes it to special ELF sections in the object
7283 file. When the object files are linked together, all the function
7284 bodies are read from these ELF sections and instantiated as if they
7285 had been part of the same translation unit.
7287 To use the link-timer optimizer, @option{-flto} needs to be specified at
7288 compile time and during the final link. For example,
7291 gcc -c -O2 -flto foo.c
7292 gcc -c -O2 -flto bar.c
7293 gcc -o myprog -flto -O2 foo.o bar.o
7296 The first two invocations to GCC will save a bytecode representation
7297 of GIMPLE into special ELF sections inside @file{foo.o} and
7298 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7299 @file{foo.o} and @file{bar.o}, merge the two files into a single
7300 internal image, and compile the result as usual. Since both
7301 @file{foo.o} and @file{bar.o} are merged into a single image, this
7302 causes all the inter-procedural analyses and optimizations in GCC to
7303 work across the two files as if they were a single one. This means,
7304 for example, that the inliner will be able to inline functions in
7305 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7307 Another (simpler) way to enable link-time optimization is,
7310 gcc -o myprog -flto -O2 foo.c bar.c
7313 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7314 merge them together into a single GIMPLE representation and optimize
7315 them as usual to produce @file{myprog}.
7317 The only important thing to keep in mind is that to enable link-time
7318 optimizations the @option{-flto} flag needs to be passed to both the
7319 compile and the link commands.
7321 Note that when a file is compiled with @option{-flto}, the generated
7322 object file will be larger than a regular object file because it will
7323 contain GIMPLE bytecodes and the usual final code. This means that
7324 object files with LTO information can be linked as a normal object
7325 file. So, in the previous example, if the final link is done with
7328 gcc -o myprog foo.o bar.o
7331 The only difference will be that no inter-procedural optimizations
7332 will be applied to produce @file{myprog}. The two object files
7333 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7336 Additionally, the optimization flags used to compile individual files
7337 are not necessarily related to those used at link-time. For instance,
7340 gcc -c -O0 -flto foo.c
7341 gcc -c -O0 -flto bar.c
7342 gcc -o myprog -flto -O3 foo.o bar.o
7345 This will produce individual object files with unoptimized assembler
7346 code, but the resulting binary @file{myprog} will be optimized at
7347 @option{-O3}. Now, if the final binary is generated without
7348 @option{-flto}, then @file{myprog} will not be optimized.
7350 When producing the final binary with @option{-flto}, GCC will only
7351 apply link-time optimizations to those files that contain bytecode.
7352 Therefore, you can mix and match object files and libraries with
7353 GIMPLE bytecodes and final object code. GCC will automatically select
7354 which files to optimize in LTO mode and which files to link without
7357 There are some code generation flags that GCC will preserve when
7358 generating bytecodes, as they need to be used during the final link
7359 stage. Currently, the following options are saved into the GIMPLE
7360 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7361 @option{-m} target flags.
7363 At link time, these options are read-in and reapplied. Note that the
7364 current implementation makes no attempt at recognizing conflicting
7365 values for these options. If two or more files have a conflicting
7366 value (e.g., one file is compiled with @option{-fPIC} and another
7367 isn't), the compiler will simply use the last value read from the
7368 bytecode files. It is recommended, then, that all the files
7369 participating in the same link be compiled with the same options.
7371 Another feature of LTO is that it is possible to apply interprocedural
7372 optimizations on files written in different languages. This requires
7373 some support in the language front end. Currently, the C, C++ and
7374 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7375 something like this should work
7380 gfortran -c -flto baz.f90
7381 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7384 Notice that the final link is done with @command{g++} to get the C++
7385 runtime libraries and @option{-lgfortran} is added to get the Fortran
7386 runtime libraries. In general, when mixing languages in LTO mode, you
7387 should use the same link command used when mixing languages in a
7388 regular (non-LTO) compilation. This means that if your build process
7389 was mixing languages before, all you need to add is @option{-flto} to
7390 all the compile and link commands.
7392 If LTO encounters objects with C linkage declared with incompatible
7393 types in separate translation units to be linked together (undefined
7394 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7395 issued. The behavior is still undefined at runtime.
7397 If object files containing GIMPLE bytecode are stored in a library
7398 archive, say @file{libfoo.a}, it is possible to extract and use them
7399 in an LTO link if you are using @command{gold} as the linker (which,
7400 in turn requires GCC to be configured with @option{--enable-gold}).
7401 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7405 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7408 With the linker plugin enabled, @command{gold} will extract the needed
7409 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7410 to make them part of the aggregated GIMPLE image to be optimized.
7412 If you are not using @command{gold} and/or do not specify
7413 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7414 will be extracted and linked as usual, but they will not participate
7415 in the LTO optimization process.
7417 Link time optimizations do not require the presence of the whole
7418 program to operate. If the program does not require any symbols to
7419 be exported, it is possible to combine @option{-flto} and
7420 @option{-fwhopr} with @option{-fwhole-program} to allow the
7421 interprocedural optimizers to use more aggressive assumptions which
7422 may lead to improved optimization opportunities.
7424 Regarding portability: the current implementation of LTO makes no
7425 attempt at generating bytecode that can be ported between different
7426 types of hosts. The bytecode files are versioned and there is a
7427 strict version check, so bytecode files generated in one version of
7428 GCC will not work with an older/newer version of GCC.
7430 Link time optimization does not play well with generating debugging
7431 information. Combining @option{-flto} or @option{-fwhopr} with
7432 @option{-g} is experimental.
7434 This option is disabled by default.
7438 This option is identical in functionality to @option{-flto} but it
7439 differs in how the final link stage is executed. Instead of loading
7440 all the function bodies in memory, the callgraph is analyzed and
7441 optimization decisions are made (whole program analysis or WPA). Once
7442 optimization decisions are made, the callgraph is partitioned and the
7443 different sections are compiled separately (local transformations or
7444 LTRANS)@. This process allows optimizations on very large programs
7445 that otherwise would not fit in memory. This option enables
7446 @option{-fwpa} and @option{-fltrans} automatically.
7448 Disabled by default.
7450 This option is experimental.
7454 This is an internal option used by GCC when compiling with
7455 @option{-fwhopr}. You should never need to use it.
7457 This option runs the link-time optimizer in the whole-program-analysis
7458 (WPA) mode, which reads in summary information from all inputs and
7459 performs a whole-program analysis based on summary information only.
7460 It generates object files for subsequent runs of the link-time
7461 optimizer where individual object files are optimized using both
7462 summary information from the WPA mode and the actual function bodies.
7463 It then drives the LTRANS phase.
7465 Disabled by default.
7469 This is an internal option used by GCC when compiling with
7470 @option{-fwhopr}. You should never need to use it.
7472 This option runs the link-time optimizer in the local-transformation (LTRANS)
7473 mode, which reads in output from a previous run of the LTO in WPA mode.
7474 In the LTRANS mode, LTO optimizes an object and produces the final assembly.
7476 Disabled by default.
7478 @item -fltrans-output-list=@var{file}
7479 @opindex fltrans-output-list
7480 This is an internal option used by GCC when compiling with
7481 @option{-fwhopr}. You should never need to use it.
7483 This option specifies a file to which the names of LTRANS output files are
7484 written. This option is only meaningful in conjunction with @option{-fwpa}.
7486 Disabled by default.
7488 @item -flto-compression-level=@var{n}
7489 This option specifies the level of compression used for intermediate
7490 language written to LTO object files, and is only meaningful in
7491 conjunction with LTO mode (@option{-fwhopr}, @option{-flto}). Valid
7492 values are 0 (no compression) to 9 (maximum compression). Values
7493 outside this range are clamped to either 0 or 9. If the option is not
7494 given, a default balanced compression setting is used.
7497 Prints a report with internal details on the workings of the link-time
7498 optimizer. The contents of this report vary from version to version,
7499 it is meant to be useful to GCC developers when processing object
7500 files in LTO mode (via @option{-fwhopr} or @option{-flto}).
7502 Disabled by default.
7504 @item -fuse-linker-plugin
7505 Enables the extraction of objects with GIMPLE bytecode information
7506 from library archives. This option relies on features available only
7507 in @command{gold}, so to use this you must configure GCC with
7508 @option{--enable-gold}. See @option{-flto} for a description on the
7509 effect of this flag and how to use it.
7511 Disabled by default.
7513 @item -fcprop-registers
7514 @opindex fcprop-registers
7515 After register allocation and post-register allocation instruction splitting,
7516 we perform a copy-propagation pass to try to reduce scheduling dependencies
7517 and occasionally eliminate the copy.
7519 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7521 @item -fprofile-correction
7522 @opindex fprofile-correction
7523 Profiles collected using an instrumented binary for multi-threaded programs may
7524 be inconsistent due to missed counter updates. When this option is specified,
7525 GCC will use heuristics to correct or smooth out such inconsistencies. By
7526 default, GCC will emit an error message when an inconsistent profile is detected.
7528 @item -fprofile-dir=@var{path}
7529 @opindex fprofile-dir
7531 Set the directory to search the profile data files in to @var{path}.
7532 This option affects only the profile data generated by
7533 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7534 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7535 and its related options.
7536 By default, GCC will use the current directory as @var{path}
7537 thus the profile data file will appear in the same directory as the object file.
7539 @item -fprofile-generate
7540 @itemx -fprofile-generate=@var{path}
7541 @opindex fprofile-generate
7543 Enable options usually used for instrumenting application to produce
7544 profile useful for later recompilation with profile feedback based
7545 optimization. You must use @option{-fprofile-generate} both when
7546 compiling and when linking your program.
7548 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7550 If @var{path} is specified, GCC will look at the @var{path} to find
7551 the profile feedback data files. See @option{-fprofile-dir}.
7554 @itemx -fprofile-use=@var{path}
7555 @opindex fprofile-use
7556 Enable profile feedback directed optimizations, and optimizations
7557 generally profitable only with profile feedback available.
7559 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7560 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7562 By default, GCC emits an error message if the feedback profiles do not
7563 match the source code. This error can be turned into a warning by using
7564 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7567 If @var{path} is specified, GCC will look at the @var{path} to find
7568 the profile feedback data files. See @option{-fprofile-dir}.
7571 The following options control compiler behavior regarding floating
7572 point arithmetic. These options trade off between speed and
7573 correctness. All must be specifically enabled.
7577 @opindex ffloat-store
7578 Do not store floating point variables in registers, and inhibit other
7579 options that might change whether a floating point value is taken from a
7582 @cindex floating point precision
7583 This option prevents undesirable excess precision on machines such as
7584 the 68000 where the floating registers (of the 68881) keep more
7585 precision than a @code{double} is supposed to have. Similarly for the
7586 x86 architecture. For most programs, the excess precision does only
7587 good, but a few programs rely on the precise definition of IEEE floating
7588 point. Use @option{-ffloat-store} for such programs, after modifying
7589 them to store all pertinent intermediate computations into variables.
7591 @item -fexcess-precision=@var{style}
7592 @opindex fexcess-precision
7593 This option allows further control over excess precision on machines
7594 where floating-point registers have more precision than the IEEE
7595 @code{float} and @code{double} types and the processor does not
7596 support operations rounding to those types. By default,
7597 @option{-fexcess-precision=fast} is in effect; this means that
7598 operations are carried out in the precision of the registers and that
7599 it is unpredictable when rounding to the types specified in the source
7600 code takes place. When compiling C, if
7601 @option{-fexcess-precision=standard} is specified then excess
7602 precision will follow the rules specified in ISO C99; in particular,
7603 both casts and assignments cause values to be rounded to their
7604 semantic types (whereas @option{-ffloat-store} only affects
7605 assignments). This option is enabled by default for C if a strict
7606 conformance option such as @option{-std=c99} is used.
7609 @option{-fexcess-precision=standard} is not implemented for languages
7610 other than C, and has no effect if
7611 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7612 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7613 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7614 semantics apply without excess precision, and in the latter, rounding
7619 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7620 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7621 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7623 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7625 This option is not turned on by any @option{-O} option since
7626 it can result in incorrect output for programs which depend on
7627 an exact implementation of IEEE or ISO rules/specifications for
7628 math functions. It may, however, yield faster code for programs
7629 that do not require the guarantees of these specifications.
7631 @item -fno-math-errno
7632 @opindex fno-math-errno
7633 Do not set ERRNO after calling math functions that are executed
7634 with a single instruction, e.g., sqrt. A program that relies on
7635 IEEE exceptions for math error handling may want to use this flag
7636 for speed while maintaining IEEE arithmetic compatibility.
7638 This option is not turned on by any @option{-O} option since
7639 it can result in incorrect output for programs which depend on
7640 an exact implementation of IEEE or ISO rules/specifications for
7641 math functions. It may, however, yield faster code for programs
7642 that do not require the guarantees of these specifications.
7644 The default is @option{-fmath-errno}.
7646 On Darwin systems, the math library never sets @code{errno}. There is
7647 therefore no reason for the compiler to consider the possibility that
7648 it might, and @option{-fno-math-errno} is the default.
7650 @item -funsafe-math-optimizations
7651 @opindex funsafe-math-optimizations
7653 Allow optimizations for floating-point arithmetic that (a) assume
7654 that arguments and results are valid and (b) may violate IEEE or
7655 ANSI standards. When used at link-time, it may include libraries
7656 or startup files that change the default FPU control word or other
7657 similar optimizations.
7659 This option is not turned on by any @option{-O} option since
7660 it can result in incorrect output for programs which depend on
7661 an exact implementation of IEEE or ISO rules/specifications for
7662 math functions. It may, however, yield faster code for programs
7663 that do not require the guarantees of these specifications.
7664 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7665 @option{-fassociative-math} and @option{-freciprocal-math}.
7667 The default is @option{-fno-unsafe-math-optimizations}.
7669 @item -fassociative-math
7670 @opindex fassociative-math
7672 Allow re-association of operands in series of floating-point operations.
7673 This violates the ISO C and C++ language standard by possibly changing
7674 computation result. NOTE: re-ordering may change the sign of zero as
7675 well as ignore NaNs and inhibit or create underflow or overflow (and
7676 thus cannot be used on a code which relies on rounding behavior like
7677 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7678 and thus may not be used when ordered comparisons are required.
7679 This option requires that both @option{-fno-signed-zeros} and
7680 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7681 much sense with @option{-frounding-math}. For Fortran the option
7682 is automatically enabled when both @option{-fno-signed-zeros} and
7683 @option{-fno-trapping-math} are in effect.
7685 The default is @option{-fno-associative-math}.
7687 @item -freciprocal-math
7688 @opindex freciprocal-math
7690 Allow the reciprocal of a value to be used instead of dividing by
7691 the value if this enables optimizations. For example @code{x / y}
7692 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7693 is subject to common subexpression elimination. Note that this loses
7694 precision and increases the number of flops operating on the value.
7696 The default is @option{-fno-reciprocal-math}.
7698 @item -ffinite-math-only
7699 @opindex ffinite-math-only
7700 Allow optimizations for floating-point arithmetic that assume
7701 that arguments and results are not NaNs or +-Infs.
7703 This option is not turned on by any @option{-O} option since
7704 it can result in incorrect output for programs which depend on
7705 an exact implementation of IEEE or ISO rules/specifications for
7706 math functions. It may, however, yield faster code for programs
7707 that do not require the guarantees of these specifications.
7709 The default is @option{-fno-finite-math-only}.
7711 @item -fno-signed-zeros
7712 @opindex fno-signed-zeros
7713 Allow optimizations for floating point arithmetic that ignore the
7714 signedness of zero. IEEE arithmetic specifies the behavior of
7715 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7716 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7717 This option implies that the sign of a zero result isn't significant.
7719 The default is @option{-fsigned-zeros}.
7721 @item -fno-trapping-math
7722 @opindex fno-trapping-math
7723 Compile code assuming that floating-point operations cannot generate
7724 user-visible traps. These traps include division by zero, overflow,
7725 underflow, inexact result and invalid operation. This option requires
7726 that @option{-fno-signaling-nans} be in effect. Setting this option may
7727 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7729 This option should never be turned on by any @option{-O} option since
7730 it can result in incorrect output for programs which depend on
7731 an exact implementation of IEEE or ISO rules/specifications for
7734 The default is @option{-ftrapping-math}.
7736 @item -frounding-math
7737 @opindex frounding-math
7738 Disable transformations and optimizations that assume default floating
7739 point rounding behavior. This is round-to-zero for all floating point
7740 to integer conversions, and round-to-nearest for all other arithmetic
7741 truncations. This option should be specified for programs that change
7742 the FP rounding mode dynamically, or that may be executed with a
7743 non-default rounding mode. This option disables constant folding of
7744 floating point expressions at compile-time (which may be affected by
7745 rounding mode) and arithmetic transformations that are unsafe in the
7746 presence of sign-dependent rounding modes.
7748 The default is @option{-fno-rounding-math}.
7750 This option is experimental and does not currently guarantee to
7751 disable all GCC optimizations that are affected by rounding mode.
7752 Future versions of GCC may provide finer control of this setting
7753 using C99's @code{FENV_ACCESS} pragma. This command line option
7754 will be used to specify the default state for @code{FENV_ACCESS}.
7756 @item -fsignaling-nans
7757 @opindex fsignaling-nans
7758 Compile code assuming that IEEE signaling NaNs may generate user-visible
7759 traps during floating-point operations. Setting this option disables
7760 optimizations that may change the number of exceptions visible with
7761 signaling NaNs. This option implies @option{-ftrapping-math}.
7763 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7766 The default is @option{-fno-signaling-nans}.
7768 This option is experimental and does not currently guarantee to
7769 disable all GCC optimizations that affect signaling NaN behavior.
7771 @item -fsingle-precision-constant
7772 @opindex fsingle-precision-constant
7773 Treat floating point constant as single precision constant instead of
7774 implicitly converting it to double precision constant.
7776 @item -fcx-limited-range
7777 @opindex fcx-limited-range
7778 When enabled, this option states that a range reduction step is not
7779 needed when performing complex division. Also, there is no checking
7780 whether the result of a complex multiplication or division is @code{NaN
7781 + I*NaN}, with an attempt to rescue the situation in that case. The
7782 default is @option{-fno-cx-limited-range}, but is enabled by
7783 @option{-ffast-math}.
7785 This option controls the default setting of the ISO C99
7786 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7789 @item -fcx-fortran-rules
7790 @opindex fcx-fortran-rules
7791 Complex multiplication and division follow Fortran rules. Range
7792 reduction is done as part of complex division, but there is no checking
7793 whether the result of a complex multiplication or division is @code{NaN
7794 + I*NaN}, with an attempt to rescue the situation in that case.
7796 The default is @option{-fno-cx-fortran-rules}.
7800 The following options control optimizations that may improve
7801 performance, but are not enabled by any @option{-O} options. This
7802 section includes experimental options that may produce broken code.
7805 @item -fbranch-probabilities
7806 @opindex fbranch-probabilities
7807 After running a program compiled with @option{-fprofile-arcs}
7808 (@pxref{Debugging Options,, Options for Debugging Your Program or
7809 @command{gcc}}), you can compile it a second time using
7810 @option{-fbranch-probabilities}, to improve optimizations based on
7811 the number of times each branch was taken. When the program
7812 compiled with @option{-fprofile-arcs} exits it saves arc execution
7813 counts to a file called @file{@var{sourcename}.gcda} for each source
7814 file. The information in this data file is very dependent on the
7815 structure of the generated code, so you must use the same source code
7816 and the same optimization options for both compilations.
7818 With @option{-fbranch-probabilities}, GCC puts a
7819 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7820 These can be used to improve optimization. Currently, they are only
7821 used in one place: in @file{reorg.c}, instead of guessing which path a
7822 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7823 exactly determine which path is taken more often.
7825 @item -fprofile-values
7826 @opindex fprofile-values
7827 If combined with @option{-fprofile-arcs}, it adds code so that some
7828 data about values of expressions in the program is gathered.
7830 With @option{-fbranch-probabilities}, it reads back the data gathered
7831 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7832 notes to instructions for their later usage in optimizations.
7834 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7838 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7839 a code to gather information about values of expressions.
7841 With @option{-fbranch-probabilities}, it reads back the data gathered
7842 and actually performs the optimizations based on them.
7843 Currently the optimizations include specialization of division operation
7844 using the knowledge about the value of the denominator.
7846 @item -frename-registers
7847 @opindex frename-registers
7848 Attempt to avoid false dependencies in scheduled code by making use
7849 of registers left over after register allocation. This optimization
7850 will most benefit processors with lots of registers. Depending on the
7851 debug information format adopted by the target, however, it can
7852 make debugging impossible, since variables will no longer stay in
7853 a ``home register''.
7855 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
7859 Perform tail duplication to enlarge superblock size. This transformation
7860 simplifies the control flow of the function allowing other optimizations to do
7863 Enabled with @option{-fprofile-use}.
7865 @item -funroll-loops
7866 @opindex funroll-loops
7867 Unroll loops whose number of iterations can be determined at compile time or
7868 upon entry to the loop. @option{-funroll-loops} implies
7869 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7870 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7871 small constant number of iterations). This option makes code larger, and may
7872 or may not make it run faster.
7874 Enabled with @option{-fprofile-use}.
7876 @item -funroll-all-loops
7877 @opindex funroll-all-loops
7878 Unroll all loops, even if their number of iterations is uncertain when
7879 the loop is entered. This usually makes programs run more slowly.
7880 @option{-funroll-all-loops} implies the same options as
7881 @option{-funroll-loops}.
7884 @opindex fpeel-loops
7885 Peels the loops for that there is enough information that they do not
7886 roll much (from profile feedback). It also turns on complete loop peeling
7887 (i.e.@: complete removal of loops with small constant number of iterations).
7889 Enabled with @option{-fprofile-use}.
7891 @item -fmove-loop-invariants
7892 @opindex fmove-loop-invariants
7893 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7894 at level @option{-O1}
7896 @item -funswitch-loops
7897 @opindex funswitch-loops
7898 Move branches with loop invariant conditions out of the loop, with duplicates
7899 of the loop on both branches (modified according to result of the condition).
7901 @item -ffunction-sections
7902 @itemx -fdata-sections
7903 @opindex ffunction-sections
7904 @opindex fdata-sections
7905 Place each function or data item into its own section in the output
7906 file if the target supports arbitrary sections. The name of the
7907 function or the name of the data item determines the section's name
7910 Use these options on systems where the linker can perform optimizations
7911 to improve locality of reference in the instruction space. Most systems
7912 using the ELF object format and SPARC processors running Solaris 2 have
7913 linkers with such optimizations. AIX may have these optimizations in
7916 Only use these options when there are significant benefits from doing
7917 so. When you specify these options, the assembler and linker will
7918 create larger object and executable files and will also be slower.
7919 You will not be able to use @code{gprof} on all systems if you
7920 specify this option and you may have problems with debugging if
7921 you specify both this option and @option{-g}.
7923 @item -fbranch-target-load-optimize
7924 @opindex fbranch-target-load-optimize
7925 Perform branch target register load optimization before prologue / epilogue
7927 The use of target registers can typically be exposed only during reload,
7928 thus hoisting loads out of loops and doing inter-block scheduling needs
7929 a separate optimization pass.
7931 @item -fbranch-target-load-optimize2
7932 @opindex fbranch-target-load-optimize2
7933 Perform branch target register load optimization after prologue / epilogue
7936 @item -fbtr-bb-exclusive
7937 @opindex fbtr-bb-exclusive
7938 When performing branch target register load optimization, don't reuse
7939 branch target registers in within any basic block.
7941 @item -fstack-protector
7942 @opindex fstack-protector
7943 Emit extra code to check for buffer overflows, such as stack smashing
7944 attacks. This is done by adding a guard variable to functions with
7945 vulnerable objects. This includes functions that call alloca, and
7946 functions with buffers larger than 8 bytes. The guards are initialized
7947 when a function is entered and then checked when the function exits.
7948 If a guard check fails, an error message is printed and the program exits.
7950 @item -fstack-protector-all
7951 @opindex fstack-protector-all
7952 Like @option{-fstack-protector} except that all functions are protected.
7954 @item -fsection-anchors
7955 @opindex fsection-anchors
7956 Try to reduce the number of symbolic address calculations by using
7957 shared ``anchor'' symbols to address nearby objects. This transformation
7958 can help to reduce the number of GOT entries and GOT accesses on some
7961 For example, the implementation of the following function @code{foo}:
7965 int foo (void) @{ return a + b + c; @}
7968 would usually calculate the addresses of all three variables, but if you
7969 compile it with @option{-fsection-anchors}, it will access the variables
7970 from a common anchor point instead. The effect is similar to the
7971 following pseudocode (which isn't valid C):
7976 register int *xr = &x;
7977 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7981 Not all targets support this option.
7983 @item --param @var{name}=@var{value}
7985 In some places, GCC uses various constants to control the amount of
7986 optimization that is done. For example, GCC will not inline functions
7987 that contain more that a certain number of instructions. You can
7988 control some of these constants on the command-line using the
7989 @option{--param} option.
7991 The names of specific parameters, and the meaning of the values, are
7992 tied to the internals of the compiler, and are subject to change
7993 without notice in future releases.
7995 In each case, the @var{value} is an integer. The allowable choices for
7996 @var{name} are given in the following table:
7999 @item struct-reorg-cold-struct-ratio
8000 The threshold ratio (as a percentage) between a structure frequency
8001 and the frequency of the hottest structure in the program. This parameter
8002 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
8003 We say that if the ratio of a structure frequency, calculated by profiling,
8004 to the hottest structure frequency in the program is less than this
8005 parameter, then structure reorganization is not applied to this structure.
8008 @item predictable-branch-outcome
8009 When branch is predicted to be taken with probability lower than this threshold
8010 (in percent), then it is considered well predictable. The default is 10.
8012 @item max-crossjump-edges
8013 The maximum number of incoming edges to consider for crossjumping.
8014 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8015 the number of edges incoming to each block. Increasing values mean
8016 more aggressive optimization, making the compile time increase with
8017 probably small improvement in executable size.
8019 @item min-crossjump-insns
8020 The minimum number of instructions which must be matched at the end
8021 of two blocks before crossjumping will be performed on them. This
8022 value is ignored in the case where all instructions in the block being
8023 crossjumped from are matched. The default value is 5.
8025 @item max-grow-copy-bb-insns
8026 The maximum code size expansion factor when copying basic blocks
8027 instead of jumping. The expansion is relative to a jump instruction.
8028 The default value is 8.
8030 @item max-goto-duplication-insns
8031 The maximum number of instructions to duplicate to a block that jumps
8032 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
8033 passes, GCC factors computed gotos early in the compilation process,
8034 and unfactors them as late as possible. Only computed jumps at the
8035 end of a basic blocks with no more than max-goto-duplication-insns are
8036 unfactored. The default value is 8.
8038 @item max-delay-slot-insn-search
8039 The maximum number of instructions to consider when looking for an
8040 instruction to fill a delay slot. If more than this arbitrary number of
8041 instructions is searched, the time savings from filling the delay slot
8042 will be minimal so stop searching. Increasing values mean more
8043 aggressive optimization, making the compile time increase with probably
8044 small improvement in executable run time.
8046 @item max-delay-slot-live-search
8047 When trying to fill delay slots, the maximum number of instructions to
8048 consider when searching for a block with valid live register
8049 information. Increasing this arbitrarily chosen value means more
8050 aggressive optimization, increasing the compile time. This parameter
8051 should be removed when the delay slot code is rewritten to maintain the
8054 @item max-gcse-memory
8055 The approximate maximum amount of memory that will be allocated in
8056 order to perform the global common subexpression elimination
8057 optimization. If more memory than specified is required, the
8058 optimization will not be done.
8060 @item max-pending-list-length
8061 The maximum number of pending dependencies scheduling will allow
8062 before flushing the current state and starting over. Large functions
8063 with few branches or calls can create excessively large lists which
8064 needlessly consume memory and resources.
8066 @item max-inline-insns-single
8067 Several parameters control the tree inliner used in gcc.
8068 This number sets the maximum number of instructions (counted in GCC's
8069 internal representation) in a single function that the tree inliner
8070 will consider for inlining. This only affects functions declared
8071 inline and methods implemented in a class declaration (C++).
8072 The default value is 300.
8074 @item max-inline-insns-auto
8075 When you use @option{-finline-functions} (included in @option{-O3}),
8076 a lot of functions that would otherwise not be considered for inlining
8077 by the compiler will be investigated. To those functions, a different
8078 (more restrictive) limit compared to functions declared inline can
8080 The default value is 50.
8082 @item large-function-insns
8083 The limit specifying really large functions. For functions larger than this
8084 limit after inlining, inlining is constrained by
8085 @option{--param large-function-growth}. This parameter is useful primarily
8086 to avoid extreme compilation time caused by non-linear algorithms used by the
8088 The default value is 2700.
8090 @item large-function-growth
8091 Specifies maximal growth of large function caused by inlining in percents.
8092 The default value is 100 which limits large function growth to 2.0 times
8095 @item large-unit-insns
8096 The limit specifying large translation unit. Growth caused by inlining of
8097 units larger than this limit is limited by @option{--param inline-unit-growth}.
8098 For small units this might be too tight (consider unit consisting of function A
8099 that is inline and B that just calls A three time. If B is small relative to
8100 A, the growth of unit is 300\% and yet such inlining is very sane. For very
8101 large units consisting of small inlineable functions however the overall unit
8102 growth limit is needed to avoid exponential explosion of code size. Thus for
8103 smaller units, the size is increased to @option{--param large-unit-insns}
8104 before applying @option{--param inline-unit-growth}. The default is 10000
8106 @item inline-unit-growth
8107 Specifies maximal overall growth of the compilation unit caused by inlining.
8108 The default value is 30 which limits unit growth to 1.3 times the original
8111 @item ipcp-unit-growth
8112 Specifies maximal overall growth of the compilation unit caused by
8113 interprocedural constant propagation. The default value is 10 which limits
8114 unit growth to 1.1 times the original size.
8116 @item large-stack-frame
8117 The limit specifying large stack frames. While inlining the algorithm is trying
8118 to not grow past this limit too much. Default value is 256 bytes.
8120 @item large-stack-frame-growth
8121 Specifies maximal growth of large stack frames caused by inlining in percents.
8122 The default value is 1000 which limits large stack frame growth to 11 times
8125 @item max-inline-insns-recursive
8126 @itemx max-inline-insns-recursive-auto
8127 Specifies maximum number of instructions out-of-line copy of self recursive inline
8128 function can grow into by performing recursive inlining.
8130 For functions declared inline @option{--param max-inline-insns-recursive} is
8131 taken into account. For function not declared inline, recursive inlining
8132 happens only when @option{-finline-functions} (included in @option{-O3}) is
8133 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8134 default value is 450.
8136 @item max-inline-recursive-depth
8137 @itemx max-inline-recursive-depth-auto
8138 Specifies maximum recursion depth used by the recursive inlining.
8140 For functions declared inline @option{--param max-inline-recursive-depth} is
8141 taken into account. For function not declared inline, recursive inlining
8142 happens only when @option{-finline-functions} (included in @option{-O3}) is
8143 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8146 @item min-inline-recursive-probability
8147 Recursive inlining is profitable only for function having deep recursion
8148 in average and can hurt for function having little recursion depth by
8149 increasing the prologue size or complexity of function body to other
8152 When profile feedback is available (see @option{-fprofile-generate}) the actual
8153 recursion depth can be guessed from probability that function will recurse via
8154 given call expression. This parameter limits inlining only to call expression
8155 whose probability exceeds given threshold (in percents). The default value is
8158 @item early-inlining-insns
8159 Specify growth that early inliner can make. In effect it increases amount of
8160 inlining for code having large abstraction penalty. The default value is 8.
8162 @item max-early-inliner-iterations
8163 @itemx max-early-inliner-iterations
8164 Limit of iterations of early inliner. This basically bounds number of nested
8165 indirect calls early inliner can resolve. Deeper chains are still handled by
8168 @item min-vect-loop-bound
8169 The minimum number of iterations under which a loop will not get vectorized
8170 when @option{-ftree-vectorize} is used. The number of iterations after
8171 vectorization needs to be greater than the value specified by this option
8172 to allow vectorization. The default value is 0.
8174 @item max-unrolled-insns
8175 The maximum number of instructions that a loop should have if that loop
8176 is unrolled, and if the loop is unrolled, it determines how many times
8177 the loop code is unrolled.
8179 @item max-average-unrolled-insns
8180 The maximum number of instructions biased by probabilities of their execution
8181 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8182 it determines how many times the loop code is unrolled.
8184 @item max-unroll-times
8185 The maximum number of unrollings of a single loop.
8187 @item max-peeled-insns
8188 The maximum number of instructions that a loop should have if that loop
8189 is peeled, and if the loop is peeled, it determines how many times
8190 the loop code is peeled.
8192 @item max-peel-times
8193 The maximum number of peelings of a single loop.
8195 @item max-completely-peeled-insns
8196 The maximum number of insns of a completely peeled loop.
8198 @item max-completely-peel-times
8199 The maximum number of iterations of a loop to be suitable for complete peeling.
8201 @item max-completely-peel-loop-nest-depth
8202 The maximum depth of a loop nest suitable for complete peeling.
8204 @item max-unswitch-insns
8205 The maximum number of insns of an unswitched loop.
8207 @item max-unswitch-level
8208 The maximum number of branches unswitched in a single loop.
8211 The minimum cost of an expensive expression in the loop invariant motion.
8213 @item iv-consider-all-candidates-bound
8214 Bound on number of candidates for induction variables below that
8215 all candidates are considered for each use in induction variable
8216 optimizations. Only the most relevant candidates are considered
8217 if there are more candidates, to avoid quadratic time complexity.
8219 @item iv-max-considered-uses
8220 The induction variable optimizations give up on loops that contain more
8221 induction variable uses.
8223 @item iv-always-prune-cand-set-bound
8224 If number of candidates in the set is smaller than this value,
8225 we always try to remove unnecessary ivs from the set during its
8226 optimization when a new iv is added to the set.
8228 @item scev-max-expr-size
8229 Bound on size of expressions used in the scalar evolutions analyzer.
8230 Large expressions slow the analyzer.
8232 @item omega-max-vars
8233 The maximum number of variables in an Omega constraint system.
8234 The default value is 128.
8236 @item omega-max-geqs
8237 The maximum number of inequalities in an Omega constraint system.
8238 The default value is 256.
8241 The maximum number of equalities in an Omega constraint system.
8242 The default value is 128.
8244 @item omega-max-wild-cards
8245 The maximum number of wildcard variables that the Omega solver will
8246 be able to insert. The default value is 18.
8248 @item omega-hash-table-size
8249 The size of the hash table in the Omega solver. The default value is
8252 @item omega-max-keys
8253 The maximal number of keys used by the Omega solver. The default
8256 @item omega-eliminate-redundant-constraints
8257 When set to 1, use expensive methods to eliminate all redundant
8258 constraints. The default value is 0.
8260 @item vect-max-version-for-alignment-checks
8261 The maximum number of runtime checks that can be performed when
8262 doing loop versioning for alignment in the vectorizer. See option
8263 ftree-vect-loop-version for more information.
8265 @item vect-max-version-for-alias-checks
8266 The maximum number of runtime checks that can be performed when
8267 doing loop versioning for alias in the vectorizer. See option
8268 ftree-vect-loop-version for more information.
8270 @item max-iterations-to-track
8272 The maximum number of iterations of a loop the brute force algorithm
8273 for analysis of # of iterations of the loop tries to evaluate.
8275 @item hot-bb-count-fraction
8276 Select fraction of the maximal count of repetitions of basic block in program
8277 given basic block needs to have to be considered hot.
8279 @item hot-bb-frequency-fraction
8280 Select fraction of the maximal frequency of executions of basic block in
8281 function given basic block needs to have to be considered hot
8283 @item max-predicted-iterations
8284 The maximum number of loop iterations we predict statically. This is useful
8285 in cases where function contain single loop with known bound and other loop
8286 with unknown. We predict the known number of iterations correctly, while
8287 the unknown number of iterations average to roughly 10. This means that the
8288 loop without bounds would appear artificially cold relative to the other one.
8290 @item align-threshold
8292 Select fraction of the maximal frequency of executions of basic block in
8293 function given basic block will get aligned.
8295 @item align-loop-iterations
8297 A loop expected to iterate at lest the selected number of iterations will get
8300 @item tracer-dynamic-coverage
8301 @itemx tracer-dynamic-coverage-feedback
8303 This value is used to limit superblock formation once the given percentage of
8304 executed instructions is covered. This limits unnecessary code size
8307 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8308 feedback is available. The real profiles (as opposed to statically estimated
8309 ones) are much less balanced allowing the threshold to be larger value.
8311 @item tracer-max-code-growth
8312 Stop tail duplication once code growth has reached given percentage. This is
8313 rather hokey argument, as most of the duplicates will be eliminated later in
8314 cross jumping, so it may be set to much higher values than is the desired code
8317 @item tracer-min-branch-ratio
8319 Stop reverse growth when the reverse probability of best edge is less than this
8320 threshold (in percent).
8322 @item tracer-min-branch-ratio
8323 @itemx tracer-min-branch-ratio-feedback
8325 Stop forward growth if the best edge do have probability lower than this
8328 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8329 compilation for profile feedback and one for compilation without. The value
8330 for compilation with profile feedback needs to be more conservative (higher) in
8331 order to make tracer effective.
8333 @item max-cse-path-length
8335 Maximum number of basic blocks on path that cse considers. The default is 10.
8338 The maximum instructions CSE process before flushing. The default is 1000.
8340 @item ggc-min-expand
8342 GCC uses a garbage collector to manage its own memory allocation. This
8343 parameter specifies the minimum percentage by which the garbage
8344 collector's heap should be allowed to expand between collections.
8345 Tuning this may improve compilation speed; it has no effect on code
8348 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8349 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8350 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8351 GCC is not able to calculate RAM on a particular platform, the lower
8352 bound of 30% is used. Setting this parameter and
8353 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8354 every opportunity. This is extremely slow, but can be useful for
8357 @item ggc-min-heapsize
8359 Minimum size of the garbage collector's heap before it begins bothering
8360 to collect garbage. The first collection occurs after the heap expands
8361 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8362 tuning this may improve compilation speed, and has no effect on code
8365 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8366 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8367 with a lower bound of 4096 (four megabytes) and an upper bound of
8368 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8369 particular platform, the lower bound is used. Setting this parameter
8370 very large effectively disables garbage collection. Setting this
8371 parameter and @option{ggc-min-expand} to zero causes a full collection
8372 to occur at every opportunity.
8374 @item max-reload-search-insns
8375 The maximum number of instruction reload should look backward for equivalent
8376 register. Increasing values mean more aggressive optimization, making the
8377 compile time increase with probably slightly better performance. The default
8380 @item max-cselib-memory-locations
8381 The maximum number of memory locations cselib should take into account.
8382 Increasing values mean more aggressive optimization, making the compile time
8383 increase with probably slightly better performance. The default value is 500.
8385 @item reorder-blocks-duplicate
8386 @itemx reorder-blocks-duplicate-feedback
8388 Used by basic block reordering pass to decide whether to use unconditional
8389 branch or duplicate the code on its destination. Code is duplicated when its
8390 estimated size is smaller than this value multiplied by the estimated size of
8391 unconditional jump in the hot spots of the program.
8393 The @option{reorder-block-duplicate-feedback} is used only when profile
8394 feedback is available and may be set to higher values than
8395 @option{reorder-block-duplicate} since information about the hot spots is more
8398 @item max-sched-ready-insns
8399 The maximum number of instructions ready to be issued the scheduler should
8400 consider at any given time during the first scheduling pass. Increasing
8401 values mean more thorough searches, making the compilation time increase
8402 with probably little benefit. The default value is 100.
8404 @item max-sched-region-blocks
8405 The maximum number of blocks in a region to be considered for
8406 interblock scheduling. The default value is 10.
8408 @item max-pipeline-region-blocks
8409 The maximum number of blocks in a region to be considered for
8410 pipelining in the selective scheduler. The default value is 15.
8412 @item max-sched-region-insns
8413 The maximum number of insns in a region to be considered for
8414 interblock scheduling. The default value is 100.
8416 @item max-pipeline-region-insns
8417 The maximum number of insns in a region to be considered for
8418 pipelining in the selective scheduler. The default value is 200.
8421 The minimum probability (in percents) of reaching a source block
8422 for interblock speculative scheduling. The default value is 40.
8424 @item max-sched-extend-regions-iters
8425 The maximum number of iterations through CFG to extend regions.
8426 0 - disable region extension,
8427 N - do at most N iterations.
8428 The default value is 0.
8430 @item max-sched-insn-conflict-delay
8431 The maximum conflict delay for an insn to be considered for speculative motion.
8432 The default value is 3.
8434 @item sched-spec-prob-cutoff
8435 The minimal probability of speculation success (in percents), so that
8436 speculative insn will be scheduled.
8437 The default value is 40.
8439 @item sched-mem-true-dep-cost
8440 Minimal distance (in CPU cycles) between store and load targeting same
8441 memory locations. The default value is 1.
8443 @item selsched-max-lookahead
8444 The maximum size of the lookahead window of selective scheduling. It is a
8445 depth of search for available instructions.
8446 The default value is 50.
8448 @item selsched-max-sched-times
8449 The maximum number of times that an instruction will be scheduled during
8450 selective scheduling. This is the limit on the number of iterations
8451 through which the instruction may be pipelined. The default value is 2.
8453 @item selsched-max-insns-to-rename
8454 The maximum number of best instructions in the ready list that are considered
8455 for renaming in the selective scheduler. The default value is 2.
8457 @item max-last-value-rtl
8458 The maximum size measured as number of RTLs that can be recorded in an expression
8459 in combiner for a pseudo register as last known value of that register. The default
8462 @item integer-share-limit
8463 Small integer constants can use a shared data structure, reducing the
8464 compiler's memory usage and increasing its speed. This sets the maximum
8465 value of a shared integer constant. The default value is 256.
8467 @item min-virtual-mappings
8468 Specifies the minimum number of virtual mappings in the incremental
8469 SSA updater that should be registered to trigger the virtual mappings
8470 heuristic defined by virtual-mappings-ratio. The default value is
8473 @item virtual-mappings-ratio
8474 If the number of virtual mappings is virtual-mappings-ratio bigger
8475 than the number of virtual symbols to be updated, then the incremental
8476 SSA updater switches to a full update for those symbols. The default
8479 @item ssp-buffer-size
8480 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8481 protection when @option{-fstack-protection} is used.
8483 @item max-jump-thread-duplication-stmts
8484 Maximum number of statements allowed in a block that needs to be
8485 duplicated when threading jumps.
8487 @item max-fields-for-field-sensitive
8488 Maximum number of fields in a structure we will treat in
8489 a field sensitive manner during pointer analysis. The default is zero
8490 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8492 @item prefetch-latency
8493 Estimate on average number of instructions that are executed before
8494 prefetch finishes. The distance we prefetch ahead is proportional
8495 to this constant. Increasing this number may also lead to less
8496 streams being prefetched (see @option{simultaneous-prefetches}).
8498 @item simultaneous-prefetches
8499 Maximum number of prefetches that can run at the same time.
8501 @item l1-cache-line-size
8502 The size of cache line in L1 cache, in bytes.
8505 The size of L1 cache, in kilobytes.
8508 The size of L2 cache, in kilobytes.
8510 @item min-insn-to-prefetch-ratio
8511 The minimum ratio between the number of instructions and the
8512 number of prefetches to enable prefetching in a loop with an
8515 @item prefetch-min-insn-to-mem-ratio
8516 The minimum ratio between the number of instructions and the
8517 number of memory references to enable prefetching in a loop.
8519 @item use-canonical-types
8520 Whether the compiler should use the ``canonical'' type system. By
8521 default, this should always be 1, which uses a more efficient internal
8522 mechanism for comparing types in C++ and Objective-C++. However, if
8523 bugs in the canonical type system are causing compilation failures,
8524 set this value to 0 to disable canonical types.
8526 @item switch-conversion-max-branch-ratio
8527 Switch initialization conversion will refuse to create arrays that are
8528 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8529 branches in the switch.
8531 @item max-partial-antic-length
8532 Maximum length of the partial antic set computed during the tree
8533 partial redundancy elimination optimization (@option{-ftree-pre}) when
8534 optimizing at @option{-O3} and above. For some sorts of source code
8535 the enhanced partial redundancy elimination optimization can run away,
8536 consuming all of the memory available on the host machine. This
8537 parameter sets a limit on the length of the sets that are computed,
8538 which prevents the runaway behavior. Setting a value of 0 for
8539 this parameter will allow an unlimited set length.
8541 @item sccvn-max-scc-size
8542 Maximum size of a strongly connected component (SCC) during SCCVN
8543 processing. If this limit is hit, SCCVN processing for the whole
8544 function will not be done and optimizations depending on it will
8545 be disabled. The default maximum SCC size is 10000.
8547 @item ira-max-loops-num
8548 IRA uses a regional register allocation by default. If a function
8549 contains loops more than number given by the parameter, only at most
8550 given number of the most frequently executed loops will form regions
8551 for the regional register allocation. The default value of the
8554 @item ira-max-conflict-table-size
8555 Although IRA uses a sophisticated algorithm of compression conflict
8556 table, the table can be still big for huge functions. If the conflict
8557 table for a function could be more than size in MB given by the
8558 parameter, the conflict table is not built and faster, simpler, and
8559 lower quality register allocation algorithm will be used. The
8560 algorithm do not use pseudo-register conflicts. The default value of
8561 the parameter is 2000.
8563 @item ira-loop-reserved-regs
8564 IRA can be used to evaluate more accurate register pressure in loops
8565 for decision to move loop invariants (see @option{-O3}). The number
8566 of available registers reserved for some other purposes is described
8567 by this parameter. The default value of the parameter is 2 which is
8568 minimal number of registers needed for execution of typical
8569 instruction. This value is the best found from numerous experiments.
8571 @item loop-invariant-max-bbs-in-loop
8572 Loop invariant motion can be very expensive, both in compile time and
8573 in amount of needed compile time memory, with very large loops. Loops
8574 with more basic blocks than this parameter won't have loop invariant
8575 motion optimization performed on them. The default value of the
8576 parameter is 1000 for -O1 and 10000 for -O2 and above.
8578 @item max-vartrack-size
8579 Sets a maximum number of hash table slots to use during variable
8580 tracking dataflow analysis of any function. If this limit is exceeded
8581 with variable tracking at assignments enabled, analysis for that
8582 function is retried without it, after removing all debug insns from
8583 the function. If the limit is exceeded even without debug insns, var
8584 tracking analysis is completely disabled for the function. Setting
8585 the parameter to zero makes it unlimited.
8587 @item min-nondebug-insn-uid
8588 Use uids starting at this parameter for nondebug insns. The range below
8589 the parameter is reserved exclusively for debug insns created by
8590 @option{-fvar-tracking-assignments}, but debug insns may get
8591 (non-overlapping) uids above it if the reserved range is exhausted.
8593 @item ipa-sra-ptr-growth-factor
8594 IPA-SRA will replace a pointer to an aggregate with one or more new
8595 parameters only when their cumulative size is less or equal to
8596 @option{ipa-sra-ptr-growth-factor} times the size of the original
8599 @item graphite-max-nb-scop-params
8600 To avoid exponential effects in the Graphite loop transforms, the
8601 number of parameters in a Static Control Part (SCoP) is bounded. The
8602 default value is 10 parameters. A variable whose value is unknown at
8603 compile time and defined outside a SCoP is a parameter of the SCoP.
8605 @item graphite-max-bbs-per-function
8606 To avoid exponential effects in the detection of SCoPs, the size of
8607 the functions analyzed by Graphite is bounded. The default value is
8610 @item loop-block-tile-size
8611 Loop blocking or strip mining transforms, enabled with
8612 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
8613 loop in the loop nest by a given number of iterations. The strip
8614 length can be changed using the @option{loop-block-tile-size}
8615 parameter. The default value is 51 iterations.
8620 @node Preprocessor Options
8621 @section Options Controlling the Preprocessor
8622 @cindex preprocessor options
8623 @cindex options, preprocessor
8625 These options control the C preprocessor, which is run on each C source
8626 file before actual compilation.
8628 If you use the @option{-E} option, nothing is done except preprocessing.
8629 Some of these options make sense only together with @option{-E} because
8630 they cause the preprocessor output to be unsuitable for actual
8634 @item -Wp,@var{option}
8636 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8637 and pass @var{option} directly through to the preprocessor. If
8638 @var{option} contains commas, it is split into multiple options at the
8639 commas. However, many options are modified, translated or interpreted
8640 by the compiler driver before being passed to the preprocessor, and
8641 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8642 interface is undocumented and subject to change, so whenever possible
8643 you should avoid using @option{-Wp} and let the driver handle the
8646 @item -Xpreprocessor @var{option}
8647 @opindex Xpreprocessor
8648 Pass @var{option} as an option to the preprocessor. You can use this to
8649 supply system-specific preprocessor options which GCC does not know how to
8652 If you want to pass an option that takes an argument, you must use
8653 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8656 @include cppopts.texi
8658 @node Assembler Options
8659 @section Passing Options to the Assembler
8661 @c prevent bad page break with this line
8662 You can pass options to the assembler.
8665 @item -Wa,@var{option}
8667 Pass @var{option} as an option to the assembler. If @var{option}
8668 contains commas, it is split into multiple options at the commas.
8670 @item -Xassembler @var{option}
8672 Pass @var{option} as an option to the assembler. You can use this to
8673 supply system-specific assembler options which GCC does not know how to
8676 If you want to pass an option that takes an argument, you must use
8677 @option{-Xassembler} twice, once for the option and once for the argument.
8682 @section Options for Linking
8683 @cindex link options
8684 @cindex options, linking
8686 These options come into play when the compiler links object files into
8687 an executable output file. They are meaningless if the compiler is
8688 not doing a link step.
8692 @item @var{object-file-name}
8693 A file name that does not end in a special recognized suffix is
8694 considered to name an object file or library. (Object files are
8695 distinguished from libraries by the linker according to the file
8696 contents.) If linking is done, these object files are used as input
8705 If any of these options is used, then the linker is not run, and
8706 object file names should not be used as arguments. @xref{Overall
8710 @item -l@var{library}
8711 @itemx -l @var{library}
8713 Search the library named @var{library} when linking. (The second
8714 alternative with the library as a separate argument is only for
8715 POSIX compliance and is not recommended.)
8717 It makes a difference where in the command you write this option; the
8718 linker searches and processes libraries and object files in the order they
8719 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8720 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8721 to functions in @samp{z}, those functions may not be loaded.
8723 The linker searches a standard list of directories for the library,
8724 which is actually a file named @file{lib@var{library}.a}. The linker
8725 then uses this file as if it had been specified precisely by name.
8727 The directories searched include several standard system directories
8728 plus any that you specify with @option{-L}.
8730 Normally the files found this way are library files---archive files
8731 whose members are object files. The linker handles an archive file by
8732 scanning through it for members which define symbols that have so far
8733 been referenced but not defined. But if the file that is found is an
8734 ordinary object file, it is linked in the usual fashion. The only
8735 difference between using an @option{-l} option and specifying a file name
8736 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8737 and searches several directories.
8741 You need this special case of the @option{-l} option in order to
8742 link an Objective-C or Objective-C++ program.
8745 @opindex nostartfiles
8746 Do not use the standard system startup files when linking.
8747 The standard system libraries are used normally, unless @option{-nostdlib}
8748 or @option{-nodefaultlibs} is used.
8750 @item -nodefaultlibs
8751 @opindex nodefaultlibs
8752 Do not use the standard system libraries when linking.
8753 Only the libraries you specify will be passed to the linker, options
8754 specifying linkage of the system libraries, such as @code{-static-libgcc}
8755 or @code{-shared-libgcc}, will be ignored.
8756 The standard startup files are used normally, unless @option{-nostartfiles}
8757 is used. The compiler may generate calls to @code{memcmp},
8758 @code{memset}, @code{memcpy} and @code{memmove}.
8759 These entries are usually resolved by entries in
8760 libc. These entry points should be supplied through some other
8761 mechanism when this option is specified.
8765 Do not use the standard system startup files or libraries when linking.
8766 No startup files and only the libraries you specify will be passed to
8767 the linker, options specifying linkage of the system libraries, such as
8768 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8769 The compiler may generate calls to @code{memcmp}, @code{memset},
8770 @code{memcpy} and @code{memmove}.
8771 These entries are usually resolved by entries in
8772 libc. These entry points should be supplied through some other
8773 mechanism when this option is specified.
8775 @cindex @option{-lgcc}, use with @option{-nostdlib}
8776 @cindex @option{-nostdlib} and unresolved references
8777 @cindex unresolved references and @option{-nostdlib}
8778 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8779 @cindex @option{-nodefaultlibs} and unresolved references
8780 @cindex unresolved references and @option{-nodefaultlibs}
8781 One of the standard libraries bypassed by @option{-nostdlib} and
8782 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8783 that GCC uses to overcome shortcomings of particular machines, or special
8784 needs for some languages.
8785 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8786 Collection (GCC) Internals},
8787 for more discussion of @file{libgcc.a}.)
8788 In most cases, you need @file{libgcc.a} even when you want to avoid
8789 other standard libraries. In other words, when you specify @option{-nostdlib}
8790 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8791 This ensures that you have no unresolved references to internal GCC
8792 library subroutines. (For example, @samp{__main}, used to ensure C++
8793 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8794 GNU Compiler Collection (GCC) Internals}.)
8798 Produce a position independent executable on targets which support it.
8799 For predictable results, you must also specify the same set of options
8800 that were used to generate code (@option{-fpie}, @option{-fPIE},
8801 or model suboptions) when you specify this option.
8805 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8806 that support it. This instructs the linker to add all symbols, not
8807 only used ones, to the dynamic symbol table. This option is needed
8808 for some uses of @code{dlopen} or to allow obtaining backtraces
8809 from within a program.
8813 Remove all symbol table and relocation information from the executable.
8817 On systems that support dynamic linking, this prevents linking with the shared
8818 libraries. On other systems, this option has no effect.
8822 Produce a shared object which can then be linked with other objects to
8823 form an executable. Not all systems support this option. For predictable
8824 results, you must also specify the same set of options that were used to
8825 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8826 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8827 needs to build supplementary stub code for constructors to work. On
8828 multi-libbed systems, @samp{gcc -shared} must select the correct support
8829 libraries to link against. Failing to supply the correct flags may lead
8830 to subtle defects. Supplying them in cases where they are not necessary
8833 @item -shared-libgcc
8834 @itemx -static-libgcc
8835 @opindex shared-libgcc
8836 @opindex static-libgcc
8837 On systems that provide @file{libgcc} as a shared library, these options
8838 force the use of either the shared or static version respectively.
8839 If no shared version of @file{libgcc} was built when the compiler was
8840 configured, these options have no effect.
8842 There are several situations in which an application should use the
8843 shared @file{libgcc} instead of the static version. The most common
8844 of these is when the application wishes to throw and catch exceptions
8845 across different shared libraries. In that case, each of the libraries
8846 as well as the application itself should use the shared @file{libgcc}.
8848 Therefore, the G++ and GCJ drivers automatically add
8849 @option{-shared-libgcc} whenever you build a shared library or a main
8850 executable, because C++ and Java programs typically use exceptions, so
8851 this is the right thing to do.
8853 If, instead, you use the GCC driver to create shared libraries, you may
8854 find that they will not always be linked with the shared @file{libgcc}.
8855 If GCC finds, at its configuration time, that you have a non-GNU linker
8856 or a GNU linker that does not support option @option{--eh-frame-hdr},
8857 it will link the shared version of @file{libgcc} into shared libraries
8858 by default. Otherwise, it will take advantage of the linker and optimize
8859 away the linking with the shared version of @file{libgcc}, linking with
8860 the static version of libgcc by default. This allows exceptions to
8861 propagate through such shared libraries, without incurring relocation
8862 costs at library load time.
8864 However, if a library or main executable is supposed to throw or catch
8865 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8866 for the languages used in the program, or using the option
8867 @option{-shared-libgcc}, such that it is linked with the shared
8870 @item -static-libstdc++
8871 When the @command{g++} program is used to link a C++ program, it will
8872 normally automatically link against @option{libstdc++}. If
8873 @file{libstdc++} is available as a shared library, and the
8874 @option{-static} option is not used, then this will link against the
8875 shared version of @file{libstdc++}. That is normally fine. However, it
8876 is sometimes useful to freeze the version of @file{libstdc++} used by
8877 the program without going all the way to a fully static link. The
8878 @option{-static-libstdc++} option directs the @command{g++} driver to
8879 link @file{libstdc++} statically, without necessarily linking other
8880 libraries statically.
8884 Bind references to global symbols when building a shared object. Warn
8885 about any unresolved references (unless overridden by the link editor
8886 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8889 @item -T @var{script}
8891 @cindex linker script
8892 Use @var{script} as the linker script. This option is supported by most
8893 systems using the GNU linker. On some targets, such as bare-board
8894 targets without an operating system, the @option{-T} option may be required
8895 when linking to avoid references to undefined symbols.
8897 @item -Xlinker @var{option}
8899 Pass @var{option} as an option to the linker. You can use this to
8900 supply system-specific linker options which GCC does not know how to
8903 If you want to pass an option that takes a separate argument, you must use
8904 @option{-Xlinker} twice, once for the option and once for the argument.
8905 For example, to pass @option{-assert definitions}, you must write
8906 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8907 @option{-Xlinker "-assert definitions"}, because this passes the entire
8908 string as a single argument, which is not what the linker expects.
8910 When using the GNU linker, it is usually more convenient to pass
8911 arguments to linker options using the @option{@var{option}=@var{value}}
8912 syntax than as separate arguments. For example, you can specify
8913 @samp{-Xlinker -Map=output.map} rather than
8914 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8915 this syntax for command-line options.
8917 @item -Wl,@var{option}
8919 Pass @var{option} as an option to the linker. If @var{option} contains
8920 commas, it is split into multiple options at the commas. You can use this
8921 syntax to pass an argument to the option.
8922 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8923 linker. When using the GNU linker, you can also get the same effect with
8924 @samp{-Wl,-Map=output.map}.
8926 @item -u @var{symbol}
8928 Pretend the symbol @var{symbol} is undefined, to force linking of
8929 library modules to define it. You can use @option{-u} multiple times with
8930 different symbols to force loading of additional library modules.
8933 @node Directory Options
8934 @section Options for Directory Search
8935 @cindex directory options
8936 @cindex options, directory search
8939 These options specify directories to search for header files, for
8940 libraries and for parts of the compiler:
8945 Add the directory @var{dir} to the head of the list of directories to be
8946 searched for header files. This can be used to override a system header
8947 file, substituting your own version, since these directories are
8948 searched before the system header file directories. However, you should
8949 not use this option to add directories that contain vendor-supplied
8950 system header files (use @option{-isystem} for that). If you use more than
8951 one @option{-I} option, the directories are scanned in left-to-right
8952 order; the standard system directories come after.
8954 If a standard system include directory, or a directory specified with
8955 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8956 option will be ignored. The directory will still be searched but as a
8957 system directory at its normal position in the system include chain.
8958 This is to ensure that GCC's procedure to fix buggy system headers and
8959 the ordering for the include_next directive are not inadvertently changed.
8960 If you really need to change the search order for system directories,
8961 use the @option{-nostdinc} and/or @option{-isystem} options.
8963 @item -iplugindir=@var{dir}
8964 Set the directory to search for plugins which are passed
8965 by @option{-fplugin=@var{name}} instead of
8966 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
8967 to be used by the user, but only passed by the driver.
8969 @item -iquote@var{dir}
8971 Add the directory @var{dir} to the head of the list of directories to
8972 be searched for header files only for the case of @samp{#include
8973 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8974 otherwise just like @option{-I}.
8978 Add directory @var{dir} to the list of directories to be searched
8981 @item -B@var{prefix}
8983 This option specifies where to find the executables, libraries,
8984 include files, and data files of the compiler itself.
8986 The compiler driver program runs one or more of the subprograms
8987 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8988 @var{prefix} as a prefix for each program it tries to run, both with and
8989 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8991 For each subprogram to be run, the compiler driver first tries the
8992 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8993 was not specified, the driver tries two standard prefixes, which are
8994 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8995 those results in a file name that is found, the unmodified program
8996 name is searched for using the directories specified in your
8997 @env{PATH} environment variable.
8999 The compiler will check to see if the path provided by the @option{-B}
9000 refers to a directory, and if necessary it will add a directory
9001 separator character at the end of the path.
9003 @option{-B} prefixes that effectively specify directory names also apply
9004 to libraries in the linker, because the compiler translates these
9005 options into @option{-L} options for the linker. They also apply to
9006 includes files in the preprocessor, because the compiler translates these
9007 options into @option{-isystem} options for the preprocessor. In this case,
9008 the compiler appends @samp{include} to the prefix.
9010 The run-time support file @file{libgcc.a} can also be searched for using
9011 the @option{-B} prefix, if needed. If it is not found there, the two
9012 standard prefixes above are tried, and that is all. The file is left
9013 out of the link if it is not found by those means.
9015 Another way to specify a prefix much like the @option{-B} prefix is to use
9016 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
9019 As a special kludge, if the path provided by @option{-B} is
9020 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
9021 9, then it will be replaced by @file{[dir/]include}. This is to help
9022 with boot-strapping the compiler.
9024 @item -specs=@var{file}
9026 Process @var{file} after the compiler reads in the standard @file{specs}
9027 file, in order to override the defaults that the @file{gcc} driver
9028 program uses when determining what switches to pass to @file{cc1},
9029 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
9030 @option{-specs=@var{file}} can be specified on the command line, and they
9031 are processed in order, from left to right.
9033 @item --sysroot=@var{dir}
9035 Use @var{dir} as the logical root directory for headers and libraries.
9036 For example, if the compiler would normally search for headers in
9037 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
9038 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
9040 If you use both this option and the @option{-isysroot} option, then
9041 the @option{--sysroot} option will apply to libraries, but the
9042 @option{-isysroot} option will apply to header files.
9044 The GNU linker (beginning with version 2.16) has the necessary support
9045 for this option. If your linker does not support this option, the
9046 header file aspect of @option{--sysroot} will still work, but the
9047 library aspect will not.
9051 This option has been deprecated. Please use @option{-iquote} instead for
9052 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
9053 Any directories you specify with @option{-I} options before the @option{-I-}
9054 option are searched only for the case of @samp{#include "@var{file}"};
9055 they are not searched for @samp{#include <@var{file}>}.
9057 If additional directories are specified with @option{-I} options after
9058 the @option{-I-}, these directories are searched for all @samp{#include}
9059 directives. (Ordinarily @emph{all} @option{-I} directories are used
9062 In addition, the @option{-I-} option inhibits the use of the current
9063 directory (where the current input file came from) as the first search
9064 directory for @samp{#include "@var{file}"}. There is no way to
9065 override this effect of @option{-I-}. With @option{-I.} you can specify
9066 searching the directory which was current when the compiler was
9067 invoked. That is not exactly the same as what the preprocessor does
9068 by default, but it is often satisfactory.
9070 @option{-I-} does not inhibit the use of the standard system directories
9071 for header files. Thus, @option{-I-} and @option{-nostdinc} are
9078 @section Specifying subprocesses and the switches to pass to them
9081 @command{gcc} is a driver program. It performs its job by invoking a
9082 sequence of other programs to do the work of compiling, assembling and
9083 linking. GCC interprets its command-line parameters and uses these to
9084 deduce which programs it should invoke, and which command-line options
9085 it ought to place on their command lines. This behavior is controlled
9086 by @dfn{spec strings}. In most cases there is one spec string for each
9087 program that GCC can invoke, but a few programs have multiple spec
9088 strings to control their behavior. The spec strings built into GCC can
9089 be overridden by using the @option{-specs=} command-line switch to specify
9092 @dfn{Spec files} are plaintext files that are used to construct spec
9093 strings. They consist of a sequence of directives separated by blank
9094 lines. The type of directive is determined by the first non-whitespace
9095 character on the line and it can be one of the following:
9098 @item %@var{command}
9099 Issues a @var{command} to the spec file processor. The commands that can
9103 @item %include <@var{file}>
9105 Search for @var{file} and insert its text at the current point in the
9108 @item %include_noerr <@var{file}>
9109 @cindex %include_noerr
9110 Just like @samp{%include}, but do not generate an error message if the include
9111 file cannot be found.
9113 @item %rename @var{old_name} @var{new_name}
9115 Rename the spec string @var{old_name} to @var{new_name}.
9119 @item *[@var{spec_name}]:
9120 This tells the compiler to create, override or delete the named spec
9121 string. All lines after this directive up to the next directive or
9122 blank line are considered to be the text for the spec string. If this
9123 results in an empty string then the spec will be deleted. (Or, if the
9124 spec did not exist, then nothing will happened.) Otherwise, if the spec
9125 does not currently exist a new spec will be created. If the spec does
9126 exist then its contents will be overridden by the text of this
9127 directive, unless the first character of that text is the @samp{+}
9128 character, in which case the text will be appended to the spec.
9130 @item [@var{suffix}]:
9131 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
9132 and up to the next directive or blank line are considered to make up the
9133 spec string for the indicated suffix. When the compiler encounters an
9134 input file with the named suffix, it will processes the spec string in
9135 order to work out how to compile that file. For example:
9142 This says that any input file whose name ends in @samp{.ZZ} should be
9143 passed to the program @samp{z-compile}, which should be invoked with the
9144 command-line switch @option{-input} and with the result of performing the
9145 @samp{%i} substitution. (See below.)
9147 As an alternative to providing a spec string, the text that follows a
9148 suffix directive can be one of the following:
9151 @item @@@var{language}
9152 This says that the suffix is an alias for a known @var{language}. This is
9153 similar to using the @option{-x} command-line switch to GCC to specify a
9154 language explicitly. For example:
9161 Says that .ZZ files are, in fact, C++ source files.
9164 This causes an error messages saying:
9167 @var{name} compiler not installed on this system.
9171 GCC already has an extensive list of suffixes built into it.
9172 This directive will add an entry to the end of the list of suffixes, but
9173 since the list is searched from the end backwards, it is effectively
9174 possible to override earlier entries using this technique.
9178 GCC has the following spec strings built into it. Spec files can
9179 override these strings or create their own. Note that individual
9180 targets can also add their own spec strings to this list.
9183 asm Options to pass to the assembler
9184 asm_final Options to pass to the assembler post-processor
9185 cpp Options to pass to the C preprocessor
9186 cc1 Options to pass to the C compiler
9187 cc1plus Options to pass to the C++ compiler
9188 endfile Object files to include at the end of the link
9189 link Options to pass to the linker
9190 lib Libraries to include on the command line to the linker
9191 libgcc Decides which GCC support library to pass to the linker
9192 linker Sets the name of the linker
9193 predefines Defines to be passed to the C preprocessor
9194 signed_char Defines to pass to CPP to say whether @code{char} is signed
9196 startfile Object files to include at the start of the link
9199 Here is a small example of a spec file:
9205 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9208 This example renames the spec called @samp{lib} to @samp{old_lib} and
9209 then overrides the previous definition of @samp{lib} with a new one.
9210 The new definition adds in some extra command-line options before
9211 including the text of the old definition.
9213 @dfn{Spec strings} are a list of command-line options to be passed to their
9214 corresponding program. In addition, the spec strings can contain
9215 @samp{%}-prefixed sequences to substitute variable text or to
9216 conditionally insert text into the command line. Using these constructs
9217 it is possible to generate quite complex command lines.
9219 Here is a table of all defined @samp{%}-sequences for spec
9220 strings. Note that spaces are not generated automatically around the
9221 results of expanding these sequences. Therefore you can concatenate them
9222 together or combine them with constant text in a single argument.
9226 Substitute one @samp{%} into the program name or argument.
9229 Substitute the name of the input file being processed.
9232 Substitute the basename of the input file being processed.
9233 This is the substring up to (and not including) the last period
9234 and not including the directory.
9237 This is the same as @samp{%b}, but include the file suffix (text after
9241 Marks the argument containing or following the @samp{%d} as a
9242 temporary file name, so that that file will be deleted if GCC exits
9243 successfully. Unlike @samp{%g}, this contributes no text to the
9246 @item %g@var{suffix}
9247 Substitute a file name that has suffix @var{suffix} and is chosen
9248 once per compilation, and mark the argument in the same way as
9249 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9250 name is now chosen in a way that is hard to predict even when previously
9251 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9252 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9253 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9254 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9255 was simply substituted with a file name chosen once per compilation,
9256 without regard to any appended suffix (which was therefore treated
9257 just like ordinary text), making such attacks more likely to succeed.
9259 @item %u@var{suffix}
9260 Like @samp{%g}, but generates a new temporary file name even if
9261 @samp{%u@var{suffix}} was already seen.
9263 @item %U@var{suffix}
9264 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9265 new one if there is no such last file name. In the absence of any
9266 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9267 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9268 would involve the generation of two distinct file names, one
9269 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9270 simply substituted with a file name chosen for the previous @samp{%u},
9271 without regard to any appended suffix.
9273 @item %j@var{suffix}
9274 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9275 writable, and if save-temps is off; otherwise, substitute the name
9276 of a temporary file, just like @samp{%u}. This temporary file is not
9277 meant for communication between processes, but rather as a junk
9280 @item %|@var{suffix}
9281 @itemx %m@var{suffix}
9282 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9283 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9284 all. These are the two most common ways to instruct a program that it
9285 should read from standard input or write to standard output. If you
9286 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9287 construct: see for example @file{f/lang-specs.h}.
9289 @item %.@var{SUFFIX}
9290 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9291 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9292 terminated by the next space or %.
9295 Marks the argument containing or following the @samp{%w} as the
9296 designated output file of this compilation. This puts the argument
9297 into the sequence of arguments that @samp{%o} will substitute later.
9300 Substitutes the names of all the output files, with spaces
9301 automatically placed around them. You should write spaces
9302 around the @samp{%o} as well or the results are undefined.
9303 @samp{%o} is for use in the specs for running the linker.
9304 Input files whose names have no recognized suffix are not compiled
9305 at all, but they are included among the output files, so they will
9309 Substitutes the suffix for object files. Note that this is
9310 handled specially when it immediately follows @samp{%g, %u, or %U},
9311 because of the need for those to form complete file names. The
9312 handling is such that @samp{%O} is treated exactly as if it had already
9313 been substituted, except that @samp{%g, %u, and %U} do not currently
9314 support additional @var{suffix} characters following @samp{%O} as they would
9315 following, for example, @samp{.o}.
9318 Substitutes the standard macro predefinitions for the
9319 current target machine. Use this when running @code{cpp}.
9322 Like @samp{%p}, but puts @samp{__} before and after the name of each
9323 predefined macro, except for macros that start with @samp{__} or with
9324 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9328 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9329 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9330 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9331 and @option{-imultilib} as necessary.
9334 Current argument is the name of a library or startup file of some sort.
9335 Search for that file in a standard list of directories and substitute
9336 the full name found. The current working directory is included in the
9337 list of directories scanned.
9340 Current argument is the name of a linker script. Search for that file
9341 in the current list of directories to scan for libraries. If the file
9342 is located insert a @option{--script} option into the command line
9343 followed by the full path name found. If the file is not found then
9344 generate an error message. Note: the current working directory is not
9348 Print @var{str} as an error message. @var{str} is terminated by a newline.
9349 Use this when inconsistent options are detected.
9352 Substitute the contents of spec string @var{name} at this point.
9355 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9357 @item %x@{@var{option}@}
9358 Accumulate an option for @samp{%X}.
9361 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9365 Output the accumulated assembler options specified by @option{-Wa}.
9368 Output the accumulated preprocessor options specified by @option{-Wp}.
9371 Process the @code{asm} spec. This is used to compute the
9372 switches to be passed to the assembler.
9375 Process the @code{asm_final} spec. This is a spec string for
9376 passing switches to an assembler post-processor, if such a program is
9380 Process the @code{link} spec. This is the spec for computing the
9381 command line passed to the linker. Typically it will make use of the
9382 @samp{%L %G %S %D and %E} sequences.
9385 Dump out a @option{-L} option for each directory that GCC believes might
9386 contain startup files. If the target supports multilibs then the
9387 current multilib directory will be prepended to each of these paths.
9390 Process the @code{lib} spec. This is a spec string for deciding which
9391 libraries should be included on the command line to the linker.
9394 Process the @code{libgcc} spec. This is a spec string for deciding
9395 which GCC support library should be included on the command line to the linker.
9398 Process the @code{startfile} spec. This is a spec for deciding which
9399 object files should be the first ones passed to the linker. Typically
9400 this might be a file named @file{crt0.o}.
9403 Process the @code{endfile} spec. This is a spec string that specifies
9404 the last object files that will be passed to the linker.
9407 Process the @code{cpp} spec. This is used to construct the arguments
9408 to be passed to the C preprocessor.
9411 Process the @code{cc1} spec. This is used to construct the options to be
9412 passed to the actual C compiler (@samp{cc1}).
9415 Process the @code{cc1plus} spec. This is used to construct the options to be
9416 passed to the actual C++ compiler (@samp{cc1plus}).
9419 Substitute the variable part of a matched option. See below.
9420 Note that each comma in the substituted string is replaced by
9424 Remove all occurrences of @code{-S} from the command line. Note---this
9425 command is position dependent. @samp{%} commands in the spec string
9426 before this one will see @code{-S}, @samp{%} commands in the spec string
9427 after this one will not.
9429 @item %:@var{function}(@var{args})
9430 Call the named function @var{function}, passing it @var{args}.
9431 @var{args} is first processed as a nested spec string, then split
9432 into an argument vector in the usual fashion. The function returns
9433 a string which is processed as if it had appeared literally as part
9434 of the current spec.
9436 The following built-in spec functions are provided:
9440 The @code{getenv} spec function takes two arguments: an environment
9441 variable name and a string. If the environment variable is not
9442 defined, a fatal error is issued. Otherwise, the return value is the
9443 value of the environment variable concatenated with the string. For
9444 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9447 %:getenv(TOPDIR /include)
9450 expands to @file{/path/to/top/include}.
9452 @item @code{if-exists}
9453 The @code{if-exists} spec function takes one argument, an absolute
9454 pathname to a file. If the file exists, @code{if-exists} returns the
9455 pathname. Here is a small example of its usage:
9459 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9462 @item @code{if-exists-else}
9463 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9464 spec function, except that it takes two arguments. The first argument is
9465 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9466 returns the pathname. If it does not exist, it returns the second argument.
9467 This way, @code{if-exists-else} can be used to select one file or another,
9468 based on the existence of the first. Here is a small example of its usage:
9472 crt0%O%s %:if-exists(crti%O%s) \
9473 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9476 @item @code{replace-outfile}
9477 The @code{replace-outfile} spec function takes two arguments. It looks for the
9478 first argument in the outfiles array and replaces it with the second argument. Here
9479 is a small example of its usage:
9482 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9485 @item @code{print-asm-header}
9486 The @code{print-asm-header} function takes no arguments and simply
9487 prints a banner like:
9493 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9496 It is used to separate compiler options from assembler options
9497 in the @option{--target-help} output.
9501 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9502 If that switch was not specified, this substitutes nothing. Note that
9503 the leading dash is omitted when specifying this option, and it is
9504 automatically inserted if the substitution is performed. Thus the spec
9505 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9506 and would output the command line option @option{-foo}.
9508 @item %W@{@code{S}@}
9509 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9512 @item %@{@code{S}*@}
9513 Substitutes all the switches specified to GCC whose names start
9514 with @code{-S}, but which also take an argument. This is used for
9515 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9516 GCC considers @option{-o foo} as being
9517 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9518 text, including the space. Thus two arguments would be generated.
9520 @item %@{@code{S}*&@code{T}*@}
9521 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9522 (the order of @code{S} and @code{T} in the spec is not significant).
9523 There can be any number of ampersand-separated variables; for each the
9524 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9526 @item %@{@code{S}:@code{X}@}
9527 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9529 @item %@{!@code{S}:@code{X}@}
9530 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9532 @item %@{@code{S}*:@code{X}@}
9533 Substitutes @code{X} if one or more switches whose names start with
9534 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9535 once, no matter how many such switches appeared. However, if @code{%*}
9536 appears somewhere in @code{X}, then @code{X} will be substituted once
9537 for each matching switch, with the @code{%*} replaced by the part of
9538 that switch that matched the @code{*}.
9540 @item %@{.@code{S}:@code{X}@}
9541 Substitutes @code{X}, if processing a file with suffix @code{S}.
9543 @item %@{!.@code{S}:@code{X}@}
9544 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9546 @item %@{,@code{S}:@code{X}@}
9547 Substitutes @code{X}, if processing a file for language @code{S}.
9549 @item %@{!,@code{S}:@code{X}@}
9550 Substitutes @code{X}, if not processing a file for language @code{S}.
9552 @item %@{@code{S}|@code{P}:@code{X}@}
9553 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9554 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9555 @code{*} sequences as well, although they have a stronger binding than
9556 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9557 alternatives must be starred, and only the first matching alternative
9560 For example, a spec string like this:
9563 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9566 will output the following command-line options from the following input
9567 command-line options:
9572 -d fred.c -foo -baz -boggle
9573 -d jim.d -bar -baz -boggle
9576 @item %@{S:X; T:Y; :D@}
9578 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9579 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9580 be as many clauses as you need. This may be combined with @code{.},
9581 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9586 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9587 construct may contain other nested @samp{%} constructs or spaces, or
9588 even newlines. They are processed as usual, as described above.
9589 Trailing white space in @code{X} is ignored. White space may also
9590 appear anywhere on the left side of the colon in these constructs,
9591 except between @code{.} or @code{*} and the corresponding word.
9593 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9594 handled specifically in these constructs. If another value of
9595 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9596 @option{-W} switch is found later in the command line, the earlier
9597 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9598 just one letter, which passes all matching options.
9600 The character @samp{|} at the beginning of the predicate text is used to
9601 indicate that a command should be piped to the following command, but
9602 only if @option{-pipe} is specified.
9604 It is built into GCC which switches take arguments and which do not.
9605 (You might think it would be useful to generalize this to allow each
9606 compiler's spec to say which switches take arguments. But this cannot
9607 be done in a consistent fashion. GCC cannot even decide which input
9608 files have been specified without knowing which switches take arguments,
9609 and it must know which input files to compile in order to tell which
9612 GCC also knows implicitly that arguments starting in @option{-l} are to be
9613 treated as compiler output files, and passed to the linker in their
9614 proper position among the other output files.
9616 @c man begin OPTIONS
9618 @node Target Options
9619 @section Specifying Target Machine and Compiler Version
9620 @cindex target options
9621 @cindex cross compiling
9622 @cindex specifying machine version
9623 @cindex specifying compiler version and target machine
9624 @cindex compiler version, specifying
9625 @cindex target machine, specifying
9627 The usual way to run GCC is to run the executable called @file{gcc}, or
9628 @file{<machine>-gcc} when cross-compiling, or
9629 @file{<machine>-gcc-<version>} to run a version other than the one that
9632 @node Submodel Options
9633 @section Hardware Models and Configurations
9634 @cindex submodel options
9635 @cindex specifying hardware config
9636 @cindex hardware models and configurations, specifying
9637 @cindex machine dependent options
9639 Each target machine types can have its own
9640 special options, starting with @samp{-m}, to choose among various
9641 hardware models or configurations---for example, 68010 vs 68020,
9642 floating coprocessor or none. A single installed version of the
9643 compiler can compile for any model or configuration, according to the
9646 Some configurations of the compiler also support additional special
9647 options, usually for compatibility with other compilers on the same
9650 @c This list is ordered alphanumerically by subsection name.
9651 @c It should be the same order and spelling as these options are listed
9652 @c in Machine Dependent Options
9658 * Blackfin Options::
9662 * DEC Alpha Options::
9663 * DEC Alpha/VMS Options::
9666 * GNU/Linux Options::
9669 * i386 and x86-64 Options::
9670 * i386 and x86-64 Windows Options::
9672 * IA-64/VMS Options::
9684 * picoChip Options::
9686 * RS/6000 and PowerPC Options::
9688 * S/390 and zSeries Options::
9693 * System V Options::
9698 * Xstormy16 Options::
9704 @subsection ARC Options
9707 These options are defined for ARC implementations:
9712 Compile code for little endian mode. This is the default.
9716 Compile code for big endian mode.
9719 @opindex mmangle-cpu
9720 Prepend the name of the cpu to all public symbol names.
9721 In multiple-processor systems, there are many ARC variants with different
9722 instruction and register set characteristics. This flag prevents code
9723 compiled for one cpu to be linked with code compiled for another.
9724 No facility exists for handling variants that are ``almost identical''.
9725 This is an all or nothing option.
9727 @item -mcpu=@var{cpu}
9729 Compile code for ARC variant @var{cpu}.
9730 Which variants are supported depend on the configuration.
9731 All variants support @option{-mcpu=base}, this is the default.
9733 @item -mtext=@var{text-section}
9734 @itemx -mdata=@var{data-section}
9735 @itemx -mrodata=@var{readonly-data-section}
9739 Put functions, data, and readonly data in @var{text-section},
9740 @var{data-section}, and @var{readonly-data-section} respectively
9741 by default. This can be overridden with the @code{section} attribute.
9742 @xref{Variable Attributes}.
9744 @item -mfix-cortex-m3-ldrd
9745 @opindex mfix-cortex-m3-ldrd
9746 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
9747 with overlapping destination and base registers are used. This option avoids
9748 generating these instructions. This option is enabled by default when
9749 @option{-mcpu=cortex-m3} is specified.
9754 @subsection ARM Options
9757 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9761 @item -mabi=@var{name}
9763 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9764 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9767 @opindex mapcs-frame
9768 Generate a stack frame that is compliant with the ARM Procedure Call
9769 Standard for all functions, even if this is not strictly necessary for
9770 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9771 with this option will cause the stack frames not to be generated for
9772 leaf functions. The default is @option{-mno-apcs-frame}.
9776 This is a synonym for @option{-mapcs-frame}.
9779 @c not currently implemented
9780 @item -mapcs-stack-check
9781 @opindex mapcs-stack-check
9782 Generate code to check the amount of stack space available upon entry to
9783 every function (that actually uses some stack space). If there is
9784 insufficient space available then either the function
9785 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9786 called, depending upon the amount of stack space required. The run time
9787 system is required to provide these functions. The default is
9788 @option{-mno-apcs-stack-check}, since this produces smaller code.
9790 @c not currently implemented
9792 @opindex mapcs-float
9793 Pass floating point arguments using the float point registers. This is
9794 one of the variants of the APCS@. This option is recommended if the
9795 target hardware has a floating point unit or if a lot of floating point
9796 arithmetic is going to be performed by the code. The default is
9797 @option{-mno-apcs-float}, since integer only code is slightly increased in
9798 size if @option{-mapcs-float} is used.
9800 @c not currently implemented
9801 @item -mapcs-reentrant
9802 @opindex mapcs-reentrant
9803 Generate reentrant, position independent code. The default is
9804 @option{-mno-apcs-reentrant}.
9807 @item -mthumb-interwork
9808 @opindex mthumb-interwork
9809 Generate code which supports calling between the ARM and Thumb
9810 instruction sets. Without this option the two instruction sets cannot
9811 be reliably used inside one program. The default is
9812 @option{-mno-thumb-interwork}, since slightly larger code is generated
9813 when @option{-mthumb-interwork} is specified.
9815 @item -mno-sched-prolog
9816 @opindex mno-sched-prolog
9817 Prevent the reordering of instructions in the function prolog, or the
9818 merging of those instruction with the instructions in the function's
9819 body. This means that all functions will start with a recognizable set
9820 of instructions (or in fact one of a choice from a small set of
9821 different function prologues), and this information can be used to
9822 locate the start if functions inside an executable piece of code. The
9823 default is @option{-msched-prolog}.
9825 @item -mfloat-abi=@var{name}
9827 Specifies which floating-point ABI to use. Permissible values
9828 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9830 Specifying @samp{soft} causes GCC to generate output containing
9831 library calls for floating-point operations.
9832 @samp{softfp} allows the generation of code using hardware floating-point
9833 instructions, but still uses the soft-float calling conventions.
9834 @samp{hard} allows generation of floating-point instructions
9835 and uses FPU-specific calling conventions.
9837 The default depends on the specific target configuration. Note that
9838 the hard-float and soft-float ABIs are not link-compatible; you must
9839 compile your entire program with the same ABI, and link with a
9840 compatible set of libraries.
9843 @opindex mhard-float
9844 Equivalent to @option{-mfloat-abi=hard}.
9847 @opindex msoft-float
9848 Equivalent to @option{-mfloat-abi=soft}.
9850 @item -mlittle-endian
9851 @opindex mlittle-endian
9852 Generate code for a processor running in little-endian mode. This is
9853 the default for all standard configurations.
9856 @opindex mbig-endian
9857 Generate code for a processor running in big-endian mode; the default is
9858 to compile code for a little-endian processor.
9860 @item -mwords-little-endian
9861 @opindex mwords-little-endian
9862 This option only applies when generating code for big-endian processors.
9863 Generate code for a little-endian word order but a big-endian byte
9864 order. That is, a byte order of the form @samp{32107654}. Note: this
9865 option should only be used if you require compatibility with code for
9866 big-endian ARM processors generated by versions of the compiler prior to
9869 @item -mcpu=@var{name}
9871 This specifies the name of the target ARM processor. GCC uses this name
9872 to determine what kind of instructions it can emit when generating
9873 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9874 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9875 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9876 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9877 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9879 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9880 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9881 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9882 @samp{strongarm1110},
9883 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9884 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9885 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9886 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9887 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9888 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9889 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9890 @samp{cortex-a5}, @samp{cortex-a8}, @samp{cortex-a9},
9891 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9894 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9896 @item -mtune=@var{name}
9898 This option is very similar to the @option{-mcpu=} option, except that
9899 instead of specifying the actual target processor type, and hence
9900 restricting which instructions can be used, it specifies that GCC should
9901 tune the performance of the code as if the target were of the type
9902 specified in this option, but still choosing the instructions that it
9903 will generate based on the cpu specified by a @option{-mcpu=} option.
9904 For some ARM implementations better performance can be obtained by using
9907 @item -march=@var{name}
9909 This specifies the name of the target ARM architecture. GCC uses this
9910 name to determine what kind of instructions it can emit when generating
9911 assembly code. This option can be used in conjunction with or instead
9912 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9913 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9914 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9915 @samp{armv6}, @samp{armv6j},
9916 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9917 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9918 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9920 @item -mfpu=@var{name}
9921 @itemx -mfpe=@var{number}
9922 @itemx -mfp=@var{number}
9926 This specifies what floating point hardware (or hardware emulation) is
9927 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9928 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
9929 @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
9930 @samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
9931 @samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
9932 @option{-mfp} and @option{-mfpe} are synonyms for
9933 @option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
9936 If @option{-msoft-float} is specified this specifies the format of
9937 floating point values.
9939 @item -mfp16-format=@var{name}
9940 @opindex mfp16-format
9941 Specify the format of the @code{__fp16} half-precision floating-point type.
9942 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
9943 the default is @samp{none}, in which case the @code{__fp16} type is not
9944 defined. @xref{Half-Precision}, for more information.
9946 @item -mstructure-size-boundary=@var{n}
9947 @opindex mstructure-size-boundary
9948 The size of all structures and unions will be rounded up to a multiple
9949 of the number of bits set by this option. Permissible values are 8, 32
9950 and 64. The default value varies for different toolchains. For the COFF
9951 targeted toolchain the default value is 8. A value of 64 is only allowed
9952 if the underlying ABI supports it.
9954 Specifying the larger number can produce faster, more efficient code, but
9955 can also increase the size of the program. Different values are potentially
9956 incompatible. Code compiled with one value cannot necessarily expect to
9957 work with code or libraries compiled with another value, if they exchange
9958 information using structures or unions.
9960 @item -mabort-on-noreturn
9961 @opindex mabort-on-noreturn
9962 Generate a call to the function @code{abort} at the end of a
9963 @code{noreturn} function. It will be executed if the function tries to
9967 @itemx -mno-long-calls
9968 @opindex mlong-calls
9969 @opindex mno-long-calls
9970 Tells the compiler to perform function calls by first loading the
9971 address of the function into a register and then performing a subroutine
9972 call on this register. This switch is needed if the target function
9973 will lie outside of the 64 megabyte addressing range of the offset based
9974 version of subroutine call instruction.
9976 Even if this switch is enabled, not all function calls will be turned
9977 into long calls. The heuristic is that static functions, functions
9978 which have the @samp{short-call} attribute, functions that are inside
9979 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9980 definitions have already been compiled within the current compilation
9981 unit, will not be turned into long calls. The exception to this rule is
9982 that weak function definitions, functions with the @samp{long-call}
9983 attribute or the @samp{section} attribute, and functions that are within
9984 the scope of a @samp{#pragma long_calls} directive, will always be
9985 turned into long calls.
9987 This feature is not enabled by default. Specifying
9988 @option{-mno-long-calls} will restore the default behavior, as will
9989 placing the function calls within the scope of a @samp{#pragma
9990 long_calls_off} directive. Note these switches have no effect on how
9991 the compiler generates code to handle function calls via function
9994 @item -msingle-pic-base
9995 @opindex msingle-pic-base
9996 Treat the register used for PIC addressing as read-only, rather than
9997 loading it in the prologue for each function. The run-time system is
9998 responsible for initializing this register with an appropriate value
9999 before execution begins.
10001 @item -mpic-register=@var{reg}
10002 @opindex mpic-register
10003 Specify the register to be used for PIC addressing. The default is R10
10004 unless stack-checking is enabled, when R9 is used.
10006 @item -mcirrus-fix-invalid-insns
10007 @opindex mcirrus-fix-invalid-insns
10008 @opindex mno-cirrus-fix-invalid-insns
10009 Insert NOPs into the instruction stream to in order to work around
10010 problems with invalid Maverick instruction combinations. This option
10011 is only valid if the @option{-mcpu=ep9312} option has been used to
10012 enable generation of instructions for the Cirrus Maverick floating
10013 point co-processor. This option is not enabled by default, since the
10014 problem is only present in older Maverick implementations. The default
10015 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
10018 @item -mpoke-function-name
10019 @opindex mpoke-function-name
10020 Write the name of each function into the text section, directly
10021 preceding the function prologue. The generated code is similar to this:
10025 .ascii "arm_poke_function_name", 0
10028 .word 0xff000000 + (t1 - t0)
10029 arm_poke_function_name
10031 stmfd sp!, @{fp, ip, lr, pc@}
10035 When performing a stack backtrace, code can inspect the value of
10036 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
10037 location @code{pc - 12} and the top 8 bits are set, then we know that
10038 there is a function name embedded immediately preceding this location
10039 and has length @code{((pc[-3]) & 0xff000000)}.
10043 Generate code for the Thumb instruction set. The default is to
10044 use the 32-bit ARM instruction set.
10045 This option automatically enables either 16-bit Thumb-1 or
10046 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
10047 and @option{-march=@var{name}} options. This option is not passed to the
10048 assembler. If you want to force assembler files to be interpreted as Thumb code,
10049 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
10050 option directly to the assembler by prefixing it with @option{-Wa}.
10053 @opindex mtpcs-frame
10054 Generate a stack frame that is compliant with the Thumb Procedure Call
10055 Standard for all non-leaf functions. (A leaf function is one that does
10056 not call any other functions.) The default is @option{-mno-tpcs-frame}.
10058 @item -mtpcs-leaf-frame
10059 @opindex mtpcs-leaf-frame
10060 Generate a stack frame that is compliant with the Thumb Procedure Call
10061 Standard for all leaf functions. (A leaf function is one that does
10062 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
10064 @item -mcallee-super-interworking
10065 @opindex mcallee-super-interworking
10066 Gives all externally visible functions in the file being compiled an ARM
10067 instruction set header which switches to Thumb mode before executing the
10068 rest of the function. This allows these functions to be called from
10069 non-interworking code. This option is not valid in AAPCS configurations
10070 because interworking is enabled by default.
10072 @item -mcaller-super-interworking
10073 @opindex mcaller-super-interworking
10074 Allows calls via function pointers (including virtual functions) to
10075 execute correctly regardless of whether the target code has been
10076 compiled for interworking or not. There is a small overhead in the cost
10077 of executing a function pointer if this option is enabled. This option
10078 is not valid in AAPCS configurations because interworking is enabled
10081 @item -mtp=@var{name}
10083 Specify the access model for the thread local storage pointer. The valid
10084 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10085 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
10086 (supported in the arm6k architecture), and @option{auto}, which uses the
10087 best available method for the selected processor. The default setting is
10090 @item -mword-relocations
10091 @opindex mword-relocations
10092 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
10093 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10094 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10100 @subsection AVR Options
10101 @cindex AVR Options
10103 These options are defined for AVR implementations:
10106 @item -mmcu=@var{mcu}
10108 Specify ATMEL AVR instruction set or MCU type.
10110 Instruction set avr1 is for the minimal AVR core, not supported by the C
10111 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10112 attiny11, attiny12, attiny15, attiny28).
10114 Instruction set avr2 (default) is for the classic AVR core with up to
10115 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10116 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10117 at90c8534, at90s8535).
10119 Instruction set avr3 is for the classic AVR core with up to 128K program
10120 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10122 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10123 memory space (MCU types: atmega8, atmega83, atmega85).
10125 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10126 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10127 atmega64, atmega128, at43usb355, at94k).
10129 @item -mno-interrupts
10130 @opindex mno-interrupts
10131 Generated code is not compatible with hardware interrupts.
10132 Code size will be smaller.
10134 @item -mcall-prologues
10135 @opindex mcall-prologues
10136 Functions prologues/epilogues expanded as call to appropriate
10137 subroutines. Code size will be smaller.
10140 @opindex mtiny-stack
10141 Change only the low 8 bits of the stack pointer.
10145 Assume int to be 8 bit integer. This affects the sizes of all types: A
10146 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10147 and long long will be 4 bytes. Please note that this option does not
10148 comply to the C standards, but it will provide you with smaller code
10152 @node Blackfin Options
10153 @subsection Blackfin Options
10154 @cindex Blackfin Options
10157 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10159 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10160 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10161 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10162 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10163 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10164 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10165 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10167 The optional @var{sirevision} specifies the silicon revision of the target
10168 Blackfin processor. Any workarounds available for the targeted silicon revision
10169 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10170 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10171 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10172 hexadecimal digits representing the major and minor numbers in the silicon
10173 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10174 is not defined. If @var{sirevision} is @samp{any}, the
10175 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10176 If this optional @var{sirevision} is not used, GCC assumes the latest known
10177 silicon revision of the targeted Blackfin processor.
10179 Support for @samp{bf561} is incomplete. For @samp{bf561},
10180 Only the processor macro is defined.
10181 Without this option, @samp{bf532} is used as the processor by default.
10182 The corresponding predefined processor macros for @var{cpu} is to
10183 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10184 provided by libgloss to be linked in if @option{-msim} is not given.
10188 Specifies that the program will be run on the simulator. This causes
10189 the simulator BSP provided by libgloss to be linked in. This option
10190 has effect only for @samp{bfin-elf} toolchain.
10191 Certain other options, such as @option{-mid-shared-library} and
10192 @option{-mfdpic}, imply @option{-msim}.
10194 @item -momit-leaf-frame-pointer
10195 @opindex momit-leaf-frame-pointer
10196 Don't keep the frame pointer in a register for leaf functions. This
10197 avoids the instructions to save, set up and restore frame pointers and
10198 makes an extra register available in leaf functions. The option
10199 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10200 which might make debugging harder.
10202 @item -mspecld-anomaly
10203 @opindex mspecld-anomaly
10204 When enabled, the compiler will ensure that the generated code does not
10205 contain speculative loads after jump instructions. If this option is used,
10206 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10208 @item -mno-specld-anomaly
10209 @opindex mno-specld-anomaly
10210 Don't generate extra code to prevent speculative loads from occurring.
10212 @item -mcsync-anomaly
10213 @opindex mcsync-anomaly
10214 When enabled, the compiler will ensure that the generated code does not
10215 contain CSYNC or SSYNC instructions too soon after conditional branches.
10216 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10218 @item -mno-csync-anomaly
10219 @opindex mno-csync-anomaly
10220 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10221 occurring too soon after a conditional branch.
10225 When enabled, the compiler is free to take advantage of the knowledge that
10226 the entire program fits into the low 64k of memory.
10229 @opindex mno-low-64k
10230 Assume that the program is arbitrarily large. This is the default.
10232 @item -mstack-check-l1
10233 @opindex mstack-check-l1
10234 Do stack checking using information placed into L1 scratchpad memory by the
10237 @item -mid-shared-library
10238 @opindex mid-shared-library
10239 Generate code that supports shared libraries via the library ID method.
10240 This allows for execute in place and shared libraries in an environment
10241 without virtual memory management. This option implies @option{-fPIC}.
10242 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10244 @item -mno-id-shared-library
10245 @opindex mno-id-shared-library
10246 Generate code that doesn't assume ID based shared libraries are being used.
10247 This is the default.
10249 @item -mleaf-id-shared-library
10250 @opindex mleaf-id-shared-library
10251 Generate code that supports shared libraries via the library ID method,
10252 but assumes that this library or executable won't link against any other
10253 ID shared libraries. That allows the compiler to use faster code for jumps
10256 @item -mno-leaf-id-shared-library
10257 @opindex mno-leaf-id-shared-library
10258 Do not assume that the code being compiled won't link against any ID shared
10259 libraries. Slower code will be generated for jump and call insns.
10261 @item -mshared-library-id=n
10262 @opindex mshared-library-id
10263 Specified the identification number of the ID based shared library being
10264 compiled. Specifying a value of 0 will generate more compact code, specifying
10265 other values will force the allocation of that number to the current
10266 library but is no more space or time efficient than omitting this option.
10270 Generate code that allows the data segment to be located in a different
10271 area of memory from the text segment. This allows for execute in place in
10272 an environment without virtual memory management by eliminating relocations
10273 against the text section.
10275 @item -mno-sep-data
10276 @opindex mno-sep-data
10277 Generate code that assumes that the data segment follows the text segment.
10278 This is the default.
10281 @itemx -mno-long-calls
10282 @opindex mlong-calls
10283 @opindex mno-long-calls
10284 Tells the compiler to perform function calls by first loading the
10285 address of the function into a register and then performing a subroutine
10286 call on this register. This switch is needed if the target function
10287 will lie outside of the 24 bit addressing range of the offset based
10288 version of subroutine call instruction.
10290 This feature is not enabled by default. Specifying
10291 @option{-mno-long-calls} will restore the default behavior. Note these
10292 switches have no effect on how the compiler generates code to handle
10293 function calls via function pointers.
10297 Link with the fast floating-point library. This library relaxes some of
10298 the IEEE floating-point standard's rules for checking inputs against
10299 Not-a-Number (NAN), in the interest of performance.
10302 @opindex minline-plt
10303 Enable inlining of PLT entries in function calls to functions that are
10304 not known to bind locally. It has no effect without @option{-mfdpic}.
10307 @opindex mmulticore
10308 Build standalone application for multicore Blackfin processor. Proper
10309 start files and link scripts will be used to support multicore.
10310 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10311 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10312 @option{-mcorea} or @option{-mcoreb}. If it's used without
10313 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10314 programming model is used. In this model, the main function of Core B
10315 should be named as coreb_main. If it's used with @option{-mcorea} or
10316 @option{-mcoreb}, one application per core programming model is used.
10317 If this option is not used, single core application programming
10322 Build standalone application for Core A of BF561 when using
10323 one application per core programming model. Proper start files
10324 and link scripts will be used to support Core A. This option
10325 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10329 Build standalone application for Core B of BF561 when using
10330 one application per core programming model. Proper start files
10331 and link scripts will be used to support Core B. This option
10332 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10333 should be used instead of main. It must be used with
10334 @option{-mmulticore}.
10338 Build standalone application for SDRAM. Proper start files and
10339 link scripts will be used to put the application into SDRAM.
10340 Loader should initialize SDRAM before loading the application
10341 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10345 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10346 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10347 are enabled; for standalone applications the default is off.
10351 @subsection CRIS Options
10352 @cindex CRIS Options
10354 These options are defined specifically for the CRIS ports.
10357 @item -march=@var{architecture-type}
10358 @itemx -mcpu=@var{architecture-type}
10361 Generate code for the specified architecture. The choices for
10362 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10363 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10364 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10367 @item -mtune=@var{architecture-type}
10369 Tune to @var{architecture-type} everything applicable about the generated
10370 code, except for the ABI and the set of available instructions. The
10371 choices for @var{architecture-type} are the same as for
10372 @option{-march=@var{architecture-type}}.
10374 @item -mmax-stack-frame=@var{n}
10375 @opindex mmax-stack-frame
10376 Warn when the stack frame of a function exceeds @var{n} bytes.
10382 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10383 @option{-march=v3} and @option{-march=v8} respectively.
10385 @item -mmul-bug-workaround
10386 @itemx -mno-mul-bug-workaround
10387 @opindex mmul-bug-workaround
10388 @opindex mno-mul-bug-workaround
10389 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10390 models where it applies. This option is active by default.
10394 Enable CRIS-specific verbose debug-related information in the assembly
10395 code. This option also has the effect to turn off the @samp{#NO_APP}
10396 formatted-code indicator to the assembler at the beginning of the
10401 Do not use condition-code results from previous instruction; always emit
10402 compare and test instructions before use of condition codes.
10404 @item -mno-side-effects
10405 @opindex mno-side-effects
10406 Do not emit instructions with side-effects in addressing modes other than
10409 @item -mstack-align
10410 @itemx -mno-stack-align
10411 @itemx -mdata-align
10412 @itemx -mno-data-align
10413 @itemx -mconst-align
10414 @itemx -mno-const-align
10415 @opindex mstack-align
10416 @opindex mno-stack-align
10417 @opindex mdata-align
10418 @opindex mno-data-align
10419 @opindex mconst-align
10420 @opindex mno-const-align
10421 These options (no-options) arranges (eliminate arrangements) for the
10422 stack-frame, individual data and constants to be aligned for the maximum
10423 single data access size for the chosen CPU model. The default is to
10424 arrange for 32-bit alignment. ABI details such as structure layout are
10425 not affected by these options.
10433 Similar to the stack- data- and const-align options above, these options
10434 arrange for stack-frame, writable data and constants to all be 32-bit,
10435 16-bit or 8-bit aligned. The default is 32-bit alignment.
10437 @item -mno-prologue-epilogue
10438 @itemx -mprologue-epilogue
10439 @opindex mno-prologue-epilogue
10440 @opindex mprologue-epilogue
10441 With @option{-mno-prologue-epilogue}, the normal function prologue and
10442 epilogue that sets up the stack-frame are omitted and no return
10443 instructions or return sequences are generated in the code. Use this
10444 option only together with visual inspection of the compiled code: no
10445 warnings or errors are generated when call-saved registers must be saved,
10446 or storage for local variable needs to be allocated.
10450 @opindex mno-gotplt
10452 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10453 instruction sequences that load addresses for functions from the PLT part
10454 of the GOT rather than (traditional on other architectures) calls to the
10455 PLT@. The default is @option{-mgotplt}.
10459 Legacy no-op option only recognized with the cris-axis-elf and
10460 cris-axis-linux-gnu targets.
10464 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10468 This option, recognized for the cris-axis-elf arranges
10469 to link with input-output functions from a simulator library. Code,
10470 initialized data and zero-initialized data are allocated consecutively.
10474 Like @option{-sim}, but pass linker options to locate initialized data at
10475 0x40000000 and zero-initialized data at 0x80000000.
10479 @subsection CRX Options
10480 @cindex CRX Options
10482 These options are defined specifically for the CRX ports.
10488 Enable the use of multiply-accumulate instructions. Disabled by default.
10491 @opindex mpush-args
10492 Push instructions will be used to pass outgoing arguments when functions
10493 are called. Enabled by default.
10496 @node Darwin Options
10497 @subsection Darwin Options
10498 @cindex Darwin options
10500 These options are defined for all architectures running the Darwin operating
10503 FSF GCC on Darwin does not create ``fat'' object files; it will create
10504 an object file for the single architecture that it was built to
10505 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10506 @option{-arch} options are used; it does so by running the compiler or
10507 linker multiple times and joining the results together with
10510 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10511 @samp{i686}) is determined by the flags that specify the ISA
10512 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10513 @option{-force_cpusubtype_ALL} option can be used to override this.
10515 The Darwin tools vary in their behavior when presented with an ISA
10516 mismatch. The assembler, @file{as}, will only permit instructions to
10517 be used that are valid for the subtype of the file it is generating,
10518 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10519 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10520 and print an error if asked to create a shared library with a less
10521 restrictive subtype than its input files (for instance, trying to put
10522 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10523 for executables, @file{ld}, will quietly give the executable the most
10524 restrictive subtype of any of its input files.
10529 Add the framework directory @var{dir} to the head of the list of
10530 directories to be searched for header files. These directories are
10531 interleaved with those specified by @option{-I} options and are
10532 scanned in a left-to-right order.
10534 A framework directory is a directory with frameworks in it. A
10535 framework is a directory with a @samp{"Headers"} and/or
10536 @samp{"PrivateHeaders"} directory contained directly in it that ends
10537 in @samp{".framework"}. The name of a framework is the name of this
10538 directory excluding the @samp{".framework"}. Headers associated with
10539 the framework are found in one of those two directories, with
10540 @samp{"Headers"} being searched first. A subframework is a framework
10541 directory that is in a framework's @samp{"Frameworks"} directory.
10542 Includes of subframework headers can only appear in a header of a
10543 framework that contains the subframework, or in a sibling subframework
10544 header. Two subframeworks are siblings if they occur in the same
10545 framework. A subframework should not have the same name as a
10546 framework, a warning will be issued if this is violated. Currently a
10547 subframework cannot have subframeworks, in the future, the mechanism
10548 may be extended to support this. The standard frameworks can be found
10549 in @samp{"/System/Library/Frameworks"} and
10550 @samp{"/Library/Frameworks"}. An example include looks like
10551 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10552 the name of the framework and header.h is found in the
10553 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10555 @item -iframework@var{dir}
10556 @opindex iframework
10557 Like @option{-F} except the directory is a treated as a system
10558 directory. The main difference between this @option{-iframework} and
10559 @option{-F} is that with @option{-iframework} the compiler does not
10560 warn about constructs contained within header files found via
10561 @var{dir}. This option is valid only for the C family of languages.
10565 Emit debugging information for symbols that are used. For STABS
10566 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10567 This is by default ON@.
10571 Emit debugging information for all symbols and types.
10573 @item -mmacosx-version-min=@var{version}
10574 The earliest version of MacOS X that this executable will run on
10575 is @var{version}. Typical values of @var{version} include @code{10.1},
10576 @code{10.2}, and @code{10.3.9}.
10578 If the compiler was built to use the system's headers by default,
10579 then the default for this option is the system version on which the
10580 compiler is running, otherwise the default is to make choices which
10581 are compatible with as many systems and code bases as possible.
10585 Enable kernel development mode. The @option{-mkernel} option sets
10586 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10587 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10588 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10589 applicable. This mode also sets @option{-mno-altivec},
10590 @option{-msoft-float}, @option{-fno-builtin} and
10591 @option{-mlong-branch} for PowerPC targets.
10593 @item -mone-byte-bool
10594 @opindex mone-byte-bool
10595 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10596 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10597 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10598 option has no effect on x86.
10600 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10601 to generate code that is not binary compatible with code generated
10602 without that switch. Using this switch may require recompiling all
10603 other modules in a program, including system libraries. Use this
10604 switch to conform to a non-default data model.
10606 @item -mfix-and-continue
10607 @itemx -ffix-and-continue
10608 @itemx -findirect-data
10609 @opindex mfix-and-continue
10610 @opindex ffix-and-continue
10611 @opindex findirect-data
10612 Generate code suitable for fast turn around development. Needed to
10613 enable gdb to dynamically load @code{.o} files into already running
10614 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10615 are provided for backwards compatibility.
10619 Loads all members of static archive libraries.
10620 See man ld(1) for more information.
10622 @item -arch_errors_fatal
10623 @opindex arch_errors_fatal
10624 Cause the errors having to do with files that have the wrong architecture
10627 @item -bind_at_load
10628 @opindex bind_at_load
10629 Causes the output file to be marked such that the dynamic linker will
10630 bind all undefined references when the file is loaded or launched.
10634 Produce a Mach-o bundle format file.
10635 See man ld(1) for more information.
10637 @item -bundle_loader @var{executable}
10638 @opindex bundle_loader
10639 This option specifies the @var{executable} that will be loading the build
10640 output file being linked. See man ld(1) for more information.
10643 @opindex dynamiclib
10644 When passed this option, GCC will produce a dynamic library instead of
10645 an executable when linking, using the Darwin @file{libtool} command.
10647 @item -force_cpusubtype_ALL
10648 @opindex force_cpusubtype_ALL
10649 This causes GCC's output file to have the @var{ALL} subtype, instead of
10650 one controlled by the @option{-mcpu} or @option{-march} option.
10652 @item -allowable_client @var{client_name}
10653 @itemx -client_name
10654 @itemx -compatibility_version
10655 @itemx -current_version
10657 @itemx -dependency-file
10659 @itemx -dylinker_install_name
10661 @itemx -exported_symbols_list
10663 @itemx -flat_namespace
10664 @itemx -force_flat_namespace
10665 @itemx -headerpad_max_install_names
10668 @itemx -install_name
10669 @itemx -keep_private_externs
10670 @itemx -multi_module
10671 @itemx -multiply_defined
10672 @itemx -multiply_defined_unused
10674 @itemx -no_dead_strip_inits_and_terms
10675 @itemx -nofixprebinding
10676 @itemx -nomultidefs
10678 @itemx -noseglinkedit
10679 @itemx -pagezero_size
10681 @itemx -prebind_all_twolevel_modules
10682 @itemx -private_bundle
10683 @itemx -read_only_relocs
10685 @itemx -sectobjectsymbols
10689 @itemx -sectobjectsymbols
10692 @itemx -segs_read_only_addr
10693 @itemx -segs_read_write_addr
10694 @itemx -seg_addr_table
10695 @itemx -seg_addr_table_filename
10696 @itemx -seglinkedit
10698 @itemx -segs_read_only_addr
10699 @itemx -segs_read_write_addr
10700 @itemx -single_module
10702 @itemx -sub_library
10703 @itemx -sub_umbrella
10704 @itemx -twolevel_namespace
10707 @itemx -unexported_symbols_list
10708 @itemx -weak_reference_mismatches
10709 @itemx -whatsloaded
10710 @opindex allowable_client
10711 @opindex client_name
10712 @opindex compatibility_version
10713 @opindex current_version
10714 @opindex dead_strip
10715 @opindex dependency-file
10716 @opindex dylib_file
10717 @opindex dylinker_install_name
10719 @opindex exported_symbols_list
10721 @opindex flat_namespace
10722 @opindex force_flat_namespace
10723 @opindex headerpad_max_install_names
10724 @opindex image_base
10726 @opindex install_name
10727 @opindex keep_private_externs
10728 @opindex multi_module
10729 @opindex multiply_defined
10730 @opindex multiply_defined_unused
10731 @opindex noall_load
10732 @opindex no_dead_strip_inits_and_terms
10733 @opindex nofixprebinding
10734 @opindex nomultidefs
10736 @opindex noseglinkedit
10737 @opindex pagezero_size
10739 @opindex prebind_all_twolevel_modules
10740 @opindex private_bundle
10741 @opindex read_only_relocs
10743 @opindex sectobjectsymbols
10746 @opindex sectcreate
10747 @opindex sectobjectsymbols
10750 @opindex segs_read_only_addr
10751 @opindex segs_read_write_addr
10752 @opindex seg_addr_table
10753 @opindex seg_addr_table_filename
10754 @opindex seglinkedit
10756 @opindex segs_read_only_addr
10757 @opindex segs_read_write_addr
10758 @opindex single_module
10760 @opindex sub_library
10761 @opindex sub_umbrella
10762 @opindex twolevel_namespace
10765 @opindex unexported_symbols_list
10766 @opindex weak_reference_mismatches
10767 @opindex whatsloaded
10768 These options are passed to the Darwin linker. The Darwin linker man page
10769 describes them in detail.
10772 @node DEC Alpha Options
10773 @subsection DEC Alpha Options
10775 These @samp{-m} options are defined for the DEC Alpha implementations:
10778 @item -mno-soft-float
10779 @itemx -msoft-float
10780 @opindex mno-soft-float
10781 @opindex msoft-float
10782 Use (do not use) the hardware floating-point instructions for
10783 floating-point operations. When @option{-msoft-float} is specified,
10784 functions in @file{libgcc.a} will be used to perform floating-point
10785 operations. Unless they are replaced by routines that emulate the
10786 floating-point operations, or compiled in such a way as to call such
10787 emulations routines, these routines will issue floating-point
10788 operations. If you are compiling for an Alpha without floating-point
10789 operations, you must ensure that the library is built so as not to call
10792 Note that Alpha implementations without floating-point operations are
10793 required to have floating-point registers.
10796 @itemx -mno-fp-regs
10798 @opindex mno-fp-regs
10799 Generate code that uses (does not use) the floating-point register set.
10800 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10801 register set is not used, floating point operands are passed in integer
10802 registers as if they were integers and floating-point results are passed
10803 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10804 so any function with a floating-point argument or return value called by code
10805 compiled with @option{-mno-fp-regs} must also be compiled with that
10808 A typical use of this option is building a kernel that does not use,
10809 and hence need not save and restore, any floating-point registers.
10813 The Alpha architecture implements floating-point hardware optimized for
10814 maximum performance. It is mostly compliant with the IEEE floating
10815 point standard. However, for full compliance, software assistance is
10816 required. This option generates code fully IEEE compliant code
10817 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10818 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10819 defined during compilation. The resulting code is less efficient but is
10820 able to correctly support denormalized numbers and exceptional IEEE
10821 values such as not-a-number and plus/minus infinity. Other Alpha
10822 compilers call this option @option{-ieee_with_no_inexact}.
10824 @item -mieee-with-inexact
10825 @opindex mieee-with-inexact
10826 This is like @option{-mieee} except the generated code also maintains
10827 the IEEE @var{inexact-flag}. Turning on this option causes the
10828 generated code to implement fully-compliant IEEE math. In addition to
10829 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10830 macro. On some Alpha implementations the resulting code may execute
10831 significantly slower than the code generated by default. Since there is
10832 very little code that depends on the @var{inexact-flag}, you should
10833 normally not specify this option. Other Alpha compilers call this
10834 option @option{-ieee_with_inexact}.
10836 @item -mfp-trap-mode=@var{trap-mode}
10837 @opindex mfp-trap-mode
10838 This option controls what floating-point related traps are enabled.
10839 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10840 The trap mode can be set to one of four values:
10844 This is the default (normal) setting. The only traps that are enabled
10845 are the ones that cannot be disabled in software (e.g., division by zero
10849 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10853 Like @samp{u}, but the instructions are marked to be safe for software
10854 completion (see Alpha architecture manual for details).
10857 Like @samp{su}, but inexact traps are enabled as well.
10860 @item -mfp-rounding-mode=@var{rounding-mode}
10861 @opindex mfp-rounding-mode
10862 Selects the IEEE rounding mode. Other Alpha compilers call this option
10863 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10868 Normal IEEE rounding mode. Floating point numbers are rounded towards
10869 the nearest machine number or towards the even machine number in case
10873 Round towards minus infinity.
10876 Chopped rounding mode. Floating point numbers are rounded towards zero.
10879 Dynamic rounding mode. A field in the floating point control register
10880 (@var{fpcr}, see Alpha architecture reference manual) controls the
10881 rounding mode in effect. The C library initializes this register for
10882 rounding towards plus infinity. Thus, unless your program modifies the
10883 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10886 @item -mtrap-precision=@var{trap-precision}
10887 @opindex mtrap-precision
10888 In the Alpha architecture, floating point traps are imprecise. This
10889 means without software assistance it is impossible to recover from a
10890 floating trap and program execution normally needs to be terminated.
10891 GCC can generate code that can assist operating system trap handlers
10892 in determining the exact location that caused a floating point trap.
10893 Depending on the requirements of an application, different levels of
10894 precisions can be selected:
10898 Program precision. This option is the default and means a trap handler
10899 can only identify which program caused a floating point exception.
10902 Function precision. The trap handler can determine the function that
10903 caused a floating point exception.
10906 Instruction precision. The trap handler can determine the exact
10907 instruction that caused a floating point exception.
10910 Other Alpha compilers provide the equivalent options called
10911 @option{-scope_safe} and @option{-resumption_safe}.
10913 @item -mieee-conformant
10914 @opindex mieee-conformant
10915 This option marks the generated code as IEEE conformant. You must not
10916 use this option unless you also specify @option{-mtrap-precision=i} and either
10917 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10918 is to emit the line @samp{.eflag 48} in the function prologue of the
10919 generated assembly file. Under DEC Unix, this has the effect that
10920 IEEE-conformant math library routines will be linked in.
10922 @item -mbuild-constants
10923 @opindex mbuild-constants
10924 Normally GCC examines a 32- or 64-bit integer constant to
10925 see if it can construct it from smaller constants in two or three
10926 instructions. If it cannot, it will output the constant as a literal and
10927 generate code to load it from the data segment at runtime.
10929 Use this option to require GCC to construct @emph{all} integer constants
10930 using code, even if it takes more instructions (the maximum is six).
10932 You would typically use this option to build a shared library dynamic
10933 loader. Itself a shared library, it must relocate itself in memory
10934 before it can find the variables and constants in its own data segment.
10940 Select whether to generate code to be assembled by the vendor-supplied
10941 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10959 Indicate whether GCC should generate code to use the optional BWX,
10960 CIX, FIX and MAX instruction sets. The default is to use the instruction
10961 sets supported by the CPU type specified via @option{-mcpu=} option or that
10962 of the CPU on which GCC was built if none was specified.
10965 @itemx -mfloat-ieee
10966 @opindex mfloat-vax
10967 @opindex mfloat-ieee
10968 Generate code that uses (does not use) VAX F and G floating point
10969 arithmetic instead of IEEE single and double precision.
10971 @item -mexplicit-relocs
10972 @itemx -mno-explicit-relocs
10973 @opindex mexplicit-relocs
10974 @opindex mno-explicit-relocs
10975 Older Alpha assemblers provided no way to generate symbol relocations
10976 except via assembler macros. Use of these macros does not allow
10977 optimal instruction scheduling. GNU binutils as of version 2.12
10978 supports a new syntax that allows the compiler to explicitly mark
10979 which relocations should apply to which instructions. This option
10980 is mostly useful for debugging, as GCC detects the capabilities of
10981 the assembler when it is built and sets the default accordingly.
10984 @itemx -mlarge-data
10985 @opindex msmall-data
10986 @opindex mlarge-data
10987 When @option{-mexplicit-relocs} is in effect, static data is
10988 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10989 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10990 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10991 16-bit relocations off of the @code{$gp} register. This limits the
10992 size of the small data area to 64KB, but allows the variables to be
10993 directly accessed via a single instruction.
10995 The default is @option{-mlarge-data}. With this option the data area
10996 is limited to just below 2GB@. Programs that require more than 2GB of
10997 data must use @code{malloc} or @code{mmap} to allocate the data in the
10998 heap instead of in the program's data segment.
11000 When generating code for shared libraries, @option{-fpic} implies
11001 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
11004 @itemx -mlarge-text
11005 @opindex msmall-text
11006 @opindex mlarge-text
11007 When @option{-msmall-text} is used, the compiler assumes that the
11008 code of the entire program (or shared library) fits in 4MB, and is
11009 thus reachable with a branch instruction. When @option{-msmall-data}
11010 is used, the compiler can assume that all local symbols share the
11011 same @code{$gp} value, and thus reduce the number of instructions
11012 required for a function call from 4 to 1.
11014 The default is @option{-mlarge-text}.
11016 @item -mcpu=@var{cpu_type}
11018 Set the instruction set and instruction scheduling parameters for
11019 machine type @var{cpu_type}. You can specify either the @samp{EV}
11020 style name or the corresponding chip number. GCC supports scheduling
11021 parameters for the EV4, EV5 and EV6 family of processors and will
11022 choose the default values for the instruction set from the processor
11023 you specify. If you do not specify a processor type, GCC will default
11024 to the processor on which the compiler was built.
11026 Supported values for @var{cpu_type} are
11032 Schedules as an EV4 and has no instruction set extensions.
11036 Schedules as an EV5 and has no instruction set extensions.
11040 Schedules as an EV5 and supports the BWX extension.
11045 Schedules as an EV5 and supports the BWX and MAX extensions.
11049 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
11053 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
11056 Native Linux/GNU toolchains also support the value @samp{native},
11057 which selects the best architecture option for the host processor.
11058 @option{-mcpu=native} has no effect if GCC does not recognize
11061 @item -mtune=@var{cpu_type}
11063 Set only the instruction scheduling parameters for machine type
11064 @var{cpu_type}. The instruction set is not changed.
11066 Native Linux/GNU toolchains also support the value @samp{native},
11067 which selects the best architecture option for the host processor.
11068 @option{-mtune=native} has no effect if GCC does not recognize
11071 @item -mmemory-latency=@var{time}
11072 @opindex mmemory-latency
11073 Sets the latency the scheduler should assume for typical memory
11074 references as seen by the application. This number is highly
11075 dependent on the memory access patterns used by the application
11076 and the size of the external cache on the machine.
11078 Valid options for @var{time} are
11082 A decimal number representing clock cycles.
11088 The compiler contains estimates of the number of clock cycles for
11089 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
11090 (also called Dcache, Scache, and Bcache), as well as to main memory.
11091 Note that L3 is only valid for EV5.
11096 @node DEC Alpha/VMS Options
11097 @subsection DEC Alpha/VMS Options
11099 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
11102 @item -mvms-return-codes
11103 @opindex mvms-return-codes
11104 Return VMS condition codes from main. The default is to return POSIX
11105 style condition (e.g.@: error) codes.
11107 @item -mdebug-main=@var{prefix}
11108 @opindex mdebug-main=@var{prefix}
11109 Flag the first routine whose name starts with @var{prefix} as the main
11110 routine for the debugger.
11114 Default to 64bit memory allocation routines.
11118 @subsection FR30 Options
11119 @cindex FR30 Options
11121 These options are defined specifically for the FR30 port.
11125 @item -msmall-model
11126 @opindex msmall-model
11127 Use the small address space model. This can produce smaller code, but
11128 it does assume that all symbolic values and addresses will fit into a
11133 Assume that run-time support has been provided and so there is no need
11134 to include the simulator library (@file{libsim.a}) on the linker
11140 @subsection FRV Options
11141 @cindex FRV Options
11147 Only use the first 32 general purpose registers.
11152 Use all 64 general purpose registers.
11157 Use only the first 32 floating point registers.
11162 Use all 64 floating point registers
11165 @opindex mhard-float
11167 Use hardware instructions for floating point operations.
11170 @opindex msoft-float
11172 Use library routines for floating point operations.
11177 Dynamically allocate condition code registers.
11182 Do not try to dynamically allocate condition code registers, only
11183 use @code{icc0} and @code{fcc0}.
11188 Change ABI to use double word insns.
11193 Do not use double word instructions.
11198 Use floating point double instructions.
11201 @opindex mno-double
11203 Do not use floating point double instructions.
11208 Use media instructions.
11213 Do not use media instructions.
11218 Use multiply and add/subtract instructions.
11221 @opindex mno-muladd
11223 Do not use multiply and add/subtract instructions.
11228 Select the FDPIC ABI, that uses function descriptors to represent
11229 pointers to functions. Without any PIC/PIE-related options, it
11230 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11231 assumes GOT entries and small data are within a 12-bit range from the
11232 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11233 are computed with 32 bits.
11234 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11237 @opindex minline-plt
11239 Enable inlining of PLT entries in function calls to functions that are
11240 not known to bind locally. It has no effect without @option{-mfdpic}.
11241 It's enabled by default if optimizing for speed and compiling for
11242 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11243 optimization option such as @option{-O3} or above is present in the
11249 Assume a large TLS segment when generating thread-local code.
11254 Do not assume a large TLS segment when generating thread-local code.
11259 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11260 that is known to be in read-only sections. It's enabled by default,
11261 except for @option{-fpic} or @option{-fpie}: even though it may help
11262 make the global offset table smaller, it trades 1 instruction for 4.
11263 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11264 one of which may be shared by multiple symbols, and it avoids the need
11265 for a GOT entry for the referenced symbol, so it's more likely to be a
11266 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11268 @item -multilib-library-pic
11269 @opindex multilib-library-pic
11271 Link with the (library, not FD) pic libraries. It's implied by
11272 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11273 @option{-fpic} without @option{-mfdpic}. You should never have to use
11277 @opindex mlinked-fp
11279 Follow the EABI requirement of always creating a frame pointer whenever
11280 a stack frame is allocated. This option is enabled by default and can
11281 be disabled with @option{-mno-linked-fp}.
11284 @opindex mlong-calls
11286 Use indirect addressing to call functions outside the current
11287 compilation unit. This allows the functions to be placed anywhere
11288 within the 32-bit address space.
11290 @item -malign-labels
11291 @opindex malign-labels
11293 Try to align labels to an 8-byte boundary by inserting nops into the
11294 previous packet. This option only has an effect when VLIW packing
11295 is enabled. It doesn't create new packets; it merely adds nops to
11298 @item -mlibrary-pic
11299 @opindex mlibrary-pic
11301 Generate position-independent EABI code.
11306 Use only the first four media accumulator registers.
11311 Use all eight media accumulator registers.
11316 Pack VLIW instructions.
11321 Do not pack VLIW instructions.
11324 @opindex mno-eflags
11326 Do not mark ABI switches in e_flags.
11329 @opindex mcond-move
11331 Enable the use of conditional-move instructions (default).
11333 This switch is mainly for debugging the compiler and will likely be removed
11334 in a future version.
11336 @item -mno-cond-move
11337 @opindex mno-cond-move
11339 Disable the use of conditional-move instructions.
11341 This switch is mainly for debugging the compiler and will likely be removed
11342 in a future version.
11347 Enable the use of conditional set instructions (default).
11349 This switch is mainly for debugging the compiler and will likely be removed
11350 in a future version.
11355 Disable the use of conditional set instructions.
11357 This switch is mainly for debugging the compiler and will likely be removed
11358 in a future version.
11361 @opindex mcond-exec
11363 Enable the use of conditional execution (default).
11365 This switch is mainly for debugging the compiler and will likely be removed
11366 in a future version.
11368 @item -mno-cond-exec
11369 @opindex mno-cond-exec
11371 Disable the use of conditional execution.
11373 This switch is mainly for debugging the compiler and will likely be removed
11374 in a future version.
11376 @item -mvliw-branch
11377 @opindex mvliw-branch
11379 Run a pass to pack branches into VLIW instructions (default).
11381 This switch is mainly for debugging the compiler and will likely be removed
11382 in a future version.
11384 @item -mno-vliw-branch
11385 @opindex mno-vliw-branch
11387 Do not run a pass to pack branches into VLIW instructions.
11389 This switch is mainly for debugging the compiler and will likely be removed
11390 in a future version.
11392 @item -mmulti-cond-exec
11393 @opindex mmulti-cond-exec
11395 Enable optimization of @code{&&} and @code{||} in conditional execution
11398 This switch is mainly for debugging the compiler and will likely be removed
11399 in a future version.
11401 @item -mno-multi-cond-exec
11402 @opindex mno-multi-cond-exec
11404 Disable optimization of @code{&&} and @code{||} in conditional execution.
11406 This switch is mainly for debugging the compiler and will likely be removed
11407 in a future version.
11409 @item -mnested-cond-exec
11410 @opindex mnested-cond-exec
11412 Enable nested conditional execution optimizations (default).
11414 This switch is mainly for debugging the compiler and will likely be removed
11415 in a future version.
11417 @item -mno-nested-cond-exec
11418 @opindex mno-nested-cond-exec
11420 Disable nested conditional execution optimizations.
11422 This switch is mainly for debugging the compiler and will likely be removed
11423 in a future version.
11425 @item -moptimize-membar
11426 @opindex moptimize-membar
11428 This switch removes redundant @code{membar} instructions from the
11429 compiler generated code. It is enabled by default.
11431 @item -mno-optimize-membar
11432 @opindex mno-optimize-membar
11434 This switch disables the automatic removal of redundant @code{membar}
11435 instructions from the generated code.
11437 @item -mtomcat-stats
11438 @opindex mtomcat-stats
11440 Cause gas to print out tomcat statistics.
11442 @item -mcpu=@var{cpu}
11445 Select the processor type for which to generate code. Possible values are
11446 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11447 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11451 @node GNU/Linux Options
11452 @subsection GNU/Linux Options
11454 These @samp{-m} options are defined for GNU/Linux targets:
11459 Use the GNU C library instead of uClibc. This is the default except
11460 on @samp{*-*-linux-*uclibc*} targets.
11464 Use uClibc instead of the GNU C library. This is the default on
11465 @samp{*-*-linux-*uclibc*} targets.
11468 @node H8/300 Options
11469 @subsection H8/300 Options
11471 These @samp{-m} options are defined for the H8/300 implementations:
11476 Shorten some address references at link time, when possible; uses the
11477 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11478 ld, Using ld}, for a fuller description.
11482 Generate code for the H8/300H@.
11486 Generate code for the H8S@.
11490 Generate code for the H8S and H8/300H in the normal mode. This switch
11491 must be used either with @option{-mh} or @option{-ms}.
11495 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11499 Make @code{int} data 32 bits by default.
11502 @opindex malign-300
11503 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11504 The default for the H8/300H and H8S is to align longs and floats on 4
11506 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11507 This option has no effect on the H8/300.
11511 @subsection HPPA Options
11512 @cindex HPPA Options
11514 These @samp{-m} options are defined for the HPPA family of computers:
11517 @item -march=@var{architecture-type}
11519 Generate code for the specified architecture. The choices for
11520 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11521 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11522 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11523 architecture option for your machine. Code compiled for lower numbered
11524 architectures will run on higher numbered architectures, but not the
11527 @item -mpa-risc-1-0
11528 @itemx -mpa-risc-1-1
11529 @itemx -mpa-risc-2-0
11530 @opindex mpa-risc-1-0
11531 @opindex mpa-risc-1-1
11532 @opindex mpa-risc-2-0
11533 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11536 @opindex mbig-switch
11537 Generate code suitable for big switch tables. Use this option only if
11538 the assembler/linker complain about out of range branches within a switch
11541 @item -mjump-in-delay
11542 @opindex mjump-in-delay
11543 Fill delay slots of function calls with unconditional jump instructions
11544 by modifying the return pointer for the function call to be the target
11545 of the conditional jump.
11547 @item -mdisable-fpregs
11548 @opindex mdisable-fpregs
11549 Prevent floating point registers from being used in any manner. This is
11550 necessary for compiling kernels which perform lazy context switching of
11551 floating point registers. If you use this option and attempt to perform
11552 floating point operations, the compiler will abort.
11554 @item -mdisable-indexing
11555 @opindex mdisable-indexing
11556 Prevent the compiler from using indexing address modes. This avoids some
11557 rather obscure problems when compiling MIG generated code under MACH@.
11559 @item -mno-space-regs
11560 @opindex mno-space-regs
11561 Generate code that assumes the target has no space registers. This allows
11562 GCC to generate faster indirect calls and use unscaled index address modes.
11564 Such code is suitable for level 0 PA systems and kernels.
11566 @item -mfast-indirect-calls
11567 @opindex mfast-indirect-calls
11568 Generate code that assumes calls never cross space boundaries. This
11569 allows GCC to emit code which performs faster indirect calls.
11571 This option will not work in the presence of shared libraries or nested
11574 @item -mfixed-range=@var{register-range}
11575 @opindex mfixed-range
11576 Generate code treating the given register range as fixed registers.
11577 A fixed register is one that the register allocator can not use. This is
11578 useful when compiling kernel code. A register range is specified as
11579 two registers separated by a dash. Multiple register ranges can be
11580 specified separated by a comma.
11582 @item -mlong-load-store
11583 @opindex mlong-load-store
11584 Generate 3-instruction load and store sequences as sometimes required by
11585 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11588 @item -mportable-runtime
11589 @opindex mportable-runtime
11590 Use the portable calling conventions proposed by HP for ELF systems.
11594 Enable the use of assembler directives only GAS understands.
11596 @item -mschedule=@var{cpu-type}
11598 Schedule code according to the constraints for the machine type
11599 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11600 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11601 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11602 proper scheduling option for your machine. The default scheduling is
11606 @opindex mlinker-opt
11607 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11608 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11609 linkers in which they give bogus error messages when linking some programs.
11612 @opindex msoft-float
11613 Generate output containing library calls for floating point.
11614 @strong{Warning:} the requisite libraries are not available for all HPPA
11615 targets. Normally the facilities of the machine's usual C compiler are
11616 used, but this cannot be done directly in cross-compilation. You must make
11617 your own arrangements to provide suitable library functions for
11620 @option{-msoft-float} changes the calling convention in the output file;
11621 therefore, it is only useful if you compile @emph{all} of a program with
11622 this option. In particular, you need to compile @file{libgcc.a}, the
11623 library that comes with GCC, with @option{-msoft-float} in order for
11628 Generate the predefine, @code{_SIO}, for server IO@. The default is
11629 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11630 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11631 options are available under HP-UX and HI-UX@.
11635 Use GNU ld specific options. This passes @option{-shared} to ld when
11636 building a shared library. It is the default when GCC is configured,
11637 explicitly or implicitly, with the GNU linker. This option does not
11638 have any affect on which ld is called, it only changes what parameters
11639 are passed to that ld. The ld that is called is determined by the
11640 @option{--with-ld} configure option, GCC's program search path, and
11641 finally by the user's @env{PATH}. The linker used by GCC can be printed
11642 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11643 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11647 Use HP ld specific options. This passes @option{-b} to ld when building
11648 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11649 links. It is the default when GCC is configured, explicitly or
11650 implicitly, with the HP linker. This option does not have any affect on
11651 which ld is called, it only changes what parameters are passed to that
11652 ld. The ld that is called is determined by the @option{--with-ld}
11653 configure option, GCC's program search path, and finally by the user's
11654 @env{PATH}. The linker used by GCC can be printed using @samp{which
11655 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11656 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11659 @opindex mno-long-calls
11660 Generate code that uses long call sequences. This ensures that a call
11661 is always able to reach linker generated stubs. The default is to generate
11662 long calls only when the distance from the call site to the beginning
11663 of the function or translation unit, as the case may be, exceeds a
11664 predefined limit set by the branch type being used. The limits for
11665 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11666 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11669 Distances are measured from the beginning of functions when using the
11670 @option{-ffunction-sections} option, or when using the @option{-mgas}
11671 and @option{-mno-portable-runtime} options together under HP-UX with
11674 It is normally not desirable to use this option as it will degrade
11675 performance. However, it may be useful in large applications,
11676 particularly when partial linking is used to build the application.
11678 The types of long calls used depends on the capabilities of the
11679 assembler and linker, and the type of code being generated. The
11680 impact on systems that support long absolute calls, and long pic
11681 symbol-difference or pc-relative calls should be relatively small.
11682 However, an indirect call is used on 32-bit ELF systems in pic code
11683 and it is quite long.
11685 @item -munix=@var{unix-std}
11687 Generate compiler predefines and select a startfile for the specified
11688 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11689 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11690 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11691 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11692 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11695 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11696 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11697 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11698 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11699 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11700 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11702 It is @emph{important} to note that this option changes the interfaces
11703 for various library routines. It also affects the operational behavior
11704 of the C library. Thus, @emph{extreme} care is needed in using this
11707 Library code that is intended to operate with more than one UNIX
11708 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11709 as appropriate. Most GNU software doesn't provide this capability.
11713 Suppress the generation of link options to search libdld.sl when the
11714 @option{-static} option is specified on HP-UX 10 and later.
11718 The HP-UX implementation of setlocale in libc has a dependency on
11719 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11720 when the @option{-static} option is specified, special link options
11721 are needed to resolve this dependency.
11723 On HP-UX 10 and later, the GCC driver adds the necessary options to
11724 link with libdld.sl when the @option{-static} option is specified.
11725 This causes the resulting binary to be dynamic. On the 64-bit port,
11726 the linkers generate dynamic binaries by default in any case. The
11727 @option{-nolibdld} option can be used to prevent the GCC driver from
11728 adding these link options.
11732 Add support for multithreading with the @dfn{dce thread} library
11733 under HP-UX@. This option sets flags for both the preprocessor and
11737 @node i386 and x86-64 Options
11738 @subsection Intel 386 and AMD x86-64 Options
11739 @cindex i386 Options
11740 @cindex x86-64 Options
11741 @cindex Intel 386 Options
11742 @cindex AMD x86-64 Options
11744 These @samp{-m} options are defined for the i386 and x86-64 family of
11748 @item -mtune=@var{cpu-type}
11750 Tune to @var{cpu-type} everything applicable about the generated code, except
11751 for the ABI and the set of available instructions. The choices for
11752 @var{cpu-type} are:
11755 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11756 If you know the CPU on which your code will run, then you should use
11757 the corresponding @option{-mtune} option instead of
11758 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11759 of your application will have, then you should use this option.
11761 As new processors are deployed in the marketplace, the behavior of this
11762 option will change. Therefore, if you upgrade to a newer version of
11763 GCC, the code generated option will change to reflect the processors
11764 that were most common when that version of GCC was released.
11766 There is no @option{-march=generic} option because @option{-march}
11767 indicates the instruction set the compiler can use, and there is no
11768 generic instruction set applicable to all processors. In contrast,
11769 @option{-mtune} indicates the processor (or, in this case, collection of
11770 processors) for which the code is optimized.
11772 This selects the CPU to tune for at compilation time by determining
11773 the processor type of the compiling machine. Using @option{-mtune=native}
11774 will produce code optimized for the local machine under the constraints
11775 of the selected instruction set. Using @option{-march=native} will
11776 enable all instruction subsets supported by the local machine (hence
11777 the result might not run on different machines).
11779 Original Intel's i386 CPU@.
11781 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11782 @item i586, pentium
11783 Intel Pentium CPU with no MMX support.
11785 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11787 Intel PentiumPro CPU@.
11789 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11790 instruction set will be used, so the code will run on all i686 family chips.
11792 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11793 @item pentium3, pentium3m
11794 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11797 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11798 support. Used by Centrino notebooks.
11799 @item pentium4, pentium4m
11800 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11802 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11805 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11806 SSE2 and SSE3 instruction set support.
11808 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11809 instruction set support.
11811 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11812 instruction set support.
11814 AMD K6 CPU with MMX instruction set support.
11816 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
11817 @item athlon, athlon-tbird
11818 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
11820 @item athlon-4, athlon-xp, athlon-mp
11821 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
11822 instruction set support.
11823 @item k8, opteron, athlon64, athlon-fx
11824 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11825 MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
11826 @item k8-sse3, opteron-sse3, athlon64-sse3
11827 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11828 @item amdfam10, barcelona
11829 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11830 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
11831 instruction set extensions.)
11833 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11836 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
11837 instruction set support.
11839 Via C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
11840 implemented for this chip.)
11842 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11843 implemented for this chip.)
11845 Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
11848 While picking a specific @var{cpu-type} will schedule things appropriately
11849 for that particular chip, the compiler will not generate any code that
11850 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11853 @item -march=@var{cpu-type}
11855 Generate instructions for the machine type @var{cpu-type}. The choices
11856 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11857 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11859 @item -mcpu=@var{cpu-type}
11861 A deprecated synonym for @option{-mtune}.
11863 @item -mfpmath=@var{unit}
11865 Generate floating point arithmetics for selected unit @var{unit}. The choices
11866 for @var{unit} are:
11870 Use the standard 387 floating point coprocessor present majority of chips and
11871 emulated otherwise. Code compiled with this option will run almost everywhere.
11872 The temporary results are computed in 80bit precision instead of precision
11873 specified by the type resulting in slightly different results compared to most
11874 of other chips. See @option{-ffloat-store} for more detailed description.
11876 This is the default choice for i386 compiler.
11879 Use scalar floating point instructions present in the SSE instruction set.
11880 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11881 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11882 instruction set supports only single precision arithmetics, thus the double and
11883 extended precision arithmetics is still done using 387. Later version, present
11884 only in Pentium4 and the future AMD x86-64 chips supports double precision
11887 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11888 or @option{-msse2} switches to enable SSE extensions and make this option
11889 effective. For the x86-64 compiler, these extensions are enabled by default.
11891 The resulting code should be considerably faster in the majority of cases and avoid
11892 the numerical instability problems of 387 code, but may break some existing
11893 code that expects temporaries to be 80bit.
11895 This is the default choice for the x86-64 compiler.
11900 Attempt to utilize both instruction sets at once. This effectively double the
11901 amount of available registers and on chips with separate execution units for
11902 387 and SSE the execution resources too. Use this option with care, as it is
11903 still experimental, because the GCC register allocator does not model separate
11904 functional units well resulting in instable performance.
11907 @item -masm=@var{dialect}
11908 @opindex masm=@var{dialect}
11909 Output asm instructions using selected @var{dialect}. Supported
11910 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11911 not support @samp{intel}.
11914 @itemx -mno-ieee-fp
11916 @opindex mno-ieee-fp
11917 Control whether or not the compiler uses IEEE floating point
11918 comparisons. These handle correctly the case where the result of a
11919 comparison is unordered.
11922 @opindex msoft-float
11923 Generate output containing library calls for floating point.
11924 @strong{Warning:} the requisite libraries are not part of GCC@.
11925 Normally the facilities of the machine's usual C compiler are used, but
11926 this can't be done directly in cross-compilation. You must make your
11927 own arrangements to provide suitable library functions for
11930 On machines where a function returns floating point results in the 80387
11931 register stack, some floating point opcodes may be emitted even if
11932 @option{-msoft-float} is used.
11934 @item -mno-fp-ret-in-387
11935 @opindex mno-fp-ret-in-387
11936 Do not use the FPU registers for return values of functions.
11938 The usual calling convention has functions return values of types
11939 @code{float} and @code{double} in an FPU register, even if there
11940 is no FPU@. The idea is that the operating system should emulate
11943 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11944 in ordinary CPU registers instead.
11946 @item -mno-fancy-math-387
11947 @opindex mno-fancy-math-387
11948 Some 387 emulators do not support the @code{sin}, @code{cos} and
11949 @code{sqrt} instructions for the 387. Specify this option to avoid
11950 generating those instructions. This option is the default on FreeBSD,
11951 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11952 indicates that the target cpu will always have an FPU and so the
11953 instruction will not need emulation. As of revision 2.6.1, these
11954 instructions are not generated unless you also use the
11955 @option{-funsafe-math-optimizations} switch.
11957 @item -malign-double
11958 @itemx -mno-align-double
11959 @opindex malign-double
11960 @opindex mno-align-double
11961 Control whether GCC aligns @code{double}, @code{long double}, and
11962 @code{long long} variables on a two word boundary or a one word
11963 boundary. Aligning @code{double} variables on a two word boundary will
11964 produce code that runs somewhat faster on a @samp{Pentium} at the
11965 expense of more memory.
11967 On x86-64, @option{-malign-double} is enabled by default.
11969 @strong{Warning:} if you use the @option{-malign-double} switch,
11970 structures containing the above types will be aligned differently than
11971 the published application binary interface specifications for the 386
11972 and will not be binary compatible with structures in code compiled
11973 without that switch.
11975 @item -m96bit-long-double
11976 @itemx -m128bit-long-double
11977 @opindex m96bit-long-double
11978 @opindex m128bit-long-double
11979 These switches control the size of @code{long double} type. The i386
11980 application binary interface specifies the size to be 96 bits,
11981 so @option{-m96bit-long-double} is the default in 32 bit mode.
11983 Modern architectures (Pentium and newer) would prefer @code{long double}
11984 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11985 conforming to the ABI, this would not be possible. So specifying a
11986 @option{-m128bit-long-double} will align @code{long double}
11987 to a 16 byte boundary by padding the @code{long double} with an additional
11990 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11991 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11993 Notice that neither of these options enable any extra precision over the x87
11994 standard of 80 bits for a @code{long double}.
11996 @strong{Warning:} if you override the default value for your target ABI, the
11997 structures and arrays containing @code{long double} variables will change
11998 their size as well as function calling convention for function taking
11999 @code{long double} will be modified. Hence they will not be binary
12000 compatible with arrays or structures in code compiled without that switch.
12002 @item -mlarge-data-threshold=@var{number}
12003 @opindex mlarge-data-threshold=@var{number}
12004 When @option{-mcmodel=medium} is specified, the data greater than
12005 @var{threshold} are placed in large data section. This value must be the
12006 same across all object linked into the binary and defaults to 65535.
12010 Use a different function-calling convention, in which functions that
12011 take a fixed number of arguments return with the @code{ret} @var{num}
12012 instruction, which pops their arguments while returning. This saves one
12013 instruction in the caller since there is no need to pop the arguments
12016 You can specify that an individual function is called with this calling
12017 sequence with the function attribute @samp{stdcall}. You can also
12018 override the @option{-mrtd} option by using the function attribute
12019 @samp{cdecl}. @xref{Function Attributes}.
12021 @strong{Warning:} this calling convention is incompatible with the one
12022 normally used on Unix, so you cannot use it if you need to call
12023 libraries compiled with the Unix compiler.
12025 Also, you must provide function prototypes for all functions that
12026 take variable numbers of arguments (including @code{printf});
12027 otherwise incorrect code will be generated for calls to those
12030 In addition, seriously incorrect code will result if you call a
12031 function with too many arguments. (Normally, extra arguments are
12032 harmlessly ignored.)
12034 @item -mregparm=@var{num}
12036 Control how many registers are used to pass integer arguments. By
12037 default, no registers are used to pass arguments, and at most 3
12038 registers can be used. You can control this behavior for a specific
12039 function by using the function attribute @samp{regparm}.
12040 @xref{Function Attributes}.
12042 @strong{Warning:} if you use this switch, and
12043 @var{num} is nonzero, then you must build all modules with the same
12044 value, including any libraries. This includes the system libraries and
12048 @opindex msseregparm
12049 Use SSE register passing conventions for float and double arguments
12050 and return values. You can control this behavior for a specific
12051 function by using the function attribute @samp{sseregparm}.
12052 @xref{Function Attributes}.
12054 @strong{Warning:} if you use this switch then you must build all
12055 modules with the same value, including any libraries. This includes
12056 the system libraries and startup modules.
12065 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
12066 is specified, the significands of results of floating-point operations are
12067 rounded to 24 bits (single precision); @option{-mpc64} rounds the
12068 significands of results of floating-point operations to 53 bits (double
12069 precision) and @option{-mpc80} rounds the significands of results of
12070 floating-point operations to 64 bits (extended double precision), which is
12071 the default. When this option is used, floating-point operations in higher
12072 precisions are not available to the programmer without setting the FPU
12073 control word explicitly.
12075 Setting the rounding of floating-point operations to less than the default
12076 80 bits can speed some programs by 2% or more. Note that some mathematical
12077 libraries assume that extended precision (80 bit) floating-point operations
12078 are enabled by default; routines in such libraries could suffer significant
12079 loss of accuracy, typically through so-called "catastrophic cancellation",
12080 when this option is used to set the precision to less than extended precision.
12082 @item -mstackrealign
12083 @opindex mstackrealign
12084 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
12085 option will generate an alternate prologue and epilogue that realigns the
12086 runtime stack if necessary. This supports mixing legacy codes that keep
12087 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
12088 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
12089 applicable to individual functions.
12091 @item -mpreferred-stack-boundary=@var{num}
12092 @opindex mpreferred-stack-boundary
12093 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
12094 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
12095 the default is 4 (16 bytes or 128 bits).
12097 @item -mincoming-stack-boundary=@var{num}
12098 @opindex mincoming-stack-boundary
12099 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
12100 boundary. If @option{-mincoming-stack-boundary} is not specified,
12101 the one specified by @option{-mpreferred-stack-boundary} will be used.
12103 On Pentium and PentiumPro, @code{double} and @code{long double} values
12104 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
12105 suffer significant run time performance penalties. On Pentium III, the
12106 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
12107 properly if it is not 16 byte aligned.
12109 To ensure proper alignment of this values on the stack, the stack boundary
12110 must be as aligned as that required by any value stored on the stack.
12111 Further, every function must be generated such that it keeps the stack
12112 aligned. Thus calling a function compiled with a higher preferred
12113 stack boundary from a function compiled with a lower preferred stack
12114 boundary will most likely misalign the stack. It is recommended that
12115 libraries that use callbacks always use the default setting.
12117 This extra alignment does consume extra stack space, and generally
12118 increases code size. Code that is sensitive to stack space usage, such
12119 as embedded systems and operating system kernels, may want to reduce the
12120 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12164 These switches enable or disable the use of instructions in the MMX,
12165 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, FMA4, XOP,
12166 LWP, ABM or 3DNow!@: extended instruction sets.
12167 These extensions are also available as built-in functions: see
12168 @ref{X86 Built-in Functions}, for details of the functions enabled and
12169 disabled by these switches.
12171 To have SSE/SSE2 instructions generated automatically from floating-point
12172 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12174 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12175 generates new AVX instructions or AVX equivalence for all SSEx instructions
12178 These options will enable GCC to use these extended instructions in
12179 generated code, even without @option{-mfpmath=sse}. Applications which
12180 perform runtime CPU detection must compile separate files for each
12181 supported architecture, using the appropriate flags. In particular,
12182 the file containing the CPU detection code should be compiled without
12186 @itemx -mno-fused-madd
12187 @opindex mfused-madd
12188 @opindex mno-fused-madd
12189 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12190 instructions. The default is to use these instructions.
12194 This option instructs GCC to emit a @code{cld} instruction in the prologue
12195 of functions that use string instructions. String instructions depend on
12196 the DF flag to select between autoincrement or autodecrement mode. While the
12197 ABI specifies the DF flag to be cleared on function entry, some operating
12198 systems violate this specification by not clearing the DF flag in their
12199 exception dispatchers. The exception handler can be invoked with the DF flag
12200 set which leads to wrong direction mode, when string instructions are used.
12201 This option can be enabled by default on 32-bit x86 targets by configuring
12202 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12203 instructions can be suppressed with the @option{-mno-cld} compiler option
12208 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12209 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12210 data types. This is useful for high resolution counters that could be updated
12211 by multiple processors (or cores). This instruction is generated as part of
12212 atomic built-in functions: see @ref{Atomic Builtins} for details.
12216 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12217 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12218 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12219 SAHF are load and store instructions, respectively, for certain status flags.
12220 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12221 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12225 This option will enable GCC to use movbe instruction to implement
12226 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12230 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12231 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12232 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12236 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12237 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12238 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12239 variants) for single precision floating point arguments. These instructions
12240 are generated only when @option{-funsafe-math-optimizations} is enabled
12241 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12242 Note that while the throughput of the sequence is higher than the throughput
12243 of the non-reciprocal instruction, the precision of the sequence can be
12244 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12246 Note that GCC implements 1.0f/sqrtf(x) in terms of RSQRTSS (or RSQRTPS)
12247 already with @option{-ffast-math} (or the above option combination), and
12248 doesn't need @option{-mrecip}.
12250 @item -mveclibabi=@var{type}
12251 @opindex mveclibabi
12252 Specifies the ABI type to use for vectorizing intrinsics using an
12253 external library. Supported types are @code{svml} for the Intel short
12254 vector math library and @code{acml} for the AMD math core library style
12255 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12256 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12257 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12258 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12259 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12260 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12261 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12262 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12263 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12264 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12265 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12266 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12267 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12268 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12269 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12270 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12271 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12272 compatible library will have to be specified at link time.
12274 @item -mabi=@var{name}
12276 Generate code for the specified calling convention. Permissible values
12277 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12278 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12279 ABI when targeting Windows. On all other systems, the default is the
12280 SYSV ABI. You can control this behavior for a specific function by
12281 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12282 @xref{Function Attributes}.
12285 @itemx -mno-push-args
12286 @opindex mpush-args
12287 @opindex mno-push-args
12288 Use PUSH operations to store outgoing parameters. This method is shorter
12289 and usually equally fast as method using SUB/MOV operations and is enabled
12290 by default. In some cases disabling it may improve performance because of
12291 improved scheduling and reduced dependencies.
12293 @item -maccumulate-outgoing-args
12294 @opindex maccumulate-outgoing-args
12295 If enabled, the maximum amount of space required for outgoing arguments will be
12296 computed in the function prologue. This is faster on most modern CPUs
12297 because of reduced dependencies, improved scheduling and reduced stack usage
12298 when preferred stack boundary is not equal to 2. The drawback is a notable
12299 increase in code size. This switch implies @option{-mno-push-args}.
12303 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12304 on thread-safe exception handling must compile and link all code with the
12305 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12306 @option{-D_MT}; when linking, it links in a special thread helper library
12307 @option{-lmingwthrd} which cleans up per thread exception handling data.
12309 @item -mno-align-stringops
12310 @opindex mno-align-stringops
12311 Do not align destination of inlined string operations. This switch reduces
12312 code size and improves performance in case the destination is already aligned,
12313 but GCC doesn't know about it.
12315 @item -minline-all-stringops
12316 @opindex minline-all-stringops
12317 By default GCC inlines string operations only when destination is known to be
12318 aligned at least to 4 byte boundary. This enables more inlining, increase code
12319 size, but may improve performance of code that depends on fast memcpy, strlen
12320 and memset for short lengths.
12322 @item -minline-stringops-dynamically
12323 @opindex minline-stringops-dynamically
12324 For string operation of unknown size, inline runtime checks so for small
12325 blocks inline code is used, while for large blocks library call is used.
12327 @item -mstringop-strategy=@var{alg}
12328 @opindex mstringop-strategy=@var{alg}
12329 Overwrite internal decision heuristic about particular algorithm to inline
12330 string operation with. The allowed values are @code{rep_byte},
12331 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12332 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12333 expanding inline loop, @code{libcall} for always expanding library call.
12335 @item -momit-leaf-frame-pointer
12336 @opindex momit-leaf-frame-pointer
12337 Don't keep the frame pointer in a register for leaf functions. This
12338 avoids the instructions to save, set up and restore frame pointers and
12339 makes an extra register available in leaf functions. The option
12340 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12341 which might make debugging harder.
12343 @item -mtls-direct-seg-refs
12344 @itemx -mno-tls-direct-seg-refs
12345 @opindex mtls-direct-seg-refs
12346 Controls whether TLS variables may be accessed with offsets from the
12347 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12348 or whether the thread base pointer must be added. Whether or not this
12349 is legal depends on the operating system, and whether it maps the
12350 segment to cover the entire TLS area.
12352 For systems that use GNU libc, the default is on.
12355 @itemx -mno-sse2avx
12357 Specify that the assembler should encode SSE instructions with VEX
12358 prefix. The option @option{-mavx} turns this on by default.
12361 These @samp{-m} switches are supported in addition to the above
12362 on AMD x86-64 processors in 64-bit environments.
12369 Generate code for a 32-bit or 64-bit environment.
12370 The 32-bit environment sets int, long and pointer to 32 bits and
12371 generates code that runs on any i386 system.
12372 The 64-bit environment sets int to 32 bits and long and pointer
12373 to 64 bits and generates code for AMD's x86-64 architecture. For
12374 darwin only the -m64 option turns off the @option{-fno-pic} and
12375 @option{-mdynamic-no-pic} options.
12377 @item -mno-red-zone
12378 @opindex mno-red-zone
12379 Do not use a so called red zone for x86-64 code. The red zone is mandated
12380 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12381 stack pointer that will not be modified by signal or interrupt handlers
12382 and therefore can be used for temporary data without adjusting the stack
12383 pointer. The flag @option{-mno-red-zone} disables this red zone.
12385 @item -mcmodel=small
12386 @opindex mcmodel=small
12387 Generate code for the small code model: the program and its symbols must
12388 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12389 Programs can be statically or dynamically linked. This is the default
12392 @item -mcmodel=kernel
12393 @opindex mcmodel=kernel
12394 Generate code for the kernel code model. The kernel runs in the
12395 negative 2 GB of the address space.
12396 This model has to be used for Linux kernel code.
12398 @item -mcmodel=medium
12399 @opindex mcmodel=medium
12400 Generate code for the medium model: The program is linked in the lower 2
12401 GB of the address space. Small symbols are also placed there. Symbols
12402 with sizes larger than @option{-mlarge-data-threshold} are put into
12403 large data or bss sections and can be located above 2GB. Programs can
12404 be statically or dynamically linked.
12406 @item -mcmodel=large
12407 @opindex mcmodel=large
12408 Generate code for the large model: This model makes no assumptions
12409 about addresses and sizes of sections.
12412 @node IA-64 Options
12413 @subsection IA-64 Options
12414 @cindex IA-64 Options
12416 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12420 @opindex mbig-endian
12421 Generate code for a big endian target. This is the default for HP-UX@.
12423 @item -mlittle-endian
12424 @opindex mlittle-endian
12425 Generate code for a little endian target. This is the default for AIX5
12431 @opindex mno-gnu-as
12432 Generate (or don't) code for the GNU assembler. This is the default.
12433 @c Also, this is the default if the configure option @option{--with-gnu-as}
12439 @opindex mno-gnu-ld
12440 Generate (or don't) code for the GNU linker. This is the default.
12441 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12446 Generate code that does not use a global pointer register. The result
12447 is not position independent code, and violates the IA-64 ABI@.
12449 @item -mvolatile-asm-stop
12450 @itemx -mno-volatile-asm-stop
12451 @opindex mvolatile-asm-stop
12452 @opindex mno-volatile-asm-stop
12453 Generate (or don't) a stop bit immediately before and after volatile asm
12456 @item -mregister-names
12457 @itemx -mno-register-names
12458 @opindex mregister-names
12459 @opindex mno-register-names
12460 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12461 the stacked registers. This may make assembler output more readable.
12467 Disable (or enable) optimizations that use the small data section. This may
12468 be useful for working around optimizer bugs.
12470 @item -mconstant-gp
12471 @opindex mconstant-gp
12472 Generate code that uses a single constant global pointer value. This is
12473 useful when compiling kernel code.
12477 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12478 This is useful when compiling firmware code.
12480 @item -minline-float-divide-min-latency
12481 @opindex minline-float-divide-min-latency
12482 Generate code for inline divides of floating point values
12483 using the minimum latency algorithm.
12485 @item -minline-float-divide-max-throughput
12486 @opindex minline-float-divide-max-throughput
12487 Generate code for inline divides of floating point values
12488 using the maximum throughput algorithm.
12490 @item -mno-inline-float-divide
12491 @opindex mno-inline-float-divide
12492 Do not generate inline code for divides of floating point values.
12494 @item -minline-int-divide-min-latency
12495 @opindex minline-int-divide-min-latency
12496 Generate code for inline divides of integer values
12497 using the minimum latency algorithm.
12499 @item -minline-int-divide-max-throughput
12500 @opindex minline-int-divide-max-throughput
12501 Generate code for inline divides of integer values
12502 using the maximum throughput algorithm.
12504 @item -mno-inline-int-divide
12505 @opindex mno-inline-int-divide
12506 Do not generate inline code for divides of integer values.
12508 @item -minline-sqrt-min-latency
12509 @opindex minline-sqrt-min-latency
12510 Generate code for inline square roots
12511 using the minimum latency algorithm.
12513 @item -minline-sqrt-max-throughput
12514 @opindex minline-sqrt-max-throughput
12515 Generate code for inline square roots
12516 using the maximum throughput algorithm.
12518 @item -mno-inline-sqrt
12519 @opindex mno-inline-sqrt
12520 Do not generate inline code for sqrt.
12523 @itemx -mno-fused-madd
12524 @opindex mfused-madd
12525 @opindex mno-fused-madd
12526 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12527 instructions. The default is to use these instructions.
12529 @item -mno-dwarf2-asm
12530 @itemx -mdwarf2-asm
12531 @opindex mno-dwarf2-asm
12532 @opindex mdwarf2-asm
12533 Don't (or do) generate assembler code for the DWARF2 line number debugging
12534 info. This may be useful when not using the GNU assembler.
12536 @item -mearly-stop-bits
12537 @itemx -mno-early-stop-bits
12538 @opindex mearly-stop-bits
12539 @opindex mno-early-stop-bits
12540 Allow stop bits to be placed earlier than immediately preceding the
12541 instruction that triggered the stop bit. This can improve instruction
12542 scheduling, but does not always do so.
12544 @item -mfixed-range=@var{register-range}
12545 @opindex mfixed-range
12546 Generate code treating the given register range as fixed registers.
12547 A fixed register is one that the register allocator can not use. This is
12548 useful when compiling kernel code. A register range is specified as
12549 two registers separated by a dash. Multiple register ranges can be
12550 specified separated by a comma.
12552 @item -mtls-size=@var{tls-size}
12554 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12557 @item -mtune=@var{cpu-type}
12559 Tune the instruction scheduling for a particular CPU, Valid values are
12560 itanium, itanium1, merced, itanium2, and mckinley.
12566 Generate code for a 32-bit or 64-bit environment.
12567 The 32-bit environment sets int, long and pointer to 32 bits.
12568 The 64-bit environment sets int to 32 bits and long and pointer
12569 to 64 bits. These are HP-UX specific flags.
12571 @item -mno-sched-br-data-spec
12572 @itemx -msched-br-data-spec
12573 @opindex mno-sched-br-data-spec
12574 @opindex msched-br-data-spec
12575 (Dis/En)able data speculative scheduling before reload.
12576 This will result in generation of the ld.a instructions and
12577 the corresponding check instructions (ld.c / chk.a).
12578 The default is 'disable'.
12580 @item -msched-ar-data-spec
12581 @itemx -mno-sched-ar-data-spec
12582 @opindex msched-ar-data-spec
12583 @opindex mno-sched-ar-data-spec
12584 (En/Dis)able data speculative scheduling after reload.
12585 This will result in generation of the ld.a instructions and
12586 the corresponding check instructions (ld.c / chk.a).
12587 The default is 'enable'.
12589 @item -mno-sched-control-spec
12590 @itemx -msched-control-spec
12591 @opindex mno-sched-control-spec
12592 @opindex msched-control-spec
12593 (Dis/En)able control speculative scheduling. This feature is
12594 available only during region scheduling (i.e.@: before reload).
12595 This will result in generation of the ld.s instructions and
12596 the corresponding check instructions chk.s .
12597 The default is 'disable'.
12599 @item -msched-br-in-data-spec
12600 @itemx -mno-sched-br-in-data-spec
12601 @opindex msched-br-in-data-spec
12602 @opindex mno-sched-br-in-data-spec
12603 (En/Dis)able speculative scheduling of the instructions that
12604 are dependent on the data speculative loads before reload.
12605 This is effective only with @option{-msched-br-data-spec} enabled.
12606 The default is 'enable'.
12608 @item -msched-ar-in-data-spec
12609 @itemx -mno-sched-ar-in-data-spec
12610 @opindex msched-ar-in-data-spec
12611 @opindex mno-sched-ar-in-data-spec
12612 (En/Dis)able speculative scheduling of the instructions that
12613 are dependent on the data speculative loads after reload.
12614 This is effective only with @option{-msched-ar-data-spec} enabled.
12615 The default is 'enable'.
12617 @item -msched-in-control-spec
12618 @itemx -mno-sched-in-control-spec
12619 @opindex msched-in-control-spec
12620 @opindex mno-sched-in-control-spec
12621 (En/Dis)able speculative scheduling of the instructions that
12622 are dependent on the control speculative loads.
12623 This is effective only with @option{-msched-control-spec} enabled.
12624 The default is 'enable'.
12626 @item -mno-sched-prefer-non-data-spec-insns
12627 @itemx -msched-prefer-non-data-spec-insns
12628 @opindex mno-sched-prefer-non-data-spec-insns
12629 @opindex msched-prefer-non-data-spec-insns
12630 If enabled, data speculative instructions will be chosen for schedule
12631 only if there are no other choices at the moment. This will make
12632 the use of the data speculation much more conservative.
12633 The default is 'disable'.
12635 @item -mno-sched-prefer-non-control-spec-insns
12636 @itemx -msched-prefer-non-control-spec-insns
12637 @opindex mno-sched-prefer-non-control-spec-insns
12638 @opindex msched-prefer-non-control-spec-insns
12639 If enabled, control speculative instructions will be chosen for schedule
12640 only if there are no other choices at the moment. This will make
12641 the use of the control speculation much more conservative.
12642 The default is 'disable'.
12644 @item -mno-sched-count-spec-in-critical-path
12645 @itemx -msched-count-spec-in-critical-path
12646 @opindex mno-sched-count-spec-in-critical-path
12647 @opindex msched-count-spec-in-critical-path
12648 If enabled, speculative dependencies will be considered during
12649 computation of the instructions priorities. This will make the use of the
12650 speculation a bit more conservative.
12651 The default is 'disable'.
12653 @item -msched-spec-ldc
12654 @opindex msched-spec-ldc
12655 Use a simple data speculation check. This option is on by default.
12657 @item -msched-control-spec-ldc
12658 @opindex msched-spec-ldc
12659 Use a simple check for control speculation. This option is on by default.
12661 @item -msched-stop-bits-after-every-cycle
12662 @opindex msched-stop-bits-after-every-cycle
12663 Place a stop bit after every cycle when scheduling. This option is on
12666 @item -msched-fp-mem-deps-zero-cost
12667 @opindex msched-fp-mem-deps-zero-cost
12668 Assume that floating-point stores and loads are not likely to cause a conflict
12669 when placed into the same instruction group. This option is disabled by
12672 @item -msel-sched-dont-check-control-spec
12673 @opindex msel-sched-dont-check-control-spec
12674 Generate checks for control speculation in selective scheduling.
12675 This flag is disabled by default.
12677 @item -msched-max-memory-insns=@var{max-insns}
12678 @opindex msched-max-memory-insns
12679 Limit on the number of memory insns per instruction group, giving lower
12680 priority to subsequent memory insns attempting to schedule in the same
12681 instruction group. Frequently useful to prevent cache bank conflicts.
12682 The default value is 1.
12684 @item -msched-max-memory-insns-hard-limit
12685 @opindex msched-max-memory-insns-hard-limit
12686 Disallow more than `msched-max-memory-insns' in instruction group.
12687 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12688 when limit is reached but may still schedule memory operations.
12692 @node IA-64/VMS Options
12693 @subsection IA-64/VMS Options
12695 These @samp{-m} options are defined for the IA-64/VMS implementations:
12698 @item -mvms-return-codes
12699 @opindex mvms-return-codes
12700 Return VMS condition codes from main. The default is to return POSIX
12701 style condition (e.g.@ error) codes.
12703 @item -mdebug-main=@var{prefix}
12704 @opindex mdebug-main=@var{prefix}
12705 Flag the first routine whose name starts with @var{prefix} as the main
12706 routine for the debugger.
12710 Default to 64bit memory allocation routines.
12714 @subsection LM32 Options
12715 @cindex LM32 options
12717 These @option{-m} options are defined for the Lattice Mico32 architecture:
12720 @item -mbarrel-shift-enabled
12721 @opindex mbarrel-shift-enabled
12722 Enable barrel-shift instructions.
12724 @item -mdivide-enabled
12725 @opindex mdivide-enabled
12726 Enable divide and modulus instructions.
12728 @item -mmultiply-enabled
12729 @opindex multiply-enabled
12730 Enable multiply instructions.
12732 @item -msign-extend-enabled
12733 @opindex msign-extend-enabled
12734 Enable sign extend instructions.
12736 @item -muser-enabled
12737 @opindex muser-enabled
12738 Enable user-defined instructions.
12743 @subsection M32C Options
12744 @cindex M32C options
12747 @item -mcpu=@var{name}
12749 Select the CPU for which code is generated. @var{name} may be one of
12750 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12751 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12752 the M32C/80 series.
12756 Specifies that the program will be run on the simulator. This causes
12757 an alternate runtime library to be linked in which supports, for
12758 example, file I/O@. You must not use this option when generating
12759 programs that will run on real hardware; you must provide your own
12760 runtime library for whatever I/O functions are needed.
12762 @item -memregs=@var{number}
12764 Specifies the number of memory-based pseudo-registers GCC will use
12765 during code generation. These pseudo-registers will be used like real
12766 registers, so there is a tradeoff between GCC's ability to fit the
12767 code into available registers, and the performance penalty of using
12768 memory instead of registers. Note that all modules in a program must
12769 be compiled with the same value for this option. Because of that, you
12770 must not use this option with the default runtime libraries gcc
12775 @node M32R/D Options
12776 @subsection M32R/D Options
12777 @cindex M32R/D options
12779 These @option{-m} options are defined for Renesas M32R/D architectures:
12784 Generate code for the M32R/2@.
12788 Generate code for the M32R/X@.
12792 Generate code for the M32R@. This is the default.
12794 @item -mmodel=small
12795 @opindex mmodel=small
12796 Assume all objects live in the lower 16MB of memory (so that their addresses
12797 can be loaded with the @code{ld24} instruction), and assume all subroutines
12798 are reachable with the @code{bl} instruction.
12799 This is the default.
12801 The addressability of a particular object can be set with the
12802 @code{model} attribute.
12804 @item -mmodel=medium
12805 @opindex mmodel=medium
12806 Assume objects may be anywhere in the 32-bit address space (the compiler
12807 will generate @code{seth/add3} instructions to load their addresses), and
12808 assume all subroutines are reachable with the @code{bl} instruction.
12810 @item -mmodel=large
12811 @opindex mmodel=large
12812 Assume objects may be anywhere in the 32-bit address space (the compiler
12813 will generate @code{seth/add3} instructions to load their addresses), and
12814 assume subroutines may not be reachable with the @code{bl} instruction
12815 (the compiler will generate the much slower @code{seth/add3/jl}
12816 instruction sequence).
12819 @opindex msdata=none
12820 Disable use of the small data area. Variables will be put into
12821 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12822 @code{section} attribute has been specified).
12823 This is the default.
12825 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12826 Objects may be explicitly put in the small data area with the
12827 @code{section} attribute using one of these sections.
12829 @item -msdata=sdata
12830 @opindex msdata=sdata
12831 Put small global and static data in the small data area, but do not
12832 generate special code to reference them.
12835 @opindex msdata=use
12836 Put small global and static data in the small data area, and generate
12837 special instructions to reference them.
12841 @cindex smaller data references
12842 Put global and static objects less than or equal to @var{num} bytes
12843 into the small data or bss sections instead of the normal data or bss
12844 sections. The default value of @var{num} is 8.
12845 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12846 for this option to have any effect.
12848 All modules should be compiled with the same @option{-G @var{num}} value.
12849 Compiling with different values of @var{num} may or may not work; if it
12850 doesn't the linker will give an error message---incorrect code will not be
12855 Makes the M32R specific code in the compiler display some statistics
12856 that might help in debugging programs.
12858 @item -malign-loops
12859 @opindex malign-loops
12860 Align all loops to a 32-byte boundary.
12862 @item -mno-align-loops
12863 @opindex mno-align-loops
12864 Do not enforce a 32-byte alignment for loops. This is the default.
12866 @item -missue-rate=@var{number}
12867 @opindex missue-rate=@var{number}
12868 Issue @var{number} instructions per cycle. @var{number} can only be 1
12871 @item -mbranch-cost=@var{number}
12872 @opindex mbranch-cost=@var{number}
12873 @var{number} can only be 1 or 2. If it is 1 then branches will be
12874 preferred over conditional code, if it is 2, then the opposite will
12877 @item -mflush-trap=@var{number}
12878 @opindex mflush-trap=@var{number}
12879 Specifies the trap number to use to flush the cache. The default is
12880 12. Valid numbers are between 0 and 15 inclusive.
12882 @item -mno-flush-trap
12883 @opindex mno-flush-trap
12884 Specifies that the cache cannot be flushed by using a trap.
12886 @item -mflush-func=@var{name}
12887 @opindex mflush-func=@var{name}
12888 Specifies the name of the operating system function to call to flush
12889 the cache. The default is @emph{_flush_cache}, but a function call
12890 will only be used if a trap is not available.
12892 @item -mno-flush-func
12893 @opindex mno-flush-func
12894 Indicates that there is no OS function for flushing the cache.
12898 @node M680x0 Options
12899 @subsection M680x0 Options
12900 @cindex M680x0 options
12902 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
12903 The default settings depend on which architecture was selected when
12904 the compiler was configured; the defaults for the most common choices
12908 @item -march=@var{arch}
12910 Generate code for a specific M680x0 or ColdFire instruction set
12911 architecture. Permissible values of @var{arch} for M680x0
12912 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
12913 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
12914 architectures are selected according to Freescale's ISA classification
12915 and the permissible values are: @samp{isaa}, @samp{isaaplus},
12916 @samp{isab} and @samp{isac}.
12918 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
12919 code for a ColdFire target. The @var{arch} in this macro is one of the
12920 @option{-march} arguments given above.
12922 When used together, @option{-march} and @option{-mtune} select code
12923 that runs on a family of similar processors but that is optimized
12924 for a particular microarchitecture.
12926 @item -mcpu=@var{cpu}
12928 Generate code for a specific M680x0 or ColdFire processor.
12929 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12930 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12931 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12932 below, which also classifies the CPUs into families:
12934 @multitable @columnfractions 0.20 0.80
12935 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12936 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
12937 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12938 @item @samp{5206e} @tab @samp{5206e}
12939 @item @samp{5208} @tab @samp{5207} @samp{5208}
12940 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12941 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12942 @item @samp{5216} @tab @samp{5214} @samp{5216}
12943 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12944 @item @samp{5225} @tab @samp{5224} @samp{5225}
12945 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
12946 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12947 @item @samp{5249} @tab @samp{5249}
12948 @item @samp{5250} @tab @samp{5250}
12949 @item @samp{5271} @tab @samp{5270} @samp{5271}
12950 @item @samp{5272} @tab @samp{5272}
12951 @item @samp{5275} @tab @samp{5274} @samp{5275}
12952 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12953 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
12954 @item @samp{5307} @tab @samp{5307}
12955 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12956 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12957 @item @samp{5407} @tab @samp{5407}
12958 @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}
12961 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12962 @var{arch} is compatible with @var{cpu}. Other combinations of
12963 @option{-mcpu} and @option{-march} are rejected.
12965 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12966 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12967 where the value of @var{family} is given by the table above.
12969 @item -mtune=@var{tune}
12971 Tune the code for a particular microarchitecture, within the
12972 constraints set by @option{-march} and @option{-mcpu}.
12973 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12974 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12975 and @samp{cpu32}. The ColdFire microarchitectures
12976 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12978 You can also use @option{-mtune=68020-40} for code that needs
12979 to run relatively well on 68020, 68030 and 68040 targets.
12980 @option{-mtune=68020-60} is similar but includes 68060 targets
12981 as well. These two options select the same tuning decisions as
12982 @option{-m68020-40} and @option{-m68020-60} respectively.
12984 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12985 when tuning for 680x0 architecture @var{arch}. It also defines
12986 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12987 option is used. If gcc is tuning for a range of architectures,
12988 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12989 it defines the macros for every architecture in the range.
12991 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12992 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12993 of the arguments given above.
12999 Generate output for a 68000. This is the default
13000 when the compiler is configured for 68000-based systems.
13001 It is equivalent to @option{-march=68000}.
13003 Use this option for microcontrollers with a 68000 or EC000 core,
13004 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
13008 Generate output for a 68010. This is the default
13009 when the compiler is configured for 68010-based systems.
13010 It is equivalent to @option{-march=68010}.
13016 Generate output for a 68020. This is the default
13017 when the compiler is configured for 68020-based systems.
13018 It is equivalent to @option{-march=68020}.
13022 Generate output for a 68030. This is the default when the compiler is
13023 configured for 68030-based systems. It is equivalent to
13024 @option{-march=68030}.
13028 Generate output for a 68040. This is the default when the compiler is
13029 configured for 68040-based systems. It is equivalent to
13030 @option{-march=68040}.
13032 This option inhibits the use of 68881/68882 instructions that have to be
13033 emulated by software on the 68040. Use this option if your 68040 does not
13034 have code to emulate those instructions.
13038 Generate output for a 68060. This is the default when the compiler is
13039 configured for 68060-based systems. It is equivalent to
13040 @option{-march=68060}.
13042 This option inhibits the use of 68020 and 68881/68882 instructions that
13043 have to be emulated by software on the 68060. Use this option if your 68060
13044 does not have code to emulate those instructions.
13048 Generate output for a CPU32. This is the default
13049 when the compiler is configured for CPU32-based systems.
13050 It is equivalent to @option{-march=cpu32}.
13052 Use this option for microcontrollers with a
13053 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
13054 68336, 68340, 68341, 68349 and 68360.
13058 Generate output for a 520X ColdFire CPU@. This is the default
13059 when the compiler is configured for 520X-based systems.
13060 It is equivalent to @option{-mcpu=5206}, and is now deprecated
13061 in favor of that option.
13063 Use this option for microcontroller with a 5200 core, including
13064 the MCF5202, MCF5203, MCF5204 and MCF5206.
13068 Generate output for a 5206e ColdFire CPU@. The option is now
13069 deprecated in favor of the equivalent @option{-mcpu=5206e}.
13073 Generate output for a member of the ColdFire 528X family.
13074 The option is now deprecated in favor of the equivalent
13075 @option{-mcpu=528x}.
13079 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
13080 in favor of the equivalent @option{-mcpu=5307}.
13084 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
13085 in favor of the equivalent @option{-mcpu=5407}.
13089 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
13090 This includes use of hardware floating point instructions.
13091 The option is equivalent to @option{-mcpu=547x}, and is now
13092 deprecated in favor of that option.
13096 Generate output for a 68040, without using any of the new instructions.
13097 This results in code which can run relatively efficiently on either a
13098 68020/68881 or a 68030 or a 68040. The generated code does use the
13099 68881 instructions that are emulated on the 68040.
13101 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
13105 Generate output for a 68060, without using any of the new instructions.
13106 This results in code which can run relatively efficiently on either a
13107 68020/68881 or a 68030 or a 68040. The generated code does use the
13108 68881 instructions that are emulated on the 68060.
13110 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
13114 @opindex mhard-float
13116 Generate floating-point instructions. This is the default for 68020
13117 and above, and for ColdFire devices that have an FPU@. It defines the
13118 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
13119 on ColdFire targets.
13122 @opindex msoft-float
13123 Do not generate floating-point instructions; use library calls instead.
13124 This is the default for 68000, 68010, and 68832 targets. It is also
13125 the default for ColdFire devices that have no FPU.
13131 Generate (do not generate) ColdFire hardware divide and remainder
13132 instructions. If @option{-march} is used without @option{-mcpu},
13133 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
13134 architectures. Otherwise, the default is taken from the target CPU
13135 (either the default CPU, or the one specified by @option{-mcpu}). For
13136 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
13137 @option{-mcpu=5206e}.
13139 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
13143 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13144 Additionally, parameters passed on the stack are also aligned to a
13145 16-bit boundary even on targets whose API mandates promotion to 32-bit.
13149 Do not consider type @code{int} to be 16 bits wide. This is the default.
13152 @itemx -mno-bitfield
13153 @opindex mnobitfield
13154 @opindex mno-bitfield
13155 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
13156 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
13160 Do use the bit-field instructions. The @option{-m68020} option implies
13161 @option{-mbitfield}. This is the default if you use a configuration
13162 designed for a 68020.
13166 Use a different function-calling convention, in which functions
13167 that take a fixed number of arguments return with the @code{rtd}
13168 instruction, which pops their arguments while returning. This
13169 saves one instruction in the caller since there is no need to pop
13170 the arguments there.
13172 This calling convention is incompatible with the one normally
13173 used on Unix, so you cannot use it if you need to call libraries
13174 compiled with the Unix compiler.
13176 Also, you must provide function prototypes for all functions that
13177 take variable numbers of arguments (including @code{printf});
13178 otherwise incorrect code will be generated for calls to those
13181 In addition, seriously incorrect code will result if you call a
13182 function with too many arguments. (Normally, extra arguments are
13183 harmlessly ignored.)
13185 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13186 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13190 Do not use the calling conventions selected by @option{-mrtd}.
13191 This is the default.
13194 @itemx -mno-align-int
13195 @opindex malign-int
13196 @opindex mno-align-int
13197 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13198 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13199 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13200 Aligning variables on 32-bit boundaries produces code that runs somewhat
13201 faster on processors with 32-bit busses at the expense of more memory.
13203 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13204 align structures containing the above types differently than
13205 most published application binary interface specifications for the m68k.
13209 Use the pc-relative addressing mode of the 68000 directly, instead of
13210 using a global offset table. At present, this option implies @option{-fpic},
13211 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13212 not presently supported with @option{-mpcrel}, though this could be supported for
13213 68020 and higher processors.
13215 @item -mno-strict-align
13216 @itemx -mstrict-align
13217 @opindex mno-strict-align
13218 @opindex mstrict-align
13219 Do not (do) assume that unaligned memory references will be handled by
13223 Generate code that allows the data segment to be located in a different
13224 area of memory from the text segment. This allows for execute in place in
13225 an environment without virtual memory management. This option implies
13228 @item -mno-sep-data
13229 Generate code that assumes that the data segment follows the text segment.
13230 This is the default.
13232 @item -mid-shared-library
13233 Generate code that supports shared libraries via the library ID method.
13234 This allows for execute in place and shared libraries in an environment
13235 without virtual memory management. This option implies @option{-fPIC}.
13237 @item -mno-id-shared-library
13238 Generate code that doesn't assume ID based shared libraries are being used.
13239 This is the default.
13241 @item -mshared-library-id=n
13242 Specified the identification number of the ID based shared library being
13243 compiled. Specifying a value of 0 will generate more compact code, specifying
13244 other values will force the allocation of that number to the current
13245 library but is no more space or time efficient than omitting this option.
13251 When generating position-independent code for ColdFire, generate code
13252 that works if the GOT has more than 8192 entries. This code is
13253 larger and slower than code generated without this option. On M680x0
13254 processors, this option is not needed; @option{-fPIC} suffices.
13256 GCC normally uses a single instruction to load values from the GOT@.
13257 While this is relatively efficient, it only works if the GOT
13258 is smaller than about 64k. Anything larger causes the linker
13259 to report an error such as:
13261 @cindex relocation truncated to fit (ColdFire)
13263 relocation truncated to fit: R_68K_GOT16O foobar
13266 If this happens, you should recompile your code with @option{-mxgot}.
13267 It should then work with very large GOTs. However, code generated with
13268 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13269 the value of a global symbol.
13271 Note that some linkers, including newer versions of the GNU linker,
13272 can create multiple GOTs and sort GOT entries. If you have such a linker,
13273 you should only need to use @option{-mxgot} when compiling a single
13274 object file that accesses more than 8192 GOT entries. Very few do.
13276 These options have no effect unless GCC is generating
13277 position-independent code.
13281 @node M68hc1x Options
13282 @subsection M68hc1x Options
13283 @cindex M68hc1x options
13285 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13286 microcontrollers. The default values for these options depends on
13287 which style of microcontroller was selected when the compiler was configured;
13288 the defaults for the most common choices are given below.
13295 Generate output for a 68HC11. This is the default
13296 when the compiler is configured for 68HC11-based systems.
13302 Generate output for a 68HC12. This is the default
13303 when the compiler is configured for 68HC12-based systems.
13309 Generate output for a 68HCS12.
13311 @item -mauto-incdec
13312 @opindex mauto-incdec
13313 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13320 Enable the use of 68HC12 min and max instructions.
13323 @itemx -mno-long-calls
13324 @opindex mlong-calls
13325 @opindex mno-long-calls
13326 Treat all calls as being far away (near). If calls are assumed to be
13327 far away, the compiler will use the @code{call} instruction to
13328 call a function and the @code{rtc} instruction for returning.
13332 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13334 @item -msoft-reg-count=@var{count}
13335 @opindex msoft-reg-count
13336 Specify the number of pseudo-soft registers which are used for the
13337 code generation. The maximum number is 32. Using more pseudo-soft
13338 register may or may not result in better code depending on the program.
13339 The default is 4 for 68HC11 and 2 for 68HC12.
13343 @node MCore Options
13344 @subsection MCore Options
13345 @cindex MCore options
13347 These are the @samp{-m} options defined for the Motorola M*Core
13353 @itemx -mno-hardlit
13355 @opindex mno-hardlit
13356 Inline constants into the code stream if it can be done in two
13357 instructions or less.
13363 Use the divide instruction. (Enabled by default).
13365 @item -mrelax-immediate
13366 @itemx -mno-relax-immediate
13367 @opindex mrelax-immediate
13368 @opindex mno-relax-immediate
13369 Allow arbitrary sized immediates in bit operations.
13371 @item -mwide-bitfields
13372 @itemx -mno-wide-bitfields
13373 @opindex mwide-bitfields
13374 @opindex mno-wide-bitfields
13375 Always treat bit-fields as int-sized.
13377 @item -m4byte-functions
13378 @itemx -mno-4byte-functions
13379 @opindex m4byte-functions
13380 @opindex mno-4byte-functions
13381 Force all functions to be aligned to a four byte boundary.
13383 @item -mcallgraph-data
13384 @itemx -mno-callgraph-data
13385 @opindex mcallgraph-data
13386 @opindex mno-callgraph-data
13387 Emit callgraph information.
13390 @itemx -mno-slow-bytes
13391 @opindex mslow-bytes
13392 @opindex mno-slow-bytes
13393 Prefer word access when reading byte quantities.
13395 @item -mlittle-endian
13396 @itemx -mbig-endian
13397 @opindex mlittle-endian
13398 @opindex mbig-endian
13399 Generate code for a little endian target.
13405 Generate code for the 210 processor.
13409 Assume that run-time support has been provided and so omit the
13410 simulator library (@file{libsim.a)} from the linker command line.
13412 @item -mstack-increment=@var{size}
13413 @opindex mstack-increment
13414 Set the maximum amount for a single stack increment operation. Large
13415 values can increase the speed of programs which contain functions
13416 that need a large amount of stack space, but they can also trigger a
13417 segmentation fault if the stack is extended too much. The default
13423 @subsection MeP Options
13424 @cindex MeP options
13430 Enables the @code{abs} instruction, which is the absolute difference
13431 between two registers.
13435 Enables all the optional instructions - average, multiply, divide, bit
13436 operations, leading zero, absolute difference, min/max, clip, and
13442 Enables the @code{ave} instruction, which computes the average of two
13445 @item -mbased=@var{n}
13447 Variables of size @var{n} bytes or smaller will be placed in the
13448 @code{.based} section by default. Based variables use the @code{$tp}
13449 register as a base register, and there is a 128 byte limit to the
13450 @code{.based} section.
13454 Enables the bit operation instructions - bit test (@code{btstm}), set
13455 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13456 test-and-set (@code{tas}).
13458 @item -mc=@var{name}
13460 Selects which section constant data will be placed in. @var{name} may
13461 be @code{tiny}, @code{near}, or @code{far}.
13465 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13466 useful unless you also provide @code{-mminmax}.
13468 @item -mconfig=@var{name}
13470 Selects one of the build-in core configurations. Each MeP chip has
13471 one or more modules in it; each module has a core CPU and a variety of
13472 coprocessors, optional instructions, and peripherals. The
13473 @code{MeP-Integrator} tool, not part of GCC, provides these
13474 configurations through this option; using this option is the same as
13475 using all the corresponding command line options. The default
13476 configuration is @code{default}.
13480 Enables the coprocessor instructions. By default, this is a 32-bit
13481 coprocessor. Note that the coprocessor is normally enabled via the
13482 @code{-mconfig=} option.
13486 Enables the 32-bit coprocessor's instructions.
13490 Enables the 64-bit coprocessor's instructions.
13494 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13498 Causes constant variables to be placed in the @code{.near} section.
13502 Enables the @code{div} and @code{divu} instructions.
13506 Generate big-endian code.
13510 Generate little-endian code.
13512 @item -mio-volatile
13513 @opindex mio-volatile
13514 Tells the compiler that any variable marked with the @code{io}
13515 attribute is to be considered volatile.
13519 Causes variables to be assigned to the @code{.far} section by default.
13523 Enables the @code{leadz} (leading zero) instruction.
13527 Causes variables to be assigned to the @code{.near} section by default.
13531 Enables the @code{min} and @code{max} instructions.
13535 Enables the multiplication and multiply-accumulate instructions.
13539 Disables all the optional instructions enabled by @code{-mall-opts}.
13543 Enables the @code{repeat} and @code{erepeat} instructions, used for
13544 low-overhead looping.
13548 Causes all variables to default to the @code{.tiny} section. Note
13549 that there is a 65536 byte limit to this section. Accesses to these
13550 variables use the @code{%gp} base register.
13554 Enables the saturation instructions. Note that the compiler does not
13555 currently generate these itself, but this option is included for
13556 compatibility with other tools, like @code{as}.
13560 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13564 Link the simulator runtime libraries.
13568 Link the simulator runtime libraries, excluding built-in support
13569 for reset and exception vectors and tables.
13573 Causes all functions to default to the @code{.far} section. Without
13574 this option, functions default to the @code{.near} section.
13576 @item -mtiny=@var{n}
13578 Variables that are @var{n} bytes or smaller will be allocated to the
13579 @code{.tiny} section. These variables use the @code{$gp} base
13580 register. The default for this option is 4, but note that there's a
13581 65536 byte limit to the @code{.tiny} section.
13586 @subsection MIPS Options
13587 @cindex MIPS options
13593 Generate big-endian code.
13597 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13600 @item -march=@var{arch}
13602 Generate code that will run on @var{arch}, which can be the name of a
13603 generic MIPS ISA, or the name of a particular processor.
13605 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13606 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13607 The processor names are:
13608 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13609 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13610 @samp{5kc}, @samp{5kf},
13612 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13613 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13614 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13615 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13616 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13617 @samp{loongson2e}, @samp{loongson2f},
13621 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13622 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13623 @samp{rm7000}, @samp{rm9000},
13624 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13627 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13628 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13630 The special value @samp{from-abi} selects the
13631 most compatible architecture for the selected ABI (that is,
13632 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13634 Native Linux/GNU toolchains also support the value @samp{native},
13635 which selects the best architecture option for the host processor.
13636 @option{-march=native} has no effect if GCC does not recognize
13639 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13640 (for example, @samp{-march=r2k}). Prefixes are optional, and
13641 @samp{vr} may be written @samp{r}.
13643 Names of the form @samp{@var{n}f2_1} refer to processors with
13644 FPUs clocked at half the rate of the core, names of the form
13645 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13646 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13647 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13648 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13649 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13650 accepted as synonyms for @samp{@var{n}f1_1}.
13652 GCC defines two macros based on the value of this option. The first
13653 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13654 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13655 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13656 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13657 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13659 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13660 above. In other words, it will have the full prefix and will not
13661 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13662 the macro names the resolved architecture (either @samp{"mips1"} or
13663 @samp{"mips3"}). It names the default architecture when no
13664 @option{-march} option is given.
13666 @item -mtune=@var{arch}
13668 Optimize for @var{arch}. Among other things, this option controls
13669 the way instructions are scheduled, and the perceived cost of arithmetic
13670 operations. The list of @var{arch} values is the same as for
13673 When this option is not used, GCC will optimize for the processor
13674 specified by @option{-march}. By using @option{-march} and
13675 @option{-mtune} together, it is possible to generate code that will
13676 run on a family of processors, but optimize the code for one
13677 particular member of that family.
13679 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13680 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13681 @samp{-march} ones described above.
13685 Equivalent to @samp{-march=mips1}.
13689 Equivalent to @samp{-march=mips2}.
13693 Equivalent to @samp{-march=mips3}.
13697 Equivalent to @samp{-march=mips4}.
13701 Equivalent to @samp{-march=mips32}.
13705 Equivalent to @samp{-march=mips32r2}.
13709 Equivalent to @samp{-march=mips64}.
13713 Equivalent to @samp{-march=mips64r2}.
13718 @opindex mno-mips16
13719 Generate (do not generate) MIPS16 code. If GCC is targetting a
13720 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13722 MIPS16 code generation can also be controlled on a per-function basis
13723 by means of @code{mips16} and @code{nomips16} attributes.
13724 @xref{Function Attributes}, for more information.
13726 @item -mflip-mips16
13727 @opindex mflip-mips16
13728 Generate MIPS16 code on alternating functions. This option is provided
13729 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13730 not intended for ordinary use in compiling user code.
13732 @item -minterlink-mips16
13733 @itemx -mno-interlink-mips16
13734 @opindex minterlink-mips16
13735 @opindex mno-interlink-mips16
13736 Require (do not require) that non-MIPS16 code be link-compatible with
13739 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13740 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13741 therefore disables direct jumps unless GCC knows that the target of the
13742 jump is not MIPS16.
13754 Generate code for the given ABI@.
13756 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13757 generates 64-bit code when you select a 64-bit architecture, but you
13758 can use @option{-mgp32} to get 32-bit code instead.
13760 For information about the O64 ABI, see
13761 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13763 GCC supports a variant of the o32 ABI in which floating-point registers
13764 are 64 rather than 32 bits wide. You can select this combination with
13765 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13766 and @samp{mfhc1} instructions and is therefore only supported for
13767 MIPS32R2 processors.
13769 The register assignments for arguments and return values remain the
13770 same, but each scalar value is passed in a single 64-bit register
13771 rather than a pair of 32-bit registers. For example, scalar
13772 floating-point values are returned in @samp{$f0} only, not a
13773 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13774 remains the same, but all 64 bits are saved.
13777 @itemx -mno-abicalls
13779 @opindex mno-abicalls
13780 Generate (do not generate) code that is suitable for SVR4-style
13781 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13786 Generate (do not generate) code that is fully position-independent,
13787 and that can therefore be linked into shared libraries. This option
13788 only affects @option{-mabicalls}.
13790 All @option{-mabicalls} code has traditionally been position-independent,
13791 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13792 as an extension, the GNU toolchain allows executables to use absolute
13793 accesses for locally-binding symbols. It can also use shorter GP
13794 initialization sequences and generate direct calls to locally-defined
13795 functions. This mode is selected by @option{-mno-shared}.
13797 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13798 objects that can only be linked by the GNU linker. However, the option
13799 does not affect the ABI of the final executable; it only affects the ABI
13800 of relocatable objects. Using @option{-mno-shared} will generally make
13801 executables both smaller and quicker.
13803 @option{-mshared} is the default.
13809 Assume (do not assume) that the static and dynamic linkers
13810 support PLTs and copy relocations. This option only affects
13811 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13812 has no effect without @samp{-msym32}.
13814 You can make @option{-mplt} the default by configuring
13815 GCC with @option{--with-mips-plt}. The default is
13816 @option{-mno-plt} otherwise.
13822 Lift (do not lift) the usual restrictions on the size of the global
13825 GCC normally uses a single instruction to load values from the GOT@.
13826 While this is relatively efficient, it will only work if the GOT
13827 is smaller than about 64k. Anything larger will cause the linker
13828 to report an error such as:
13830 @cindex relocation truncated to fit (MIPS)
13832 relocation truncated to fit: R_MIPS_GOT16 foobar
13835 If this happens, you should recompile your code with @option{-mxgot}.
13836 It should then work with very large GOTs, although it will also be
13837 less efficient, since it will take three instructions to fetch the
13838 value of a global symbol.
13840 Note that some linkers can create multiple GOTs. If you have such a
13841 linker, you should only need to use @option{-mxgot} when a single object
13842 file accesses more than 64k's worth of GOT entries. Very few do.
13844 These options have no effect unless GCC is generating position
13849 Assume that general-purpose registers are 32 bits wide.
13853 Assume that general-purpose registers are 64 bits wide.
13857 Assume that floating-point registers are 32 bits wide.
13861 Assume that floating-point registers are 64 bits wide.
13864 @opindex mhard-float
13865 Use floating-point coprocessor instructions.
13868 @opindex msoft-float
13869 Do not use floating-point coprocessor instructions. Implement
13870 floating-point calculations using library calls instead.
13872 @item -msingle-float
13873 @opindex msingle-float
13874 Assume that the floating-point coprocessor only supports single-precision
13877 @item -mdouble-float
13878 @opindex mdouble-float
13879 Assume that the floating-point coprocessor supports double-precision
13880 operations. This is the default.
13886 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13887 implement atomic memory built-in functions. When neither option is
13888 specified, GCC will use the instructions if the target architecture
13891 @option{-mllsc} is useful if the runtime environment can emulate the
13892 instructions and @option{-mno-llsc} can be useful when compiling for
13893 nonstandard ISAs. You can make either option the default by
13894 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
13895 respectively. @option{--with-llsc} is the default for some
13896 configurations; see the installation documentation for details.
13902 Use (do not use) revision 1 of the MIPS DSP ASE@.
13903 @xref{MIPS DSP Built-in Functions}. This option defines the
13904 preprocessor macro @samp{__mips_dsp}. It also defines
13905 @samp{__mips_dsp_rev} to 1.
13911 Use (do not use) revision 2 of the MIPS DSP ASE@.
13912 @xref{MIPS DSP Built-in Functions}. This option defines the
13913 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
13914 It also defines @samp{__mips_dsp_rev} to 2.
13917 @itemx -mno-smartmips
13918 @opindex msmartmips
13919 @opindex mno-smartmips
13920 Use (do not use) the MIPS SmartMIPS ASE.
13922 @item -mpaired-single
13923 @itemx -mno-paired-single
13924 @opindex mpaired-single
13925 @opindex mno-paired-single
13926 Use (do not use) paired-single floating-point instructions.
13927 @xref{MIPS Paired-Single Support}. This option requires
13928 hardware floating-point support to be enabled.
13934 Use (do not use) MIPS Digital Media Extension instructions.
13935 This option can only be used when generating 64-bit code and requires
13936 hardware floating-point support to be enabled.
13941 @opindex mno-mips3d
13942 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
13943 The option @option{-mips3d} implies @option{-mpaired-single}.
13949 Use (do not use) MT Multithreading instructions.
13953 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
13954 an explanation of the default and the way that the pointer size is
13959 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
13961 The default size of @code{int}s, @code{long}s and pointers depends on
13962 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
13963 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
13964 32-bit @code{long}s. Pointers are the same size as @code{long}s,
13965 or the same size as integer registers, whichever is smaller.
13971 Assume (do not assume) that all symbols have 32-bit values, regardless
13972 of the selected ABI@. This option is useful in combination with
13973 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
13974 to generate shorter and faster references to symbolic addresses.
13978 Put definitions of externally-visible data in a small data section
13979 if that data is no bigger than @var{num} bytes. GCC can then access
13980 the data more efficiently; see @option{-mgpopt} for details.
13982 The default @option{-G} option depends on the configuration.
13984 @item -mlocal-sdata
13985 @itemx -mno-local-sdata
13986 @opindex mlocal-sdata
13987 @opindex mno-local-sdata
13988 Extend (do not extend) the @option{-G} behavior to local data too,
13989 such as to static variables in C@. @option{-mlocal-sdata} is the
13990 default for all configurations.
13992 If the linker complains that an application is using too much small data,
13993 you might want to try rebuilding the less performance-critical parts with
13994 @option{-mno-local-sdata}. You might also want to build large
13995 libraries with @option{-mno-local-sdata}, so that the libraries leave
13996 more room for the main program.
13998 @item -mextern-sdata
13999 @itemx -mno-extern-sdata
14000 @opindex mextern-sdata
14001 @opindex mno-extern-sdata
14002 Assume (do not assume) that externally-defined data will be in
14003 a small data section if that data is within the @option{-G} limit.
14004 @option{-mextern-sdata} is the default for all configurations.
14006 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
14007 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
14008 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
14009 is placed in a small data section. If @var{Var} is defined by another
14010 module, you must either compile that module with a high-enough
14011 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
14012 definition. If @var{Var} is common, you must link the application
14013 with a high-enough @option{-G} setting.
14015 The easiest way of satisfying these restrictions is to compile
14016 and link every module with the same @option{-G} option. However,
14017 you may wish to build a library that supports several different
14018 small data limits. You can do this by compiling the library with
14019 the highest supported @option{-G} setting and additionally using
14020 @option{-mno-extern-sdata} to stop the library from making assumptions
14021 about externally-defined data.
14027 Use (do not use) GP-relative accesses for symbols that are known to be
14028 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
14029 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
14032 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
14033 might not hold the value of @code{_gp}. For example, if the code is
14034 part of a library that might be used in a boot monitor, programs that
14035 call boot monitor routines will pass an unknown value in @code{$gp}.
14036 (In such situations, the boot monitor itself would usually be compiled
14037 with @option{-G0}.)
14039 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
14040 @option{-mno-extern-sdata}.
14042 @item -membedded-data
14043 @itemx -mno-embedded-data
14044 @opindex membedded-data
14045 @opindex mno-embedded-data
14046 Allocate variables to the read-only data section first if possible, then
14047 next in the small data section if possible, otherwise in data. This gives
14048 slightly slower code than the default, but reduces the amount of RAM required
14049 when executing, and thus may be preferred for some embedded systems.
14051 @item -muninit-const-in-rodata
14052 @itemx -mno-uninit-const-in-rodata
14053 @opindex muninit-const-in-rodata
14054 @opindex mno-uninit-const-in-rodata
14055 Put uninitialized @code{const} variables in the read-only data section.
14056 This option is only meaningful in conjunction with @option{-membedded-data}.
14058 @item -mcode-readable=@var{setting}
14059 @opindex mcode-readable
14060 Specify whether GCC may generate code that reads from executable sections.
14061 There are three possible settings:
14064 @item -mcode-readable=yes
14065 Instructions may freely access executable sections. This is the
14068 @item -mcode-readable=pcrel
14069 MIPS16 PC-relative load instructions can access executable sections,
14070 but other instructions must not do so. This option is useful on 4KSc
14071 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
14072 It is also useful on processors that can be configured to have a dual
14073 instruction/data SRAM interface and that, like the M4K, automatically
14074 redirect PC-relative loads to the instruction RAM.
14076 @item -mcode-readable=no
14077 Instructions must not access executable sections. This option can be
14078 useful on targets that are configured to have a dual instruction/data
14079 SRAM interface but that (unlike the M4K) do not automatically redirect
14080 PC-relative loads to the instruction RAM.
14083 @item -msplit-addresses
14084 @itemx -mno-split-addresses
14085 @opindex msplit-addresses
14086 @opindex mno-split-addresses
14087 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
14088 relocation operators. This option has been superseded by
14089 @option{-mexplicit-relocs} but is retained for backwards compatibility.
14091 @item -mexplicit-relocs
14092 @itemx -mno-explicit-relocs
14093 @opindex mexplicit-relocs
14094 @opindex mno-explicit-relocs
14095 Use (do not use) assembler relocation operators when dealing with symbolic
14096 addresses. The alternative, selected by @option{-mno-explicit-relocs},
14097 is to use assembler macros instead.
14099 @option{-mexplicit-relocs} is the default if GCC was configured
14100 to use an assembler that supports relocation operators.
14102 @item -mcheck-zero-division
14103 @itemx -mno-check-zero-division
14104 @opindex mcheck-zero-division
14105 @opindex mno-check-zero-division
14106 Trap (do not trap) on integer division by zero.
14108 The default is @option{-mcheck-zero-division}.
14110 @item -mdivide-traps
14111 @itemx -mdivide-breaks
14112 @opindex mdivide-traps
14113 @opindex mdivide-breaks
14114 MIPS systems check for division by zero by generating either a
14115 conditional trap or a break instruction. Using traps results in
14116 smaller code, but is only supported on MIPS II and later. Also, some
14117 versions of the Linux kernel have a bug that prevents trap from
14118 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
14119 allow conditional traps on architectures that support them and
14120 @option{-mdivide-breaks} to force the use of breaks.
14122 The default is usually @option{-mdivide-traps}, but this can be
14123 overridden at configure time using @option{--with-divide=breaks}.
14124 Divide-by-zero checks can be completely disabled using
14125 @option{-mno-check-zero-division}.
14130 @opindex mno-memcpy
14131 Force (do not force) the use of @code{memcpy()} for non-trivial block
14132 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
14133 most constant-sized copies.
14136 @itemx -mno-long-calls
14137 @opindex mlong-calls
14138 @opindex mno-long-calls
14139 Disable (do not disable) use of the @code{jal} instruction. Calling
14140 functions using @code{jal} is more efficient but requires the caller
14141 and callee to be in the same 256 megabyte segment.
14143 This option has no effect on abicalls code. The default is
14144 @option{-mno-long-calls}.
14150 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
14151 instructions, as provided by the R4650 ISA@.
14154 @itemx -mno-fused-madd
14155 @opindex mfused-madd
14156 @opindex mno-fused-madd
14157 Enable (disable) use of the floating point multiply-accumulate
14158 instructions, when they are available. The default is
14159 @option{-mfused-madd}.
14161 When multiply-accumulate instructions are used, the intermediate
14162 product is calculated to infinite precision and is not subject to
14163 the FCSR Flush to Zero bit. This may be undesirable in some
14168 Tell the MIPS assembler to not run its preprocessor over user
14169 assembler files (with a @samp{.s} suffix) when assembling them.
14172 @itemx -mno-fix-r4000
14173 @opindex mfix-r4000
14174 @opindex mno-fix-r4000
14175 Work around certain R4000 CPU errata:
14178 A double-word or a variable shift may give an incorrect result if executed
14179 immediately after starting an integer division.
14181 A double-word or a variable shift may give an incorrect result if executed
14182 while an integer multiplication is in progress.
14184 An integer division may give an incorrect result if started in a delay slot
14185 of a taken branch or a jump.
14189 @itemx -mno-fix-r4400
14190 @opindex mfix-r4400
14191 @opindex mno-fix-r4400
14192 Work around certain R4400 CPU errata:
14195 A double-word or a variable shift may give an incorrect result if executed
14196 immediately after starting an integer division.
14200 @itemx -mno-fix-r10000
14201 @opindex mfix-r10000
14202 @opindex mno-fix-r10000
14203 Work around certain R10000 errata:
14206 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14207 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14210 This option can only be used if the target architecture supports
14211 branch-likely instructions. @option{-mfix-r10000} is the default when
14212 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14216 @itemx -mno-fix-vr4120
14217 @opindex mfix-vr4120
14218 Work around certain VR4120 errata:
14221 @code{dmultu} does not always produce the correct result.
14223 @code{div} and @code{ddiv} do not always produce the correct result if one
14224 of the operands is negative.
14226 The workarounds for the division errata rely on special functions in
14227 @file{libgcc.a}. At present, these functions are only provided by
14228 the @code{mips64vr*-elf} configurations.
14230 Other VR4120 errata require a nop to be inserted between certain pairs of
14231 instructions. These errata are handled by the assembler, not by GCC itself.
14234 @opindex mfix-vr4130
14235 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14236 workarounds are implemented by the assembler rather than by GCC,
14237 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14238 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14239 instructions are available instead.
14242 @itemx -mno-fix-sb1
14244 Work around certain SB-1 CPU core errata.
14245 (This flag currently works around the SB-1 revision 2
14246 ``F1'' and ``F2'' floating point errata.)
14248 @item -mr10k-cache-barrier=@var{setting}
14249 @opindex mr10k-cache-barrier
14250 Specify whether GCC should insert cache barriers to avoid the
14251 side-effects of speculation on R10K processors.
14253 In common with many processors, the R10K tries to predict the outcome
14254 of a conditional branch and speculatively executes instructions from
14255 the ``taken'' branch. It later aborts these instructions if the
14256 predicted outcome was wrong. However, on the R10K, even aborted
14257 instructions can have side effects.
14259 This problem only affects kernel stores and, depending on the system,
14260 kernel loads. As an example, a speculatively-executed store may load
14261 the target memory into cache and mark the cache line as dirty, even if
14262 the store itself is later aborted. If a DMA operation writes to the
14263 same area of memory before the ``dirty'' line is flushed, the cached
14264 data will overwrite the DMA-ed data. See the R10K processor manual
14265 for a full description, including other potential problems.
14267 One workaround is to insert cache barrier instructions before every memory
14268 access that might be speculatively executed and that might have side
14269 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14270 controls GCC's implementation of this workaround. It assumes that
14271 aborted accesses to any byte in the following regions will not have
14276 the memory occupied by the current function's stack frame;
14279 the memory occupied by an incoming stack argument;
14282 the memory occupied by an object with a link-time-constant address.
14285 It is the kernel's responsibility to ensure that speculative
14286 accesses to these regions are indeed safe.
14288 If the input program contains a function declaration such as:
14294 then the implementation of @code{foo} must allow @code{j foo} and
14295 @code{jal foo} to be executed speculatively. GCC honors this
14296 restriction for functions it compiles itself. It expects non-GCC
14297 functions (such as hand-written assembly code) to do the same.
14299 The option has three forms:
14302 @item -mr10k-cache-barrier=load-store
14303 Insert a cache barrier before a load or store that might be
14304 speculatively executed and that might have side effects even
14307 @item -mr10k-cache-barrier=store
14308 Insert a cache barrier before a store that might be speculatively
14309 executed and that might have side effects even if aborted.
14311 @item -mr10k-cache-barrier=none
14312 Disable the insertion of cache barriers. This is the default setting.
14315 @item -mflush-func=@var{func}
14316 @itemx -mno-flush-func
14317 @opindex mflush-func
14318 Specifies the function to call to flush the I and D caches, or to not
14319 call any such function. If called, the function must take the same
14320 arguments as the common @code{_flush_func()}, that is, the address of the
14321 memory range for which the cache is being flushed, the size of the
14322 memory range, and the number 3 (to flush both caches). The default
14323 depends on the target GCC was configured for, but commonly is either
14324 @samp{_flush_func} or @samp{__cpu_flush}.
14326 @item mbranch-cost=@var{num}
14327 @opindex mbranch-cost
14328 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14329 This cost is only a heuristic and is not guaranteed to produce
14330 consistent results across releases. A zero cost redundantly selects
14331 the default, which is based on the @option{-mtune} setting.
14333 @item -mbranch-likely
14334 @itemx -mno-branch-likely
14335 @opindex mbranch-likely
14336 @opindex mno-branch-likely
14337 Enable or disable use of Branch Likely instructions, regardless of the
14338 default for the selected architecture. By default, Branch Likely
14339 instructions may be generated if they are supported by the selected
14340 architecture. An exception is for the MIPS32 and MIPS64 architectures
14341 and processors which implement those architectures; for those, Branch
14342 Likely instructions will not be generated by default because the MIPS32
14343 and MIPS64 architectures specifically deprecate their use.
14345 @item -mfp-exceptions
14346 @itemx -mno-fp-exceptions
14347 @opindex mfp-exceptions
14348 Specifies whether FP exceptions are enabled. This affects how we schedule
14349 FP instructions for some processors. The default is that FP exceptions are
14352 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14353 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14356 @item -mvr4130-align
14357 @itemx -mno-vr4130-align
14358 @opindex mvr4130-align
14359 The VR4130 pipeline is two-way superscalar, but can only issue two
14360 instructions together if the first one is 8-byte aligned. When this
14361 option is enabled, GCC will align pairs of instructions that it
14362 thinks should execute in parallel.
14364 This option only has an effect when optimizing for the VR4130.
14365 It normally makes code faster, but at the expense of making it bigger.
14366 It is enabled by default at optimization level @option{-O3}.
14371 Enable (disable) generation of @code{synci} instructions on
14372 architectures that support it. The @code{synci} instructions (if
14373 enabled) will be generated when @code{__builtin___clear_cache()} is
14376 This option defaults to @code{-mno-synci}, but the default can be
14377 overridden by configuring with @code{--with-synci}.
14379 When compiling code for single processor systems, it is generally safe
14380 to use @code{synci}. However, on many multi-core (SMP) systems, it
14381 will not invalidate the instruction caches on all cores and may lead
14382 to undefined behavior.
14384 @item -mrelax-pic-calls
14385 @itemx -mno-relax-pic-calls
14386 @opindex mrelax-pic-calls
14387 Try to turn PIC calls that are normally dispatched via register
14388 @code{$25} into direct calls. This is only possible if the linker can
14389 resolve the destination at link-time and if the destination is within
14390 range for a direct call.
14392 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14393 an assembler and a linker that supports the @code{.reloc} assembly
14394 directive and @code{-mexplicit-relocs} is in effect. With
14395 @code{-mno-explicit-relocs}, this optimization can be performed by the
14396 assembler and the linker alone without help from the compiler.
14398 @item -mmcount-ra-address
14399 @itemx -mno-mcount-ra-address
14400 @opindex mmcount-ra-address
14401 @opindex mno-mcount-ra-address
14402 Emit (do not emit) code that allows @code{_mcount} to modify the
14403 calling function's return address. When enabled, this option extends
14404 the usual @code{_mcount} interface with a new @var{ra-address}
14405 parameter, which has type @code{intptr_t *} and is passed in register
14406 @code{$12}. @code{_mcount} can then modify the return address by
14407 doing both of the following:
14410 Returning the new address in register @code{$31}.
14412 Storing the new address in @code{*@var{ra-address}},
14413 if @var{ra-address} is nonnull.
14416 The default is @option{-mno-mcount-ra-address}.
14421 @subsection MMIX Options
14422 @cindex MMIX Options
14424 These options are defined for the MMIX:
14428 @itemx -mno-libfuncs
14430 @opindex mno-libfuncs
14431 Specify that intrinsic library functions are being compiled, passing all
14432 values in registers, no matter the size.
14435 @itemx -mno-epsilon
14437 @opindex mno-epsilon
14438 Generate floating-point comparison instructions that compare with respect
14439 to the @code{rE} epsilon register.
14441 @item -mabi=mmixware
14443 @opindex mabi=mmixware
14445 Generate code that passes function parameters and return values that (in
14446 the called function) are seen as registers @code{$0} and up, as opposed to
14447 the GNU ABI which uses global registers @code{$231} and up.
14449 @item -mzero-extend
14450 @itemx -mno-zero-extend
14451 @opindex mzero-extend
14452 @opindex mno-zero-extend
14453 When reading data from memory in sizes shorter than 64 bits, use (do not
14454 use) zero-extending load instructions by default, rather than
14455 sign-extending ones.
14458 @itemx -mno-knuthdiv
14460 @opindex mno-knuthdiv
14461 Make the result of a division yielding a remainder have the same sign as
14462 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14463 remainder follows the sign of the dividend. Both methods are
14464 arithmetically valid, the latter being almost exclusively used.
14466 @item -mtoplevel-symbols
14467 @itemx -mno-toplevel-symbols
14468 @opindex mtoplevel-symbols
14469 @opindex mno-toplevel-symbols
14470 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14471 code can be used with the @code{PREFIX} assembly directive.
14475 Generate an executable in the ELF format, rather than the default
14476 @samp{mmo} format used by the @command{mmix} simulator.
14478 @item -mbranch-predict
14479 @itemx -mno-branch-predict
14480 @opindex mbranch-predict
14481 @opindex mno-branch-predict
14482 Use (do not use) the probable-branch instructions, when static branch
14483 prediction indicates a probable branch.
14485 @item -mbase-addresses
14486 @itemx -mno-base-addresses
14487 @opindex mbase-addresses
14488 @opindex mno-base-addresses
14489 Generate (do not generate) code that uses @emph{base addresses}. Using a
14490 base address automatically generates a request (handled by the assembler
14491 and the linker) for a constant to be set up in a global register. The
14492 register is used for one or more base address requests within the range 0
14493 to 255 from the value held in the register. The generally leads to short
14494 and fast code, but the number of different data items that can be
14495 addressed is limited. This means that a program that uses lots of static
14496 data may require @option{-mno-base-addresses}.
14498 @item -msingle-exit
14499 @itemx -mno-single-exit
14500 @opindex msingle-exit
14501 @opindex mno-single-exit
14502 Force (do not force) generated code to have a single exit point in each
14506 @node MN10300 Options
14507 @subsection MN10300 Options
14508 @cindex MN10300 options
14510 These @option{-m} options are defined for Matsushita MN10300 architectures:
14515 Generate code to avoid bugs in the multiply instructions for the MN10300
14516 processors. This is the default.
14518 @item -mno-mult-bug
14519 @opindex mno-mult-bug
14520 Do not generate code to avoid bugs in the multiply instructions for the
14521 MN10300 processors.
14525 Generate code which uses features specific to the AM33 processor.
14529 Do not generate code which uses features specific to the AM33 processor. This
14532 @item -mreturn-pointer-on-d0
14533 @opindex mreturn-pointer-on-d0
14534 When generating a function which returns a pointer, return the pointer
14535 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14536 only in a0, and attempts to call such functions without a prototype
14537 would result in errors. Note that this option is on by default; use
14538 @option{-mno-return-pointer-on-d0} to disable it.
14542 Do not link in the C run-time initialization object file.
14546 Indicate to the linker that it should perform a relaxation optimization pass
14547 to shorten branches, calls and absolute memory addresses. This option only
14548 has an effect when used on the command line for the final link step.
14550 This option makes symbolic debugging impossible.
14553 @node PDP-11 Options
14554 @subsection PDP-11 Options
14555 @cindex PDP-11 Options
14557 These options are defined for the PDP-11:
14562 Use hardware FPP floating point. This is the default. (FIS floating
14563 point on the PDP-11/40 is not supported.)
14566 @opindex msoft-float
14567 Do not use hardware floating point.
14571 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14575 Return floating-point results in memory. This is the default.
14579 Generate code for a PDP-11/40.
14583 Generate code for a PDP-11/45. This is the default.
14587 Generate code for a PDP-11/10.
14589 @item -mbcopy-builtin
14590 @opindex mbcopy-builtin
14591 Use inline @code{movmemhi} patterns for copying memory. This is the
14596 Do not use inline @code{movmemhi} patterns for copying memory.
14602 Use 16-bit @code{int}. This is the default.
14608 Use 32-bit @code{int}.
14611 @itemx -mno-float32
14613 @opindex mno-float32
14614 Use 64-bit @code{float}. This is the default.
14617 @itemx -mno-float64
14619 @opindex mno-float64
14620 Use 32-bit @code{float}.
14624 Use @code{abshi2} pattern. This is the default.
14628 Do not use @code{abshi2} pattern.
14630 @item -mbranch-expensive
14631 @opindex mbranch-expensive
14632 Pretend that branches are expensive. This is for experimenting with
14633 code generation only.
14635 @item -mbranch-cheap
14636 @opindex mbranch-cheap
14637 Do not pretend that branches are expensive. This is the default.
14641 Generate code for a system with split I&D@.
14645 Generate code for a system without split I&D@. This is the default.
14649 Use Unix assembler syntax. This is the default when configured for
14650 @samp{pdp11-*-bsd}.
14654 Use DEC assembler syntax. This is the default when configured for any
14655 PDP-11 target other than @samp{pdp11-*-bsd}.
14658 @node picoChip Options
14659 @subsection picoChip Options
14660 @cindex picoChip options
14662 These @samp{-m} options are defined for picoChip implementations:
14666 @item -mae=@var{ae_type}
14668 Set the instruction set, register set, and instruction scheduling
14669 parameters for array element type @var{ae_type}. Supported values
14670 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14672 @option{-mae=ANY} selects a completely generic AE type. Code
14673 generated with this option will run on any of the other AE types. The
14674 code will not be as efficient as it would be if compiled for a specific
14675 AE type, and some types of operation (e.g., multiplication) will not
14676 work properly on all types of AE.
14678 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14679 for compiled code, and is the default.
14681 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14682 option may suffer from poor performance of byte (char) manipulation,
14683 since the DSP AE does not provide hardware support for byte load/stores.
14685 @item -msymbol-as-address
14686 Enable the compiler to directly use a symbol name as an address in a
14687 load/store instruction, without first loading it into a
14688 register. Typically, the use of this option will generate larger
14689 programs, which run faster than when the option isn't used. However, the
14690 results vary from program to program, so it is left as a user option,
14691 rather than being permanently enabled.
14693 @item -mno-inefficient-warnings
14694 Disables warnings about the generation of inefficient code. These
14695 warnings can be generated, for example, when compiling code which
14696 performs byte-level memory operations on the MAC AE type. The MAC AE has
14697 no hardware support for byte-level memory operations, so all byte
14698 load/stores must be synthesized from word load/store operations. This is
14699 inefficient and a warning will be generated indicating to the programmer
14700 that they should rewrite the code to avoid byte operations, or to target
14701 an AE type which has the necessary hardware support. This option enables
14702 the warning to be turned off.
14706 @node PowerPC Options
14707 @subsection PowerPC Options
14708 @cindex PowerPC options
14710 These are listed under @xref{RS/6000 and PowerPC Options}.
14712 @node RS/6000 and PowerPC Options
14713 @subsection IBM RS/6000 and PowerPC Options
14714 @cindex RS/6000 and PowerPC Options
14715 @cindex IBM RS/6000 and PowerPC Options
14717 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14724 @itemx -mno-powerpc
14725 @itemx -mpowerpc-gpopt
14726 @itemx -mno-powerpc-gpopt
14727 @itemx -mpowerpc-gfxopt
14728 @itemx -mno-powerpc-gfxopt
14730 @itemx -mno-powerpc64
14734 @itemx -mno-popcntb
14736 @itemx -mno-popcntd
14744 @itemx -mno-hard-dfp
14748 @opindex mno-power2
14750 @opindex mno-powerpc
14751 @opindex mpowerpc-gpopt
14752 @opindex mno-powerpc-gpopt
14753 @opindex mpowerpc-gfxopt
14754 @opindex mno-powerpc-gfxopt
14755 @opindex mpowerpc64
14756 @opindex mno-powerpc64
14760 @opindex mno-popcntb
14762 @opindex mno-popcntd
14768 @opindex mno-mfpgpr
14770 @opindex mno-hard-dfp
14771 GCC supports two related instruction set architectures for the
14772 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14773 instructions supported by the @samp{rios} chip set used in the original
14774 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14775 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14776 the IBM 4xx, 6xx, and follow-on microprocessors.
14778 Neither architecture is a subset of the other. However there is a
14779 large common subset of instructions supported by both. An MQ
14780 register is included in processors supporting the POWER architecture.
14782 You use these options to specify which instructions are available on the
14783 processor you are using. The default value of these options is
14784 determined when configuring GCC@. Specifying the
14785 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14786 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14787 rather than the options listed above.
14789 The @option{-mpower} option allows GCC to generate instructions that
14790 are found only in the POWER architecture and to use the MQ register.
14791 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14792 to generate instructions that are present in the POWER2 architecture but
14793 not the original POWER architecture.
14795 The @option{-mpowerpc} option allows GCC to generate instructions that
14796 are found only in the 32-bit subset of the PowerPC architecture.
14797 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14798 GCC to use the optional PowerPC architecture instructions in the
14799 General Purpose group, including floating-point square root. Specifying
14800 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14801 use the optional PowerPC architecture instructions in the Graphics
14802 group, including floating-point select.
14804 The @option{-mmfcrf} option allows GCC to generate the move from
14805 condition register field instruction implemented on the POWER4
14806 processor and other processors that support the PowerPC V2.01
14808 The @option{-mpopcntb} option allows GCC to generate the popcount and
14809 double precision FP reciprocal estimate instruction implemented on the
14810 POWER5 processor and other processors that support the PowerPC V2.02
14812 The @option{-mpopcntd} option allows GCC to generate the popcount
14813 instruction implemented on the POWER7 processor and other processors
14814 that support the PowerPC V2.06 architecture.
14815 The @option{-mfprnd} option allows GCC to generate the FP round to
14816 integer instructions implemented on the POWER5+ processor and other
14817 processors that support the PowerPC V2.03 architecture.
14818 The @option{-mcmpb} option allows GCC to generate the compare bytes
14819 instruction implemented on the POWER6 processor and other processors
14820 that support the PowerPC V2.05 architecture.
14821 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14822 general purpose register instructions implemented on the POWER6X
14823 processor and other processors that support the extended PowerPC V2.05
14825 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14826 point instructions implemented on some POWER processors.
14828 The @option{-mpowerpc64} option allows GCC to generate the additional
14829 64-bit instructions that are found in the full PowerPC64 architecture
14830 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14831 @option{-mno-powerpc64}.
14833 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14834 will use only the instructions in the common subset of both
14835 architectures plus some special AIX common-mode calls, and will not use
14836 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14837 permits GCC to use any instruction from either architecture and to
14838 allow use of the MQ register; specify this for the Motorola MPC601.
14840 @item -mnew-mnemonics
14841 @itemx -mold-mnemonics
14842 @opindex mnew-mnemonics
14843 @opindex mold-mnemonics
14844 Select which mnemonics to use in the generated assembler code. With
14845 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14846 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14847 assembler mnemonics defined for the POWER architecture. Instructions
14848 defined in only one architecture have only one mnemonic; GCC uses that
14849 mnemonic irrespective of which of these options is specified.
14851 GCC defaults to the mnemonics appropriate for the architecture in
14852 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14853 value of these option. Unless you are building a cross-compiler, you
14854 should normally not specify either @option{-mnew-mnemonics} or
14855 @option{-mold-mnemonics}, but should instead accept the default.
14857 @item -mcpu=@var{cpu_type}
14859 Set architecture type, register usage, choice of mnemonics, and
14860 instruction scheduling parameters for machine type @var{cpu_type}.
14861 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14862 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14863 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
14864 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
14865 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14866 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
14867 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
14868 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
14869 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
14870 @samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14871 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14873 @option{-mcpu=common} selects a completely generic processor. Code
14874 generated under this option will run on any POWER or PowerPC processor.
14875 GCC will use only the instructions in the common subset of both
14876 architectures, and will not use the MQ register. GCC assumes a generic
14877 processor model for scheduling purposes.
14879 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14880 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14881 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14882 types, with an appropriate, generic processor model assumed for
14883 scheduling purposes.
14885 The other options specify a specific processor. Code generated under
14886 those options will run best on that processor, and may not run at all on
14889 The @option{-mcpu} options automatically enable or disable the
14892 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
14893 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
14894 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
14895 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
14897 The particular options set for any particular CPU will vary between
14898 compiler versions, depending on what setting seems to produce optimal
14899 code for that CPU; it doesn't necessarily reflect the actual hardware's
14900 capabilities. If you wish to set an individual option to a particular
14901 value, you may specify it after the @option{-mcpu} option, like
14902 @samp{-mcpu=970 -mno-altivec}.
14904 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
14905 not enabled or disabled by the @option{-mcpu} option at present because
14906 AIX does not have full support for these options. You may still
14907 enable or disable them individually if you're sure it'll work in your
14910 @item -mtune=@var{cpu_type}
14912 Set the instruction scheduling parameters for machine type
14913 @var{cpu_type}, but do not set the architecture type, register usage, or
14914 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
14915 values for @var{cpu_type} are used for @option{-mtune} as for
14916 @option{-mcpu}. If both are specified, the code generated will use the
14917 architecture, registers, and mnemonics set by @option{-mcpu}, but the
14918 scheduling parameters set by @option{-mtune}.
14924 Generate code to compute division as reciprocal estimate and iterative
14925 refinement, creating opportunities for increased throughput. This
14926 feature requires: optional PowerPC Graphics instruction set for single
14927 precision and FRE instruction for double precision, assuming divides
14928 cannot generate user-visible traps, and the domain values not include
14929 Infinities, denormals or zero denominator.
14932 @itemx -mno-altivec
14934 @opindex mno-altivec
14935 Generate code that uses (does not use) AltiVec instructions, and also
14936 enable the use of built-in functions that allow more direct access to
14937 the AltiVec instruction set. You may also need to set
14938 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
14944 @opindex mno-vrsave
14945 Generate VRSAVE instructions when generating AltiVec code.
14947 @item -mgen-cell-microcode
14948 @opindex mgen-cell-microcode
14949 Generate Cell microcode instructions
14951 @item -mwarn-cell-microcode
14952 @opindex mwarn-cell-microcode
14953 Warning when a Cell microcode instruction is going to emitted. An example
14954 of a Cell microcode instruction is a variable shift.
14957 @opindex msecure-plt
14958 Generate code that allows ld and ld.so to build executables and shared
14959 libraries with non-exec .plt and .got sections. This is a PowerPC
14960 32-bit SYSV ABI option.
14964 Generate code that uses a BSS .plt section that ld.so fills in, and
14965 requires .plt and .got sections that are both writable and executable.
14966 This is a PowerPC 32-bit SYSV ABI option.
14972 This switch enables or disables the generation of ISEL instructions.
14974 @item -misel=@var{yes/no}
14975 This switch has been deprecated. Use @option{-misel} and
14976 @option{-mno-isel} instead.
14982 This switch enables or disables the generation of SPE simd
14988 @opindex mno-paired
14989 This switch enables or disables the generation of PAIRED simd
14992 @item -mspe=@var{yes/no}
14993 This option has been deprecated. Use @option{-mspe} and
14994 @option{-mno-spe} instead.
15000 Generate code that uses (does not use) vector/scalar (VSX)
15001 instructions, and also enable the use of built-in functions that allow
15002 more direct access to the VSX instruction set.
15004 @item -mfloat-gprs=@var{yes/single/double/no}
15005 @itemx -mfloat-gprs
15006 @opindex mfloat-gprs
15007 This switch enables or disables the generation of floating point
15008 operations on the general purpose registers for architectures that
15011 The argument @var{yes} or @var{single} enables the use of
15012 single-precision floating point operations.
15014 The argument @var{double} enables the use of single and
15015 double-precision floating point operations.
15017 The argument @var{no} disables floating point operations on the
15018 general purpose registers.
15020 This option is currently only available on the MPC854x.
15026 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
15027 targets (including GNU/Linux). The 32-bit environment sets int, long
15028 and pointer to 32 bits and generates code that runs on any PowerPC
15029 variant. The 64-bit environment sets int to 32 bits and long and
15030 pointer to 64 bits, and generates code for PowerPC64, as for
15031 @option{-mpowerpc64}.
15034 @itemx -mno-fp-in-toc
15035 @itemx -mno-sum-in-toc
15036 @itemx -mminimal-toc
15038 @opindex mno-fp-in-toc
15039 @opindex mno-sum-in-toc
15040 @opindex mminimal-toc
15041 Modify generation of the TOC (Table Of Contents), which is created for
15042 every executable file. The @option{-mfull-toc} option is selected by
15043 default. In that case, GCC will allocate at least one TOC entry for
15044 each unique non-automatic variable reference in your program. GCC
15045 will also place floating-point constants in the TOC@. However, only
15046 16,384 entries are available in the TOC@.
15048 If you receive a linker error message that saying you have overflowed
15049 the available TOC space, you can reduce the amount of TOC space used
15050 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
15051 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
15052 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
15053 generate code to calculate the sum of an address and a constant at
15054 run-time instead of putting that sum into the TOC@. You may specify one
15055 or both of these options. Each causes GCC to produce very slightly
15056 slower and larger code at the expense of conserving TOC space.
15058 If you still run out of space in the TOC even when you specify both of
15059 these options, specify @option{-mminimal-toc} instead. This option causes
15060 GCC to make only one TOC entry for every file. When you specify this
15061 option, GCC will produce code that is slower and larger but which
15062 uses extremely little TOC space. You may wish to use this option
15063 only on files that contain less frequently executed code.
15069 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
15070 @code{long} type, and the infrastructure needed to support them.
15071 Specifying @option{-maix64} implies @option{-mpowerpc64} and
15072 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
15073 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
15076 @itemx -mno-xl-compat
15077 @opindex mxl-compat
15078 @opindex mno-xl-compat
15079 Produce code that conforms more closely to IBM XL compiler semantics
15080 when using AIX-compatible ABI@. Pass floating-point arguments to
15081 prototyped functions beyond the register save area (RSA) on the stack
15082 in addition to argument FPRs. Do not assume that most significant
15083 double in 128-bit long double value is properly rounded when comparing
15084 values and converting to double. Use XL symbol names for long double
15087 The AIX calling convention was extended but not initially documented to
15088 handle an obscure K&R C case of calling a function that takes the
15089 address of its arguments with fewer arguments than declared. IBM XL
15090 compilers access floating point arguments which do not fit in the
15091 RSA from the stack when a subroutine is compiled without
15092 optimization. Because always storing floating-point arguments on the
15093 stack is inefficient and rarely needed, this option is not enabled by
15094 default and only is necessary when calling subroutines compiled by IBM
15095 XL compilers without optimization.
15099 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
15100 application written to use message passing with special startup code to
15101 enable the application to run. The system must have PE installed in the
15102 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
15103 must be overridden with the @option{-specs=} option to specify the
15104 appropriate directory location. The Parallel Environment does not
15105 support threads, so the @option{-mpe} option and the @option{-pthread}
15106 option are incompatible.
15108 @item -malign-natural
15109 @itemx -malign-power
15110 @opindex malign-natural
15111 @opindex malign-power
15112 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
15113 @option{-malign-natural} overrides the ABI-defined alignment of larger
15114 types, such as floating-point doubles, on their natural size-based boundary.
15115 The option @option{-malign-power} instructs GCC to follow the ABI-specified
15116 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
15118 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
15122 @itemx -mhard-float
15123 @opindex msoft-float
15124 @opindex mhard-float
15125 Generate code that does not use (uses) the floating-point register set.
15126 Software floating point emulation is provided if you use the
15127 @option{-msoft-float} option, and pass the option to GCC when linking.
15129 @item -msingle-float
15130 @itemx -mdouble-float
15131 @opindex msingle-float
15132 @opindex mdouble-float
15133 Generate code for single or double-precision floating point operations.
15134 @option{-mdouble-float} implies @option{-msingle-float}.
15137 @opindex msimple-fpu
15138 Do not generate sqrt and div instructions for hardware floating point unit.
15142 Specify type of floating point unit. Valid values are @var{sp_lite}
15143 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
15144 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
15145 and @var{dp_full} (equivalent to -mdouble-float).
15148 @opindex mxilinx-fpu
15149 Perform optimizations for floating point unit on Xilinx PPC 405/440.
15152 @itemx -mno-multiple
15154 @opindex mno-multiple
15155 Generate code that uses (does not use) the load multiple word
15156 instructions and the store multiple word instructions. These
15157 instructions are generated by default on POWER systems, and not
15158 generated on PowerPC systems. Do not use @option{-mmultiple} on little
15159 endian PowerPC systems, since those instructions do not work when the
15160 processor is in little endian mode. The exceptions are PPC740 and
15161 PPC750 which permit the instructions usage in little endian mode.
15166 @opindex mno-string
15167 Generate code that uses (does not use) the load string instructions
15168 and the store string word instructions to save multiple registers and
15169 do small block moves. These instructions are generated by default on
15170 POWER systems, and not generated on PowerPC systems. Do not use
15171 @option{-mstring} on little endian PowerPC systems, since those
15172 instructions do not work when the processor is in little endian mode.
15173 The exceptions are PPC740 and PPC750 which permit the instructions
15174 usage in little endian mode.
15179 @opindex mno-update
15180 Generate code that uses (does not use) the load or store instructions
15181 that update the base register to the address of the calculated memory
15182 location. These instructions are generated by default. If you use
15183 @option{-mno-update}, there is a small window between the time that the
15184 stack pointer is updated and the address of the previous frame is
15185 stored, which means code that walks the stack frame across interrupts or
15186 signals may get corrupted data.
15188 @item -mavoid-indexed-addresses
15189 @itemx -mno-avoid-indexed-addresses
15190 @opindex mavoid-indexed-addresses
15191 @opindex mno-avoid-indexed-addresses
15192 Generate code that tries to avoid (not avoid) the use of indexed load
15193 or store instructions. These instructions can incur a performance
15194 penalty on Power6 processors in certain situations, such as when
15195 stepping through large arrays that cross a 16M boundary. This option
15196 is enabled by default when targetting Power6 and disabled otherwise.
15199 @itemx -mno-fused-madd
15200 @opindex mfused-madd
15201 @opindex mno-fused-madd
15202 Generate code that uses (does not use) the floating point multiply and
15203 accumulate instructions. These instructions are generated by default if
15204 hardware floating is used.
15210 Generate code that uses (does not use) the half-word multiply and
15211 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15212 These instructions are generated by default when targetting those
15219 Generate code that uses (does not use) the string-search @samp{dlmzb}
15220 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15221 generated by default when targetting those processors.
15223 @item -mno-bit-align
15225 @opindex mno-bit-align
15226 @opindex mbit-align
15227 On System V.4 and embedded PowerPC systems do not (do) force structures
15228 and unions that contain bit-fields to be aligned to the base type of the
15231 For example, by default a structure containing nothing but 8
15232 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15233 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15234 the structure would be aligned to a 1 byte boundary and be one byte in
15237 @item -mno-strict-align
15238 @itemx -mstrict-align
15239 @opindex mno-strict-align
15240 @opindex mstrict-align
15241 On System V.4 and embedded PowerPC systems do not (do) assume that
15242 unaligned memory references will be handled by the system.
15244 @item -mrelocatable
15245 @itemx -mno-relocatable
15246 @opindex mrelocatable
15247 @opindex mno-relocatable
15248 On embedded PowerPC systems generate code that allows (does not allow)
15249 the program to be relocated to a different address at runtime. If you
15250 use @option{-mrelocatable} on any module, all objects linked together must
15251 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
15253 @item -mrelocatable-lib
15254 @itemx -mno-relocatable-lib
15255 @opindex mrelocatable-lib
15256 @opindex mno-relocatable-lib
15257 On embedded PowerPC systems generate code that allows (does not allow)
15258 the program to be relocated to a different address at runtime. Modules
15259 compiled with @option{-mrelocatable-lib} can be linked with either modules
15260 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
15261 with modules compiled with the @option{-mrelocatable} options.
15267 On System V.4 and embedded PowerPC systems do not (do) assume that
15268 register 2 contains a pointer to a global area pointing to the addresses
15269 used in the program.
15272 @itemx -mlittle-endian
15274 @opindex mlittle-endian
15275 On System V.4 and embedded PowerPC systems compile code for the
15276 processor in little endian mode. The @option{-mlittle-endian} option is
15277 the same as @option{-mlittle}.
15280 @itemx -mbig-endian
15282 @opindex mbig-endian
15283 On System V.4 and embedded PowerPC systems compile code for the
15284 processor in big endian mode. The @option{-mbig-endian} option is
15285 the same as @option{-mbig}.
15287 @item -mdynamic-no-pic
15288 @opindex mdynamic-no-pic
15289 On Darwin and Mac OS X systems, compile code so that it is not
15290 relocatable, but that its external references are relocatable. The
15291 resulting code is suitable for applications, but not shared
15294 @item -mprioritize-restricted-insns=@var{priority}
15295 @opindex mprioritize-restricted-insns
15296 This option controls the priority that is assigned to
15297 dispatch-slot restricted instructions during the second scheduling
15298 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15299 @var{no/highest/second-highest} priority to dispatch slot restricted
15302 @item -msched-costly-dep=@var{dependence_type}
15303 @opindex msched-costly-dep
15304 This option controls which dependences are considered costly
15305 by the target during instruction scheduling. The argument
15306 @var{dependence_type} takes one of the following values:
15307 @var{no}: no dependence is costly,
15308 @var{all}: all dependences are costly,
15309 @var{true_store_to_load}: a true dependence from store to load is costly,
15310 @var{store_to_load}: any dependence from store to load is costly,
15311 @var{number}: any dependence which latency >= @var{number} is costly.
15313 @item -minsert-sched-nops=@var{scheme}
15314 @opindex minsert-sched-nops
15315 This option controls which nop insertion scheme will be used during
15316 the second scheduling pass. The argument @var{scheme} takes one of the
15318 @var{no}: Don't insert nops.
15319 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15320 according to the scheduler's grouping.
15321 @var{regroup_exact}: Insert nops to force costly dependent insns into
15322 separate groups. Insert exactly as many nops as needed to force an insn
15323 to a new group, according to the estimated processor grouping.
15324 @var{number}: Insert nops to force costly dependent insns into
15325 separate groups. Insert @var{number} nops to force an insn to a new group.
15328 @opindex mcall-sysv
15329 On System V.4 and embedded PowerPC systems compile code using calling
15330 conventions that adheres to the March 1995 draft of the System V
15331 Application Binary Interface, PowerPC processor supplement. This is the
15332 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15334 @item -mcall-sysv-eabi
15336 @opindex mcall-sysv-eabi
15337 @opindex mcall-eabi
15338 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15340 @item -mcall-sysv-noeabi
15341 @opindex mcall-sysv-noeabi
15342 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15344 @item -mcall-aixdesc
15346 On System V.4 and embedded PowerPC systems compile code for the AIX
15350 @opindex mcall-linux
15351 On System V.4 and embedded PowerPC systems compile code for the
15352 Linux-based GNU system.
15356 On System V.4 and embedded PowerPC systems compile code for the
15357 Hurd-based GNU system.
15359 @item -mcall-freebsd
15360 @opindex mcall-freebsd
15361 On System V.4 and embedded PowerPC systems compile code for the
15362 FreeBSD operating system.
15364 @item -mcall-netbsd
15365 @opindex mcall-netbsd
15366 On System V.4 and embedded PowerPC systems compile code for the
15367 NetBSD operating system.
15369 @item -mcall-openbsd
15370 @opindex mcall-netbsd
15371 On System V.4 and embedded PowerPC systems compile code for the
15372 OpenBSD operating system.
15374 @item -maix-struct-return
15375 @opindex maix-struct-return
15376 Return all structures in memory (as specified by the AIX ABI)@.
15378 @item -msvr4-struct-return
15379 @opindex msvr4-struct-return
15380 Return structures smaller than 8 bytes in registers (as specified by the
15383 @item -mabi=@var{abi-type}
15385 Extend the current ABI with a particular extension, or remove such extension.
15386 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15387 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15391 Extend the current ABI with SPE ABI extensions. This does not change
15392 the default ABI, instead it adds the SPE ABI extensions to the current
15396 @opindex mabi=no-spe
15397 Disable Booke SPE ABI extensions for the current ABI@.
15399 @item -mabi=ibmlongdouble
15400 @opindex mabi=ibmlongdouble
15401 Change the current ABI to use IBM extended precision long double.
15402 This is a PowerPC 32-bit SYSV ABI option.
15404 @item -mabi=ieeelongdouble
15405 @opindex mabi=ieeelongdouble
15406 Change the current ABI to use IEEE extended precision long double.
15407 This is a PowerPC 32-bit Linux ABI option.
15410 @itemx -mno-prototype
15411 @opindex mprototype
15412 @opindex mno-prototype
15413 On System V.4 and embedded PowerPC systems assume that all calls to
15414 variable argument functions are properly prototyped. Otherwise, the
15415 compiler must insert an instruction before every non prototyped call to
15416 set or clear bit 6 of the condition code register (@var{CR}) to
15417 indicate whether floating point values were passed in the floating point
15418 registers in case the function takes a variable arguments. With
15419 @option{-mprototype}, only calls to prototyped variable argument functions
15420 will set or clear the bit.
15424 On embedded PowerPC systems, assume that the startup module is called
15425 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15426 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15431 On embedded PowerPC systems, assume that the startup module is called
15432 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15437 On embedded PowerPC systems, assume that the startup module is called
15438 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15441 @item -myellowknife
15442 @opindex myellowknife
15443 On embedded PowerPC systems, assume that the startup module is called
15444 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15449 On System V.4 and embedded PowerPC systems, specify that you are
15450 compiling for a VxWorks system.
15454 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15455 header to indicate that @samp{eabi} extended relocations are used.
15461 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15462 Embedded Applications Binary Interface (eabi) which is a set of
15463 modifications to the System V.4 specifications. Selecting @option{-meabi}
15464 means that the stack is aligned to an 8 byte boundary, a function
15465 @code{__eabi} is called to from @code{main} to set up the eabi
15466 environment, and the @option{-msdata} option can use both @code{r2} and
15467 @code{r13} to point to two separate small data areas. Selecting
15468 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15469 do not call an initialization function from @code{main}, and the
15470 @option{-msdata} option will only use @code{r13} to point to a single
15471 small data area. The @option{-meabi} option is on by default if you
15472 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15475 @opindex msdata=eabi
15476 On System V.4 and embedded PowerPC systems, put small initialized
15477 @code{const} global and static data in the @samp{.sdata2} section, which
15478 is pointed to by register @code{r2}. Put small initialized
15479 non-@code{const} global and static data in the @samp{.sdata} section,
15480 which is pointed to by register @code{r13}. Put small uninitialized
15481 global and static data in the @samp{.sbss} section, which is adjacent to
15482 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15483 incompatible with the @option{-mrelocatable} option. The
15484 @option{-msdata=eabi} option also sets the @option{-memb} option.
15487 @opindex msdata=sysv
15488 On System V.4 and embedded PowerPC systems, put small global and static
15489 data in the @samp{.sdata} section, which is pointed to by register
15490 @code{r13}. Put small uninitialized global and static data in the
15491 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15492 The @option{-msdata=sysv} option is incompatible with the
15493 @option{-mrelocatable} option.
15495 @item -msdata=default
15497 @opindex msdata=default
15499 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15500 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15501 same as @option{-msdata=sysv}.
15504 @opindex msdata=data
15505 On System V.4 and embedded PowerPC systems, put small global
15506 data in the @samp{.sdata} section. Put small uninitialized global
15507 data in the @samp{.sbss} section. Do not use register @code{r13}
15508 to address small data however. This is the default behavior unless
15509 other @option{-msdata} options are used.
15513 @opindex msdata=none
15515 On embedded PowerPC systems, put all initialized global and static data
15516 in the @samp{.data} section, and all uninitialized data in the
15517 @samp{.bss} section.
15521 @cindex smaller data references (PowerPC)
15522 @cindex .sdata/.sdata2 references (PowerPC)
15523 On embedded PowerPC systems, put global and static items less than or
15524 equal to @var{num} bytes into the small data or bss sections instead of
15525 the normal data or bss section. By default, @var{num} is 8. The
15526 @option{-G @var{num}} switch is also passed to the linker.
15527 All modules should be compiled with the same @option{-G @var{num}} value.
15530 @itemx -mno-regnames
15532 @opindex mno-regnames
15533 On System V.4 and embedded PowerPC systems do (do not) emit register
15534 names in the assembly language output using symbolic forms.
15537 @itemx -mno-longcall
15539 @opindex mno-longcall
15540 By default assume that all calls are far away so that a longer more
15541 expensive calling sequence is required. This is required for calls
15542 further than 32 megabytes (33,554,432 bytes) from the current location.
15543 A short call will be generated if the compiler knows
15544 the call cannot be that far away. This setting can be overridden by
15545 the @code{shortcall} function attribute, or by @code{#pragma
15548 Some linkers are capable of detecting out-of-range calls and generating
15549 glue code on the fly. On these systems, long calls are unnecessary and
15550 generate slower code. As of this writing, the AIX linker can do this,
15551 as can the GNU linker for PowerPC/64. It is planned to add this feature
15552 to the GNU linker for 32-bit PowerPC systems as well.
15554 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15555 callee, L42'', plus a ``branch island'' (glue code). The two target
15556 addresses represent the callee and the ``branch island''. The
15557 Darwin/PPC linker will prefer the first address and generate a ``bl
15558 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15559 otherwise, the linker will generate ``bl L42'' to call the ``branch
15560 island''. The ``branch island'' is appended to the body of the
15561 calling function; it computes the full 32-bit address of the callee
15564 On Mach-O (Darwin) systems, this option directs the compiler emit to
15565 the glue for every direct call, and the Darwin linker decides whether
15566 to use or discard it.
15568 In the future, we may cause GCC to ignore all longcall specifications
15569 when the linker is known to generate glue.
15571 @item -mtls-markers
15572 @itemx -mno-tls-markers
15573 @opindex mtls-markers
15574 @opindex mno-tls-markers
15575 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15576 specifying the function argument. The relocation allows ld to
15577 reliably associate function call with argument setup instructions for
15578 TLS optimization, which in turn allows gcc to better schedule the
15583 Adds support for multithreading with the @dfn{pthreads} library.
15584 This option sets flags for both the preprocessor and linker.
15589 @subsection RX Options
15592 These command line options are defined for RX targets:
15595 @item -m64bit-doubles
15596 @itemx -m32bit-doubles
15597 @opindex m64bit-doubles
15598 @opindex m32bit-doubles
15599 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
15600 or 32-bits (@option{-m32bit-doubles}) in size. The default is
15601 @option{-m32bit-doubles}. @emph{Note} RX floating point hardware only
15602 works on 32-bit values, which is why the default is
15603 @option{-m32bit-doubles}.
15609 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
15610 floating point hardware. The default is enabled for the @var{RX600}
15611 series and disabled for the @var{RX200} series.
15613 Floating point instructions will only be generated for 32-bit floating
15614 point values however, so if the @option{-m64bit-doubles} option is in
15615 use then the FPU hardware will not be used for doubles.
15617 @emph{Note} If the @option{-fpu} option is enabled then
15618 @option{-funsafe-math-optimizations} is also enabled automatically.
15619 This is because the RX FPU instructions are themselves unsafe.
15621 @item -mcpu=@var{name}
15622 @itemx -patch=@var{name}
15625 Selects the type of RX CPU to be targeted. Currently three types are
15626 supported, the generic @var{RX600} and @var{RX200} series hardware and
15627 the specific @var{RX610} cpu. The default is @var{RX600}.
15629 The only difference between @var{RX600} and @var{RX610} is that the
15630 @var{RX610} does not support the @code{MVTIPL} instruction.
15632 The @var{RX200} series does not have a hardware floating point unit
15633 and so @option{-nofpu} is enabled by default when this type is
15636 @item -mbig-endian-data
15637 @itemx -mlittle-endian-data
15638 @opindex mbig-endian-data
15639 @opindex mlittle-endian-data
15640 Store data (but not code) in the big-endian format. The default is
15641 @option{-mlittle-endian-data}, ie to store data in the little endian
15644 @item -msmall-data-limit=@var{N}
15645 @opindex msmall-data-limit
15646 Specifies the maximum size in bytes of global and static variables
15647 which can be placed into the small data area. Using the small data
15648 area can lead to smaller and faster code, but the size of area is
15649 limited and it is up to the programmer to ensure that the area does
15650 not overflow. Also when the small data area is used one of the RX's
15651 registers (@code{r13}) is reserved for use pointing to this area, so
15652 it is no longer available for use by the compiler. This could result
15653 in slower and/or larger code if variables which once could have been
15654 held in @code{r13} are now pushed onto the stack.
15656 Note, common variables (variables which have not been initialised) and
15657 constants are not placed into the small data area as they are assigned
15658 to other sections in the output executable.
15660 The default value is zero, which disables this feature. Note, this
15661 feature is not enabled by default with higher optimization levels
15662 (@option{-O2} etc) because of the potentially detrimental effects of
15663 reserving register @code{r13}. It is up to the programmer to
15664 experiment and discover whether this feature is of benefit to their
15671 Use the simulator runtime. The default is to use the libgloss board
15674 @item -mas100-syntax
15675 @itemx -mno-as100-syntax
15676 @opindex mas100-syntax
15677 @opindex mno-as100-syntax
15678 When generating assembler output use a syntax that is compatible with
15679 Renesas's AS100 assembler. This syntax can also be handled by the GAS
15680 assembler but it has some restrictions so generating it is not the
15683 @item -mmax-constant-size=@var{N}
15684 @opindex mmax-constant-size
15685 Specifies the maximum size, in bytes, of a constant that can be used as
15686 an operand in a RX instruction. Although the RX instruction set does
15687 allow constants of up to 4 bytes in length to be used in instructions,
15688 a longer value equates to a longer instruction. Thus in some
15689 circumstances it can be beneficial to restrict the size of constants
15690 that are used in instructions. Constants that are too big are instead
15691 placed into a constant pool and referenced via register indirection.
15693 The value @var{N} can be between 0 and 4. A value of 0 (the default)
15694 or 4 means that constants of any size are allowed.
15698 Enable linker relaxation. Linker relaxation is a process whereby the
15699 linker will attempt to reduce the size of a program by finding shorter
15700 versions of various instructions. Disabled by default.
15702 @item -mint-register=@var{N}
15703 @opindex mint-register
15704 Specify the number of registers to reserve for fast interrupt handler
15705 functions. The value @var{N} can be between 0 and 4. A value of 1
15706 means that register @code{r13} will be reserved for the exclusive use
15707 of fast interrupt handlers. A value of 2 reserves @code{r13} and
15708 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
15709 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
15710 A value of 0, the default, does not reserve any registers.
15712 @item -msave-acc-in-interrupts
15713 @opindex msave-acc-in-interrupts
15714 Specifies that interrupt handler functions should preserve the
15715 accumulator register. This is only necessary if normal code might use
15716 the accumulator register, for example because it performs 64-bit
15717 multiplications. The default is to ignore the accumulator as this
15718 makes the interrupt handlers faster.
15722 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
15723 has special significance to the RX port when used with the
15724 @code{interrupt} function attribute. This attribute indicates a
15725 function intended to process fast interrupts. GCC will will ensure
15726 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
15727 and/or @code{r13} and only provided that the normal use of the
15728 corresponding registers have been restricted via the
15729 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
15732 @node S/390 and zSeries Options
15733 @subsection S/390 and zSeries Options
15734 @cindex S/390 and zSeries Options
15736 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15740 @itemx -msoft-float
15741 @opindex mhard-float
15742 @opindex msoft-float
15743 Use (do not use) the hardware floating-point instructions and registers
15744 for floating-point operations. When @option{-msoft-float} is specified,
15745 functions in @file{libgcc.a} will be used to perform floating-point
15746 operations. When @option{-mhard-float} is specified, the compiler
15747 generates IEEE floating-point instructions. This is the default.
15750 @itemx -mno-hard-dfp
15752 @opindex mno-hard-dfp
15753 Use (do not use) the hardware decimal-floating-point instructions for
15754 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15755 specified, functions in @file{libgcc.a} will be used to perform
15756 decimal-floating-point operations. When @option{-mhard-dfp} is
15757 specified, the compiler generates decimal-floating-point hardware
15758 instructions. This is the default for @option{-march=z9-ec} or higher.
15760 @item -mlong-double-64
15761 @itemx -mlong-double-128
15762 @opindex mlong-double-64
15763 @opindex mlong-double-128
15764 These switches control the size of @code{long double} type. A size
15765 of 64bit makes the @code{long double} type equivalent to the @code{double}
15766 type. This is the default.
15769 @itemx -mno-backchain
15770 @opindex mbackchain
15771 @opindex mno-backchain
15772 Store (do not store) the address of the caller's frame as backchain pointer
15773 into the callee's stack frame.
15774 A backchain may be needed to allow debugging using tools that do not understand
15775 DWARF-2 call frame information.
15776 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15777 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15778 the backchain is placed into the topmost word of the 96/160 byte register
15781 In general, code compiled with @option{-mbackchain} is call-compatible with
15782 code compiled with @option{-mmo-backchain}; however, use of the backchain
15783 for debugging purposes usually requires that the whole binary is built with
15784 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15785 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15786 to build a linux kernel use @option{-msoft-float}.
15788 The default is to not maintain the backchain.
15790 @item -mpacked-stack
15791 @itemx -mno-packed-stack
15792 @opindex mpacked-stack
15793 @opindex mno-packed-stack
15794 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15795 specified, the compiler uses the all fields of the 96/160 byte register save
15796 area only for their default purpose; unused fields still take up stack space.
15797 When @option{-mpacked-stack} is specified, register save slots are densely
15798 packed at the top of the register save area; unused space is reused for other
15799 purposes, allowing for more efficient use of the available stack space.
15800 However, when @option{-mbackchain} is also in effect, the topmost word of
15801 the save area is always used to store the backchain, and the return address
15802 register is always saved two words below the backchain.
15804 As long as the stack frame backchain is not used, code generated with
15805 @option{-mpacked-stack} is call-compatible with code generated with
15806 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15807 S/390 or zSeries generated code that uses the stack frame backchain at run
15808 time, not just for debugging purposes. Such code is not call-compatible
15809 with code compiled with @option{-mpacked-stack}. Also, note that the
15810 combination of @option{-mbackchain},
15811 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15812 to build a linux kernel use @option{-msoft-float}.
15814 The default is to not use the packed stack layout.
15817 @itemx -mno-small-exec
15818 @opindex msmall-exec
15819 @opindex mno-small-exec
15820 Generate (or do not generate) code using the @code{bras} instruction
15821 to do subroutine calls.
15822 This only works reliably if the total executable size does not
15823 exceed 64k. The default is to use the @code{basr} instruction instead,
15824 which does not have this limitation.
15830 When @option{-m31} is specified, generate code compliant to the
15831 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15832 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15833 particular to generate 64-bit instructions. For the @samp{s390}
15834 targets, the default is @option{-m31}, while the @samp{s390x}
15835 targets default to @option{-m64}.
15841 When @option{-mzarch} is specified, generate code using the
15842 instructions available on z/Architecture.
15843 When @option{-mesa} is specified, generate code using the
15844 instructions available on ESA/390. Note that @option{-mesa} is
15845 not possible with @option{-m64}.
15846 When generating code compliant to the GNU/Linux for S/390 ABI,
15847 the default is @option{-mesa}. When generating code compliant
15848 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
15854 Generate (or do not generate) code using the @code{mvcle} instruction
15855 to perform block moves. When @option{-mno-mvcle} is specified,
15856 use a @code{mvc} loop instead. This is the default unless optimizing for
15863 Print (or do not print) additional debug information when compiling.
15864 The default is to not print debug information.
15866 @item -march=@var{cpu-type}
15868 Generate code that will run on @var{cpu-type}, which is the name of a system
15869 representing a certain processor type. Possible values for
15870 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
15871 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
15872 When generating code using the instructions available on z/Architecture,
15873 the default is @option{-march=z900}. Otherwise, the default is
15874 @option{-march=g5}.
15876 @item -mtune=@var{cpu-type}
15878 Tune to @var{cpu-type} everything applicable about the generated code,
15879 except for the ABI and the set of available instructions.
15880 The list of @var{cpu-type} values is the same as for @option{-march}.
15881 The default is the value used for @option{-march}.
15884 @itemx -mno-tpf-trace
15885 @opindex mtpf-trace
15886 @opindex mno-tpf-trace
15887 Generate code that adds (does not add) in TPF OS specific branches to trace
15888 routines in the operating system. This option is off by default, even
15889 when compiling for the TPF OS@.
15892 @itemx -mno-fused-madd
15893 @opindex mfused-madd
15894 @opindex mno-fused-madd
15895 Generate code that uses (does not use) the floating point multiply and
15896 accumulate instructions. These instructions are generated by default if
15897 hardware floating point is used.
15899 @item -mwarn-framesize=@var{framesize}
15900 @opindex mwarn-framesize
15901 Emit a warning if the current function exceeds the given frame size. Because
15902 this is a compile time check it doesn't need to be a real problem when the program
15903 runs. It is intended to identify functions which most probably cause
15904 a stack overflow. It is useful to be used in an environment with limited stack
15905 size e.g.@: the linux kernel.
15907 @item -mwarn-dynamicstack
15908 @opindex mwarn-dynamicstack
15909 Emit a warning if the function calls alloca or uses dynamically
15910 sized arrays. This is generally a bad idea with a limited stack size.
15912 @item -mstack-guard=@var{stack-guard}
15913 @itemx -mstack-size=@var{stack-size}
15914 @opindex mstack-guard
15915 @opindex mstack-size
15916 If these options are provided the s390 back end emits additional instructions in
15917 the function prologue which trigger a trap if the stack size is @var{stack-guard}
15918 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
15919 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
15920 the frame size of the compiled function is chosen.
15921 These options are intended to be used to help debugging stack overflow problems.
15922 The additionally emitted code causes only little overhead and hence can also be
15923 used in production like systems without greater performance degradation. The given
15924 values have to be exact powers of 2 and @var{stack-size} has to be greater than
15925 @var{stack-guard} without exceeding 64k.
15926 In order to be efficient the extra code makes the assumption that the stack starts
15927 at an address aligned to the value given by @var{stack-size}.
15928 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
15931 @node Score Options
15932 @subsection Score Options
15933 @cindex Score Options
15935 These options are defined for Score implementations:
15940 Compile code for big endian mode. This is the default.
15944 Compile code for little endian mode.
15948 Disable generate bcnz instruction.
15952 Enable generate unaligned load and store instruction.
15956 Enable the use of multiply-accumulate instructions. Disabled by default.
15960 Specify the SCORE5 as the target architecture.
15964 Specify the SCORE5U of the target architecture.
15968 Specify the SCORE7 as the target architecture. This is the default.
15972 Specify the SCORE7D as the target architecture.
15976 @subsection SH Options
15978 These @samp{-m} options are defined for the SH implementations:
15983 Generate code for the SH1.
15987 Generate code for the SH2.
15990 Generate code for the SH2e.
15994 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
15995 that the floating-point unit is not used.
15997 @item -m2a-single-only
15998 @opindex m2a-single-only
15999 Generate code for the SH2a-FPU, in such a way that no double-precision
16000 floating point operations are used.
16003 @opindex m2a-single
16004 Generate code for the SH2a-FPU assuming the floating-point unit is in
16005 single-precision mode by default.
16009 Generate code for the SH2a-FPU assuming the floating-point unit is in
16010 double-precision mode by default.
16014 Generate code for the SH3.
16018 Generate code for the SH3e.
16022 Generate code for the SH4 without a floating-point unit.
16024 @item -m4-single-only
16025 @opindex m4-single-only
16026 Generate code for the SH4 with a floating-point unit that only
16027 supports single-precision arithmetic.
16031 Generate code for the SH4 assuming the floating-point unit is in
16032 single-precision mode by default.
16036 Generate code for the SH4.
16040 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
16041 floating-point unit is not used.
16043 @item -m4a-single-only
16044 @opindex m4a-single-only
16045 Generate code for the SH4a, in such a way that no double-precision
16046 floating point operations are used.
16049 @opindex m4a-single
16050 Generate code for the SH4a assuming the floating-point unit is in
16051 single-precision mode by default.
16055 Generate code for the SH4a.
16059 Same as @option{-m4a-nofpu}, except that it implicitly passes
16060 @option{-dsp} to the assembler. GCC doesn't generate any DSP
16061 instructions at the moment.
16065 Compile code for the processor in big endian mode.
16069 Compile code for the processor in little endian mode.
16073 Align doubles at 64-bit boundaries. Note that this changes the calling
16074 conventions, and thus some functions from the standard C library will
16075 not work unless you recompile it first with @option{-mdalign}.
16079 Shorten some address references at link time, when possible; uses the
16080 linker option @option{-relax}.
16084 Use 32-bit offsets in @code{switch} tables. The default is to use
16089 Enable the use of bit manipulation instructions on SH2A.
16093 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
16094 alignment constraints.
16098 Comply with the calling conventions defined by Renesas.
16102 Comply with the calling conventions defined by Renesas.
16106 Comply with the calling conventions defined for GCC before the Renesas
16107 conventions were available. This option is the default for all
16108 targets of the SH toolchain except for @samp{sh-symbianelf}.
16111 @opindex mnomacsave
16112 Mark the @code{MAC} register as call-clobbered, even if
16113 @option{-mhitachi} is given.
16117 Increase IEEE-compliance of floating-point code.
16118 At the moment, this is equivalent to @option{-fno-finite-math-only}.
16119 When generating 16 bit SH opcodes, getting IEEE-conforming results for
16120 comparisons of NANs / infinities incurs extra overhead in every
16121 floating point comparison, therefore the default is set to
16122 @option{-ffinite-math-only}.
16124 @item -minline-ic_invalidate
16125 @opindex minline-ic_invalidate
16126 Inline code to invalidate instruction cache entries after setting up
16127 nested function trampolines.
16128 This option has no effect if -musermode is in effect and the selected
16129 code generation option (e.g. -m4) does not allow the use of the icbi
16131 If the selected code generation option does not allow the use of the icbi
16132 instruction, and -musermode is not in effect, the inlined code will
16133 manipulate the instruction cache address array directly with an associative
16134 write. This not only requires privileged mode, but it will also
16135 fail if the cache line had been mapped via the TLB and has become unmapped.
16139 Dump instruction size and location in the assembly code.
16142 @opindex mpadstruct
16143 This option is deprecated. It pads structures to multiple of 4 bytes,
16144 which is incompatible with the SH ABI@.
16148 Optimize for space instead of speed. Implied by @option{-Os}.
16151 @opindex mprefergot
16152 When generating position-independent code, emit function calls using
16153 the Global Offset Table instead of the Procedure Linkage Table.
16157 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
16158 if the inlined code would not work in user mode.
16159 This is the default when the target is @code{sh-*-linux*}.
16161 @item -multcost=@var{number}
16162 @opindex multcost=@var{number}
16163 Set the cost to assume for a multiply insn.
16165 @item -mdiv=@var{strategy}
16166 @opindex mdiv=@var{strategy}
16167 Set the division strategy to use for SHmedia code. @var{strategy} must be
16168 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
16169 inv:call2, inv:fp .
16170 "fp" performs the operation in floating point. This has a very high latency,
16171 but needs only a few instructions, so it might be a good choice if
16172 your code has enough easily exploitable ILP to allow the compiler to
16173 schedule the floating point instructions together with other instructions.
16174 Division by zero causes a floating point exception.
16175 "inv" uses integer operations to calculate the inverse of the divisor,
16176 and then multiplies the dividend with the inverse. This strategy allows
16177 cse and hoisting of the inverse calculation. Division by zero calculates
16178 an unspecified result, but does not trap.
16179 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
16180 have been found, or if the entire operation has been hoisted to the same
16181 place, the last stages of the inverse calculation are intertwined with the
16182 final multiply to reduce the overall latency, at the expense of using a few
16183 more instructions, and thus offering fewer scheduling opportunities with
16185 "call" calls a library function that usually implements the inv:minlat
16187 This gives high code density for m5-*media-nofpu compilations.
16188 "call2" uses a different entry point of the same library function, where it
16189 assumes that a pointer to a lookup table has already been set up, which
16190 exposes the pointer load to cse / code hoisting optimizations.
16191 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
16192 code generation, but if the code stays unoptimized, revert to the "call",
16193 "call2", or "fp" strategies, respectively. Note that the
16194 potentially-trapping side effect of division by zero is carried by a
16195 separate instruction, so it is possible that all the integer instructions
16196 are hoisted out, but the marker for the side effect stays where it is.
16197 A recombination to fp operations or a call is not possible in that case.
16198 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
16199 that the inverse calculation was nor separated from the multiply, they speed
16200 up division where the dividend fits into 20 bits (plus sign where applicable),
16201 by inserting a test to skip a number of operations in this case; this test
16202 slows down the case of larger dividends. inv20u assumes the case of a such
16203 a small dividend to be unlikely, and inv20l assumes it to be likely.
16205 @item -maccumulate-outgoing-args
16206 @opindex maccumulate-outgoing-args
16207 Reserve space once for outgoing arguments in the function prologue rather
16208 than around each call. Generally beneficial for performance and size. Also
16209 needed for unwinding to avoid changing the stack frame around conditional code.
16211 @item -mdivsi3_libfunc=@var{name}
16212 @opindex mdivsi3_libfunc=@var{name}
16213 Set the name of the library function used for 32 bit signed division to
16214 @var{name}. This only affect the name used in the call and inv:call
16215 division strategies, and the compiler will still expect the same
16216 sets of input/output/clobbered registers as if this option was not present.
16218 @item -mfixed-range=@var{register-range}
16219 @opindex mfixed-range
16220 Generate code treating the given register range as fixed registers.
16221 A fixed register is one that the register allocator can not use. This is
16222 useful when compiling kernel code. A register range is specified as
16223 two registers separated by a dash. Multiple register ranges can be
16224 specified separated by a comma.
16226 @item -madjust-unroll
16227 @opindex madjust-unroll
16228 Throttle unrolling to avoid thrashing target registers.
16229 This option only has an effect if the gcc code base supports the
16230 TARGET_ADJUST_UNROLL_MAX target hook.
16232 @item -mindexed-addressing
16233 @opindex mindexed-addressing
16234 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16235 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16236 semantics for the indexed addressing mode. The architecture allows the
16237 implementation of processors with 64 bit MMU, which the OS could use to
16238 get 32 bit addressing, but since no current hardware implementation supports
16239 this or any other way to make the indexed addressing mode safe to use in
16240 the 32 bit ABI, the default is -mno-indexed-addressing.
16242 @item -mgettrcost=@var{number}
16243 @opindex mgettrcost=@var{number}
16244 Set the cost assumed for the gettr instruction to @var{number}.
16245 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16249 Assume pt* instructions won't trap. This will generally generate better
16250 scheduled code, but is unsafe on current hardware. The current architecture
16251 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16252 This has the unintentional effect of making it unsafe to schedule ptabs /
16253 ptrel before a branch, or hoist it out of a loop. For example,
16254 __do_global_ctors, a part of libgcc that runs constructors at program
16255 startup, calls functions in a list which is delimited by @minus{}1. With the
16256 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16257 That means that all the constructors will be run a bit quicker, but when
16258 the loop comes to the end of the list, the program crashes because ptabs
16259 loads @minus{}1 into a target register. Since this option is unsafe for any
16260 hardware implementing the current architecture specification, the default
16261 is -mno-pt-fixed. Unless the user specifies a specific cost with
16262 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16263 this deters register allocation using target registers for storing
16266 @item -minvalid-symbols
16267 @opindex minvalid-symbols
16268 Assume symbols might be invalid. Ordinary function symbols generated by
16269 the compiler will always be valid to load with movi/shori/ptabs or
16270 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16271 to generate symbols that will cause ptabs / ptrel to trap.
16272 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16273 It will then prevent cross-basic-block cse, hoisting and most scheduling
16274 of symbol loads. The default is @option{-mno-invalid-symbols}.
16277 @node SPARC Options
16278 @subsection SPARC Options
16279 @cindex SPARC options
16281 These @samp{-m} options are supported on the SPARC:
16284 @item -mno-app-regs
16286 @opindex mno-app-regs
16288 Specify @option{-mapp-regs} to generate output using the global registers
16289 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16292 To be fully SVR4 ABI compliant at the cost of some performance loss,
16293 specify @option{-mno-app-regs}. You should compile libraries and system
16294 software with this option.
16297 @itemx -mhard-float
16299 @opindex mhard-float
16300 Generate output containing floating point instructions. This is the
16304 @itemx -msoft-float
16306 @opindex msoft-float
16307 Generate output containing library calls for floating point.
16308 @strong{Warning:} the requisite libraries are not available for all SPARC
16309 targets. Normally the facilities of the machine's usual C compiler are
16310 used, but this cannot be done directly in cross-compilation. You must make
16311 your own arrangements to provide suitable library functions for
16312 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16313 @samp{sparclite-*-*} do provide software floating point support.
16315 @option{-msoft-float} changes the calling convention in the output file;
16316 therefore, it is only useful if you compile @emph{all} of a program with
16317 this option. In particular, you need to compile @file{libgcc.a}, the
16318 library that comes with GCC, with @option{-msoft-float} in order for
16321 @item -mhard-quad-float
16322 @opindex mhard-quad-float
16323 Generate output containing quad-word (long double) floating point
16326 @item -msoft-quad-float
16327 @opindex msoft-quad-float
16328 Generate output containing library calls for quad-word (long double)
16329 floating point instructions. The functions called are those specified
16330 in the SPARC ABI@. This is the default.
16332 As of this writing, there are no SPARC implementations that have hardware
16333 support for the quad-word floating point instructions. They all invoke
16334 a trap handler for one of these instructions, and then the trap handler
16335 emulates the effect of the instruction. Because of the trap handler overhead,
16336 this is much slower than calling the ABI library routines. Thus the
16337 @option{-msoft-quad-float} option is the default.
16339 @item -mno-unaligned-doubles
16340 @itemx -munaligned-doubles
16341 @opindex mno-unaligned-doubles
16342 @opindex munaligned-doubles
16343 Assume that doubles have 8 byte alignment. This is the default.
16345 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16346 alignment only if they are contained in another type, or if they have an
16347 absolute address. Otherwise, it assumes they have 4 byte alignment.
16348 Specifying this option avoids some rare compatibility problems with code
16349 generated by other compilers. It is not the default because it results
16350 in a performance loss, especially for floating point code.
16352 @item -mno-faster-structs
16353 @itemx -mfaster-structs
16354 @opindex mno-faster-structs
16355 @opindex mfaster-structs
16356 With @option{-mfaster-structs}, the compiler assumes that structures
16357 should have 8 byte alignment. This enables the use of pairs of
16358 @code{ldd} and @code{std} instructions for copies in structure
16359 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16360 However, the use of this changed alignment directly violates the SPARC
16361 ABI@. Thus, it's intended only for use on targets where the developer
16362 acknowledges that their resulting code will not be directly in line with
16363 the rules of the ABI@.
16365 @item -mimpure-text
16366 @opindex mimpure-text
16367 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16368 the compiler to not pass @option{-z text} to the linker when linking a
16369 shared object. Using this option, you can link position-dependent
16370 code into a shared object.
16372 @option{-mimpure-text} suppresses the ``relocations remain against
16373 allocatable but non-writable sections'' linker error message.
16374 However, the necessary relocations will trigger copy-on-write, and the
16375 shared object is not actually shared across processes. Instead of
16376 using @option{-mimpure-text}, you should compile all source code with
16377 @option{-fpic} or @option{-fPIC}.
16379 This option is only available on SunOS and Solaris.
16381 @item -mcpu=@var{cpu_type}
16383 Set the instruction set, register set, and instruction scheduling parameters
16384 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16385 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16386 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16387 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16388 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16390 Default instruction scheduling parameters are used for values that select
16391 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16392 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16394 Here is a list of each supported architecture and their supported
16399 v8: supersparc, hypersparc
16400 sparclite: f930, f934, sparclite86x
16402 v9: ultrasparc, ultrasparc3, niagara, niagara2
16405 By default (unless configured otherwise), GCC generates code for the V7
16406 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16407 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16408 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16409 SPARCStation 1, 2, IPX etc.
16411 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16412 architecture. The only difference from V7 code is that the compiler emits
16413 the integer multiply and integer divide instructions which exist in SPARC-V8
16414 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16415 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16418 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16419 the SPARC architecture. This adds the integer multiply, integer divide step
16420 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16421 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16422 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16423 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16424 MB86934 chip, which is the more recent SPARClite with FPU@.
16426 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16427 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16428 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16429 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16430 optimizes it for the TEMIC SPARClet chip.
16432 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16433 architecture. This adds 64-bit integer and floating-point move instructions,
16434 3 additional floating-point condition code registers and conditional move
16435 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16436 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16437 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
16438 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
16439 @option{-mcpu=niagara}, the compiler additionally optimizes it for
16440 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
16441 additionally optimizes it for Sun UltraSPARC T2 chips.
16443 @item -mtune=@var{cpu_type}
16445 Set the instruction scheduling parameters for machine type
16446 @var{cpu_type}, but do not set the instruction set or register set that the
16447 option @option{-mcpu=@var{cpu_type}} would.
16449 The same values for @option{-mcpu=@var{cpu_type}} can be used for
16450 @option{-mtune=@var{cpu_type}}, but the only useful values are those
16451 that select a particular cpu implementation. Those are @samp{cypress},
16452 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
16453 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
16454 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
16459 @opindex mno-v8plus
16460 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
16461 difference from the V8 ABI is that the global and out registers are
16462 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
16463 mode for all SPARC-V9 processors.
16469 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
16470 Visual Instruction Set extensions. The default is @option{-mno-vis}.
16473 These @samp{-m} options are supported in addition to the above
16474 on SPARC-V9 processors in 64-bit environments:
16477 @item -mlittle-endian
16478 @opindex mlittle-endian
16479 Generate code for a processor running in little-endian mode. It is only
16480 available for a few configurations and most notably not on Solaris and Linux.
16486 Generate code for a 32-bit or 64-bit environment.
16487 The 32-bit environment sets int, long and pointer to 32 bits.
16488 The 64-bit environment sets int to 32 bits and long and pointer
16491 @item -mcmodel=medlow
16492 @opindex mcmodel=medlow
16493 Generate code for the Medium/Low code model: 64-bit addresses, programs
16494 must be linked in the low 32 bits of memory. Programs can be statically
16495 or dynamically linked.
16497 @item -mcmodel=medmid
16498 @opindex mcmodel=medmid
16499 Generate code for the Medium/Middle code model: 64-bit addresses, programs
16500 must be linked in the low 44 bits of memory, the text and data segments must
16501 be less than 2GB in size and the data segment must be located within 2GB of
16504 @item -mcmodel=medany
16505 @opindex mcmodel=medany
16506 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
16507 may be linked anywhere in memory, the text and data segments must be less
16508 than 2GB in size and the data segment must be located within 2GB of the
16511 @item -mcmodel=embmedany
16512 @opindex mcmodel=embmedany
16513 Generate code for the Medium/Anywhere code model for embedded systems:
16514 64-bit addresses, the text and data segments must be less than 2GB in
16515 size, both starting anywhere in memory (determined at link time). The
16516 global register %g4 points to the base of the data segment. Programs
16517 are statically linked and PIC is not supported.
16520 @itemx -mno-stack-bias
16521 @opindex mstack-bias
16522 @opindex mno-stack-bias
16523 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
16524 frame pointer if present, are offset by @minus{}2047 which must be added back
16525 when making stack frame references. This is the default in 64-bit mode.
16526 Otherwise, assume no such offset is present.
16529 These switches are supported in addition to the above on Solaris:
16534 Add support for multithreading using the Solaris threads library. This
16535 option sets flags for both the preprocessor and linker. This option does
16536 not affect the thread safety of object code produced by the compiler or
16537 that of libraries supplied with it.
16541 Add support for multithreading using the POSIX threads library. This
16542 option sets flags for both the preprocessor and linker. This option does
16543 not affect the thread safety of object code produced by the compiler or
16544 that of libraries supplied with it.
16548 This is a synonym for @option{-pthreads}.
16552 @subsection SPU Options
16553 @cindex SPU options
16555 These @samp{-m} options are supported on the SPU:
16559 @itemx -merror-reloc
16560 @opindex mwarn-reloc
16561 @opindex merror-reloc
16563 The loader for SPU does not handle dynamic relocations. By default, GCC
16564 will give an error when it generates code that requires a dynamic
16565 relocation. @option{-mno-error-reloc} disables the error,
16566 @option{-mwarn-reloc} will generate a warning instead.
16569 @itemx -munsafe-dma
16571 @opindex munsafe-dma
16573 Instructions which initiate or test completion of DMA must not be
16574 reordered with respect to loads and stores of the memory which is being
16575 accessed. Users typically address this problem using the volatile
16576 keyword, but that can lead to inefficient code in places where the
16577 memory is known to not change. Rather than mark the memory as volatile
16578 we treat the DMA instructions as potentially effecting all memory. With
16579 @option{-munsafe-dma} users must use the volatile keyword to protect
16582 @item -mbranch-hints
16583 @opindex mbranch-hints
16585 By default, GCC will generate a branch hint instruction to avoid
16586 pipeline stalls for always taken or probably taken branches. A hint
16587 will not be generated closer than 8 instructions away from its branch.
16588 There is little reason to disable them, except for debugging purposes,
16589 or to make an object a little bit smaller.
16593 @opindex msmall-mem
16594 @opindex mlarge-mem
16596 By default, GCC generates code assuming that addresses are never larger
16597 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
16598 a full 32 bit address.
16603 By default, GCC links against startup code that assumes the SPU-style
16604 main function interface (which has an unconventional parameter list).
16605 With @option{-mstdmain}, GCC will link your program against startup
16606 code that assumes a C99-style interface to @code{main}, including a
16607 local copy of @code{argv} strings.
16609 @item -mfixed-range=@var{register-range}
16610 @opindex mfixed-range
16611 Generate code treating the given register range as fixed registers.
16612 A fixed register is one that the register allocator can not use. This is
16613 useful when compiling kernel code. A register range is specified as
16614 two registers separated by a dash. Multiple register ranges can be
16615 specified separated by a comma.
16621 Compile code assuming that pointers to the PPU address space accessed
16622 via the @code{__ea} named address space qualifier are either 32 or 64
16623 bits wide. The default is 32 bits. As this is an ABI changing option,
16624 all object code in an executable must be compiled with the same setting.
16626 @item -maddress-space-conversion
16627 @itemx -mno-address-space-conversion
16628 @opindex maddress-space-conversion
16629 @opindex mno-address-space-conversion
16630 Allow/disallow treating the @code{__ea} address space as superset
16631 of the generic address space. This enables explicit type casts
16632 between @code{__ea} and generic pointer as well as implicit
16633 conversions of generic pointers to @code{__ea} pointers. The
16634 default is to allow address space pointer conversions.
16636 @item -mcache-size=@var{cache-size}
16637 @opindex mcache-size
16638 This option controls the version of libgcc that the compiler links to an
16639 executable and selects a software-managed cache for accessing variables
16640 in the @code{__ea} address space with a particular cache size. Possible
16641 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
16642 and @samp{128}. The default cache size is 64KB.
16644 @item -matomic-updates
16645 @itemx -mno-atomic-updates
16646 @opindex matomic-updates
16647 @opindex mno-atomic-updates
16648 This option controls the version of libgcc that the compiler links to an
16649 executable and selects whether atomic updates to the software-managed
16650 cache of PPU-side variables are used. If you use atomic updates, changes
16651 to a PPU variable from SPU code using the @code{__ea} named address space
16652 qualifier will not interfere with changes to other PPU variables residing
16653 in the same cache line from PPU code. If you do not use atomic updates,
16654 such interference may occur; however, writing back cache lines will be
16655 more efficient. The default behavior is to use atomic updates.
16658 @itemx -mdual-nops=@var{n}
16659 @opindex mdual-nops
16660 By default, GCC will insert nops to increase dual issue when it expects
16661 it to increase performance. @var{n} can be a value from 0 to 10. A
16662 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
16663 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
16665 @item -mhint-max-nops=@var{n}
16666 @opindex mhint-max-nops
16667 Maximum number of nops to insert for a branch hint. A branch hint must
16668 be at least 8 instructions away from the branch it is effecting. GCC
16669 will insert up to @var{n} nops to enforce this, otherwise it will not
16670 generate the branch hint.
16672 @item -mhint-max-distance=@var{n}
16673 @opindex mhint-max-distance
16674 The encoding of the branch hint instruction limits the hint to be within
16675 256 instructions of the branch it is effecting. By default, GCC makes
16676 sure it is within 125.
16679 @opindex msafe-hints
16680 Work around a hardware bug which causes the SPU to stall indefinitely.
16681 By default, GCC will insert the @code{hbrp} instruction to make sure
16682 this stall won't happen.
16686 @node System V Options
16687 @subsection Options for System V
16689 These additional options are available on System V Release 4 for
16690 compatibility with other compilers on those systems:
16695 Create a shared object.
16696 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16700 Identify the versions of each tool used by the compiler, in a
16701 @code{.ident} assembler directive in the output.
16705 Refrain from adding @code{.ident} directives to the output file (this is
16708 @item -YP,@var{dirs}
16710 Search the directories @var{dirs}, and no others, for libraries
16711 specified with @option{-l}.
16713 @item -Ym,@var{dir}
16715 Look in the directory @var{dir} to find the M4 preprocessor.
16716 The assembler uses this option.
16717 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16718 @c the generic assembler that comes with Solaris takes just -Ym.
16722 @subsection V850 Options
16723 @cindex V850 Options
16725 These @samp{-m} options are defined for V850 implementations:
16729 @itemx -mno-long-calls
16730 @opindex mlong-calls
16731 @opindex mno-long-calls
16732 Treat all calls as being far away (near). If calls are assumed to be
16733 far away, the compiler will always load the functions address up into a
16734 register, and call indirect through the pointer.
16740 Do not optimize (do optimize) basic blocks that use the same index
16741 pointer 4 or more times to copy pointer into the @code{ep} register, and
16742 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16743 option is on by default if you optimize.
16745 @item -mno-prolog-function
16746 @itemx -mprolog-function
16747 @opindex mno-prolog-function
16748 @opindex mprolog-function
16749 Do not use (do use) external functions to save and restore registers
16750 at the prologue and epilogue of a function. The external functions
16751 are slower, but use less code space if more than one function saves
16752 the same number of registers. The @option{-mprolog-function} option
16753 is on by default if you optimize.
16757 Try to make the code as small as possible. At present, this just turns
16758 on the @option{-mep} and @option{-mprolog-function} options.
16760 @item -mtda=@var{n}
16762 Put static or global variables whose size is @var{n} bytes or less into
16763 the tiny data area that register @code{ep} points to. The tiny data
16764 area can hold up to 256 bytes in total (128 bytes for byte references).
16766 @item -msda=@var{n}
16768 Put static or global variables whose size is @var{n} bytes or less into
16769 the small data area that register @code{gp} points to. The small data
16770 area can hold up to 64 kilobytes.
16772 @item -mzda=@var{n}
16774 Put static or global variables whose size is @var{n} bytes or less into
16775 the first 32 kilobytes of memory.
16779 Specify that the target processor is the V850.
16782 @opindex mbig-switch
16783 Generate code suitable for big switch tables. Use this option only if
16784 the assembler/linker complain about out of range branches within a switch
16789 This option will cause r2 and r5 to be used in the code generated by
16790 the compiler. This setting is the default.
16792 @item -mno-app-regs
16793 @opindex mno-app-regs
16794 This option will cause r2 and r5 to be treated as fixed registers.
16798 Specify that the target processor is the V850E1. The preprocessor
16799 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16800 this option is used.
16804 Specify that the target processor is the V850E@. The preprocessor
16805 constant @samp{__v850e__} will be defined if this option is used.
16807 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16808 are defined then a default target processor will be chosen and the
16809 relevant @samp{__v850*__} preprocessor constant will be defined.
16811 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16812 defined, regardless of which processor variant is the target.
16814 @item -mdisable-callt
16815 @opindex mdisable-callt
16816 This option will suppress generation of the CALLT instruction for the
16817 v850e and v850e1 flavors of the v850 architecture. The default is
16818 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16823 @subsection VAX Options
16824 @cindex VAX options
16826 These @samp{-m} options are defined for the VAX:
16831 Do not output certain jump instructions (@code{aobleq} and so on)
16832 that the Unix assembler for the VAX cannot handle across long
16837 Do output those jump instructions, on the assumption that you
16838 will assemble with the GNU assembler.
16842 Output code for g-format floating point numbers instead of d-format.
16845 @node VxWorks Options
16846 @subsection VxWorks Options
16847 @cindex VxWorks Options
16849 The options in this section are defined for all VxWorks targets.
16850 Options specific to the target hardware are listed with the other
16851 options for that target.
16856 GCC can generate code for both VxWorks kernels and real time processes
16857 (RTPs). This option switches from the former to the latter. It also
16858 defines the preprocessor macro @code{__RTP__}.
16861 @opindex non-static
16862 Link an RTP executable against shared libraries rather than static
16863 libraries. The options @option{-static} and @option{-shared} can
16864 also be used for RTPs (@pxref{Link Options}); @option{-static}
16871 These options are passed down to the linker. They are defined for
16872 compatibility with Diab.
16875 @opindex Xbind-lazy
16876 Enable lazy binding of function calls. This option is equivalent to
16877 @option{-Wl,-z,now} and is defined for compatibility with Diab.
16881 Disable lazy binding of function calls. This option is the default and
16882 is defined for compatibility with Diab.
16885 @node x86-64 Options
16886 @subsection x86-64 Options
16887 @cindex x86-64 options
16889 These are listed under @xref{i386 and x86-64 Options}.
16891 @node i386 and x86-64 Windows Options
16892 @subsection i386 and x86-64 Windows Options
16893 @cindex i386 and x86-64 Windows Options
16895 These additional options are available for Windows targets:
16900 This option is available for Cygwin and MinGW targets. It
16901 specifies that a console application is to be generated, by
16902 instructing the linker to set the PE header subsystem type
16903 required for console applications.
16904 This is the default behavior for Cygwin and MinGW targets.
16908 This option is available for Cygwin targets. It specifies that
16909 the Cygwin internal interface is to be used for predefined
16910 preprocessor macros, C runtime libraries and related linker
16911 paths and options. For Cygwin targets this is the default behavior.
16912 This option is deprecated and will be removed in a future release.
16915 @opindex mno-cygwin
16916 This option is available for Cygwin targets. It specifies that
16917 the MinGW internal interface is to be used instead of Cygwin's, by
16918 setting MinGW-related predefined macros and linker paths and default
16920 This option is deprecated and will be removed in a future release.
16924 This option is available for Cygwin and MinGW targets. It
16925 specifies that a DLL - a dynamic link library - is to be
16926 generated, enabling the selection of the required runtime
16927 startup object and entry point.
16929 @item -mnop-fun-dllimport
16930 @opindex mnop-fun-dllimport
16931 This option is available for Cygwin and MinGW targets. It
16932 specifies that the dllimport attribute should be ignored.
16936 This option is available for MinGW targets. It specifies
16937 that MinGW-specific thread support is to be used.
16941 This option is available for mingw-w64 targets. It specifies
16942 that the UNICODE macro is getting pre-defined and that the
16943 unicode capable runtime startup code is chosen.
16947 This option is available for Cygwin and MinGW targets. It
16948 specifies that the typical Windows pre-defined macros are to
16949 be set in the pre-processor, but does not influence the choice
16950 of runtime library/startup code.
16954 This option is available for Cygwin and MinGW targets. It
16955 specifies that a GUI application is to be generated by
16956 instructing the linker to set the PE header subsystem type
16959 @item -fno-set-stack-executable
16960 @opindex fno-set-stack-executable
16961 This option is available for MinGW targets. It specifies that
16962 the executable flag for stack used by nested functions isn't
16963 set. This is necessary for binaries running in kernel mode of
16964 Windows, as there the user32 API, which is used to set executable
16965 privileges, isn't available.
16967 @item -mpe-aligned-commons
16968 @opindex mpe-aligned-commons
16969 This option is available for Cygwin and MinGW targets. It
16970 specifies that the GNU extension to the PE file format that
16971 permits the correct alignment of COMMON variables should be
16972 used when generating code. It will be enabled by default if
16973 GCC detects that the target assembler found during configuration
16974 supports the feature.
16977 See also under @ref{i386 and x86-64 Options} for standard options.
16979 @node Xstormy16 Options
16980 @subsection Xstormy16 Options
16981 @cindex Xstormy16 Options
16983 These options are defined for Xstormy16:
16988 Choose startup files and linker script suitable for the simulator.
16991 @node Xtensa Options
16992 @subsection Xtensa Options
16993 @cindex Xtensa Options
16995 These options are supported for Xtensa targets:
16999 @itemx -mno-const16
17001 @opindex mno-const16
17002 Enable or disable use of @code{CONST16} instructions for loading
17003 constant values. The @code{CONST16} instruction is currently not a
17004 standard option from Tensilica. When enabled, @code{CONST16}
17005 instructions are always used in place of the standard @code{L32R}
17006 instructions. The use of @code{CONST16} is enabled by default only if
17007 the @code{L32R} instruction is not available.
17010 @itemx -mno-fused-madd
17011 @opindex mfused-madd
17012 @opindex mno-fused-madd
17013 Enable or disable use of fused multiply/add and multiply/subtract
17014 instructions in the floating-point option. This has no effect if the
17015 floating-point option is not also enabled. Disabling fused multiply/add
17016 and multiply/subtract instructions forces the compiler to use separate
17017 instructions for the multiply and add/subtract operations. This may be
17018 desirable in some cases where strict IEEE 754-compliant results are
17019 required: the fused multiply add/subtract instructions do not round the
17020 intermediate result, thereby producing results with @emph{more} bits of
17021 precision than specified by the IEEE standard. Disabling fused multiply
17022 add/subtract instructions also ensures that the program output is not
17023 sensitive to the compiler's ability to combine multiply and add/subtract
17026 @item -mserialize-volatile
17027 @itemx -mno-serialize-volatile
17028 @opindex mserialize-volatile
17029 @opindex mno-serialize-volatile
17030 When this option is enabled, GCC inserts @code{MEMW} instructions before
17031 @code{volatile} memory references to guarantee sequential consistency.
17032 The default is @option{-mserialize-volatile}. Use
17033 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
17035 @item -mtext-section-literals
17036 @itemx -mno-text-section-literals
17037 @opindex mtext-section-literals
17038 @opindex mno-text-section-literals
17039 Control the treatment of literal pools. The default is
17040 @option{-mno-text-section-literals}, which places literals in a separate
17041 section in the output file. This allows the literal pool to be placed
17042 in a data RAM/ROM, and it also allows the linker to combine literal
17043 pools from separate object files to remove redundant literals and
17044 improve code size. With @option{-mtext-section-literals}, the literals
17045 are interspersed in the text section in order to keep them as close as
17046 possible to their references. This may be necessary for large assembly
17049 @item -mtarget-align
17050 @itemx -mno-target-align
17051 @opindex mtarget-align
17052 @opindex mno-target-align
17053 When this option is enabled, GCC instructs the assembler to
17054 automatically align instructions to reduce branch penalties at the
17055 expense of some code density. The assembler attempts to widen density
17056 instructions to align branch targets and the instructions following call
17057 instructions. If there are not enough preceding safe density
17058 instructions to align a target, no widening will be performed. The
17059 default is @option{-mtarget-align}. These options do not affect the
17060 treatment of auto-aligned instructions like @code{LOOP}, which the
17061 assembler will always align, either by widening density instructions or
17062 by inserting no-op instructions.
17065 @itemx -mno-longcalls
17066 @opindex mlongcalls
17067 @opindex mno-longcalls
17068 When this option is enabled, GCC instructs the assembler to translate
17069 direct calls to indirect calls unless it can determine that the target
17070 of a direct call is in the range allowed by the call instruction. This
17071 translation typically occurs for calls to functions in other source
17072 files. Specifically, the assembler translates a direct @code{CALL}
17073 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
17074 The default is @option{-mno-longcalls}. This option should be used in
17075 programs where the call target can potentially be out of range. This
17076 option is implemented in the assembler, not the compiler, so the
17077 assembly code generated by GCC will still show direct call
17078 instructions---look at the disassembled object code to see the actual
17079 instructions. Note that the assembler will use an indirect call for
17080 every cross-file call, not just those that really will be out of range.
17083 @node zSeries Options
17084 @subsection zSeries Options
17085 @cindex zSeries options
17087 These are listed under @xref{S/390 and zSeries Options}.
17089 @node Code Gen Options
17090 @section Options for Code Generation Conventions
17091 @cindex code generation conventions
17092 @cindex options, code generation
17093 @cindex run-time options
17095 These machine-independent options control the interface conventions
17096 used in code generation.
17098 Most of them have both positive and negative forms; the negative form
17099 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
17100 one of the forms is listed---the one which is not the default. You
17101 can figure out the other form by either removing @samp{no-} or adding
17105 @item -fbounds-check
17106 @opindex fbounds-check
17107 For front-ends that support it, generate additional code to check that
17108 indices used to access arrays are within the declared range. This is
17109 currently only supported by the Java and Fortran front-ends, where
17110 this option defaults to true and false respectively.
17114 This option generates traps for signed overflow on addition, subtraction,
17115 multiplication operations.
17119 This option instructs the compiler to assume that signed arithmetic
17120 overflow of addition, subtraction and multiplication wraps around
17121 using twos-complement representation. This flag enables some optimizations
17122 and disables others. This option is enabled by default for the Java
17123 front-end, as required by the Java language specification.
17126 @opindex fexceptions
17127 Enable exception handling. Generates extra code needed to propagate
17128 exceptions. For some targets, this implies GCC will generate frame
17129 unwind information for all functions, which can produce significant data
17130 size overhead, although it does not affect execution. If you do not
17131 specify this option, GCC will enable it by default for languages like
17132 C++ which normally require exception handling, and disable it for
17133 languages like C that do not normally require it. However, you may need
17134 to enable this option when compiling C code that needs to interoperate
17135 properly with exception handlers written in C++. You may also wish to
17136 disable this option if you are compiling older C++ programs that don't
17137 use exception handling.
17139 @item -fnon-call-exceptions
17140 @opindex fnon-call-exceptions
17141 Generate code that allows trapping instructions to throw exceptions.
17142 Note that this requires platform-specific runtime support that does
17143 not exist everywhere. Moreover, it only allows @emph{trapping}
17144 instructions to throw exceptions, i.e.@: memory references or floating
17145 point instructions. It does not allow exceptions to be thrown from
17146 arbitrary signal handlers such as @code{SIGALRM}.
17148 @item -funwind-tables
17149 @opindex funwind-tables
17150 Similar to @option{-fexceptions}, except that it will just generate any needed
17151 static data, but will not affect the generated code in any other way.
17152 You will normally not enable this option; instead, a language processor
17153 that needs this handling would enable it on your behalf.
17155 @item -fasynchronous-unwind-tables
17156 @opindex fasynchronous-unwind-tables
17157 Generate unwind table in dwarf2 format, if supported by target machine. The
17158 table is exact at each instruction boundary, so it can be used for stack
17159 unwinding from asynchronous events (such as debugger or garbage collector).
17161 @item -fpcc-struct-return
17162 @opindex fpcc-struct-return
17163 Return ``short'' @code{struct} and @code{union} values in memory like
17164 longer ones, rather than in registers. This convention is less
17165 efficient, but it has the advantage of allowing intercallability between
17166 GCC-compiled files and files compiled with other compilers, particularly
17167 the Portable C Compiler (pcc).
17169 The precise convention for returning structures in memory depends
17170 on the target configuration macros.
17172 Short structures and unions are those whose size and alignment match
17173 that of some integer type.
17175 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
17176 switch is not binary compatible with code compiled with the
17177 @option{-freg-struct-return} switch.
17178 Use it to conform to a non-default application binary interface.
17180 @item -freg-struct-return
17181 @opindex freg-struct-return
17182 Return @code{struct} and @code{union} values in registers when possible.
17183 This is more efficient for small structures than
17184 @option{-fpcc-struct-return}.
17186 If you specify neither @option{-fpcc-struct-return} nor
17187 @option{-freg-struct-return}, GCC defaults to whichever convention is
17188 standard for the target. If there is no standard convention, GCC
17189 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
17190 the principal compiler. In those cases, we can choose the standard, and
17191 we chose the more efficient register return alternative.
17193 @strong{Warning:} code compiled with the @option{-freg-struct-return}
17194 switch is not binary compatible with code compiled with the
17195 @option{-fpcc-struct-return} switch.
17196 Use it to conform to a non-default application binary interface.
17198 @item -fshort-enums
17199 @opindex fshort-enums
17200 Allocate to an @code{enum} type only as many bytes as it needs for the
17201 declared range of possible values. Specifically, the @code{enum} type
17202 will be equivalent to the smallest integer type which has enough room.
17204 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
17205 code that is not binary compatible with code generated without that switch.
17206 Use it to conform to a non-default application binary interface.
17208 @item -fshort-double
17209 @opindex fshort-double
17210 Use the same size for @code{double} as for @code{float}.
17212 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
17213 code that is not binary compatible with code generated without that switch.
17214 Use it to conform to a non-default application binary interface.
17216 @item -fshort-wchar
17217 @opindex fshort-wchar
17218 Override the underlying type for @samp{wchar_t} to be @samp{short
17219 unsigned int} instead of the default for the target. This option is
17220 useful for building programs to run under WINE@.
17222 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
17223 code that is not binary compatible with code generated without that switch.
17224 Use it to conform to a non-default application binary interface.
17227 @opindex fno-common
17228 In C code, controls the placement of uninitialized global variables.
17229 Unix C compilers have traditionally permitted multiple definitions of
17230 such variables in different compilation units by placing the variables
17232 This is the behavior specified by @option{-fcommon}, and is the default
17233 for GCC on most targets.
17234 On the other hand, this behavior is not required by ISO C, and on some
17235 targets may carry a speed or code size penalty on variable references.
17236 The @option{-fno-common} option specifies that the compiler should place
17237 uninitialized global variables in the data section of the object file,
17238 rather than generating them as common blocks.
17239 This has the effect that if the same variable is declared
17240 (without @code{extern}) in two different compilations,
17241 you will get a multiple-definition error when you link them.
17242 In this case, you must compile with @option{-fcommon} instead.
17243 Compiling with @option{-fno-common} is useful on targets for which
17244 it provides better performance, or if you wish to verify that the
17245 program will work on other systems which always treat uninitialized
17246 variable declarations this way.
17250 Ignore the @samp{#ident} directive.
17252 @item -finhibit-size-directive
17253 @opindex finhibit-size-directive
17254 Don't output a @code{.size} assembler directive, or anything else that
17255 would cause trouble if the function is split in the middle, and the
17256 two halves are placed at locations far apart in memory. This option is
17257 used when compiling @file{crtstuff.c}; you should not need to use it
17260 @item -fverbose-asm
17261 @opindex fverbose-asm
17262 Put extra commentary information in the generated assembly code to
17263 make it more readable. This option is generally only of use to those
17264 who actually need to read the generated assembly code (perhaps while
17265 debugging the compiler itself).
17267 @option{-fno-verbose-asm}, the default, causes the
17268 extra information to be omitted and is useful when comparing two assembler
17271 @item -frecord-gcc-switches
17272 @opindex frecord-gcc-switches
17273 This switch causes the command line that was used to invoke the
17274 compiler to be recorded into the object file that is being created.
17275 This switch is only implemented on some targets and the exact format
17276 of the recording is target and binary file format dependent, but it
17277 usually takes the form of a section containing ASCII text. This
17278 switch is related to the @option{-fverbose-asm} switch, but that
17279 switch only records information in the assembler output file as
17280 comments, so it never reaches the object file.
17284 @cindex global offset table
17286 Generate position-independent code (PIC) suitable for use in a shared
17287 library, if supported for the target machine. Such code accesses all
17288 constant addresses through a global offset table (GOT)@. The dynamic
17289 loader resolves the GOT entries when the program starts (the dynamic
17290 loader is not part of GCC; it is part of the operating system). If
17291 the GOT size for the linked executable exceeds a machine-specific
17292 maximum size, you get an error message from the linker indicating that
17293 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17294 instead. (These maximums are 8k on the SPARC and 32k
17295 on the m68k and RS/6000. The 386 has no such limit.)
17297 Position-independent code requires special support, and therefore works
17298 only on certain machines. For the 386, GCC supports PIC for System V
17299 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17300 position-independent.
17302 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17307 If supported for the target machine, emit position-independent code,
17308 suitable for dynamic linking and avoiding any limit on the size of the
17309 global offset table. This option makes a difference on the m68k,
17310 PowerPC and SPARC@.
17312 Position-independent code requires special support, and therefore works
17313 only on certain machines.
17315 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17322 These options are similar to @option{-fpic} and @option{-fPIC}, but
17323 generated position independent code can be only linked into executables.
17324 Usually these options are used when @option{-pie} GCC option will be
17325 used during linking.
17327 @option{-fpie} and @option{-fPIE} both define the macros
17328 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17329 for @option{-fpie} and 2 for @option{-fPIE}.
17331 @item -fno-jump-tables
17332 @opindex fno-jump-tables
17333 Do not use jump tables for switch statements even where it would be
17334 more efficient than other code generation strategies. This option is
17335 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17336 building code which forms part of a dynamic linker and cannot
17337 reference the address of a jump table. On some targets, jump tables
17338 do not require a GOT and this option is not needed.
17340 @item -ffixed-@var{reg}
17342 Treat the register named @var{reg} as a fixed register; generated code
17343 should never refer to it (except perhaps as a stack pointer, frame
17344 pointer or in some other fixed role).
17346 @var{reg} must be the name of a register. The register names accepted
17347 are machine-specific and are defined in the @code{REGISTER_NAMES}
17348 macro in the machine description macro file.
17350 This flag does not have a negative form, because it specifies a
17353 @item -fcall-used-@var{reg}
17354 @opindex fcall-used
17355 Treat the register named @var{reg} as an allocable register that is
17356 clobbered by function calls. It may be allocated for temporaries or
17357 variables that do not live across a call. Functions compiled this way
17358 will not save and restore the register @var{reg}.
17360 It is an error to used this flag with the frame pointer or stack pointer.
17361 Use of this flag for other registers that have fixed pervasive roles in
17362 the machine's execution model will produce disastrous results.
17364 This flag does not have a negative form, because it specifies a
17367 @item -fcall-saved-@var{reg}
17368 @opindex fcall-saved
17369 Treat the register named @var{reg} as an allocable register saved by
17370 functions. It may be allocated even for temporaries or variables that
17371 live across a call. Functions compiled this way will save and restore
17372 the register @var{reg} if they use it.
17374 It is an error to used this flag with the frame pointer or stack pointer.
17375 Use of this flag for other registers that have fixed pervasive roles in
17376 the machine's execution model will produce disastrous results.
17378 A different sort of disaster will result from the use of this flag for
17379 a register in which function values may be returned.
17381 This flag does not have a negative form, because it specifies a
17384 @item -fpack-struct[=@var{n}]
17385 @opindex fpack-struct
17386 Without a value specified, pack all structure members together without
17387 holes. When a value is specified (which must be a small power of two), pack
17388 structure members according to this value, representing the maximum
17389 alignment (that is, objects with default alignment requirements larger than
17390 this will be output potentially unaligned at the next fitting location.
17392 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17393 code that is not binary compatible with code generated without that switch.
17394 Additionally, it makes the code suboptimal.
17395 Use it to conform to a non-default application binary interface.
17397 @item -finstrument-functions
17398 @opindex finstrument-functions
17399 Generate instrumentation calls for entry and exit to functions. Just
17400 after function entry and just before function exit, the following
17401 profiling functions will be called with the address of the current
17402 function and its call site. (On some platforms,
17403 @code{__builtin_return_address} does not work beyond the current
17404 function, so the call site information may not be available to the
17405 profiling functions otherwise.)
17408 void __cyg_profile_func_enter (void *this_fn,
17410 void __cyg_profile_func_exit (void *this_fn,
17414 The first argument is the address of the start of the current function,
17415 which may be looked up exactly in the symbol table.
17417 This instrumentation is also done for functions expanded inline in other
17418 functions. The profiling calls will indicate where, conceptually, the
17419 inline function is entered and exited. This means that addressable
17420 versions of such functions must be available. If all your uses of a
17421 function are expanded inline, this may mean an additional expansion of
17422 code size. If you use @samp{extern inline} in your C code, an
17423 addressable version of such functions must be provided. (This is
17424 normally the case anyways, but if you get lucky and the optimizer always
17425 expands the functions inline, you might have gotten away without
17426 providing static copies.)
17428 A function may be given the attribute @code{no_instrument_function}, in
17429 which case this instrumentation will not be done. This can be used, for
17430 example, for the profiling functions listed above, high-priority
17431 interrupt routines, and any functions from which the profiling functions
17432 cannot safely be called (perhaps signal handlers, if the profiling
17433 routines generate output or allocate memory).
17435 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17436 @opindex finstrument-functions-exclude-file-list
17438 Set the list of functions that are excluded from instrumentation (see
17439 the description of @code{-finstrument-functions}). If the file that
17440 contains a function definition matches with one of @var{file}, then
17441 that function is not instrumented. The match is done on substrings:
17442 if the @var{file} parameter is a substring of the file name, it is
17443 considered to be a match.
17446 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
17447 will exclude any inline function defined in files whose pathnames
17448 contain @code{/bits/stl} or @code{include/sys}.
17450 If, for some reason, you want to include letter @code{','} in one of
17451 @var{sym}, write @code{'\,'}. For example,
17452 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17453 (note the single quote surrounding the option).
17455 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
17456 @opindex finstrument-functions-exclude-function-list
17458 This is similar to @code{-finstrument-functions-exclude-file-list},
17459 but this option sets the list of function names to be excluded from
17460 instrumentation. The function name to be matched is its user-visible
17461 name, such as @code{vector<int> blah(const vector<int> &)}, not the
17462 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
17463 match is done on substrings: if the @var{sym} parameter is a substring
17464 of the function name, it is considered to be a match. For C99 and C++
17465 extended identifiers, the function name must be given in UTF-8, not
17466 using universal character names.
17468 @item -fstack-check
17469 @opindex fstack-check
17470 Generate code to verify that you do not go beyond the boundary of the
17471 stack. You should specify this flag if you are running in an
17472 environment with multiple threads, but only rarely need to specify it in
17473 a single-threaded environment since stack overflow is automatically
17474 detected on nearly all systems if there is only one stack.
17476 Note that this switch does not actually cause checking to be done; the
17477 operating system or the language runtime must do that. The switch causes
17478 generation of code to ensure that they see the stack being extended.
17480 You can additionally specify a string parameter: @code{no} means no
17481 checking, @code{generic} means force the use of old-style checking,
17482 @code{specific} means use the best checking method and is equivalent
17483 to bare @option{-fstack-check}.
17485 Old-style checking is a generic mechanism that requires no specific
17486 target support in the compiler but comes with the following drawbacks:
17490 Modified allocation strategy for large objects: they will always be
17491 allocated dynamically if their size exceeds a fixed threshold.
17494 Fixed limit on the size of the static frame of functions: when it is
17495 topped by a particular function, stack checking is not reliable and
17496 a warning is issued by the compiler.
17499 Inefficiency: because of both the modified allocation strategy and the
17500 generic implementation, the performances of the code are hampered.
17503 Note that old-style stack checking is also the fallback method for
17504 @code{specific} if no target support has been added in the compiler.
17506 @item -fstack-limit-register=@var{reg}
17507 @itemx -fstack-limit-symbol=@var{sym}
17508 @itemx -fno-stack-limit
17509 @opindex fstack-limit-register
17510 @opindex fstack-limit-symbol
17511 @opindex fno-stack-limit
17512 Generate code to ensure that the stack does not grow beyond a certain value,
17513 either the value of a register or the address of a symbol. If the stack
17514 would grow beyond the value, a signal is raised. For most targets,
17515 the signal is raised before the stack overruns the boundary, so
17516 it is possible to catch the signal without taking special precautions.
17518 For instance, if the stack starts at absolute address @samp{0x80000000}
17519 and grows downwards, you can use the flags
17520 @option{-fstack-limit-symbol=__stack_limit} and
17521 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
17522 of 128KB@. Note that this may only work with the GNU linker.
17524 @item -fleading-underscore
17525 @opindex fleading-underscore
17526 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
17527 change the way C symbols are represented in the object file. One use
17528 is to help link with legacy assembly code.
17530 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
17531 generate code that is not binary compatible with code generated without that
17532 switch. Use it to conform to a non-default application binary interface.
17533 Not all targets provide complete support for this switch.
17535 @item -ftls-model=@var{model}
17536 @opindex ftls-model
17537 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
17538 The @var{model} argument should be one of @code{global-dynamic},
17539 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
17541 The default without @option{-fpic} is @code{initial-exec}; with
17542 @option{-fpic} the default is @code{global-dynamic}.
17544 @item -fvisibility=@var{default|internal|hidden|protected}
17545 @opindex fvisibility
17546 Set the default ELF image symbol visibility to the specified option---all
17547 symbols will be marked with this unless overridden within the code.
17548 Using this feature can very substantially improve linking and
17549 load times of shared object libraries, produce more optimized
17550 code, provide near-perfect API export and prevent symbol clashes.
17551 It is @strong{strongly} recommended that you use this in any shared objects
17554 Despite the nomenclature, @code{default} always means public ie;
17555 available to be linked against from outside the shared object.
17556 @code{protected} and @code{internal} are pretty useless in real-world
17557 usage so the only other commonly used option will be @code{hidden}.
17558 The default if @option{-fvisibility} isn't specified is
17559 @code{default}, i.e., make every
17560 symbol public---this causes the same behavior as previous versions of
17563 A good explanation of the benefits offered by ensuring ELF
17564 symbols have the correct visibility is given by ``How To Write
17565 Shared Libraries'' by Ulrich Drepper (which can be found at
17566 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
17567 solution made possible by this option to marking things hidden when
17568 the default is public is to make the default hidden and mark things
17569 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
17570 and @code{__attribute__ ((visibility("default")))} instead of
17571 @code{__declspec(dllexport)} you get almost identical semantics with
17572 identical syntax. This is a great boon to those working with
17573 cross-platform projects.
17575 For those adding visibility support to existing code, you may find
17576 @samp{#pragma GCC visibility} of use. This works by you enclosing
17577 the declarations you wish to set visibility for with (for example)
17578 @samp{#pragma GCC visibility push(hidden)} and
17579 @samp{#pragma GCC visibility pop}.
17580 Bear in mind that symbol visibility should be viewed @strong{as
17581 part of the API interface contract} and thus all new code should
17582 always specify visibility when it is not the default ie; declarations
17583 only for use within the local DSO should @strong{always} be marked explicitly
17584 as hidden as so to avoid PLT indirection overheads---making this
17585 abundantly clear also aids readability and self-documentation of the code.
17586 Note that due to ISO C++ specification requirements, operator new and
17587 operator delete must always be of default visibility.
17589 Be aware that headers from outside your project, in particular system
17590 headers and headers from any other library you use, may not be
17591 expecting to be compiled with visibility other than the default. You
17592 may need to explicitly say @samp{#pragma GCC visibility push(default)}
17593 before including any such headers.
17595 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
17596 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
17597 no modifications. However, this means that calls to @samp{extern}
17598 functions with no explicit visibility will use the PLT, so it is more
17599 effective to use @samp{__attribute ((visibility))} and/or
17600 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
17601 declarations should be treated as hidden.
17603 Note that @samp{-fvisibility} does affect C++ vague linkage
17604 entities. This means that, for instance, an exception class that will
17605 be thrown between DSOs must be explicitly marked with default
17606 visibility so that the @samp{type_info} nodes will be unified between
17609 An overview of these techniques, their benefits and how to use them
17610 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
17616 @node Environment Variables
17617 @section Environment Variables Affecting GCC
17618 @cindex environment variables
17620 @c man begin ENVIRONMENT
17621 This section describes several environment variables that affect how GCC
17622 operates. Some of them work by specifying directories or prefixes to use
17623 when searching for various kinds of files. Some are used to specify other
17624 aspects of the compilation environment.
17626 Note that you can also specify places to search using options such as
17627 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
17628 take precedence over places specified using environment variables, which
17629 in turn take precedence over those specified by the configuration of GCC@.
17630 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
17631 GNU Compiler Collection (GCC) Internals}.
17636 @c @itemx LC_COLLATE
17638 @c @itemx LC_MONETARY
17639 @c @itemx LC_NUMERIC
17644 @c @findex LC_COLLATE
17645 @findex LC_MESSAGES
17646 @c @findex LC_MONETARY
17647 @c @findex LC_NUMERIC
17651 These environment variables control the way that GCC uses
17652 localization information that allow GCC to work with different
17653 national conventions. GCC inspects the locale categories
17654 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
17655 so. These locale categories can be set to any value supported by your
17656 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
17657 Kingdom encoded in UTF-8.
17659 The @env{LC_CTYPE} environment variable specifies character
17660 classification. GCC uses it to determine the character boundaries in
17661 a string; this is needed for some multibyte encodings that contain quote
17662 and escape characters that would otherwise be interpreted as a string
17665 The @env{LC_MESSAGES} environment variable specifies the language to
17666 use in diagnostic messages.
17668 If the @env{LC_ALL} environment variable is set, it overrides the value
17669 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17670 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17671 environment variable. If none of these variables are set, GCC
17672 defaults to traditional C English behavior.
17676 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17677 files. GCC uses temporary files to hold the output of one stage of
17678 compilation which is to be used as input to the next stage: for example,
17679 the output of the preprocessor, which is the input to the compiler
17682 @item GCC_EXEC_PREFIX
17683 @findex GCC_EXEC_PREFIX
17684 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17685 names of the subprograms executed by the compiler. No slash is added
17686 when this prefix is combined with the name of a subprogram, but you can
17687 specify a prefix that ends with a slash if you wish.
17689 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17690 an appropriate prefix to use based on the pathname it was invoked with.
17692 If GCC cannot find the subprogram using the specified prefix, it
17693 tries looking in the usual places for the subprogram.
17695 The default value of @env{GCC_EXEC_PREFIX} is
17696 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17697 the installed compiler. In many cases @var{prefix} is the value
17698 of @code{prefix} when you ran the @file{configure} script.
17700 Other prefixes specified with @option{-B} take precedence over this prefix.
17702 This prefix is also used for finding files such as @file{crt0.o} that are
17705 In addition, the prefix is used in an unusual way in finding the
17706 directories to search for header files. For each of the standard
17707 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17708 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17709 replacing that beginning with the specified prefix to produce an
17710 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17711 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17712 These alternate directories are searched first; the standard directories
17713 come next. If a standard directory begins with the configured
17714 @var{prefix} then the value of @var{prefix} is replaced by
17715 @env{GCC_EXEC_PREFIX} when looking for header files.
17717 @item COMPILER_PATH
17718 @findex COMPILER_PATH
17719 The value of @env{COMPILER_PATH} is a colon-separated list of
17720 directories, much like @env{PATH}. GCC tries the directories thus
17721 specified when searching for subprograms, if it can't find the
17722 subprograms using @env{GCC_EXEC_PREFIX}.
17725 @findex LIBRARY_PATH
17726 The value of @env{LIBRARY_PATH} is a colon-separated list of
17727 directories, much like @env{PATH}. When configured as a native compiler,
17728 GCC tries the directories thus specified when searching for special
17729 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17730 using GCC also uses these directories when searching for ordinary
17731 libraries for the @option{-l} option (but directories specified with
17732 @option{-L} come first).
17736 @cindex locale definition
17737 This variable is used to pass locale information to the compiler. One way in
17738 which this information is used is to determine the character set to be used
17739 when character literals, string literals and comments are parsed in C and C++.
17740 When the compiler is configured to allow multibyte characters,
17741 the following values for @env{LANG} are recognized:
17745 Recognize JIS characters.
17747 Recognize SJIS characters.
17749 Recognize EUCJP characters.
17752 If @env{LANG} is not defined, or if it has some other value, then the
17753 compiler will use mblen and mbtowc as defined by the default locale to
17754 recognize and translate multibyte characters.
17758 Some additional environments variables affect the behavior of the
17761 @include cppenv.texi
17765 @node Precompiled Headers
17766 @section Using Precompiled Headers
17767 @cindex precompiled headers
17768 @cindex speed of compilation
17770 Often large projects have many header files that are included in every
17771 source file. The time the compiler takes to process these header files
17772 over and over again can account for nearly all of the time required to
17773 build the project. To make builds faster, GCC allows users to
17774 `precompile' a header file; then, if builds can use the precompiled
17775 header file they will be much faster.
17777 To create a precompiled header file, simply compile it as you would any
17778 other file, if necessary using the @option{-x} option to make the driver
17779 treat it as a C or C++ header file. You will probably want to use a
17780 tool like @command{make} to keep the precompiled header up-to-date when
17781 the headers it contains change.
17783 A precompiled header file will be searched for when @code{#include} is
17784 seen in the compilation. As it searches for the included file
17785 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17786 compiler looks for a precompiled header in each directory just before it
17787 looks for the include file in that directory. The name searched for is
17788 the name specified in the @code{#include} with @samp{.gch} appended. If
17789 the precompiled header file can't be used, it is ignored.
17791 For instance, if you have @code{#include "all.h"}, and you have
17792 @file{all.h.gch} in the same directory as @file{all.h}, then the
17793 precompiled header file will be used if possible, and the original
17794 header will be used otherwise.
17796 Alternatively, you might decide to put the precompiled header file in a
17797 directory and use @option{-I} to ensure that directory is searched
17798 before (or instead of) the directory containing the original header.
17799 Then, if you want to check that the precompiled header file is always
17800 used, you can put a file of the same name as the original header in this
17801 directory containing an @code{#error} command.
17803 This also works with @option{-include}. So yet another way to use
17804 precompiled headers, good for projects not designed with precompiled
17805 header files in mind, is to simply take most of the header files used by
17806 a project, include them from another header file, precompile that header
17807 file, and @option{-include} the precompiled header. If the header files
17808 have guards against multiple inclusion, they will be skipped because
17809 they've already been included (in the precompiled header).
17811 If you need to precompile the same header file for different
17812 languages, targets, or compiler options, you can instead make a
17813 @emph{directory} named like @file{all.h.gch}, and put each precompiled
17814 header in the directory, perhaps using @option{-o}. It doesn't matter
17815 what you call the files in the directory, every precompiled header in
17816 the directory will be considered. The first precompiled header
17817 encountered in the directory that is valid for this compilation will
17818 be used; they're searched in no particular order.
17820 There are many other possibilities, limited only by your imagination,
17821 good sense, and the constraints of your build system.
17823 A precompiled header file can be used only when these conditions apply:
17827 Only one precompiled header can be used in a particular compilation.
17830 A precompiled header can't be used once the first C token is seen. You
17831 can have preprocessor directives before a precompiled header; you can
17832 even include a precompiled header from inside another header, so long as
17833 there are no C tokens before the @code{#include}.
17836 The precompiled header file must be produced for the same language as
17837 the current compilation. You can't use a C precompiled header for a C++
17841 The precompiled header file must have been produced by the same compiler
17842 binary as the current compilation is using.
17845 Any macros defined before the precompiled header is included must
17846 either be defined in the same way as when the precompiled header was
17847 generated, or must not affect the precompiled header, which usually
17848 means that they don't appear in the precompiled header at all.
17850 The @option{-D} option is one way to define a macro before a
17851 precompiled header is included; using a @code{#define} can also do it.
17852 There are also some options that define macros implicitly, like
17853 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
17856 @item If debugging information is output when using the precompiled
17857 header, using @option{-g} or similar, the same kind of debugging information
17858 must have been output when building the precompiled header. However,
17859 a precompiled header built using @option{-g} can be used in a compilation
17860 when no debugging information is being output.
17862 @item The same @option{-m} options must generally be used when building
17863 and using the precompiled header. @xref{Submodel Options},
17864 for any cases where this rule is relaxed.
17866 @item Each of the following options must be the same when building and using
17867 the precompiled header:
17869 @gccoptlist{-fexceptions}
17872 Some other command-line options starting with @option{-f},
17873 @option{-p}, or @option{-O} must be defined in the same way as when
17874 the precompiled header was generated. At present, it's not clear
17875 which options are safe to change and which are not; the safest choice
17876 is to use exactly the same options when generating and using the
17877 precompiled header. The following are known to be safe:
17879 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
17880 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
17881 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
17886 For all of these except the last, the compiler will automatically
17887 ignore the precompiled header if the conditions aren't met. If you
17888 find an option combination that doesn't work and doesn't cause the
17889 precompiled header to be ignored, please consider filing a bug report,
17892 If you do use differing options when generating and using the
17893 precompiled header, the actual behavior will be a mixture of the
17894 behavior for the options. For instance, if you use @option{-g} to
17895 generate the precompiled header but not when using it, you may or may
17896 not get debugging information for routines in the precompiled header.