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 -fnothrow-opt -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 -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[=@var{n}] -fwpa -fuse-linker-plugin @gol
395 --param @var{name}=@var{value}
396 -O -O0 -O1 -O2 -O3 -Os -Ofast}
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 arguments
1176 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1177 This is useful for shell scripts to capture the 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.
1933 @opindex fnothrow-opt
1934 Treat a @code{throw()} exception specification as though it were a
1935 @code{noexcept} specification to reduce or eliminate the text size
1936 overhead relative to a function with no exception specification. If
1937 the function has local variables of types with non-trivial
1938 destructors, the exception specification will actually make the
1939 function smaller because the EH cleanups for those variables can be
1940 optimized away. The semantic effect is that an exception thrown out of
1941 a function with such an exception specification will result in a call
1942 to @code{terminate} rather than @code{unexpected}.
1944 @item -fno-operator-names
1945 @opindex fno-operator-names
1946 Do not treat the operator name keywords @code{and}, @code{bitand},
1947 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1948 synonyms as keywords.
1950 @item -fno-optional-diags
1951 @opindex fno-optional-diags
1952 Disable diagnostics that the standard says a compiler does not need to
1953 issue. Currently, the only such diagnostic issued by G++ is the one for
1954 a name having multiple meanings within a class.
1957 @opindex fpermissive
1958 Downgrade some diagnostics about nonconformant code from errors to
1959 warnings. Thus, using @option{-fpermissive} will allow some
1960 nonconforming code to compile.
1962 @item -fno-pretty-templates
1963 @opindex fno-pretty-templates
1964 When an error message refers to a specialization of a function
1965 template, the compiler will normally print the signature of the
1966 template followed by the template arguments and any typedefs or
1967 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1968 rather than @code{void f(int)}) so that it's clear which template is
1969 involved. When an error message refers to a specialization of a class
1970 template, the compiler will omit any template arguments which match
1971 the default template arguments for that template. If either of these
1972 behaviors make it harder to understand the error message rather than
1973 easier, using @option{-fno-pretty-templates} will disable them.
1977 Enable automatic template instantiation at link time. This option also
1978 implies @option{-fno-implicit-templates}. @xref{Template
1979 Instantiation}, for more information.
1983 Disable generation of information about every class with virtual
1984 functions for use by the C++ runtime type identification features
1985 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1986 of the language, you can save some space by using this flag. Note that
1987 exception handling uses the same information, but it will generate it as
1988 needed. The @samp{dynamic_cast} operator can still be used for casts that
1989 do not require runtime type information, i.e.@: casts to @code{void *} or to
1990 unambiguous base classes.
1994 Emit statistics about front-end processing at the end of the compilation.
1995 This information is generally only useful to the G++ development team.
1997 @item -fstrict-enums
1998 @opindex fstrict-enums
1999 Allow the compiler to optimize using the assumption that a value of
2000 enumeration type can only be one of the values of the enumeration (as
2001 defined in the C++ standard; basically, a value which can be
2002 represented in the minimum number of bits needed to represent all the
2003 enumerators). This assumption may not be valid if the program uses a
2004 cast to convert an arbitrary integer value to the enumeration type.
2006 @item -ftemplate-depth=@var{n}
2007 @opindex ftemplate-depth
2008 Set the maximum instantiation depth for template classes to @var{n}.
2009 A limit on the template instantiation depth is needed to detect
2010 endless recursions during template class instantiation. ANSI/ISO C++
2011 conforming programs must not rely on a maximum depth greater than 17
2012 (changed to 1024 in C++0x).
2014 @item -fno-threadsafe-statics
2015 @opindex fno-threadsafe-statics
2016 Do not emit the extra code to use the routines specified in the C++
2017 ABI for thread-safe initialization of local statics. You can use this
2018 option to reduce code size slightly in code that doesn't need to be
2021 @item -fuse-cxa-atexit
2022 @opindex fuse-cxa-atexit
2023 Register destructors for objects with static storage duration with the
2024 @code{__cxa_atexit} function rather than the @code{atexit} function.
2025 This option is required for fully standards-compliant handling of static
2026 destructors, but will only work if your C library supports
2027 @code{__cxa_atexit}.
2029 @item -fno-use-cxa-get-exception-ptr
2030 @opindex fno-use-cxa-get-exception-ptr
2031 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2032 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2033 if the runtime routine is not available.
2035 @item -fvisibility-inlines-hidden
2036 @opindex fvisibility-inlines-hidden
2037 This switch declares that the user does not attempt to compare
2038 pointers to inline methods where the addresses of the two functions
2039 were taken in different shared objects.
2041 The effect of this is that GCC may, effectively, mark inline methods with
2042 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2043 appear in the export table of a DSO and do not require a PLT indirection
2044 when used within the DSO@. Enabling this option can have a dramatic effect
2045 on load and link times of a DSO as it massively reduces the size of the
2046 dynamic export table when the library makes heavy use of templates.
2048 The behavior of this switch is not quite the same as marking the
2049 methods as hidden directly, because it does not affect static variables
2050 local to the function or cause the compiler to deduce that
2051 the function is defined in only one shared object.
2053 You may mark a method as having a visibility explicitly to negate the
2054 effect of the switch for that method. For example, if you do want to
2055 compare pointers to a particular inline method, you might mark it as
2056 having default visibility. Marking the enclosing class with explicit
2057 visibility will have no effect.
2059 Explicitly instantiated inline methods are unaffected by this option
2060 as their linkage might otherwise cross a shared library boundary.
2061 @xref{Template Instantiation}.
2063 @item -fvisibility-ms-compat
2064 @opindex fvisibility-ms-compat
2065 This flag attempts to use visibility settings to make GCC's C++
2066 linkage model compatible with that of Microsoft Visual Studio.
2068 The flag makes these changes to GCC's linkage model:
2072 It sets the default visibility to @code{hidden}, like
2073 @option{-fvisibility=hidden}.
2076 Types, but not their members, are not hidden by default.
2079 The One Definition Rule is relaxed for types without explicit
2080 visibility specifications which are defined in more than one different
2081 shared object: those declarations are permitted if they would have
2082 been permitted when this option was not used.
2085 In new code it is better to use @option{-fvisibility=hidden} and
2086 export those classes which are intended to be externally visible.
2087 Unfortunately it is possible for code to rely, perhaps accidentally,
2088 on the Visual Studio behavior.
2090 Among the consequences of these changes are that static data members
2091 of the same type with the same name but defined in different shared
2092 objects will be different, so changing one will not change the other;
2093 and that pointers to function members defined in different shared
2094 objects may not compare equal. When this flag is given, it is a
2095 violation of the ODR to define types with the same name differently.
2099 Do not use weak symbol support, even if it is provided by the linker.
2100 By default, G++ will use weak symbols if they are available. This
2101 option exists only for testing, and should not be used by end-users;
2102 it will result in inferior code and has no benefits. This option may
2103 be removed in a future release of G++.
2107 Do not search for header files in the standard directories specific to
2108 C++, but do still search the other standard directories. (This option
2109 is used when building the C++ library.)
2112 In addition, these optimization, warning, and code generation options
2113 have meanings only for C++ programs:
2116 @item -fno-default-inline
2117 @opindex fno-default-inline
2118 Do not assume @samp{inline} for functions defined inside a class scope.
2119 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2120 functions will have linkage like inline functions; they just won't be
2123 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2126 Warn when G++ generates code that is probably not compatible with the
2127 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2128 all such cases, there are probably some cases that are not warned about,
2129 even though G++ is generating incompatible code. There may also be
2130 cases where warnings are emitted even though the code that is generated
2133 You should rewrite your code to avoid these warnings if you are
2134 concerned about the fact that code generated by G++ may not be binary
2135 compatible with code generated by other compilers.
2137 The known incompatibilities in @option{-fabi-version=2} (the default) include:
2142 A template with a non-type template parameter of reference type is
2143 mangled incorrectly:
2146 template <int &> struct S @{@};
2150 This is fixed in @option{-fabi-version=3}.
2153 SIMD vector types declared using @code{__attribute ((vector_size))} are
2154 mangled in a non-standard way that does not allow for overloading of
2155 functions taking vectors of different sizes.
2157 The mangling is changed in @option{-fabi-version=4}.
2160 The known incompatibilities in @option{-fabi-version=1} include:
2165 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2166 pack data into the same byte as a base class. For example:
2169 struct A @{ virtual void f(); int f1 : 1; @};
2170 struct B : public A @{ int f2 : 1; @};
2174 In this case, G++ will place @code{B::f2} into the same byte
2175 as@code{A::f1}; other compilers will not. You can avoid this problem
2176 by explicitly padding @code{A} so that its size is a multiple of the
2177 byte size on your platform; that will cause G++ and other compilers to
2178 layout @code{B} identically.
2181 Incorrect handling of tail-padding for virtual bases. G++ does not use
2182 tail padding when laying out virtual bases. For example:
2185 struct A @{ virtual void f(); char c1; @};
2186 struct B @{ B(); char c2; @};
2187 struct C : public A, public virtual B @{@};
2191 In this case, G++ will not place @code{B} into the tail-padding for
2192 @code{A}; other compilers will. You can avoid this problem by
2193 explicitly padding @code{A} so that its size is a multiple of its
2194 alignment (ignoring virtual base classes); that will cause G++ and other
2195 compilers to layout @code{C} identically.
2198 Incorrect handling of bit-fields with declared widths greater than that
2199 of their underlying types, when the bit-fields appear in a union. For
2203 union U @{ int i : 4096; @};
2207 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2208 union too small by the number of bits in an @code{int}.
2211 Empty classes can be placed at incorrect offsets. For example:
2221 struct C : public B, public A @{@};
2225 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2226 it should be placed at offset zero. G++ mistakenly believes that the
2227 @code{A} data member of @code{B} is already at offset zero.
2230 Names of template functions whose types involve @code{typename} or
2231 template template parameters can be mangled incorrectly.
2234 template <typename Q>
2235 void f(typename Q::X) @{@}
2237 template <template <typename> class Q>
2238 void f(typename Q<int>::X) @{@}
2242 Instantiations of these templates may be mangled incorrectly.
2246 It also warns psABI related changes. The known psABI changes at this
2252 For SYSV/x86-64, when passing union with long double, it is changed to
2253 pass in memory as specified in psABI. For example:
2263 @code{union U} will always be passed in memory.
2267 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2268 @opindex Wctor-dtor-privacy
2269 @opindex Wno-ctor-dtor-privacy
2270 Warn when a class seems unusable because all the constructors or
2271 destructors in that class are private, and it has neither friends nor
2272 public static member functions.
2274 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2275 @opindex Wnon-virtual-dtor
2276 @opindex Wno-non-virtual-dtor
2277 Warn when a class has virtual functions and accessible non-virtual
2278 destructor, in which case it would be possible but unsafe to delete
2279 an instance of a derived class through a pointer to the base class.
2280 This warning is also enabled if -Weffc++ is specified.
2282 @item -Wreorder @r{(C++ and Objective-C++ only)}
2284 @opindex Wno-reorder
2285 @cindex reordering, warning
2286 @cindex warning for reordering of member initializers
2287 Warn when the order of member initializers given in the code does not
2288 match the order in which they must be executed. For instance:
2294 A(): j (0), i (1) @{ @}
2298 The compiler will rearrange the member initializers for @samp{i}
2299 and @samp{j} to match the declaration order of the members, emitting
2300 a warning to that effect. This warning is enabled by @option{-Wall}.
2303 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2306 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2309 Warn about violations of the following style guidelines from Scott Meyers'
2310 @cite{Effective C++} book:
2314 Item 11: Define a copy constructor and an assignment operator for classes
2315 with dynamically allocated memory.
2318 Item 12: Prefer initialization to assignment in constructors.
2321 Item 14: Make destructors virtual in base classes.
2324 Item 15: Have @code{operator=} return a reference to @code{*this}.
2327 Item 23: Don't try to return a reference when you must return an object.
2331 Also warn about violations of the following style guidelines from
2332 Scott Meyers' @cite{More Effective C++} book:
2336 Item 6: Distinguish between prefix and postfix forms of increment and
2337 decrement operators.
2340 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2344 When selecting this option, be aware that the standard library
2345 headers do not obey all of these guidelines; use @samp{grep -v}
2346 to filter out those warnings.
2348 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2349 @opindex Wstrict-null-sentinel
2350 @opindex Wno-strict-null-sentinel
2351 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2352 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2353 to @code{__null}. Although it is a null pointer constant not a null pointer,
2354 it is guaranteed to be of the same size as a pointer. But this use is
2355 not portable across different compilers.
2357 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2358 @opindex Wno-non-template-friend
2359 @opindex Wnon-template-friend
2360 Disable warnings when non-templatized friend functions are declared
2361 within a template. Since the advent of explicit template specification
2362 support in G++, if the name of the friend is an unqualified-id (i.e.,
2363 @samp{friend foo(int)}), the C++ language specification demands that the
2364 friend declare or define an ordinary, nontemplate function. (Section
2365 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2366 could be interpreted as a particular specialization of a templatized
2367 function. Because this non-conforming behavior is no longer the default
2368 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2369 check existing code for potential trouble spots and is on by default.
2370 This new compiler behavior can be turned off with
2371 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2372 but disables the helpful warning.
2374 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2375 @opindex Wold-style-cast
2376 @opindex Wno-old-style-cast
2377 Warn if an old-style (C-style) cast to a non-void type is used within
2378 a C++ program. The new-style casts (@samp{dynamic_cast},
2379 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2380 less vulnerable to unintended effects and much easier to search for.
2382 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2383 @opindex Woverloaded-virtual
2384 @opindex Wno-overloaded-virtual
2385 @cindex overloaded virtual fn, warning
2386 @cindex warning for overloaded virtual fn
2387 Warn when a function declaration hides virtual functions from a
2388 base class. For example, in:
2395 struct B: public A @{
2400 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2408 will fail to compile.
2410 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2411 @opindex Wno-pmf-conversions
2412 @opindex Wpmf-conversions
2413 Disable the diagnostic for converting a bound pointer to member function
2416 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2417 @opindex Wsign-promo
2418 @opindex Wno-sign-promo
2419 Warn when overload resolution chooses a promotion from unsigned or
2420 enumerated type to a signed type, over a conversion to an unsigned type of
2421 the same size. Previous versions of G++ would try to preserve
2422 unsignedness, but the standard mandates the current behavior.
2427 A& operator = (int);
2437 In this example, G++ will synthesize a default @samp{A& operator =
2438 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2441 @node Objective-C and Objective-C++ Dialect Options
2442 @section Options Controlling Objective-C and Objective-C++ Dialects
2444 @cindex compiler options, Objective-C and Objective-C++
2445 @cindex Objective-C and Objective-C++ options, command line
2446 @cindex options, Objective-C and Objective-C++
2447 (NOTE: This manual does not describe the Objective-C and Objective-C++
2448 languages themselves. See @xref{Standards,,Language Standards
2449 Supported by GCC}, for references.)
2451 This section describes the command-line options that are only meaningful
2452 for Objective-C and Objective-C++ programs, but you can also use most of
2453 the language-independent GNU compiler options.
2454 For example, you might compile a file @code{some_class.m} like this:
2457 gcc -g -fgnu-runtime -O -c some_class.m
2461 In this example, @option{-fgnu-runtime} is an option meant only for
2462 Objective-C and Objective-C++ programs; you can use the other options with
2463 any language supported by GCC@.
2465 Note that since Objective-C is an extension of the C language, Objective-C
2466 compilations may also use options specific to the C front-end (e.g.,
2467 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2468 C++-specific options (e.g., @option{-Wabi}).
2470 Here is a list of options that are @emph{only} for compiling Objective-C
2471 and Objective-C++ programs:
2474 @item -fconstant-string-class=@var{class-name}
2475 @opindex fconstant-string-class
2476 Use @var{class-name} as the name of the class to instantiate for each
2477 literal string specified with the syntax @code{@@"@dots{}"}. The default
2478 class name is @code{NXConstantString} if the GNU runtime is being used, and
2479 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2480 @option{-fconstant-cfstrings} option, if also present, will override the
2481 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2482 to be laid out as constant CoreFoundation strings.
2485 @opindex fgnu-runtime
2486 Generate object code compatible with the standard GNU Objective-C
2487 runtime. This is the default for most types of systems.
2489 @item -fnext-runtime
2490 @opindex fnext-runtime
2491 Generate output compatible with the NeXT runtime. This is the default
2492 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2493 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2496 @item -fno-nil-receivers
2497 @opindex fno-nil-receivers
2498 Assume that all Objective-C message dispatches (e.g.,
2499 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2500 is not @code{nil}. This allows for more efficient entry points in the runtime
2501 to be used. Currently, this option is only available in conjunction with
2502 the NeXT runtime on Mac OS X 10.3 and later.
2504 @item -fobjc-call-cxx-cdtors
2505 @opindex fobjc-call-cxx-cdtors
2506 For each Objective-C class, check if any of its instance variables is a
2507 C++ object with a non-trivial default constructor. If so, synthesize a
2508 special @code{- (id) .cxx_construct} instance method that will run
2509 non-trivial default constructors on any such instance variables, in order,
2510 and then return @code{self}. Similarly, check if any instance variable
2511 is a C++ object with a non-trivial destructor, and if so, synthesize a
2512 special @code{- (void) .cxx_destruct} method that will run
2513 all such default destructors, in reverse order.
2515 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2516 thusly generated will only operate on instance variables declared in the
2517 current Objective-C class, and not those inherited from superclasses. It
2518 is the responsibility of the Objective-C runtime to invoke all such methods
2519 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2520 will be invoked by the runtime immediately after a new object
2521 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2522 be invoked immediately before the runtime deallocates an object instance.
2524 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2525 support for invoking the @code{- (id) .cxx_construct} and
2526 @code{- (void) .cxx_destruct} methods.
2528 @item -fobjc-direct-dispatch
2529 @opindex fobjc-direct-dispatch
2530 Allow fast jumps to the message dispatcher. On Darwin this is
2531 accomplished via the comm page.
2533 @item -fobjc-exceptions
2534 @opindex fobjc-exceptions
2535 Enable syntactic support for structured exception handling in Objective-C,
2536 similar to what is offered by C++ and Java. This option is
2537 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2546 @@catch (AnObjCClass *exc) @{
2553 @@catch (AnotherClass *exc) @{
2556 @@catch (id allOthers) @{
2566 The @code{@@throw} statement may appear anywhere in an Objective-C or
2567 Objective-C++ program; when used inside of a @code{@@catch} block, the
2568 @code{@@throw} may appear without an argument (as shown above), in which case
2569 the object caught by the @code{@@catch} will be rethrown.
2571 Note that only (pointers to) Objective-C objects may be thrown and
2572 caught using this scheme. When an object is thrown, it will be caught
2573 by the nearest @code{@@catch} clause capable of handling objects of that type,
2574 analogously to how @code{catch} blocks work in C++ and Java. A
2575 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2576 any and all Objective-C exceptions not caught by previous @code{@@catch}
2579 The @code{@@finally} clause, if present, will be executed upon exit from the
2580 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2581 regardless of whether any exceptions are thrown, caught or rethrown
2582 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2583 of the @code{finally} clause in Java.
2585 There are several caveats to using the new exception mechanism:
2589 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2590 idioms provided by the @code{NSException} class, the new
2591 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2592 systems, due to additional functionality needed in the (NeXT) Objective-C
2596 As mentioned above, the new exceptions do not support handling
2597 types other than Objective-C objects. Furthermore, when used from
2598 Objective-C++, the Objective-C exception model does not interoperate with C++
2599 exceptions at this time. This means you cannot @code{@@throw} an exception
2600 from Objective-C and @code{catch} it in C++, or vice versa
2601 (i.e., @code{throw @dots{} @@catch}).
2604 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2605 blocks for thread-safe execution:
2608 @@synchronized (ObjCClass *guard) @{
2613 Upon entering the @code{@@synchronized} block, a thread of execution shall
2614 first check whether a lock has been placed on the corresponding @code{guard}
2615 object by another thread. If it has, the current thread shall wait until
2616 the other thread relinquishes its lock. Once @code{guard} becomes available,
2617 the current thread will place its own lock on it, execute the code contained in
2618 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2619 making @code{guard} available to other threads).
2621 Unlike Java, Objective-C does not allow for entire methods to be marked
2622 @code{@@synchronized}. Note that throwing exceptions out of
2623 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2624 to be unlocked properly.
2628 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2630 @item -freplace-objc-classes
2631 @opindex freplace-objc-classes
2632 Emit a special marker instructing @command{ld(1)} not to statically link in
2633 the resulting object file, and allow @command{dyld(1)} to load it in at
2634 run time instead. This is used in conjunction with the Fix-and-Continue
2635 debugging mode, where the object file in question may be recompiled and
2636 dynamically reloaded in the course of program execution, without the need
2637 to restart the program itself. Currently, Fix-and-Continue functionality
2638 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2643 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2644 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2645 compile time) with static class references that get initialized at load time,
2646 which improves run-time performance. Specifying the @option{-fzero-link} flag
2647 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2648 to be retained. This is useful in Zero-Link debugging mode, since it allows
2649 for individual class implementations to be modified during program execution.
2653 Dump interface declarations for all classes seen in the source file to a
2654 file named @file{@var{sourcename}.decl}.
2656 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2657 @opindex Wassign-intercept
2658 @opindex Wno-assign-intercept
2659 Warn whenever an Objective-C assignment is being intercepted by the
2662 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2663 @opindex Wno-protocol
2665 If a class is declared to implement a protocol, a warning is issued for
2666 every method in the protocol that is not implemented by the class. The
2667 default behavior is to issue a warning for every method not explicitly
2668 implemented in the class, even if a method implementation is inherited
2669 from the superclass. If you use the @option{-Wno-protocol} option, then
2670 methods inherited from the superclass are considered to be implemented,
2671 and no warning is issued for them.
2673 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2675 @opindex Wno-selector
2676 Warn if multiple methods of different types for the same selector are
2677 found during compilation. The check is performed on the list of methods
2678 in the final stage of compilation. Additionally, a check is performed
2679 for each selector appearing in a @code{@@selector(@dots{})}
2680 expression, and a corresponding method for that selector has been found
2681 during compilation. Because these checks scan the method table only at
2682 the end of compilation, these warnings are not produced if the final
2683 stage of compilation is not reached, for example because an error is
2684 found during compilation, or because the @option{-fsyntax-only} option is
2687 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2688 @opindex Wstrict-selector-match
2689 @opindex Wno-strict-selector-match
2690 Warn if multiple methods with differing argument and/or return types are
2691 found for a given selector when attempting to send a message using this
2692 selector to a receiver of type @code{id} or @code{Class}. When this flag
2693 is off (which is the default behavior), the compiler will omit such warnings
2694 if any differences found are confined to types which share the same size
2697 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2698 @opindex Wundeclared-selector
2699 @opindex Wno-undeclared-selector
2700 Warn if a @code{@@selector(@dots{})} expression referring to an
2701 undeclared selector is found. A selector is considered undeclared if no
2702 method with that name has been declared before the
2703 @code{@@selector(@dots{})} expression, either explicitly in an
2704 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2705 an @code{@@implementation} section. This option always performs its
2706 checks as soon as a @code{@@selector(@dots{})} expression is found,
2707 while @option{-Wselector} only performs its checks in the final stage of
2708 compilation. This also enforces the coding style convention
2709 that methods and selectors must be declared before being used.
2711 @item -print-objc-runtime-info
2712 @opindex print-objc-runtime-info
2713 Generate C header describing the largest structure that is passed by
2718 @node Language Independent Options
2719 @section Options to Control Diagnostic Messages Formatting
2720 @cindex options to control diagnostics formatting
2721 @cindex diagnostic messages
2722 @cindex message formatting
2724 Traditionally, diagnostic messages have been formatted irrespective of
2725 the output device's aspect (e.g.@: its width, @dots{}). The options described
2726 below can be used to control the diagnostic messages formatting
2727 algorithm, e.g.@: how many characters per line, how often source location
2728 information should be reported. Right now, only the C++ front end can
2729 honor these options. However it is expected, in the near future, that
2730 the remaining front ends would be able to digest them correctly.
2733 @item -fmessage-length=@var{n}
2734 @opindex fmessage-length
2735 Try to format error messages so that they fit on lines of about @var{n}
2736 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2737 the front ends supported by GCC@. If @var{n} is zero, then no
2738 line-wrapping will be done; each error message will appear on a single
2741 @opindex fdiagnostics-show-location
2742 @item -fdiagnostics-show-location=once
2743 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2744 reporter to emit @emph{once} source location information; that is, in
2745 case the message is too long to fit on a single physical line and has to
2746 be wrapped, the source location won't be emitted (as prefix) again,
2747 over and over, in subsequent continuation lines. This is the default
2750 @item -fdiagnostics-show-location=every-line
2751 Only meaningful in line-wrapping mode. Instructs the diagnostic
2752 messages reporter to emit the same source location information (as
2753 prefix) for physical lines that result from the process of breaking
2754 a message which is too long to fit on a single line.
2756 @item -fdiagnostics-show-option
2757 @opindex fdiagnostics-show-option
2758 This option instructs the diagnostic machinery to add text to each
2759 diagnostic emitted, which indicates which command line option directly
2760 controls that diagnostic, when such an option is known to the
2761 diagnostic machinery.
2763 @item -Wcoverage-mismatch
2764 @opindex Wcoverage-mismatch
2765 Warn if feedback profiles do not match when using the
2766 @option{-fprofile-use} option.
2767 If a source file was changed between @option{-fprofile-gen} and
2768 @option{-fprofile-use}, the files with the profile feedback can fail
2769 to match the source file and GCC can not use the profile feedback
2770 information. By default, this warning is enabled and is treated as an
2771 error. @option{-Wno-coverage-mismatch} can be used to disable the
2772 warning or @option{-Wno-error=coverage-mismatch} can be used to
2773 disable the error. Disable the error for this warning can result in
2774 poorly optimized code, so disabling the error is useful only in the
2775 case of very minor changes such as bug fixes to an existing code-base.
2776 Completely disabling the warning is not recommended.
2780 @node Warning Options
2781 @section Options to Request or Suppress Warnings
2782 @cindex options to control warnings
2783 @cindex warning messages
2784 @cindex messages, warning
2785 @cindex suppressing warnings
2787 Warnings are diagnostic messages that report constructions which
2788 are not inherently erroneous but which are risky or suggest there
2789 may have been an error.
2791 The following language-independent options do not enable specific
2792 warnings but control the kinds of diagnostics produced by GCC.
2795 @cindex syntax checking
2797 @opindex fsyntax-only
2798 Check the code for syntax errors, but don't do anything beyond that.
2802 Inhibit all warning messages.
2807 Make all warnings into errors.
2812 Make the specified warning into an error. The specifier for a warning
2813 is appended, for example @option{-Werror=switch} turns the warnings
2814 controlled by @option{-Wswitch} into errors. This switch takes a
2815 negative form, to be used to negate @option{-Werror} for specific
2816 warnings, for example @option{-Wno-error=switch} makes
2817 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2818 is in effect. You can use the @option{-fdiagnostics-show-option}
2819 option to have each controllable warning amended with the option which
2820 controls it, to determine what to use with this option.
2822 Note that specifying @option{-Werror=}@var{foo} automatically implies
2823 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2826 @item -Wfatal-errors
2827 @opindex Wfatal-errors
2828 @opindex Wno-fatal-errors
2829 This option causes the compiler to abort compilation on the first error
2830 occurred rather than trying to keep going and printing further error
2835 You can request many specific warnings with options beginning
2836 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2837 implicit declarations. Each of these specific warning options also
2838 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2839 example, @option{-Wno-implicit}. This manual lists only one of the
2840 two forms, whichever is not the default. For further,
2841 language-specific options also refer to @ref{C++ Dialect Options} and
2842 @ref{Objective-C and Objective-C++ Dialect Options}.
2844 When an unrecognized warning label is requested (e.g.,
2845 @option{-Wunknown-warning}), GCC will emit a diagnostic stating
2846 that the option is not recognized. However, if the @samp{-Wno-} form
2847 is used, the behavior is slightly different: No diagnostic will be
2848 produced for @option{-Wno-unknown-warning} unless other diagnostics
2849 are being produced. This allows the use of new @option{-Wno-} options
2850 with old compilers, but if something goes wrong, the compiler will
2851 warn that an unrecognized option was used.
2856 Issue all the warnings demanded by strict ISO C and ISO C++;
2857 reject all programs that use forbidden extensions, and some other
2858 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2859 version of the ISO C standard specified by any @option{-std} option used.
2861 Valid ISO C and ISO C++ programs should compile properly with or without
2862 this option (though a rare few will require @option{-ansi} or a
2863 @option{-std} option specifying the required version of ISO C)@. However,
2864 without this option, certain GNU extensions and traditional C and C++
2865 features are supported as well. With this option, they are rejected.
2867 @option{-pedantic} does not cause warning messages for use of the
2868 alternate keywords whose names begin and end with @samp{__}. Pedantic
2869 warnings are also disabled in the expression that follows
2870 @code{__extension__}. However, only system header files should use
2871 these escape routes; application programs should avoid them.
2872 @xref{Alternate Keywords}.
2874 Some users try to use @option{-pedantic} to check programs for strict ISO
2875 C conformance. They soon find that it does not do quite what they want:
2876 it finds some non-ISO practices, but not all---only those for which
2877 ISO C @emph{requires} a diagnostic, and some others for which
2878 diagnostics have been added.
2880 A feature to report any failure to conform to ISO C might be useful in
2881 some instances, but would require considerable additional work and would
2882 be quite different from @option{-pedantic}. We don't have plans to
2883 support such a feature in the near future.
2885 Where the standard specified with @option{-std} represents a GNU
2886 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
2887 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2888 extended dialect is based. Warnings from @option{-pedantic} are given
2889 where they are required by the base standard. (It would not make sense
2890 for such warnings to be given only for features not in the specified GNU
2891 C dialect, since by definition the GNU dialects of C include all
2892 features the compiler supports with the given option, and there would be
2893 nothing to warn about.)
2895 @item -pedantic-errors
2896 @opindex pedantic-errors
2897 Like @option{-pedantic}, except that errors are produced rather than
2903 This enables all the warnings about constructions that some users
2904 consider questionable, and that are easy to avoid (or modify to
2905 prevent the warning), even in conjunction with macros. This also
2906 enables some language-specific warnings described in @ref{C++ Dialect
2907 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2909 @option{-Wall} turns on the following warning flags:
2911 @gccoptlist{-Waddress @gol
2912 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2914 -Wchar-subscripts @gol
2915 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2916 -Wimplicit-int @r{(C and Objective-C only)} @gol
2917 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
2920 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2921 -Wmissing-braces @gol
2927 -Wsequence-point @gol
2928 -Wsign-compare @r{(only in C++)} @gol
2929 -Wstrict-aliasing @gol
2930 -Wstrict-overflow=1 @gol
2933 -Wuninitialized @gol
2934 -Wunknown-pragmas @gol
2935 -Wunused-function @gol
2938 -Wunused-variable @gol
2939 -Wvolatile-register-var @gol
2942 Note that some warning flags are not implied by @option{-Wall}. Some of
2943 them warn about constructions that users generally do not consider
2944 questionable, but which occasionally you might wish to check for;
2945 others warn about constructions that are necessary or hard to avoid in
2946 some cases, and there is no simple way to modify the code to suppress
2947 the warning. Some of them are enabled by @option{-Wextra} but many of
2948 them must be enabled individually.
2954 This enables some extra warning flags that are not enabled by
2955 @option{-Wall}. (This option used to be called @option{-W}. The older
2956 name is still supported, but the newer name is more descriptive.)
2958 @gccoptlist{-Wclobbered @gol
2960 -Wignored-qualifiers @gol
2961 -Wmissing-field-initializers @gol
2962 -Wmissing-parameter-type @r{(C only)} @gol
2963 -Wold-style-declaration @r{(C only)} @gol
2964 -Woverride-init @gol
2967 -Wuninitialized @gol
2968 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2969 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2972 The option @option{-Wextra} also prints warning messages for the
2978 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2979 @samp{>}, or @samp{>=}.
2982 (C++ only) An enumerator and a non-enumerator both appear in a
2983 conditional expression.
2986 (C++ only) Ambiguous virtual bases.
2989 (C++ only) Subscripting an array which has been declared @samp{register}.
2992 (C++ only) Taking the address of a variable which has been declared
2996 (C++ only) A base class is not initialized in a derived class' copy
3001 @item -Wchar-subscripts
3002 @opindex Wchar-subscripts
3003 @opindex Wno-char-subscripts
3004 Warn if an array subscript has type @code{char}. This is a common cause
3005 of error, as programmers often forget that this type is signed on some
3007 This warning is enabled by @option{-Wall}.
3011 @opindex Wno-comment
3012 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3013 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3014 This warning is enabled by @option{-Wall}.
3017 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3019 Suppress warning messages emitted by @code{#warning} directives.
3024 @opindex ffreestanding
3025 @opindex fno-builtin
3026 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3027 the arguments supplied have types appropriate to the format string
3028 specified, and that the conversions specified in the format string make
3029 sense. This includes standard functions, and others specified by format
3030 attributes (@pxref{Function Attributes}), in the @code{printf},
3031 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3032 not in the C standard) families (or other target-specific families).
3033 Which functions are checked without format attributes having been
3034 specified depends on the standard version selected, and such checks of
3035 functions without the attribute specified are disabled by
3036 @option{-ffreestanding} or @option{-fno-builtin}.
3038 The formats are checked against the format features supported by GNU
3039 libc version 2.2. These include all ISO C90 and C99 features, as well
3040 as features from the Single Unix Specification and some BSD and GNU
3041 extensions. Other library implementations may not support all these
3042 features; GCC does not support warning about features that go beyond a
3043 particular library's limitations. However, if @option{-pedantic} is used
3044 with @option{-Wformat}, warnings will be given about format features not
3045 in the selected standard version (but not for @code{strfmon} formats,
3046 since those are not in any version of the C standard). @xref{C Dialect
3047 Options,,Options Controlling C Dialect}.
3049 Since @option{-Wformat} also checks for null format arguments for
3050 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3052 @option{-Wformat} is included in @option{-Wall}. For more control over some
3053 aspects of format checking, the options @option{-Wformat-y2k},
3054 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3055 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3056 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
3059 @opindex Wformat-y2k
3060 @opindex Wno-format-y2k
3061 If @option{-Wformat} is specified, also warn about @code{strftime}
3062 formats which may yield only a two-digit year.
3064 @item -Wno-format-contains-nul
3065 @opindex Wno-format-contains-nul
3066 @opindex Wformat-contains-nul
3067 If @option{-Wformat} is specified, do not warn about format strings that
3070 @item -Wno-format-extra-args
3071 @opindex Wno-format-extra-args
3072 @opindex Wformat-extra-args
3073 If @option{-Wformat} is specified, do not warn about excess arguments to a
3074 @code{printf} or @code{scanf} format function. The C standard specifies
3075 that such arguments are ignored.
3077 Where the unused arguments lie between used arguments that are
3078 specified with @samp{$} operand number specifications, normally
3079 warnings are still given, since the implementation could not know what
3080 type to pass to @code{va_arg} to skip the unused arguments. However,
3081 in the case of @code{scanf} formats, this option will suppress the
3082 warning if the unused arguments are all pointers, since the Single
3083 Unix Specification says that such unused arguments are allowed.
3085 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3086 @opindex Wno-format-zero-length
3087 @opindex Wformat-zero-length
3088 If @option{-Wformat} is specified, do not warn about zero-length formats.
3089 The C standard specifies that zero-length formats are allowed.
3091 @item -Wformat-nonliteral
3092 @opindex Wformat-nonliteral
3093 @opindex Wno-format-nonliteral
3094 If @option{-Wformat} is specified, also warn if the format string is not a
3095 string literal and so cannot be checked, unless the format function
3096 takes its format arguments as a @code{va_list}.
3098 @item -Wformat-security
3099 @opindex Wformat-security
3100 @opindex Wno-format-security
3101 If @option{-Wformat} is specified, also warn about uses of format
3102 functions that represent possible security problems. At present, this
3103 warns about calls to @code{printf} and @code{scanf} functions where the
3104 format string is not a string literal and there are no format arguments,
3105 as in @code{printf (foo);}. This may be a security hole if the format
3106 string came from untrusted input and contains @samp{%n}. (This is
3107 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3108 in future warnings may be added to @option{-Wformat-security} that are not
3109 included in @option{-Wformat-nonliteral}.)
3113 @opindex Wno-format=2
3114 Enable @option{-Wformat} plus format checks not included in
3115 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3116 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3118 @item -Wnonnull @r{(C and Objective-C only)}
3120 @opindex Wno-nonnull
3121 Warn about passing a null pointer for arguments marked as
3122 requiring a non-null value by the @code{nonnull} function attribute.
3124 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3125 can be disabled with the @option{-Wno-nonnull} option.
3127 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3129 @opindex Wno-init-self
3130 Warn about uninitialized variables which are initialized with themselves.
3131 Note this option can only be used with the @option{-Wuninitialized} option.
3133 For example, GCC will warn about @code{i} being uninitialized in the
3134 following snippet only when @option{-Winit-self} has been specified:
3145 @item -Wimplicit-int @r{(C and Objective-C only)}
3146 @opindex Wimplicit-int
3147 @opindex Wno-implicit-int
3148 Warn when a declaration does not specify a type.
3149 This warning is enabled by @option{-Wall}.
3151 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3152 @opindex Wimplicit-function-declaration
3153 @opindex Wno-implicit-function-declaration
3154 Give a warning whenever a function is used before being declared. In
3155 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3156 enabled by default and it is made into an error by
3157 @option{-pedantic-errors}. This warning is also enabled by
3160 @item -Wimplicit @r{(C and Objective-C only)}
3162 @opindex Wno-implicit
3163 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3164 This warning is enabled by @option{-Wall}.
3166 @item -Wignored-qualifiers @r{(C and C++ only)}
3167 @opindex Wignored-qualifiers
3168 @opindex Wno-ignored-qualifiers
3169 Warn if the return type of a function has a type qualifier
3170 such as @code{const}. For ISO C such a type qualifier has no effect,
3171 since the value returned by a function is not an lvalue.
3172 For C++, the warning is only emitted for scalar types or @code{void}.
3173 ISO C prohibits qualified @code{void} return types on function
3174 definitions, so such return types always receive a warning
3175 even without this option.
3177 This warning is also enabled by @option{-Wextra}.
3182 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3183 a function with external linkage, returning int, taking either zero
3184 arguments, two, or three arguments of appropriate types. This warning
3185 is enabled by default in C++ and is enabled by either @option{-Wall}
3186 or @option{-pedantic}.
3188 @item -Wmissing-braces
3189 @opindex Wmissing-braces
3190 @opindex Wno-missing-braces
3191 Warn if an aggregate or union initializer is not fully bracketed. In
3192 the following example, the initializer for @samp{a} is not fully
3193 bracketed, but that for @samp{b} is fully bracketed.
3196 int a[2][2] = @{ 0, 1, 2, 3 @};
3197 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3200 This warning is enabled by @option{-Wall}.
3202 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3203 @opindex Wmissing-include-dirs
3204 @opindex Wno-missing-include-dirs
3205 Warn if a user-supplied include directory does not exist.
3208 @opindex Wparentheses
3209 @opindex Wno-parentheses
3210 Warn if parentheses are omitted in certain contexts, such
3211 as when there is an assignment in a context where a truth value
3212 is expected, or when operators are nested whose precedence people
3213 often get confused about.
3215 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3216 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3217 interpretation from that of ordinary mathematical notation.
3219 Also warn about constructions where there may be confusion to which
3220 @code{if} statement an @code{else} branch belongs. Here is an example of
3235 In C/C++, every @code{else} branch belongs to the innermost possible
3236 @code{if} statement, which in this example is @code{if (b)}. This is
3237 often not what the programmer expected, as illustrated in the above
3238 example by indentation the programmer chose. When there is the
3239 potential for this confusion, GCC will issue a warning when this flag
3240 is specified. To eliminate the warning, add explicit braces around
3241 the innermost @code{if} statement so there is no way the @code{else}
3242 could belong to the enclosing @code{if}. The resulting code would
3259 This warning is enabled by @option{-Wall}.
3261 @item -Wsequence-point
3262 @opindex Wsequence-point
3263 @opindex Wno-sequence-point
3264 Warn about code that may have undefined semantics because of violations
3265 of sequence point rules in the C and C++ standards.
3267 The C and C++ standards defines the order in which expressions in a C/C++
3268 program are evaluated in terms of @dfn{sequence points}, which represent
3269 a partial ordering between the execution of parts of the program: those
3270 executed before the sequence point, and those executed after it. These
3271 occur after the evaluation of a full expression (one which is not part
3272 of a larger expression), after the evaluation of the first operand of a
3273 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3274 function is called (but after the evaluation of its arguments and the
3275 expression denoting the called function), and in certain other places.
3276 Other than as expressed by the sequence point rules, the order of
3277 evaluation of subexpressions of an expression is not specified. All
3278 these rules describe only a partial order rather than a total order,
3279 since, for example, if two functions are called within one expression
3280 with no sequence point between them, the order in which the functions
3281 are called is not specified. However, the standards committee have
3282 ruled that function calls do not overlap.
3284 It is not specified when between sequence points modifications to the
3285 values of objects take effect. Programs whose behavior depends on this
3286 have undefined behavior; the C and C++ standards specify that ``Between
3287 the previous and next sequence point an object shall have its stored
3288 value modified at most once by the evaluation of an expression.
3289 Furthermore, the prior value shall be read only to determine the value
3290 to be stored.''. If a program breaks these rules, the results on any
3291 particular implementation are entirely unpredictable.
3293 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3294 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3295 diagnosed by this option, and it may give an occasional false positive
3296 result, but in general it has been found fairly effective at detecting
3297 this sort of problem in programs.
3299 The standard is worded confusingly, therefore there is some debate
3300 over the precise meaning of the sequence point rules in subtle cases.
3301 Links to discussions of the problem, including proposed formal
3302 definitions, may be found on the GCC readings page, at
3303 @w{@uref{http://gcc.gnu.org/readings.html}}.
3305 This warning is enabled by @option{-Wall} for C and C++.
3308 @opindex Wreturn-type
3309 @opindex Wno-return-type
3310 Warn whenever a function is defined with a return-type that defaults
3311 to @code{int}. Also warn about any @code{return} statement with no
3312 return-value in a function whose return-type is not @code{void}
3313 (falling off the end of the function body is considered returning
3314 without a value), and about a @code{return} statement with an
3315 expression in a function whose return-type is @code{void}.
3317 For C++, a function without return type always produces a diagnostic
3318 message, even when @option{-Wno-return-type} is specified. The only
3319 exceptions are @samp{main} and functions defined in system headers.
3321 This warning is enabled by @option{-Wall}.
3326 Warn whenever a @code{switch} statement has an index of enumerated type
3327 and lacks a @code{case} for one or more of the named codes of that
3328 enumeration. (The presence of a @code{default} label prevents this
3329 warning.) @code{case} labels outside the enumeration range also
3330 provoke warnings when this option is used (even if there is a
3331 @code{default} label).
3332 This warning is enabled by @option{-Wall}.
3334 @item -Wswitch-default
3335 @opindex Wswitch-default
3336 @opindex Wno-switch-default
3337 Warn whenever a @code{switch} statement does not have a @code{default}
3341 @opindex Wswitch-enum
3342 @opindex Wno-switch-enum
3343 Warn whenever a @code{switch} statement has an index of enumerated type
3344 and lacks a @code{case} for one or more of the named codes of that
3345 enumeration. @code{case} labels outside the enumeration range also
3346 provoke warnings when this option is used. The only difference
3347 between @option{-Wswitch} and this option is that this option gives a
3348 warning about an omitted enumeration code even if there is a
3349 @code{default} label.
3351 @item -Wsync-nand @r{(C and C++ only)}
3353 @opindex Wno-sync-nand
3354 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3355 built-in functions are used. These functions changed semantics in GCC 4.4.
3359 @opindex Wno-trigraphs
3360 Warn if any trigraphs are encountered that might change the meaning of
3361 the program (trigraphs within comments are not warned about).
3362 This warning is enabled by @option{-Wall}.
3364 @item -Wunused-but-set-parameter
3365 @opindex Wunused-but-set-parameter
3366 @opindex Wno-unused-but-set-parameter
3367 Warn whenever a function parameter is assigned to, but otherwise unused
3368 (aside from its declaration).
3370 To suppress this warning use the @samp{unused} attribute
3371 (@pxref{Variable Attributes}).
3373 This warning is also enabled by @option{-Wunused} together with
3376 @item -Wunused-but-set-variable
3377 @opindex Wunused-but-set-variable
3378 @opindex Wno-unused-but-set-variable
3379 Warn whenever a local variable is assigned to, but otherwise unused
3380 (aside from its declaration).
3381 This warning is enabled by @option{-Wall}.
3383 To suppress this warning use the @samp{unused} attribute
3384 (@pxref{Variable Attributes}).
3386 This warning is also enabled by @option{-Wunused}, which is enabled
3389 @item -Wunused-function
3390 @opindex Wunused-function
3391 @opindex Wno-unused-function
3392 Warn whenever a static function is declared but not defined or a
3393 non-inline static function is unused.
3394 This warning is enabled by @option{-Wall}.
3396 @item -Wunused-label
3397 @opindex Wunused-label
3398 @opindex Wno-unused-label
3399 Warn whenever a label is declared but not used.
3400 This warning is enabled by @option{-Wall}.
3402 To suppress this warning use the @samp{unused} attribute
3403 (@pxref{Variable Attributes}).
3405 @item -Wunused-parameter
3406 @opindex Wunused-parameter
3407 @opindex Wno-unused-parameter
3408 Warn whenever a function parameter is unused aside from its declaration.
3410 To suppress this warning use the @samp{unused} attribute
3411 (@pxref{Variable Attributes}).
3413 @item -Wno-unused-result
3414 @opindex Wunused-result
3415 @opindex Wno-unused-result
3416 Do not warn if a caller of a function marked with attribute
3417 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3418 its return value. The default is @option{-Wunused-result}.
3420 @item -Wunused-variable
3421 @opindex Wunused-variable
3422 @opindex Wno-unused-variable
3423 Warn whenever a local variable or non-constant static variable is unused
3424 aside from its declaration.
3425 This warning is enabled by @option{-Wall}.
3427 To suppress this warning use the @samp{unused} attribute
3428 (@pxref{Variable Attributes}).
3430 @item -Wunused-value
3431 @opindex Wunused-value
3432 @opindex Wno-unused-value
3433 Warn whenever a statement computes a result that is explicitly not
3434 used. To suppress this warning cast the unused expression to
3435 @samp{void}. This includes an expression-statement or the left-hand
3436 side of a comma expression that contains no side effects. For example,
3437 an expression such as @samp{x[i,j]} will cause a warning, while
3438 @samp{x[(void)i,j]} will not.
3440 This warning is enabled by @option{-Wall}.
3445 All the above @option{-Wunused} options combined.
3447 In order to get a warning about an unused function parameter, you must
3448 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3449 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3451 @item -Wuninitialized
3452 @opindex Wuninitialized
3453 @opindex Wno-uninitialized
3454 Warn if an automatic variable is used without first being initialized
3455 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3456 warn if a non-static reference or non-static @samp{const} member
3457 appears in a class without constructors.
3459 If you want to warn about code which uses the uninitialized value of the
3460 variable in its own initializer, use the @option{-Winit-self} option.
3462 These warnings occur for individual uninitialized or clobbered
3463 elements of structure, union or array variables as well as for
3464 variables which are uninitialized or clobbered as a whole. They do
3465 not occur for variables or elements declared @code{volatile}. Because
3466 these warnings depend on optimization, the exact variables or elements
3467 for which there are warnings will depend on the precise optimization
3468 options and version of GCC used.
3470 Note that there may be no warning about a variable that is used only
3471 to compute a value that itself is never used, because such
3472 computations may be deleted by data flow analysis before the warnings
3475 These warnings are made optional because GCC is not smart
3476 enough to see all the reasons why the code might be correct
3477 despite appearing to have an error. Here is one example of how
3498 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3499 always initialized, but GCC doesn't know this. Here is
3500 another common case:
3505 if (change_y) save_y = y, y = new_y;
3507 if (change_y) y = save_y;
3512 This has no bug because @code{save_y} is used only if it is set.
3514 @cindex @code{longjmp} warnings
3515 This option also warns when a non-volatile automatic variable might be
3516 changed by a call to @code{longjmp}. These warnings as well are possible
3517 only in optimizing compilation.
3519 The compiler sees only the calls to @code{setjmp}. It cannot know
3520 where @code{longjmp} will be called; in fact, a signal handler could
3521 call it at any point in the code. As a result, you may get a warning
3522 even when there is in fact no problem because @code{longjmp} cannot
3523 in fact be called at the place which would cause a problem.
3525 Some spurious warnings can be avoided if you declare all the functions
3526 you use that never return as @code{noreturn}. @xref{Function
3529 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3531 @item -Wunknown-pragmas
3532 @opindex Wunknown-pragmas
3533 @opindex Wno-unknown-pragmas
3534 @cindex warning for unknown pragmas
3535 @cindex unknown pragmas, warning
3536 @cindex pragmas, warning of unknown
3537 Warn when a #pragma directive is encountered which is not understood by
3538 GCC@. If this command line option is used, warnings will even be issued
3539 for unknown pragmas in system header files. This is not the case if
3540 the warnings were only enabled by the @option{-Wall} command line option.
3543 @opindex Wno-pragmas
3545 Do not warn about misuses of pragmas, such as incorrect parameters,
3546 invalid syntax, or conflicts between pragmas. See also
3547 @samp{-Wunknown-pragmas}.
3549 @item -Wstrict-aliasing
3550 @opindex Wstrict-aliasing
3551 @opindex Wno-strict-aliasing
3552 This option is only active when @option{-fstrict-aliasing} is active.
3553 It warns about code which might break the strict aliasing rules that the
3554 compiler is using for optimization. The warning does not catch all
3555 cases, but does attempt to catch the more common pitfalls. It is
3556 included in @option{-Wall}.
3557 It is equivalent to @option{-Wstrict-aliasing=3}
3559 @item -Wstrict-aliasing=n
3560 @opindex Wstrict-aliasing=n
3561 @opindex Wno-strict-aliasing=n
3562 This option is only active when @option{-fstrict-aliasing} is active.
3563 It warns about code which might break the strict aliasing rules that the
3564 compiler is using for optimization.
3565 Higher levels correspond to higher accuracy (fewer false positives).
3566 Higher levels also correspond to more effort, similar to the way -O works.
3567 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3570 Level 1: Most aggressive, quick, least accurate.
3571 Possibly useful when higher levels
3572 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3573 false negatives. However, it has many false positives.
3574 Warns for all pointer conversions between possibly incompatible types,
3575 even if never dereferenced. Runs in the frontend only.
3577 Level 2: Aggressive, quick, not too precise.
3578 May still have many false positives (not as many as level 1 though),
3579 and few false negatives (but possibly more than level 1).
3580 Unlike level 1, it only warns when an address is taken. Warns about
3581 incomplete types. Runs in the frontend only.
3583 Level 3 (default for @option{-Wstrict-aliasing}):
3584 Should have very few false positives and few false
3585 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3586 Takes care of the common pun+dereference pattern in the frontend:
3587 @code{*(int*)&some_float}.
3588 If optimization is enabled, it also runs in the backend, where it deals
3589 with multiple statement cases using flow-sensitive points-to information.
3590 Only warns when the converted pointer is dereferenced.
3591 Does not warn about incomplete types.
3593 @item -Wstrict-overflow
3594 @itemx -Wstrict-overflow=@var{n}
3595 @opindex Wstrict-overflow
3596 @opindex Wno-strict-overflow
3597 This option is only active when @option{-fstrict-overflow} is active.
3598 It warns about cases where the compiler optimizes based on the
3599 assumption that signed overflow does not occur. Note that it does not
3600 warn about all cases where the code might overflow: it only warns
3601 about cases where the compiler implements some optimization. Thus
3602 this warning depends on the optimization level.
3604 An optimization which assumes that signed overflow does not occur is
3605 perfectly safe if the values of the variables involved are such that
3606 overflow never does, in fact, occur. Therefore this warning can
3607 easily give a false positive: a warning about code which is not
3608 actually a problem. To help focus on important issues, several
3609 warning levels are defined. No warnings are issued for the use of
3610 undefined signed overflow when estimating how many iterations a loop
3611 will require, in particular when determining whether a loop will be
3615 @item -Wstrict-overflow=1
3616 Warn about cases which are both questionable and easy to avoid. For
3617 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3618 compiler will simplify this to @code{1}. This level of
3619 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3620 are not, and must be explicitly requested.
3622 @item -Wstrict-overflow=2
3623 Also warn about other cases where a comparison is simplified to a
3624 constant. For example: @code{abs (x) >= 0}. This can only be
3625 simplified when @option{-fstrict-overflow} is in effect, because
3626 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3627 zero. @option{-Wstrict-overflow} (with no level) is the same as
3628 @option{-Wstrict-overflow=2}.
3630 @item -Wstrict-overflow=3
3631 Also warn about other cases where a comparison is simplified. For
3632 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3634 @item -Wstrict-overflow=4
3635 Also warn about other simplifications not covered by the above cases.
3636 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3638 @item -Wstrict-overflow=5
3639 Also warn about cases where the compiler reduces the magnitude of a
3640 constant involved in a comparison. For example: @code{x + 2 > y} will
3641 be simplified to @code{x + 1 >= y}. This is reported only at the
3642 highest warning level because this simplification applies to many
3643 comparisons, so this warning level will give a very large number of
3647 @item -Wsuggest-attribute=@r{[}const@r{|}pure@r{]}
3648 @opindex Wsuggest-attribute=
3649 @opindex Wno-suggest-attribute=
3650 Warn for cases where adding an attribute may be beneficial. The
3651 attributes currently supported are listed below.
3654 @item -Wsuggest-attribute=pure
3655 @itemx -Wsuggest-attribute=const
3656 @opindex Wsuggest-attribute=pure
3657 @opindex Wno-suggest-attribute=pure
3658 @opindex Wsuggest-attribute=const
3659 @opindex Wno-suggest-attribute=const
3661 Warn about functions which might be candidates for attributes
3662 @code{pure} or @code{const}. The compiler only warns for functions
3663 visible in other compilation units or if it cannot prove that the
3664 function returns normally. A function returns normally if it doesn't
3665 contain an infinite loop nor returns abnormally by throwing, calling
3666 @code{abort()} or trapping. This analysis requires option
3667 @option{-fipa-pure-const}, which is enabled by default at @option{-O}
3668 and higher. Higher optimization levels improve the accuracy of the
3672 @item -Warray-bounds
3673 @opindex Wno-array-bounds
3674 @opindex Warray-bounds
3675 This option is only active when @option{-ftree-vrp} is active
3676 (default for @option{-O2} and above). It warns about subscripts to arrays
3677 that are always out of bounds. This warning is enabled by @option{-Wall}.
3679 @item -Wno-div-by-zero
3680 @opindex Wno-div-by-zero
3681 @opindex Wdiv-by-zero
3682 Do not warn about compile-time integer division by zero. Floating point
3683 division by zero is not warned about, as it can be a legitimate way of
3684 obtaining infinities and NaNs.
3686 @item -Wsystem-headers
3687 @opindex Wsystem-headers
3688 @opindex Wno-system-headers
3689 @cindex warnings from system headers
3690 @cindex system headers, warnings from
3691 Print warning messages for constructs found in system header files.
3692 Warnings from system headers are normally suppressed, on the assumption
3693 that they usually do not indicate real problems and would only make the
3694 compiler output harder to read. Using this command line option tells
3695 GCC to emit warnings from system headers as if they occurred in user
3696 code. However, note that using @option{-Wall} in conjunction with this
3697 option will @emph{not} warn about unknown pragmas in system
3698 headers---for that, @option{-Wunknown-pragmas} must also be used.
3701 @opindex Wfloat-equal
3702 @opindex Wno-float-equal
3703 Warn if floating point values are used in equality comparisons.
3705 The idea behind this is that sometimes it is convenient (for the
3706 programmer) to consider floating-point values as approximations to
3707 infinitely precise real numbers. If you are doing this, then you need
3708 to compute (by analyzing the code, or in some other way) the maximum or
3709 likely maximum error that the computation introduces, and allow for it
3710 when performing comparisons (and when producing output, but that's a
3711 different problem). In particular, instead of testing for equality, you
3712 would check to see whether the two values have ranges that overlap; and
3713 this is done with the relational operators, so equality comparisons are
3716 @item -Wtraditional @r{(C and Objective-C only)}
3717 @opindex Wtraditional
3718 @opindex Wno-traditional
3719 Warn about certain constructs that behave differently in traditional and
3720 ISO C@. Also warn about ISO C constructs that have no traditional C
3721 equivalent, and/or problematic constructs which should be avoided.
3725 Macro parameters that appear within string literals in the macro body.
3726 In traditional C macro replacement takes place within string literals,
3727 but does not in ISO C@.
3730 In traditional C, some preprocessor directives did not exist.
3731 Traditional preprocessors would only consider a line to be a directive
3732 if the @samp{#} appeared in column 1 on the line. Therefore
3733 @option{-Wtraditional} warns about directives that traditional C
3734 understands but would ignore because the @samp{#} does not appear as the
3735 first character on the line. It also suggests you hide directives like
3736 @samp{#pragma} not understood by traditional C by indenting them. Some
3737 traditional implementations would not recognize @samp{#elif}, so it
3738 suggests avoiding it altogether.
3741 A function-like macro that appears without arguments.
3744 The unary plus operator.
3747 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3748 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3749 constants.) Note, these suffixes appear in macros defined in the system
3750 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3751 Use of these macros in user code might normally lead to spurious
3752 warnings, however GCC's integrated preprocessor has enough context to
3753 avoid warning in these cases.
3756 A function declared external in one block and then used after the end of
3760 A @code{switch} statement has an operand of type @code{long}.
3763 A non-@code{static} function declaration follows a @code{static} one.
3764 This construct is not accepted by some traditional C compilers.
3767 The ISO type of an integer constant has a different width or
3768 signedness from its traditional type. This warning is only issued if
3769 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3770 typically represent bit patterns, are not warned about.
3773 Usage of ISO string concatenation is detected.
3776 Initialization of automatic aggregates.
3779 Identifier conflicts with labels. Traditional C lacks a separate
3780 namespace for labels.
3783 Initialization of unions. If the initializer is zero, the warning is
3784 omitted. This is done under the assumption that the zero initializer in
3785 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3786 initializer warnings and relies on default initialization to zero in the
3790 Conversions by prototypes between fixed/floating point values and vice
3791 versa. The absence of these prototypes when compiling with traditional
3792 C would cause serious problems. This is a subset of the possible
3793 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3796 Use of ISO C style function definitions. This warning intentionally is
3797 @emph{not} issued for prototype declarations or variadic functions
3798 because these ISO C features will appear in your code when using
3799 libiberty's traditional C compatibility macros, @code{PARAMS} and
3800 @code{VPARAMS}. This warning is also bypassed for nested functions
3801 because that feature is already a GCC extension and thus not relevant to
3802 traditional C compatibility.
3805 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3806 @opindex Wtraditional-conversion
3807 @opindex Wno-traditional-conversion
3808 Warn if a prototype causes a type conversion that is different from what
3809 would happen to the same argument in the absence of a prototype. This
3810 includes conversions of fixed point to floating and vice versa, and
3811 conversions changing the width or signedness of a fixed point argument
3812 except when the same as the default promotion.
3814 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3815 @opindex Wdeclaration-after-statement
3816 @opindex Wno-declaration-after-statement
3817 Warn when a declaration is found after a statement in a block. This
3818 construct, known from C++, was introduced with ISO C99 and is by default
3819 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3820 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3825 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3827 @item -Wno-endif-labels
3828 @opindex Wno-endif-labels
3829 @opindex Wendif-labels
3830 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3835 Warn whenever a local variable shadows another local variable, parameter or
3836 global variable or whenever a built-in function is shadowed.
3838 @item -Wlarger-than=@var{len}
3839 @opindex Wlarger-than=@var{len}
3840 @opindex Wlarger-than-@var{len}
3841 Warn whenever an object of larger than @var{len} bytes is defined.
3843 @item -Wframe-larger-than=@var{len}
3844 @opindex Wframe-larger-than
3845 Warn if the size of a function frame is larger than @var{len} bytes.
3846 The computation done to determine the stack frame size is approximate
3847 and not conservative.
3848 The actual requirements may be somewhat greater than @var{len}
3849 even if you do not get a warning. In addition, any space allocated
3850 via @code{alloca}, variable-length arrays, or related constructs
3851 is not included by the compiler when determining
3852 whether or not to issue a warning.
3854 @item -Wunsafe-loop-optimizations
3855 @opindex Wunsafe-loop-optimizations
3856 @opindex Wno-unsafe-loop-optimizations
3857 Warn if the loop cannot be optimized because the compiler could not
3858 assume anything on the bounds of the loop indices. With
3859 @option{-funsafe-loop-optimizations} warn if the compiler made
3862 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3863 @opindex Wno-pedantic-ms-format
3864 @opindex Wpedantic-ms-format
3865 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3866 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3867 depending on the MS runtime, when you are using the options @option{-Wformat}
3868 and @option{-pedantic} without gnu-extensions.
3870 @item -Wpointer-arith
3871 @opindex Wpointer-arith
3872 @opindex Wno-pointer-arith
3873 Warn about anything that depends on the ``size of'' a function type or
3874 of @code{void}. GNU C assigns these types a size of 1, for
3875 convenience in calculations with @code{void *} pointers and pointers
3876 to functions. In C++, warn also when an arithmetic operation involves
3877 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3880 @opindex Wtype-limits
3881 @opindex Wno-type-limits
3882 Warn if a comparison is always true or always false due to the limited
3883 range of the data type, but do not warn for constant expressions. For
3884 example, warn if an unsigned variable is compared against zero with
3885 @samp{<} or @samp{>=}. This warning is also enabled by
3888 @item -Wbad-function-cast @r{(C and Objective-C only)}
3889 @opindex Wbad-function-cast
3890 @opindex Wno-bad-function-cast
3891 Warn whenever a function call is cast to a non-matching type.
3892 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3894 @item -Wc++-compat @r{(C and Objective-C only)}
3895 Warn about ISO C constructs that are outside of the common subset of
3896 ISO C and ISO C++, e.g.@: request for implicit conversion from
3897 @code{void *} to a pointer to non-@code{void} type.
3899 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3900 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3901 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3902 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3906 @opindex Wno-cast-qual
3907 Warn whenever a pointer is cast so as to remove a type qualifier from
3908 the target type. For example, warn if a @code{const char *} is cast
3909 to an ordinary @code{char *}.
3911 Also warn when making a cast which introduces a type qualifier in an
3912 unsafe way. For example, casting @code{char **} to @code{const char **}
3913 is unsafe, as in this example:
3916 /* p is char ** value. */
3917 const char **q = (const char **) p;
3918 /* Assignment of readonly string to const char * is OK. */
3920 /* Now char** pointer points to read-only memory. */
3925 @opindex Wcast-align
3926 @opindex Wno-cast-align
3927 Warn whenever a pointer is cast such that the required alignment of the
3928 target is increased. For example, warn if a @code{char *} is cast to
3929 an @code{int *} on machines where integers can only be accessed at
3930 two- or four-byte boundaries.
3932 @item -Wwrite-strings
3933 @opindex Wwrite-strings
3934 @opindex Wno-write-strings
3935 When compiling C, give string constants the type @code{const
3936 char[@var{length}]} so that copying the address of one into a
3937 non-@code{const} @code{char *} pointer will get a warning. These
3938 warnings will help you find at compile time code that can try to write
3939 into a string constant, but only if you have been very careful about
3940 using @code{const} in declarations and prototypes. Otherwise, it will
3941 just be a nuisance. This is why we did not make @option{-Wall} request
3944 When compiling C++, warn about the deprecated conversion from string
3945 literals to @code{char *}. This warning is enabled by default for C++
3950 @opindex Wno-clobbered
3951 Warn for variables that might be changed by @samp{longjmp} or
3952 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3955 @opindex Wconversion
3956 @opindex Wno-conversion
3957 Warn for implicit conversions that may alter a value. This includes
3958 conversions between real and integer, like @code{abs (x)} when
3959 @code{x} is @code{double}; conversions between signed and unsigned,
3960 like @code{unsigned ui = -1}; and conversions to smaller types, like
3961 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3962 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3963 changed by the conversion like in @code{abs (2.0)}. Warnings about
3964 conversions between signed and unsigned integers can be disabled by
3965 using @option{-Wno-sign-conversion}.
3967 For C++, also warn for confusing overload resolution for user-defined
3968 conversions; and conversions that will never use a type conversion
3969 operator: conversions to @code{void}, the same type, a base class or a
3970 reference to them. Warnings about conversions between signed and
3971 unsigned integers are disabled by default in C++ unless
3972 @option{-Wsign-conversion} is explicitly enabled.
3974 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
3975 @opindex Wconversion-null
3976 @opindex Wno-conversion-null
3977 Do not warn for conversions between @code{NULL} and non-pointer
3978 types. @option{-Wconversion-null} is enabled by default.
3981 @opindex Wempty-body
3982 @opindex Wno-empty-body
3983 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3984 while} statement. This warning is also enabled by @option{-Wextra}.
3986 @item -Wenum-compare
3987 @opindex Wenum-compare
3988 @opindex Wno-enum-compare
3989 Warn about a comparison between values of different enum types. In C++
3990 this warning is enabled by default. In C this warning is enabled by
3993 @item -Wjump-misses-init @r{(C, Objective-C only)}
3994 @opindex Wjump-misses-init
3995 @opindex Wno-jump-misses-init
3996 Warn if a @code{goto} statement or a @code{switch} statement jumps
3997 forward across the initialization of a variable, or jumps backward to a
3998 label after the variable has been initialized. This only warns about
3999 variables which are initialized when they are declared. This warning is
4000 only supported for C and Objective C; in C++ this sort of branch is an
4003 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
4004 can be disabled with the @option{-Wno-jump-misses-init} option.
4006 @item -Wsign-compare
4007 @opindex Wsign-compare
4008 @opindex Wno-sign-compare
4009 @cindex warning for comparison of signed and unsigned values
4010 @cindex comparison of signed and unsigned values, warning
4011 @cindex signed and unsigned values, comparison warning
4012 Warn when a comparison between signed and unsigned values could produce
4013 an incorrect result when the signed value is converted to unsigned.
4014 This warning is also enabled by @option{-Wextra}; to get the other warnings
4015 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
4017 @item -Wsign-conversion
4018 @opindex Wsign-conversion
4019 @opindex Wno-sign-conversion
4020 Warn for implicit conversions that may change the sign of an integer
4021 value, like assigning a signed integer expression to an unsigned
4022 integer variable. An explicit cast silences the warning. In C, this
4023 option is enabled also by @option{-Wconversion}.
4027 @opindex Wno-address
4028 Warn about suspicious uses of memory addresses. These include using
4029 the address of a function in a conditional expression, such as
4030 @code{void func(void); if (func)}, and comparisons against the memory
4031 address of a string literal, such as @code{if (x == "abc")}. Such
4032 uses typically indicate a programmer error: the address of a function
4033 always evaluates to true, so their use in a conditional usually
4034 indicate that the programmer forgot the parentheses in a function
4035 call; and comparisons against string literals result in unspecified
4036 behavior and are not portable in C, so they usually indicate that the
4037 programmer intended to use @code{strcmp}. This warning is enabled by
4041 @opindex Wlogical-op
4042 @opindex Wno-logical-op
4043 Warn about suspicious uses of logical operators in expressions.
4044 This includes using logical operators in contexts where a
4045 bit-wise operator is likely to be expected.
4047 @item -Waggregate-return
4048 @opindex Waggregate-return
4049 @opindex Wno-aggregate-return
4050 Warn if any functions that return structures or unions are defined or
4051 called. (In languages where you can return an array, this also elicits
4054 @item -Wno-attributes
4055 @opindex Wno-attributes
4056 @opindex Wattributes
4057 Do not warn if an unexpected @code{__attribute__} is used, such as
4058 unrecognized attributes, function attributes applied to variables,
4059 etc. This will not stop errors for incorrect use of supported
4062 @item -Wno-builtin-macro-redefined
4063 @opindex Wno-builtin-macro-redefined
4064 @opindex Wbuiltin-macro-redefined
4065 Do not warn if certain built-in macros are redefined. This suppresses
4066 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4067 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4069 @item -Wstrict-prototypes @r{(C and Objective-C only)}
4070 @opindex Wstrict-prototypes
4071 @opindex Wno-strict-prototypes
4072 Warn if a function is declared or defined without specifying the
4073 argument types. (An old-style function definition is permitted without
4074 a warning if preceded by a declaration which specifies the argument
4077 @item -Wold-style-declaration @r{(C and Objective-C only)}
4078 @opindex Wold-style-declaration
4079 @opindex Wno-old-style-declaration
4080 Warn for obsolescent usages, according to the C Standard, in a
4081 declaration. For example, warn if storage-class specifiers like
4082 @code{static} are not the first things in a declaration. This warning
4083 is also enabled by @option{-Wextra}.
4085 @item -Wold-style-definition @r{(C and Objective-C only)}
4086 @opindex Wold-style-definition
4087 @opindex Wno-old-style-definition
4088 Warn if an old-style function definition is used. A warning is given
4089 even if there is a previous prototype.
4091 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
4092 @opindex Wmissing-parameter-type
4093 @opindex Wno-missing-parameter-type
4094 A function parameter is declared without a type specifier in K&R-style
4101 This warning is also enabled by @option{-Wextra}.
4103 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4104 @opindex Wmissing-prototypes
4105 @opindex Wno-missing-prototypes
4106 Warn if a global function is defined without a previous prototype
4107 declaration. This warning is issued even if the definition itself
4108 provides a prototype. The aim is to detect global functions that fail
4109 to be declared in header files.
4111 @item -Wmissing-declarations
4112 @opindex Wmissing-declarations
4113 @opindex Wno-missing-declarations
4114 Warn if a global function is defined without a previous declaration.
4115 Do so even if the definition itself provides a prototype.
4116 Use this option to detect global functions that are not declared in
4117 header files. In C++, no warnings are issued for function templates,
4118 or for inline functions, or for functions in anonymous namespaces.
4120 @item -Wmissing-field-initializers
4121 @opindex Wmissing-field-initializers
4122 @opindex Wno-missing-field-initializers
4126 Warn if a structure's initializer has some fields missing. For
4127 example, the following code would cause such a warning, because
4128 @code{x.h} is implicitly zero:
4131 struct s @{ int f, g, h; @};
4132 struct s x = @{ 3, 4 @};
4135 This option does not warn about designated initializers, so the following
4136 modification would not trigger a warning:
4139 struct s @{ int f, g, h; @};
4140 struct s x = @{ .f = 3, .g = 4 @};
4143 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4144 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4146 @item -Wmissing-noreturn
4147 @opindex Wmissing-noreturn
4148 @opindex Wno-missing-noreturn
4149 Warn about functions which might be candidates for attribute @code{noreturn}.
4150 Note these are only possible candidates, not absolute ones. Care should
4151 be taken to manually verify functions actually do not ever return before
4152 adding the @code{noreturn} attribute, otherwise subtle code generation
4153 bugs could be introduced. You will not get a warning for @code{main} in
4154 hosted C environments.
4156 @item -Wmissing-format-attribute
4157 @opindex Wmissing-format-attribute
4158 @opindex Wno-missing-format-attribute
4161 Warn about function pointers which might be candidates for @code{format}
4162 attributes. Note these are only possible candidates, not absolute ones.
4163 GCC will guess that function pointers with @code{format} attributes that
4164 are used in assignment, initialization, parameter passing or return
4165 statements should have a corresponding @code{format} attribute in the
4166 resulting type. I.e.@: the left-hand side of the assignment or
4167 initialization, the type of the parameter variable, or the return type
4168 of the containing function respectively should also have a @code{format}
4169 attribute to avoid the warning.
4171 GCC will also warn about function definitions which might be
4172 candidates for @code{format} attributes. Again, these are only
4173 possible candidates. GCC will guess that @code{format} attributes
4174 might be appropriate for any function that calls a function like
4175 @code{vprintf} or @code{vscanf}, but this might not always be the
4176 case, and some functions for which @code{format} attributes are
4177 appropriate may not be detected.
4179 @item -Wno-multichar
4180 @opindex Wno-multichar
4182 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4183 Usually they indicate a typo in the user's code, as they have
4184 implementation-defined values, and should not be used in portable code.
4186 @item -Wnormalized=<none|id|nfc|nfkc>
4187 @opindex Wnormalized=
4190 @cindex character set, input normalization
4191 In ISO C and ISO C++, two identifiers are different if they are
4192 different sequences of characters. However, sometimes when characters
4193 outside the basic ASCII character set are used, you can have two
4194 different character sequences that look the same. To avoid confusion,
4195 the ISO 10646 standard sets out some @dfn{normalization rules} which
4196 when applied ensure that two sequences that look the same are turned into
4197 the same sequence. GCC can warn you if you are using identifiers which
4198 have not been normalized; this option controls that warning.
4200 There are four levels of warning that GCC supports. The default is
4201 @option{-Wnormalized=nfc}, which warns about any identifier which is
4202 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4203 recommended form for most uses.
4205 Unfortunately, there are some characters which ISO C and ISO C++ allow
4206 in identifiers that when turned into NFC aren't allowable as
4207 identifiers. That is, there's no way to use these symbols in portable
4208 ISO C or C++ and have all your identifiers in NFC@.
4209 @option{-Wnormalized=id} suppresses the warning for these characters.
4210 It is hoped that future versions of the standards involved will correct
4211 this, which is why this option is not the default.
4213 You can switch the warning off for all characters by writing
4214 @option{-Wnormalized=none}. You would only want to do this if you
4215 were using some other normalization scheme (like ``D''), because
4216 otherwise you can easily create bugs that are literally impossible to see.
4218 Some characters in ISO 10646 have distinct meanings but look identical
4219 in some fonts or display methodologies, especially once formatting has
4220 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4221 LETTER N'', will display just like a regular @code{n} which has been
4222 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4223 normalization scheme to convert all these into a standard form as
4224 well, and GCC will warn if your code is not in NFKC if you use
4225 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4226 about every identifier that contains the letter O because it might be
4227 confused with the digit 0, and so is not the default, but may be
4228 useful as a local coding convention if the programming environment is
4229 unable to be fixed to display these characters distinctly.
4231 @item -Wno-deprecated
4232 @opindex Wno-deprecated
4233 @opindex Wdeprecated
4234 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4236 @item -Wno-deprecated-declarations
4237 @opindex Wno-deprecated-declarations
4238 @opindex Wdeprecated-declarations
4239 Do not warn about uses of functions (@pxref{Function Attributes}),
4240 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4241 Attributes}) marked as deprecated by using the @code{deprecated}
4245 @opindex Wno-overflow
4247 Do not warn about compile-time overflow in constant expressions.
4249 @item -Woverride-init @r{(C and Objective-C only)}
4250 @opindex Woverride-init
4251 @opindex Wno-override-init
4255 Warn if an initialized field without side effects is overridden when
4256 using designated initializers (@pxref{Designated Inits, , Designated
4259 This warning is included in @option{-Wextra}. To get other
4260 @option{-Wextra} warnings without this one, use @samp{-Wextra
4261 -Wno-override-init}.
4266 Warn if a structure is given the packed attribute, but the packed
4267 attribute has no effect on the layout or size of the structure.
4268 Such structures may be mis-aligned for little benefit. For
4269 instance, in this code, the variable @code{f.x} in @code{struct bar}
4270 will be misaligned even though @code{struct bar} does not itself
4271 have the packed attribute:
4278 @} __attribute__((packed));
4286 @item -Wpacked-bitfield-compat
4287 @opindex Wpacked-bitfield-compat
4288 @opindex Wno-packed-bitfield-compat
4289 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4290 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4291 the change can lead to differences in the structure layout. GCC
4292 informs you when the offset of such a field has changed in GCC 4.4.
4293 For example there is no longer a 4-bit padding between field @code{a}
4294 and @code{b} in this structure:
4301 @} __attribute__ ((packed));
4304 This warning is enabled by default. Use
4305 @option{-Wno-packed-bitfield-compat} to disable this warning.
4310 Warn if padding is included in a structure, either to align an element
4311 of the structure or to align the whole structure. Sometimes when this
4312 happens it is possible to rearrange the fields of the structure to
4313 reduce the padding and so make the structure smaller.
4315 @item -Wredundant-decls
4316 @opindex Wredundant-decls
4317 @opindex Wno-redundant-decls
4318 Warn if anything is declared more than once in the same scope, even in
4319 cases where multiple declaration is valid and changes nothing.
4321 @item -Wnested-externs @r{(C and Objective-C only)}
4322 @opindex Wnested-externs
4323 @opindex Wno-nested-externs
4324 Warn if an @code{extern} declaration is encountered within a function.
4329 Warn if a function can not be inlined and it was declared as inline.
4330 Even with this option, the compiler will not warn about failures to
4331 inline functions declared in system headers.
4333 The compiler uses a variety of heuristics to determine whether or not
4334 to inline a function. For example, the compiler takes into account
4335 the size of the function being inlined and the amount of inlining
4336 that has already been done in the current function. Therefore,
4337 seemingly insignificant changes in the source program can cause the
4338 warnings produced by @option{-Winline} to appear or disappear.
4340 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4341 @opindex Wno-invalid-offsetof
4342 @opindex Winvalid-offsetof
4343 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4344 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4345 to a non-POD type is undefined. In existing C++ implementations,
4346 however, @samp{offsetof} typically gives meaningful results even when
4347 applied to certain kinds of non-POD types. (Such as a simple
4348 @samp{struct} that fails to be a POD type only by virtue of having a
4349 constructor.) This flag is for users who are aware that they are
4350 writing nonportable code and who have deliberately chosen to ignore the
4353 The restrictions on @samp{offsetof} may be relaxed in a future version
4354 of the C++ standard.
4356 @item -Wno-int-to-pointer-cast
4357 @opindex Wno-int-to-pointer-cast
4358 @opindex Wint-to-pointer-cast
4359 Suppress warnings from casts to pointer type of an integer of a
4360 different size. In C++, casting to a pointer type of smaller size is
4361 an error. @option{Wint-to-pointer-cast} is enabled by default.
4364 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4365 @opindex Wno-pointer-to-int-cast
4366 @opindex Wpointer-to-int-cast
4367 Suppress warnings from casts from a pointer to an integer type of a
4371 @opindex Winvalid-pch
4372 @opindex Wno-invalid-pch
4373 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4374 the search path but can't be used.
4378 @opindex Wno-long-long
4379 Warn if @samp{long long} type is used. This is enabled by either
4380 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4381 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4383 @item -Wvariadic-macros
4384 @opindex Wvariadic-macros
4385 @opindex Wno-variadic-macros
4386 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4387 alternate syntax when in pedantic ISO C99 mode. This is default.
4388 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4393 Warn if variable length array is used in the code.
4394 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4395 the variable length array.
4397 @item -Wvolatile-register-var
4398 @opindex Wvolatile-register-var
4399 @opindex Wno-volatile-register-var
4400 Warn if a register variable is declared volatile. The volatile
4401 modifier does not inhibit all optimizations that may eliminate reads
4402 and/or writes to register variables. This warning is enabled by
4405 @item -Wdisabled-optimization
4406 @opindex Wdisabled-optimization
4407 @opindex Wno-disabled-optimization
4408 Warn if a requested optimization pass is disabled. This warning does
4409 not generally indicate that there is anything wrong with your code; it
4410 merely indicates that GCC's optimizers were unable to handle the code
4411 effectively. Often, the problem is that your code is too big or too
4412 complex; GCC will refuse to optimize programs when the optimization
4413 itself is likely to take inordinate amounts of time.
4415 @item -Wpointer-sign @r{(C and Objective-C only)}
4416 @opindex Wpointer-sign
4417 @opindex Wno-pointer-sign
4418 Warn for pointer argument passing or assignment with different signedness.
4419 This option is only supported for C and Objective-C@. It is implied by
4420 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4421 @option{-Wno-pointer-sign}.
4423 @item -Wstack-protector
4424 @opindex Wstack-protector
4425 @opindex Wno-stack-protector
4426 This option is only active when @option{-fstack-protector} is active. It
4427 warns about functions that will not be protected against stack smashing.
4430 @opindex Wno-mudflap
4431 Suppress warnings about constructs that cannot be instrumented by
4434 @item -Woverlength-strings
4435 @opindex Woverlength-strings
4436 @opindex Wno-overlength-strings
4437 Warn about string constants which are longer than the ``minimum
4438 maximum'' length specified in the C standard. Modern compilers
4439 generally allow string constants which are much longer than the
4440 standard's minimum limit, but very portable programs should avoid
4441 using longer strings.
4443 The limit applies @emph{after} string constant concatenation, and does
4444 not count the trailing NUL@. In C90, the limit was 509 characters; in
4445 C99, it was raised to 4095. C++98 does not specify a normative
4446 minimum maximum, so we do not diagnose overlength strings in C++@.
4448 This option is implied by @option{-pedantic}, and can be disabled with
4449 @option{-Wno-overlength-strings}.
4451 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4452 @opindex Wunsuffixed-float-constants
4454 GCC will issue a warning for any floating constant that does not have
4455 a suffix. When used together with @option{-Wsystem-headers} it will
4456 warn about such constants in system header files. This can be useful
4457 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4458 from the decimal floating-point extension to C99.
4461 @node Debugging Options
4462 @section Options for Debugging Your Program or GCC
4463 @cindex options, debugging
4464 @cindex debugging information options
4466 GCC has various special options that are used for debugging
4467 either your program or GCC:
4472 Produce debugging information in the operating system's native format
4473 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4476 On most systems that use stabs format, @option{-g} enables use of extra
4477 debugging information that only GDB can use; this extra information
4478 makes debugging work better in GDB but will probably make other debuggers
4480 refuse to read the program. If you want to control for certain whether
4481 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4482 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4484 GCC allows you to use @option{-g} with
4485 @option{-O}. The shortcuts taken by optimized code may occasionally
4486 produce surprising results: some variables you declared may not exist
4487 at all; flow of control may briefly move where you did not expect it;
4488 some statements may not be executed because they compute constant
4489 results or their values were already at hand; some statements may
4490 execute in different places because they were moved out of loops.
4492 Nevertheless it proves possible to debug optimized output. This makes
4493 it reasonable to use the optimizer for programs that might have bugs.
4495 The following options are useful when GCC is generated with the
4496 capability for more than one debugging format.
4500 Produce debugging information for use by GDB@. This means to use the
4501 most expressive format available (DWARF 2, stabs, or the native format
4502 if neither of those are supported), including GDB extensions if at all
4507 Produce debugging information in stabs format (if that is supported),
4508 without GDB extensions. This is the format used by DBX on most BSD
4509 systems. On MIPS, Alpha and System V Release 4 systems this option
4510 produces stabs debugging output which is not understood by DBX or SDB@.
4511 On System V Release 4 systems this option requires the GNU assembler.
4513 @item -feliminate-unused-debug-symbols
4514 @opindex feliminate-unused-debug-symbols
4515 Produce debugging information in stabs format (if that is supported),
4516 for only symbols that are actually used.
4518 @item -femit-class-debug-always
4519 Instead of emitting debugging information for a C++ class in only one
4520 object file, emit it in all object files using the class. This option
4521 should be used only with debuggers that are unable to handle the way GCC
4522 normally emits debugging information for classes because using this
4523 option will increase the size of debugging information by as much as a
4528 Produce debugging information in stabs format (if that is supported),
4529 using GNU extensions understood only by the GNU debugger (GDB)@. The
4530 use of these extensions is likely to make other debuggers crash or
4531 refuse to read the program.
4535 Produce debugging information in COFF format (if that is supported).
4536 This is the format used by SDB on most System V systems prior to
4541 Produce debugging information in XCOFF format (if that is supported).
4542 This is the format used by the DBX debugger on IBM RS/6000 systems.
4546 Produce debugging information in XCOFF format (if that is supported),
4547 using GNU extensions understood only by the GNU debugger (GDB)@. The
4548 use of these extensions is likely to make other debuggers crash or
4549 refuse to read the program, and may cause assemblers other than the GNU
4550 assembler (GAS) to fail with an error.
4552 @item -gdwarf-@var{version}
4553 @opindex gdwarf-@var{version}
4554 Produce debugging information in DWARF format (if that is
4555 supported). This is the format used by DBX on IRIX 6. The value
4556 of @var{version} may be either 2, 3 or 4; the default version is 2.
4558 Note that with DWARF version 2 some ports require, and will always
4559 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4561 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4562 for maximum benefit.
4564 @item -gstrict-dwarf
4565 @opindex gstrict-dwarf
4566 Disallow using extensions of later DWARF standard version than selected
4567 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4568 DWARF extensions from later standard versions is allowed.
4570 @item -gno-strict-dwarf
4571 @opindex gno-strict-dwarf
4572 Allow using extensions of later DWARF standard version than selected with
4573 @option{-gdwarf-@var{version}}.
4577 Produce debugging information in VMS debug format (if that is
4578 supported). This is the format used by DEBUG on VMS systems.
4581 @itemx -ggdb@var{level}
4582 @itemx -gstabs@var{level}
4583 @itemx -gcoff@var{level}
4584 @itemx -gxcoff@var{level}
4585 @itemx -gvms@var{level}
4586 Request debugging information and also use @var{level} to specify how
4587 much information. The default level is 2.
4589 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4592 Level 1 produces minimal information, enough for making backtraces in
4593 parts of the program that you don't plan to debug. This includes
4594 descriptions of functions and external variables, but no information
4595 about local variables and no line numbers.
4597 Level 3 includes extra information, such as all the macro definitions
4598 present in the program. Some debuggers support macro expansion when
4599 you use @option{-g3}.
4601 @option{-gdwarf-2} does not accept a concatenated debug level, because
4602 GCC used to support an option @option{-gdwarf} that meant to generate
4603 debug information in version 1 of the DWARF format (which is very
4604 different from version 2), and it would have been too confusing. That
4605 debug format is long obsolete, but the option cannot be changed now.
4606 Instead use an additional @option{-g@var{level}} option to change the
4607 debug level for DWARF.
4611 Turn off generation of debug info, if leaving out this option would have
4612 generated it, or turn it on at level 2 otherwise. The position of this
4613 argument in the command line does not matter, it takes effect after all
4614 other options are processed, and it does so only once, no matter how
4615 many times it is given. This is mainly intended to be used with
4616 @option{-fcompare-debug}.
4618 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4619 @opindex fdump-final-insns
4620 Dump the final internal representation (RTL) to @var{file}. If the
4621 optional argument is omitted (or if @var{file} is @code{.}), the name
4622 of the dump file will be determined by appending @code{.gkd} to the
4623 compilation output file name.
4625 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4626 @opindex fcompare-debug
4627 @opindex fno-compare-debug
4628 If no error occurs during compilation, run the compiler a second time,
4629 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4630 passed to the second compilation. Dump the final internal
4631 representation in both compilations, and print an error if they differ.
4633 If the equal sign is omitted, the default @option{-gtoggle} is used.
4635 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4636 and nonzero, implicitly enables @option{-fcompare-debug}. If
4637 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4638 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4641 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4642 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4643 of the final representation and the second compilation, preventing even
4644 @env{GCC_COMPARE_DEBUG} from taking effect.
4646 To verify full coverage during @option{-fcompare-debug} testing, set
4647 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4648 which GCC will reject as an invalid option in any actual compilation
4649 (rather than preprocessing, assembly or linking). To get just a
4650 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4651 not overridden} will do.
4653 @item -fcompare-debug-second
4654 @opindex fcompare-debug-second
4655 This option is implicitly passed to the compiler for the second
4656 compilation requested by @option{-fcompare-debug}, along with options to
4657 silence warnings, and omitting other options that would cause
4658 side-effect compiler outputs to files or to the standard output. Dump
4659 files and preserved temporary files are renamed so as to contain the
4660 @code{.gk} additional extension during the second compilation, to avoid
4661 overwriting those generated by the first.
4663 When this option is passed to the compiler driver, it causes the
4664 @emph{first} compilation to be skipped, which makes it useful for little
4665 other than debugging the compiler proper.
4667 @item -feliminate-dwarf2-dups
4668 @opindex feliminate-dwarf2-dups
4669 Compress DWARF2 debugging information by eliminating duplicated
4670 information about each symbol. This option only makes sense when
4671 generating DWARF2 debugging information with @option{-gdwarf-2}.
4673 @item -femit-struct-debug-baseonly
4674 Emit debug information for struct-like types
4675 only when the base name of the compilation source file
4676 matches the base name of file in which the struct was defined.
4678 This option substantially reduces the size of debugging information,
4679 but at significant potential loss in type information to the debugger.
4680 See @option{-femit-struct-debug-reduced} for a less 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-reduced
4686 Emit debug information for struct-like types
4687 only when the base name of the compilation source file
4688 matches the base name of file in which the type was defined,
4689 unless the struct is a template or defined in a system header.
4691 This option significantly reduces the size of debugging information,
4692 with some potential loss in type information to the debugger.
4693 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4694 See @option{-femit-struct-debug-detailed} for more detailed control.
4696 This option works only with DWARF 2.
4698 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4699 Specify the struct-like types
4700 for which the compiler will generate debug information.
4701 The intent is to reduce duplicate struct debug information
4702 between different object files within the same program.
4704 This option is a detailed version of
4705 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4706 which will serve for most needs.
4708 A specification has the syntax
4709 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4711 The optional first word limits the specification to
4712 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4713 A struct type is used directly when it is the type of a variable, member.
4714 Indirect uses arise through pointers to structs.
4715 That is, when use of an incomplete struct would be legal, the use is indirect.
4717 @samp{struct one direct; struct two * indirect;}.
4719 The optional second word limits the specification to
4720 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4721 Generic structs are a bit complicated to explain.
4722 For C++, these are non-explicit specializations of template classes,
4723 or non-template classes within the above.
4724 Other programming languages have generics,
4725 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4727 The third word specifies the source files for those
4728 structs for which the compiler will emit debug information.
4729 The values @samp{none} and @samp{any} have the normal meaning.
4730 The value @samp{base} means that
4731 the base of name of the file in which the type declaration appears
4732 must match the base of the name of the main compilation file.
4733 In practice, this means that
4734 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4735 but types declared in other header will not.
4736 The value @samp{sys} means those types satisfying @samp{base}
4737 or declared in system or compiler headers.
4739 You may need to experiment to determine the best settings for your application.
4741 The default is @samp{-femit-struct-debug-detailed=all}.
4743 This option works only with DWARF 2.
4745 @item -fenable-icf-debug
4746 @opindex fenable-icf-debug
4747 Generate additional debug information to support identical code folding (ICF).
4748 This option only works with DWARF version 2 or higher.
4750 @item -fno-merge-debug-strings
4751 @opindex fmerge-debug-strings
4752 @opindex fno-merge-debug-strings
4753 Direct the linker to not merge together strings in the debugging
4754 information which are identical in different object files. Merging is
4755 not supported by all assemblers or linkers. Merging decreases the size
4756 of the debug information in the output file at the cost of increasing
4757 link processing time. Merging is enabled by default.
4759 @item -fdebug-prefix-map=@var{old}=@var{new}
4760 @opindex fdebug-prefix-map
4761 When compiling files in directory @file{@var{old}}, record debugging
4762 information describing them as in @file{@var{new}} instead.
4764 @item -fno-dwarf2-cfi-asm
4765 @opindex fdwarf2-cfi-asm
4766 @opindex fno-dwarf2-cfi-asm
4767 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4768 instead of using GAS @code{.cfi_*} directives.
4770 @cindex @command{prof}
4773 Generate extra code to write profile information suitable for the
4774 analysis program @command{prof}. You must use this option when compiling
4775 the source files you want data about, and you must also use it when
4778 @cindex @command{gprof}
4781 Generate extra code to write profile information suitable for the
4782 analysis program @command{gprof}. You must use this option when compiling
4783 the source files you want data about, and you must also use it when
4788 Makes the compiler print out each function name as it is compiled, and
4789 print some statistics about each pass when it finishes.
4792 @opindex ftime-report
4793 Makes the compiler print some statistics about the time consumed by each
4794 pass when it finishes.
4797 @opindex fmem-report
4798 Makes the compiler print some statistics about permanent memory
4799 allocation when it finishes.
4801 @item -fpre-ipa-mem-report
4802 @opindex fpre-ipa-mem-report
4803 @item -fpost-ipa-mem-report
4804 @opindex fpost-ipa-mem-report
4805 Makes the compiler print some statistics about permanent memory
4806 allocation before or after interprocedural optimization.
4808 @item -fprofile-arcs
4809 @opindex fprofile-arcs
4810 Add code so that program flow @dfn{arcs} are instrumented. During
4811 execution the program records how many times each branch and call is
4812 executed and how many times it is taken or returns. When the compiled
4813 program exits it saves this data to a file called
4814 @file{@var{auxname}.gcda} for each source file. The data may be used for
4815 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4816 test coverage analysis (@option{-ftest-coverage}). Each object file's
4817 @var{auxname} is generated from the name of the output file, if
4818 explicitly specified and it is not the final executable, otherwise it is
4819 the basename of the source file. In both cases any suffix is removed
4820 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4821 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4822 @xref{Cross-profiling}.
4824 @cindex @command{gcov}
4828 This option is used to compile and link code instrumented for coverage
4829 analysis. The option is a synonym for @option{-fprofile-arcs}
4830 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4831 linking). See the documentation for those options for more details.
4836 Compile the source files with @option{-fprofile-arcs} plus optimization
4837 and code generation options. For test coverage analysis, use the
4838 additional @option{-ftest-coverage} option. You do not need to profile
4839 every source file in a program.
4842 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4843 (the latter implies the former).
4846 Run the program on a representative workload to generate the arc profile
4847 information. This may be repeated any number of times. You can run
4848 concurrent instances of your program, and provided that the file system
4849 supports locking, the data files will be correctly updated. Also
4850 @code{fork} calls are detected and correctly handled (double counting
4854 For profile-directed optimizations, compile the source files again with
4855 the same optimization and code generation options plus
4856 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4857 Control Optimization}).
4860 For test coverage analysis, use @command{gcov} to produce human readable
4861 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4862 @command{gcov} documentation for further information.
4866 With @option{-fprofile-arcs}, for each function of your program GCC
4867 creates a program flow graph, then finds a spanning tree for the graph.
4868 Only arcs that are not on the spanning tree have to be instrumented: the
4869 compiler adds code to count the number of times that these arcs are
4870 executed. When an arc is the only exit or only entrance to a block, the
4871 instrumentation code can be added to the block; otherwise, a new basic
4872 block must be created to hold the instrumentation code.
4875 @item -ftest-coverage
4876 @opindex ftest-coverage
4877 Produce a notes file that the @command{gcov} code-coverage utility
4878 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4879 show program coverage. Each source file's note file is called
4880 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4881 above for a description of @var{auxname} and instructions on how to
4882 generate test coverage data. Coverage data will match the source files
4883 more closely, if you do not optimize.
4885 @item -fdbg-cnt-list
4886 @opindex fdbg-cnt-list
4887 Print the name and the counter upperbound for all debug counters.
4889 @item -fdbg-cnt=@var{counter-value-list}
4891 Set the internal debug counter upperbound. @var{counter-value-list}
4892 is a comma-separated list of @var{name}:@var{value} pairs
4893 which sets the upperbound of each debug counter @var{name} to @var{value}.
4894 All debug counters have the initial upperbound of @var{UINT_MAX},
4895 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4896 e.g. With -fdbg-cnt=dce:10,tail_call:0
4897 dbg_cnt(dce) will return true only for first 10 invocations
4898 and dbg_cnt(tail_call) will return false always.
4900 @item -d@var{letters}
4901 @itemx -fdump-rtl-@var{pass}
4903 Says to make debugging dumps during compilation at times specified by
4904 @var{letters}. This is used for debugging the RTL-based passes of the
4905 compiler. The file names for most of the dumps are made by appending
4906 a pass number and a word to the @var{dumpname}, and the files are
4907 created in the directory of the output file. @var{dumpname} is
4908 generated from the name of the output file, if explicitly specified
4909 and it is not an executable, otherwise it is the basename of the
4910 source file. These switches may have different effects when
4911 @option{-E} is used for preprocessing.
4913 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4914 @option{-d} option @var{letters}. Here are the possible
4915 letters for use in @var{pass} and @var{letters}, and their meanings:
4919 @item -fdump-rtl-alignments
4920 @opindex fdump-rtl-alignments
4921 Dump after branch alignments have been computed.
4923 @item -fdump-rtl-asmcons
4924 @opindex fdump-rtl-asmcons
4925 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4927 @item -fdump-rtl-auto_inc_dec
4928 @opindex fdump-rtl-auto_inc_dec
4929 Dump after auto-inc-dec discovery. This pass is only run on
4930 architectures that have auto inc or auto dec instructions.
4932 @item -fdump-rtl-barriers
4933 @opindex fdump-rtl-barriers
4934 Dump after cleaning up the barrier instructions.
4936 @item -fdump-rtl-bbpart
4937 @opindex fdump-rtl-bbpart
4938 Dump after partitioning hot and cold basic blocks.
4940 @item -fdump-rtl-bbro
4941 @opindex fdump-rtl-bbro
4942 Dump after block reordering.
4944 @item -fdump-rtl-btl1
4945 @itemx -fdump-rtl-btl2
4946 @opindex fdump-rtl-btl2
4947 @opindex fdump-rtl-btl2
4948 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4949 after the two branch
4950 target load optimization passes.
4952 @item -fdump-rtl-bypass
4953 @opindex fdump-rtl-bypass
4954 Dump after jump bypassing and control flow optimizations.
4956 @item -fdump-rtl-combine
4957 @opindex fdump-rtl-combine
4958 Dump after the RTL instruction combination pass.
4960 @item -fdump-rtl-compgotos
4961 @opindex fdump-rtl-compgotos
4962 Dump after duplicating the computed gotos.
4964 @item -fdump-rtl-ce1
4965 @itemx -fdump-rtl-ce2
4966 @itemx -fdump-rtl-ce3
4967 @opindex fdump-rtl-ce1
4968 @opindex fdump-rtl-ce2
4969 @opindex fdump-rtl-ce3
4970 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4971 @option{-fdump-rtl-ce3} enable dumping after the three
4972 if conversion passes.
4974 @itemx -fdump-rtl-cprop_hardreg
4975 @opindex fdump-rtl-cprop_hardreg
4976 Dump after hard register copy propagation.
4978 @itemx -fdump-rtl-csa
4979 @opindex fdump-rtl-csa
4980 Dump after combining stack adjustments.
4982 @item -fdump-rtl-cse1
4983 @itemx -fdump-rtl-cse2
4984 @opindex fdump-rtl-cse1
4985 @opindex fdump-rtl-cse2
4986 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4987 the two common sub-expression elimination passes.
4989 @itemx -fdump-rtl-dce
4990 @opindex fdump-rtl-dce
4991 Dump after the standalone dead code elimination passes.
4993 @itemx -fdump-rtl-dbr
4994 @opindex fdump-rtl-dbr
4995 Dump after delayed branch scheduling.
4997 @item -fdump-rtl-dce1
4998 @itemx -fdump-rtl-dce2
4999 @opindex fdump-rtl-dce1
5000 @opindex fdump-rtl-dce2
5001 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
5002 the two dead store elimination passes.
5005 @opindex fdump-rtl-eh
5006 Dump after finalization of EH handling code.
5008 @item -fdump-rtl-eh_ranges
5009 @opindex fdump-rtl-eh_ranges
5010 Dump after conversion of EH handling range regions.
5012 @item -fdump-rtl-expand
5013 @opindex fdump-rtl-expand
5014 Dump after RTL generation.
5016 @item -fdump-rtl-fwprop1
5017 @itemx -fdump-rtl-fwprop2
5018 @opindex fdump-rtl-fwprop1
5019 @opindex fdump-rtl-fwprop2
5020 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
5021 dumping after the two forward propagation passes.
5023 @item -fdump-rtl-gcse1
5024 @itemx -fdump-rtl-gcse2
5025 @opindex fdump-rtl-gcse1
5026 @opindex fdump-rtl-gcse2
5027 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
5028 after global common subexpression elimination.
5030 @item -fdump-rtl-init-regs
5031 @opindex fdump-rtl-init-regs
5032 Dump after the initialization of the registers.
5034 @item -fdump-rtl-initvals
5035 @opindex fdump-rtl-initvals
5036 Dump after the computation of the initial value sets.
5038 @itemx -fdump-rtl-into_cfglayout
5039 @opindex fdump-rtl-into_cfglayout
5040 Dump after converting to cfglayout mode.
5042 @item -fdump-rtl-ira
5043 @opindex fdump-rtl-ira
5044 Dump after iterated register allocation.
5046 @item -fdump-rtl-jump
5047 @opindex fdump-rtl-jump
5048 Dump after the second jump optimization.
5050 @item -fdump-rtl-loop2
5051 @opindex fdump-rtl-loop2
5052 @option{-fdump-rtl-loop2} enables dumping after the rtl
5053 loop optimization passes.
5055 @item -fdump-rtl-mach
5056 @opindex fdump-rtl-mach
5057 Dump after performing the machine dependent reorganization pass, if that
5060 @item -fdump-rtl-mode_sw
5061 @opindex fdump-rtl-mode_sw
5062 Dump after removing redundant mode switches.
5064 @item -fdump-rtl-rnreg
5065 @opindex fdump-rtl-rnreg
5066 Dump after register renumbering.
5068 @itemx -fdump-rtl-outof_cfglayout
5069 @opindex fdump-rtl-outof_cfglayout
5070 Dump after converting from cfglayout mode.
5072 @item -fdump-rtl-peephole2
5073 @opindex fdump-rtl-peephole2
5074 Dump after the peephole pass.
5076 @item -fdump-rtl-postreload
5077 @opindex fdump-rtl-postreload
5078 Dump after post-reload optimizations.
5080 @itemx -fdump-rtl-pro_and_epilogue
5081 @opindex fdump-rtl-pro_and_epilogue
5082 Dump after generating the function pro and epilogues.
5084 @item -fdump-rtl-regmove
5085 @opindex fdump-rtl-regmove
5086 Dump after the register move pass.
5088 @item -fdump-rtl-sched1
5089 @itemx -fdump-rtl-sched2
5090 @opindex fdump-rtl-sched1
5091 @opindex fdump-rtl-sched2
5092 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
5093 after the basic block scheduling passes.
5095 @item -fdump-rtl-see
5096 @opindex fdump-rtl-see
5097 Dump after sign extension elimination.
5099 @item -fdump-rtl-seqabstr
5100 @opindex fdump-rtl-seqabstr
5101 Dump after common sequence discovery.
5103 @item -fdump-rtl-shorten
5104 @opindex fdump-rtl-shorten
5105 Dump after shortening branches.
5107 @item -fdump-rtl-sibling
5108 @opindex fdump-rtl-sibling
5109 Dump after sibling call optimizations.
5111 @item -fdump-rtl-split1
5112 @itemx -fdump-rtl-split2
5113 @itemx -fdump-rtl-split3
5114 @itemx -fdump-rtl-split4
5115 @itemx -fdump-rtl-split5
5116 @opindex fdump-rtl-split1
5117 @opindex fdump-rtl-split2
5118 @opindex fdump-rtl-split3
5119 @opindex fdump-rtl-split4
5120 @opindex fdump-rtl-split5
5121 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5122 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5123 @option{-fdump-rtl-split5} enable dumping after five rounds of
5124 instruction splitting.
5126 @item -fdump-rtl-sms
5127 @opindex fdump-rtl-sms
5128 Dump after modulo scheduling. This pass is only run on some
5131 @item -fdump-rtl-stack
5132 @opindex fdump-rtl-stack
5133 Dump after conversion from GCC's "flat register file" registers to the
5134 x87's stack-like registers. This pass is only run on x86 variants.
5136 @item -fdump-rtl-subreg1
5137 @itemx -fdump-rtl-subreg2
5138 @opindex fdump-rtl-subreg1
5139 @opindex fdump-rtl-subreg2
5140 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5141 the two subreg expansion passes.
5143 @item -fdump-rtl-unshare
5144 @opindex fdump-rtl-unshare
5145 Dump after all rtl has been unshared.
5147 @item -fdump-rtl-vartrack
5148 @opindex fdump-rtl-vartrack
5149 Dump after variable tracking.
5151 @item -fdump-rtl-vregs
5152 @opindex fdump-rtl-vregs
5153 Dump after converting virtual registers to hard registers.
5155 @item -fdump-rtl-web
5156 @opindex fdump-rtl-web
5157 Dump after live range splitting.
5159 @item -fdump-rtl-regclass
5160 @itemx -fdump-rtl-subregs_of_mode_init
5161 @itemx -fdump-rtl-subregs_of_mode_finish
5162 @itemx -fdump-rtl-dfinit
5163 @itemx -fdump-rtl-dfinish
5164 @opindex fdump-rtl-regclass
5165 @opindex fdump-rtl-subregs_of_mode_init
5166 @opindex fdump-rtl-subregs_of_mode_finish
5167 @opindex fdump-rtl-dfinit
5168 @opindex fdump-rtl-dfinish
5169 These dumps are defined but always produce empty files.
5171 @item -fdump-rtl-all
5172 @opindex fdump-rtl-all
5173 Produce all the dumps listed above.
5177 Annotate the assembler output with miscellaneous debugging information.
5181 Dump all macro definitions, at the end of preprocessing, in addition to
5186 Produce a core dump whenever an error occurs.
5190 Print statistics on memory usage, at the end of the run, to
5195 Annotate the assembler output with a comment indicating which
5196 pattern and alternative was used. The length of each instruction is
5201 Dump the RTL in the assembler output as a comment before each instruction.
5202 Also turns on @option{-dp} annotation.
5206 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5207 dump a representation of the control flow graph suitable for viewing with VCG
5208 to @file{@var{file}.@var{pass}.vcg}.
5212 Just generate RTL for a function instead of compiling it. Usually used
5213 with @option{-fdump-rtl-expand}.
5217 Dump debugging information during parsing, to standard error.
5221 @opindex fdump-noaddr
5222 When doing debugging dumps, suppress address output. This makes it more
5223 feasible to use diff on debugging dumps for compiler invocations with
5224 different compiler binaries and/or different
5225 text / bss / data / heap / stack / dso start locations.
5227 @item -fdump-unnumbered
5228 @opindex fdump-unnumbered
5229 When doing debugging dumps, suppress instruction numbers and address output.
5230 This makes it more feasible to use diff on debugging dumps for compiler
5231 invocations with different options, in particular with and without
5234 @item -fdump-unnumbered-links
5235 @opindex fdump-unnumbered-links
5236 When doing debugging dumps (see @option{-d} option above), suppress
5237 instruction numbers for the links to the previous and next instructions
5240 @item -fdump-translation-unit @r{(C++ only)}
5241 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5242 @opindex fdump-translation-unit
5243 Dump a representation of the tree structure for the entire translation
5244 unit to a file. The file name is made by appending @file{.tu} to the
5245 source file name, and the file is created in the same directory as the
5246 output file. If the @samp{-@var{options}} form is used, @var{options}
5247 controls the details of the dump as described for the
5248 @option{-fdump-tree} options.
5250 @item -fdump-class-hierarchy @r{(C++ only)}
5251 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5252 @opindex fdump-class-hierarchy
5253 Dump a representation of each class's hierarchy and virtual function
5254 table layout to a file. The file name is made by appending
5255 @file{.class} to the source file name, and the file is created in the
5256 same directory as the output file. If the @samp{-@var{options}} form
5257 is used, @var{options} controls the details of the dump as described
5258 for the @option{-fdump-tree} options.
5260 @item -fdump-ipa-@var{switch}
5262 Control the dumping at various stages of inter-procedural analysis
5263 language tree to a file. The file name is generated by appending a
5264 switch specific suffix to the source file name, and the file is created
5265 in the same directory as the output file. The following dumps are
5270 Enables all inter-procedural analysis dumps.
5273 Dumps information about call-graph optimization, unused function removal,
5274 and inlining decisions.
5277 Dump after function inlining.
5281 @item -fdump-statistics-@var{option}
5282 @opindex fdump-statistics
5283 Enable and control dumping of pass statistics in a separate file. The
5284 file name is generated by appending a suffix ending in
5285 @samp{.statistics} to the source file name, and the file is created in
5286 the same directory as the output file. If the @samp{-@var{option}}
5287 form is used, @samp{-stats} will cause counters to be summed over the
5288 whole compilation unit while @samp{-details} will dump every event as
5289 the passes generate them. The default with no option is to sum
5290 counters for each function compiled.
5292 @item -fdump-tree-@var{switch}
5293 @itemx -fdump-tree-@var{switch}-@var{options}
5295 Control the dumping at various stages of processing the intermediate
5296 language tree to a file. The file name is generated by appending a
5297 switch specific suffix to the source file name, and the file is
5298 created in the same directory as the output file. If the
5299 @samp{-@var{options}} form is used, @var{options} is a list of
5300 @samp{-} separated options that control the details of the dump. Not
5301 all options are applicable to all dumps, those which are not
5302 meaningful will be ignored. The following options are available
5306 Print the address of each node. Usually this is not meaningful as it
5307 changes according to the environment and source file. Its primary use
5308 is for tying up a dump file with a debug environment.
5310 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5311 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5312 use working backward from mangled names in the assembly file.
5314 Inhibit dumping of members of a scope or body of a function merely
5315 because that scope has been reached. Only dump such items when they
5316 are directly reachable by some other path. When dumping pretty-printed
5317 trees, this option inhibits dumping the bodies of control structures.
5319 Print a raw representation of the tree. By default, trees are
5320 pretty-printed into a C-like representation.
5322 Enable more detailed dumps (not honored by every dump option).
5324 Enable dumping various statistics about the pass (not honored by every dump
5327 Enable showing basic block boundaries (disabled in raw dumps).
5329 Enable showing virtual operands for every statement.
5331 Enable showing line numbers for statements.
5333 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5335 Enable showing the tree dump for each statement.
5337 Enable showing the EH region number holding each statement.
5339 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5340 and @option{lineno}.
5343 The following tree dumps are possible:
5347 @opindex fdump-tree-original
5348 Dump before any tree based optimization, to @file{@var{file}.original}.
5351 @opindex fdump-tree-optimized
5352 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5355 @opindex fdump-tree-gimple
5356 Dump each function before and after the gimplification pass to a file. The
5357 file name is made by appending @file{.gimple} to the source file name.
5360 @opindex fdump-tree-cfg
5361 Dump the control flow graph of each function to a file. The file name is
5362 made by appending @file{.cfg} to the source file name.
5365 @opindex fdump-tree-vcg
5366 Dump the control flow graph of each function to a file in VCG format. The
5367 file name is made by appending @file{.vcg} to the source file name. Note
5368 that if the file contains more than one function, the generated file cannot
5369 be used directly by VCG@. You will need to cut and paste each function's
5370 graph into its own separate file first.
5373 @opindex fdump-tree-ch
5374 Dump each function after copying loop headers. The file name is made by
5375 appending @file{.ch} to the source file name.
5378 @opindex fdump-tree-ssa
5379 Dump SSA related information to a file. The file name is made by appending
5380 @file{.ssa} to the source file name.
5383 @opindex fdump-tree-alias
5384 Dump aliasing information for each function. The file name is made by
5385 appending @file{.alias} to the source file name.
5388 @opindex fdump-tree-ccp
5389 Dump each function after CCP@. The file name is made by appending
5390 @file{.ccp} to the source file name.
5393 @opindex fdump-tree-storeccp
5394 Dump each function after STORE-CCP@. The file name is made by appending
5395 @file{.storeccp} to the source file name.
5398 @opindex fdump-tree-pre
5399 Dump trees after partial redundancy elimination. The file name is made
5400 by appending @file{.pre} to the source file name.
5403 @opindex fdump-tree-fre
5404 Dump trees after full redundancy elimination. The file name is made
5405 by appending @file{.fre} to the source file name.
5408 @opindex fdump-tree-copyprop
5409 Dump trees after copy propagation. The file name is made
5410 by appending @file{.copyprop} to the source file name.
5412 @item store_copyprop
5413 @opindex fdump-tree-store_copyprop
5414 Dump trees after store copy-propagation. The file name is made
5415 by appending @file{.store_copyprop} to the source file name.
5418 @opindex fdump-tree-dce
5419 Dump each function after dead code elimination. The file name is made by
5420 appending @file{.dce} to the source file name.
5423 @opindex fdump-tree-mudflap
5424 Dump each function after adding mudflap instrumentation. The file name is
5425 made by appending @file{.mudflap} to the source file name.
5428 @opindex fdump-tree-sra
5429 Dump each function after performing scalar replacement of aggregates. The
5430 file name is made by appending @file{.sra} to the source file name.
5433 @opindex fdump-tree-sink
5434 Dump each function after performing code sinking. The file name is made
5435 by appending @file{.sink} to the source file name.
5438 @opindex fdump-tree-dom
5439 Dump each function after applying dominator tree optimizations. The file
5440 name is made by appending @file{.dom} to the source file name.
5443 @opindex fdump-tree-dse
5444 Dump each function after applying dead store elimination. The file
5445 name is made by appending @file{.dse} to the source file name.
5448 @opindex fdump-tree-phiopt
5449 Dump each function after optimizing PHI nodes into straightline code. The file
5450 name is made by appending @file{.phiopt} to the source file name.
5453 @opindex fdump-tree-forwprop
5454 Dump each function after forward propagating single use variables. The file
5455 name is made by appending @file{.forwprop} to the source file name.
5458 @opindex fdump-tree-copyrename
5459 Dump each function after applying the copy rename optimization. The file
5460 name is made by appending @file{.copyrename} to the source file name.
5463 @opindex fdump-tree-nrv
5464 Dump each function after applying the named return value optimization on
5465 generic trees. The file name is made by appending @file{.nrv} to the source
5469 @opindex fdump-tree-vect
5470 Dump each function after applying vectorization of loops. The file name is
5471 made by appending @file{.vect} to the source file name.
5474 @opindex fdump-tree-slp
5475 Dump each function after applying vectorization of basic blocks. The file name
5476 is made by appending @file{.slp} to the source file name.
5479 @opindex fdump-tree-vrp
5480 Dump each function after Value Range Propagation (VRP). The file name
5481 is made by appending @file{.vrp} to the source file name.
5484 @opindex fdump-tree-all
5485 Enable all the available tree dumps with the flags provided in this option.
5488 @item -ftree-vectorizer-verbose=@var{n}
5489 @opindex ftree-vectorizer-verbose
5490 This option controls the amount of debugging output the vectorizer prints.
5491 This information is written to standard error, unless
5492 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5493 in which case it is output to the usual dump listing file, @file{.vect}.
5494 For @var{n}=0 no diagnostic information is reported.
5495 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5496 and the total number of loops that got vectorized.
5497 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5498 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5499 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5500 level that @option{-fdump-tree-vect-stats} uses.
5501 Higher verbosity levels mean either more information dumped for each
5502 reported loop, or same amount of information reported for more loops:
5503 if @var{n}=3, vectorizer cost model information is reported.
5504 If @var{n}=4, alignment related information is added to the reports.
5505 If @var{n}=5, data-references related information (e.g.@: memory dependences,
5506 memory access-patterns) is added to the reports.
5507 If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5508 that did not pass the first analysis phase (i.e., may not be countable, or
5509 may have complicated control-flow).
5510 If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5511 If @var{n}=8, SLP related information is added to the reports.
5512 For @var{n}=9, all the information the vectorizer generates during its
5513 analysis and transformation is reported. This is the same verbosity level
5514 that @option{-fdump-tree-vect-details} uses.
5516 @item -frandom-seed=@var{string}
5517 @opindex frandom-seed
5518 This option provides a seed that GCC uses when it would otherwise use
5519 random numbers. It is used to generate certain symbol names
5520 that have to be different in every compiled file. It is also used to
5521 place unique stamps in coverage data files and the object files that
5522 produce them. You can use the @option{-frandom-seed} option to produce
5523 reproducibly identical object files.
5525 The @var{string} should be different for every file you compile.
5527 @item -fsched-verbose=@var{n}
5528 @opindex fsched-verbose
5529 On targets that use instruction scheduling, this option controls the
5530 amount of debugging output the scheduler prints. This information is
5531 written to standard error, unless @option{-fdump-rtl-sched1} or
5532 @option{-fdump-rtl-sched2} is specified, in which case it is output
5533 to the usual dump listing file, @file{.sched1} or @file{.sched2}
5534 respectively. However for @var{n} greater than nine, the output is
5535 always printed to standard error.
5537 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5538 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5539 For @var{n} greater than one, it also output basic block probabilities,
5540 detailed ready list information and unit/insn info. For @var{n} greater
5541 than two, it includes RTL at abort point, control-flow and regions info.
5542 And for @var{n} over four, @option{-fsched-verbose} also includes
5546 @itemx -save-temps=cwd
5548 Store the usual ``temporary'' intermediate files permanently; place them
5549 in the current directory and name them based on the source file. Thus,
5550 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5551 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5552 preprocessed @file{foo.i} output file even though the compiler now
5553 normally uses an integrated preprocessor.
5555 When used in combination with the @option{-x} command line option,
5556 @option{-save-temps} is sensible enough to avoid over writing an
5557 input source file with the same extension as an intermediate file.
5558 The corresponding intermediate file may be obtained by renaming the
5559 source file before using @option{-save-temps}.
5561 If you invoke GCC in parallel, compiling several different source
5562 files that share a common base name in different subdirectories or the
5563 same source file compiled for multiple output destinations, it is
5564 likely that the different parallel compilers will interfere with each
5565 other, and overwrite the temporary files. For instance:
5568 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5569 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5572 may result in @file{foo.i} and @file{foo.o} being written to
5573 simultaneously by both compilers.
5575 @item -save-temps=obj
5576 @opindex save-temps=obj
5577 Store the usual ``temporary'' intermediate files permanently. If the
5578 @option{-o} option is used, the temporary files are based on the
5579 object file. If the @option{-o} option is not used, the
5580 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5585 gcc -save-temps=obj -c foo.c
5586 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5587 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5590 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5591 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5592 @file{dir2/yfoobar.o}.
5594 @item -time@r{[}=@var{file}@r{]}
5596 Report the CPU time taken by each subprocess in the compilation
5597 sequence. For C source files, this is the compiler proper and assembler
5598 (plus the linker if linking is done).
5600 Without the specification of an output file, the output looks like this:
5607 The first number on each line is the ``user time'', that is time spent
5608 executing the program itself. The second number is ``system time'',
5609 time spent executing operating system routines on behalf of the program.
5610 Both numbers are in seconds.
5612 With the specification of an output file, the output is appended to the
5613 named file, and it looks like this:
5616 0.12 0.01 cc1 @var{options}
5617 0.00 0.01 as @var{options}
5620 The ``user time'' and the ``system time'' are moved before the program
5621 name, and the options passed to the program are displayed, so that one
5622 can later tell what file was being compiled, and with which options.
5624 @item -fvar-tracking
5625 @opindex fvar-tracking
5626 Run variable tracking pass. It computes where variables are stored at each
5627 position in code. Better debugging information is then generated
5628 (if the debugging information format supports this information).
5630 It is enabled by default when compiling with optimization (@option{-Os},
5631 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5632 the debug info format supports it.
5634 @item -fvar-tracking-assignments
5635 @opindex fvar-tracking-assignments
5636 @opindex fno-var-tracking-assignments
5637 Annotate assignments to user variables early in the compilation and
5638 attempt to carry the annotations over throughout the compilation all the
5639 way to the end, in an attempt to improve debug information while
5640 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5642 It can be enabled even if var-tracking is disabled, in which case
5643 annotations will be created and maintained, but discarded at the end.
5645 @item -fvar-tracking-assignments-toggle
5646 @opindex fvar-tracking-assignments-toggle
5647 @opindex fno-var-tracking-assignments-toggle
5648 Toggle @option{-fvar-tracking-assignments}, in the same way that
5649 @option{-gtoggle} toggles @option{-g}.
5651 @item -print-file-name=@var{library}
5652 @opindex print-file-name
5653 Print the full absolute name of the library file @var{library} that
5654 would be used when linking---and don't do anything else. With this
5655 option, GCC does not compile or link anything; it just prints the
5658 @item -print-multi-directory
5659 @opindex print-multi-directory
5660 Print the directory name corresponding to the multilib selected by any
5661 other switches present in the command line. This directory is supposed
5662 to exist in @env{GCC_EXEC_PREFIX}.
5664 @item -print-multi-lib
5665 @opindex print-multi-lib
5666 Print the mapping from multilib directory names to compiler switches
5667 that enable them. The directory name is separated from the switches by
5668 @samp{;}, and each switch starts with an @samp{@@} instead of the
5669 @samp{-}, without spaces between multiple switches. This is supposed to
5670 ease shell-processing.
5672 @item -print-multi-os-directory
5673 @opindex print-multi-os-directory
5674 Print the path to OS libraries for the selected
5675 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5676 present in the @file{lib} subdirectory and no multilibs are used, this is
5677 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5678 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5679 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5680 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5682 @item -print-prog-name=@var{program}
5683 @opindex print-prog-name
5684 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5686 @item -print-libgcc-file-name
5687 @opindex print-libgcc-file-name
5688 Same as @option{-print-file-name=libgcc.a}.
5690 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5691 but you do want to link with @file{libgcc.a}. You can do
5694 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5697 @item -print-search-dirs
5698 @opindex print-search-dirs
5699 Print the name of the configured installation directory and a list of
5700 program and library directories @command{gcc} will search---and don't do anything else.
5702 This is useful when @command{gcc} prints the error message
5703 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5704 To resolve this you either need to put @file{cpp0} and the other compiler
5705 components where @command{gcc} expects to find them, or you can set the environment
5706 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5707 Don't forget the trailing @samp{/}.
5708 @xref{Environment Variables}.
5710 @item -print-sysroot
5711 @opindex print-sysroot
5712 Print the target sysroot directory that will be used during
5713 compilation. This is the target sysroot specified either at configure
5714 time or using the @option{--sysroot} option, possibly with an extra
5715 suffix that depends on compilation options. If no target sysroot is
5716 specified, the option prints nothing.
5718 @item -print-sysroot-headers-suffix
5719 @opindex print-sysroot-headers-suffix
5720 Print the suffix added to the target sysroot when searching for
5721 headers, or give an error if the compiler is not configured with such
5722 a suffix---and don't do anything else.
5725 @opindex dumpmachine
5726 Print the compiler's target machine (for example,
5727 @samp{i686-pc-linux-gnu})---and don't do anything else.
5730 @opindex dumpversion
5731 Print the compiler version (for example, @samp{3.0})---and don't do
5736 Print the compiler's built-in specs---and don't do anything else. (This
5737 is used when GCC itself is being built.) @xref{Spec Files}.
5739 @item -feliminate-unused-debug-types
5740 @opindex feliminate-unused-debug-types
5741 Normally, when producing DWARF2 output, GCC will emit debugging
5742 information for all types declared in a compilation
5743 unit, regardless of whether or not they are actually used
5744 in that compilation unit. Sometimes this is useful, such as
5745 if, in the debugger, you want to cast a value to a type that is
5746 not actually used in your program (but is declared). More often,
5747 however, this results in a significant amount of wasted space.
5748 With this option, GCC will avoid producing debug symbol output
5749 for types that are nowhere used in the source file being compiled.
5752 @node Optimize Options
5753 @section Options That Control Optimization
5754 @cindex optimize options
5755 @cindex options, optimization
5757 These options control various sorts of optimizations.
5759 Without any optimization option, the compiler's goal is to reduce the
5760 cost of compilation and to make debugging produce the expected
5761 results. Statements are independent: if you stop the program with a
5762 breakpoint between statements, you can then assign a new value to any
5763 variable or change the program counter to any other statement in the
5764 function and get exactly the results you would expect from the source
5767 Turning on optimization flags makes the compiler attempt to improve
5768 the performance and/or code size at the expense of compilation time
5769 and possibly the ability to debug the program.
5771 The compiler performs optimization based on the knowledge it has of the
5772 program. Compiling multiple files at once to a single output file mode allows
5773 the compiler to use information gained from all of the files when compiling
5776 Not all optimizations are controlled directly by a flag. Only
5777 optimizations that have a flag are listed in this section.
5779 Most optimizations are only enabled if an @option{-O} level is set on
5780 the command line. Otherwise they are disabled, even if individual
5781 optimization flags are specified.
5783 Depending on the target and how GCC was configured, a slightly different
5784 set of optimizations may be enabled at each @option{-O} level than
5785 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5786 to find out the exact set of optimizations that are enabled at each level.
5787 @xref{Overall Options}, for examples.
5794 Optimize. Optimizing compilation takes somewhat more time, and a lot
5795 more memory for a large function.
5797 With @option{-O}, the compiler tries to reduce code size and execution
5798 time, without performing any optimizations that take a great deal of
5801 @option{-O} turns on the following optimization flags:
5804 -fcprop-registers @gol
5807 -fdelayed-branch @gol
5809 -fguess-branch-probability @gol
5810 -fif-conversion2 @gol
5811 -fif-conversion @gol
5812 -fipa-pure-const @gol
5814 -fipa-reference @gol
5816 -fsplit-wide-types @gol
5817 -ftree-builtin-call-dce @gol
5820 -ftree-copyrename @gol
5822 -ftree-dominator-opts @gol
5824 -ftree-forwprop @gol
5832 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5833 where doing so does not interfere with debugging.
5837 Optimize even more. GCC performs nearly all supported optimizations
5838 that do not involve a space-speed tradeoff.
5839 As compared to @option{-O}, this option increases both compilation time
5840 and the performance of the generated code.
5842 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5843 also turns on the following optimization flags:
5844 @gccoptlist{-fthread-jumps @gol
5845 -falign-functions -falign-jumps @gol
5846 -falign-loops -falign-labels @gol
5849 -fcse-follow-jumps -fcse-skip-blocks @gol
5850 -fdelete-null-pointer-checks @gol
5851 -fexpensive-optimizations @gol
5852 -fgcse -fgcse-lm @gol
5853 -finline-small-functions @gol
5854 -findirect-inlining @gol
5856 -foptimize-sibling-calls @gol
5859 -freorder-blocks -freorder-functions @gol
5860 -frerun-cse-after-loop @gol
5861 -fsched-interblock -fsched-spec @gol
5862 -fschedule-insns -fschedule-insns2 @gol
5863 -fstrict-aliasing -fstrict-overflow @gol
5864 -ftree-switch-conversion @gol
5868 Please note the warning under @option{-fgcse} about
5869 invoking @option{-O2} on programs that use computed gotos.
5873 Optimize yet more. @option{-O3} turns on all optimizations specified
5874 by @option{-O2} and also turns on the @option{-finline-functions},
5875 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5876 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5880 Reduce compilation time and make debugging produce the expected
5881 results. This is the default.
5885 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5886 do not typically increase code size. It also performs further
5887 optimizations designed to reduce code size.
5889 @option{-Os} disables the following optimization flags:
5890 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5891 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5892 -fprefetch-loop-arrays -ftree-vect-loop-version}
5896 Disregard strict standards compliance. @option{-Ofast} enables all
5897 @option{-O3} optimizations. It also enables optimizations that are not
5898 valid for all standard compliant programs.
5899 It turns on @option{-ffast-math}.
5901 If you use multiple @option{-O} options, with or without level numbers,
5902 the last such option is the one that is effective.
5905 Options of the form @option{-f@var{flag}} specify machine-independent
5906 flags. Most flags have both positive and negative forms; the negative
5907 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5908 below, only one of the forms is listed---the one you typically will
5909 use. You can figure out the other form by either removing @samp{no-}
5912 The following options control specific optimizations. They are either
5913 activated by @option{-O} options or are related to ones that are. You
5914 can use the following flags in the rare cases when ``fine-tuning'' of
5915 optimizations to be performed is desired.
5918 @item -fno-default-inline
5919 @opindex fno-default-inline
5920 Do not make member functions inline by default merely because they are
5921 defined inside the class scope (C++ only). Otherwise, when you specify
5922 @w{@option{-O}}, member functions defined inside class scope are compiled
5923 inline by default; i.e., you don't need to add @samp{inline} in front of
5924 the member function name.
5926 @item -fno-defer-pop
5927 @opindex fno-defer-pop
5928 Always pop the arguments to each function call as soon as that function
5929 returns. For machines which must pop arguments after a function call,
5930 the compiler normally lets arguments accumulate on the stack for several
5931 function calls and pops them all at once.
5933 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5935 @item -fforward-propagate
5936 @opindex fforward-propagate
5937 Perform a forward propagation pass on RTL@. The pass tries to combine two
5938 instructions and checks if the result can be simplified. If loop unrolling
5939 is active, two passes are performed and the second is scheduled after
5942 This option is enabled by default at optimization levels @option{-O},
5943 @option{-O2}, @option{-O3}, @option{-Os}.
5945 @item -fomit-frame-pointer
5946 @opindex fomit-frame-pointer
5947 Don't keep the frame pointer in a register for functions that
5948 don't need one. This avoids the instructions to save, set up and
5949 restore frame pointers; it also makes an extra register available
5950 in many functions. @strong{It also makes debugging impossible on
5953 On some machines, such as the VAX, this flag has no effect, because
5954 the standard calling sequence automatically handles the frame pointer
5955 and nothing is saved by pretending it doesn't exist. The
5956 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5957 whether a target machine supports this flag. @xref{Registers,,Register
5958 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5960 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5962 @item -foptimize-sibling-calls
5963 @opindex foptimize-sibling-calls
5964 Optimize sibling and tail recursive calls.
5966 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5970 Don't pay attention to the @code{inline} keyword. Normally this option
5971 is used to keep the compiler from expanding any functions inline.
5972 Note that if you are not optimizing, no functions can be expanded inline.
5974 @item -finline-small-functions
5975 @opindex finline-small-functions
5976 Integrate functions into their callers when their body is smaller than expected
5977 function call code (so overall size of program gets smaller). The compiler
5978 heuristically decides which functions are simple enough to be worth integrating
5981 Enabled at level @option{-O2}.
5983 @item -findirect-inlining
5984 @opindex findirect-inlining
5985 Inline also indirect calls that are discovered to be known at compile
5986 time thanks to previous inlining. This option has any effect only
5987 when inlining itself is turned on by the @option{-finline-functions}
5988 or @option{-finline-small-functions} options.
5990 Enabled at level @option{-O2}.
5992 @item -finline-functions
5993 @opindex finline-functions
5994 Integrate all simple functions into their callers. The compiler
5995 heuristically decides which functions are simple enough to be worth
5996 integrating in this way.
5998 If all calls to a given function are integrated, and the function is
5999 declared @code{static}, then the function is normally not output as
6000 assembler code in its own right.
6002 Enabled at level @option{-O3}.
6004 @item -finline-functions-called-once
6005 @opindex finline-functions-called-once
6006 Consider all @code{static} functions called once for inlining into their
6007 caller even if they are not marked @code{inline}. If a call to a given
6008 function is integrated, then the function is not output as assembler code
6011 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
6013 @item -fearly-inlining
6014 @opindex fearly-inlining
6015 Inline functions marked by @code{always_inline} and functions whose body seems
6016 smaller than the function call overhead early before doing
6017 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
6018 makes profiling significantly cheaper and usually inlining faster on programs
6019 having large chains of nested wrapper functions.
6025 Perform interprocedural scalar replacement of aggregates, removal of
6026 unused parameters and replacement of parameters passed by reference
6027 by parameters passed by value.
6029 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6031 @item -finline-limit=@var{n}
6032 @opindex finline-limit
6033 By default, GCC limits the size of functions that can be inlined. This flag
6034 allows coarse control of this limit. @var{n} is the size of functions that
6035 can be inlined in number of pseudo instructions.
6037 Inlining is actually controlled by a number of parameters, which may be
6038 specified individually by using @option{--param @var{name}=@var{value}}.
6039 The @option{-finline-limit=@var{n}} option sets some of these parameters
6043 @item max-inline-insns-single
6044 is set to @var{n}/2.
6045 @item max-inline-insns-auto
6046 is set to @var{n}/2.
6049 See below for a documentation of the individual
6050 parameters controlling inlining and for the defaults of these parameters.
6052 @emph{Note:} there may be no value to @option{-finline-limit} that results
6053 in default behavior.
6055 @emph{Note:} pseudo instruction represents, in this particular context, an
6056 abstract measurement of function's size. In no way does it represent a count
6057 of assembly instructions and as such its exact meaning might change from one
6058 release to an another.
6060 @item -fkeep-inline-functions
6061 @opindex fkeep-inline-functions
6062 In C, emit @code{static} functions that are declared @code{inline}
6063 into the object file, even if the function has been inlined into all
6064 of its callers. This switch does not affect functions using the
6065 @code{extern inline} extension in GNU C90@. In C++, emit any and all
6066 inline functions into the object file.
6068 @item -fkeep-static-consts
6069 @opindex fkeep-static-consts
6070 Emit variables declared @code{static const} when optimization isn't turned
6071 on, even if the variables aren't referenced.
6073 GCC enables this option by default. If you want to force the compiler to
6074 check if the variable was referenced, regardless of whether or not
6075 optimization is turned on, use the @option{-fno-keep-static-consts} option.
6077 @item -fmerge-constants
6078 @opindex fmerge-constants
6079 Attempt to merge identical constants (string constants and floating point
6080 constants) across compilation units.
6082 This option is the default for optimized compilation if the assembler and
6083 linker support it. Use @option{-fno-merge-constants} to inhibit this
6086 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6088 @item -fmerge-all-constants
6089 @opindex fmerge-all-constants
6090 Attempt to merge identical constants and identical variables.
6092 This option implies @option{-fmerge-constants}. In addition to
6093 @option{-fmerge-constants} this considers e.g.@: even constant initialized
6094 arrays or initialized constant variables with integral or floating point
6095 types. Languages like C or C++ require each variable, including multiple
6096 instances of the same variable in recursive calls, to have distinct locations,
6097 so using this option will result in non-conforming
6100 @item -fmodulo-sched
6101 @opindex fmodulo-sched
6102 Perform swing modulo scheduling immediately before the first scheduling
6103 pass. This pass looks at innermost loops and reorders their
6104 instructions by overlapping different iterations.
6106 @item -fmodulo-sched-allow-regmoves
6107 @opindex fmodulo-sched-allow-regmoves
6108 Perform more aggressive SMS based modulo scheduling with register moves
6109 allowed. By setting this flag certain anti-dependences edges will be
6110 deleted which will trigger the generation of reg-moves based on the
6111 life-range analysis. This option is effective only with
6112 @option{-fmodulo-sched} enabled.
6114 @item -fno-branch-count-reg
6115 @opindex fno-branch-count-reg
6116 Do not use ``decrement and branch'' instructions on a count register,
6117 but instead generate a sequence of instructions that decrement a
6118 register, compare it against zero, then branch based upon the result.
6119 This option is only meaningful on architectures that support such
6120 instructions, which include x86, PowerPC, IA-64 and S/390.
6122 The default is @option{-fbranch-count-reg}.
6124 @item -fno-function-cse
6125 @opindex fno-function-cse
6126 Do not put function addresses in registers; make each instruction that
6127 calls a constant function contain the function's address explicitly.
6129 This option results in less efficient code, but some strange hacks
6130 that alter the assembler output may be confused by the optimizations
6131 performed when this option is not used.
6133 The default is @option{-ffunction-cse}
6135 @item -fno-zero-initialized-in-bss
6136 @opindex fno-zero-initialized-in-bss
6137 If the target supports a BSS section, GCC by default puts variables that
6138 are initialized to zero into BSS@. This can save space in the resulting
6141 This option turns off this behavior because some programs explicitly
6142 rely on variables going to the data section. E.g., so that the
6143 resulting executable can find the beginning of that section and/or make
6144 assumptions based on that.
6146 The default is @option{-fzero-initialized-in-bss}.
6148 @item -fmudflap -fmudflapth -fmudflapir
6152 @cindex bounds checking
6154 For front-ends that support it (C and C++), instrument all risky
6155 pointer/array dereferencing operations, some standard library
6156 string/heap functions, and some other associated constructs with
6157 range/validity tests. Modules so instrumented should be immune to
6158 buffer overflows, invalid heap use, and some other classes of C/C++
6159 programming errors. The instrumentation relies on a separate runtime
6160 library (@file{libmudflap}), which will be linked into a program if
6161 @option{-fmudflap} is given at link time. Run-time behavior of the
6162 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6163 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6166 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6167 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6168 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6169 instrumentation should ignore pointer reads. This produces less
6170 instrumentation (and therefore faster execution) and still provides
6171 some protection against outright memory corrupting writes, but allows
6172 erroneously read data to propagate within a program.
6174 @item -fthread-jumps
6175 @opindex fthread-jumps
6176 Perform optimizations where we check to see if a jump branches to a
6177 location where another comparison subsumed by the first is found. If
6178 so, the first branch is redirected to either the destination of the
6179 second branch or a point immediately following it, depending on whether
6180 the condition is known to be true or false.
6182 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6184 @item -fsplit-wide-types
6185 @opindex fsplit-wide-types
6186 When using a type that occupies multiple registers, such as @code{long
6187 long} on a 32-bit system, split the registers apart and allocate them
6188 independently. This normally generates better code for those types,
6189 but may make debugging more difficult.
6191 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6194 @item -fcse-follow-jumps
6195 @opindex fcse-follow-jumps
6196 In common subexpression elimination (CSE), scan through jump instructions
6197 when the target of the jump is not reached by any other path. For
6198 example, when CSE encounters an @code{if} statement with an
6199 @code{else} clause, CSE will follow the jump when the condition
6202 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6204 @item -fcse-skip-blocks
6205 @opindex fcse-skip-blocks
6206 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6207 follow jumps which conditionally skip over blocks. When CSE
6208 encounters a simple @code{if} statement with no else clause,
6209 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6210 body of the @code{if}.
6212 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6214 @item -frerun-cse-after-loop
6215 @opindex frerun-cse-after-loop
6216 Re-run common subexpression elimination after loop optimizations has been
6219 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6223 Perform a global common subexpression elimination pass.
6224 This pass also performs global constant and copy propagation.
6226 @emph{Note:} When compiling a program using computed gotos, a GCC
6227 extension, you may get better runtime performance if you disable
6228 the global common subexpression elimination pass by adding
6229 @option{-fno-gcse} to the command line.
6231 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6235 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6236 attempt to move loads which are only killed by stores into themselves. This
6237 allows a loop containing a load/store sequence to be changed to a load outside
6238 the loop, and a copy/store within the loop.
6240 Enabled by default when gcse is enabled.
6244 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6245 global common subexpression elimination. This pass will attempt to move
6246 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6247 loops containing a load/store sequence can be changed to a load before
6248 the loop and a store after the loop.
6250 Not enabled at any optimization level.
6254 When @option{-fgcse-las} is enabled, the global common subexpression
6255 elimination pass eliminates redundant loads that come after stores to the
6256 same memory location (both partial and full redundancies).
6258 Not enabled at any optimization level.
6260 @item -fgcse-after-reload
6261 @opindex fgcse-after-reload
6262 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6263 pass is performed after reload. The purpose of this pass is to cleanup
6266 @item -funsafe-loop-optimizations
6267 @opindex funsafe-loop-optimizations
6268 If given, the loop optimizer will assume that loop indices do not
6269 overflow, and that the loops with nontrivial exit condition are not
6270 infinite. This enables a wider range of loop optimizations even if
6271 the loop optimizer itself cannot prove that these assumptions are valid.
6272 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6273 if it finds this kind of loop.
6275 @item -fcrossjumping
6276 @opindex fcrossjumping
6277 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6278 resulting code may or may not perform better than without cross-jumping.
6280 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6282 @item -fauto-inc-dec
6283 @opindex fauto-inc-dec
6284 Combine increments or decrements of addresses with memory accesses.
6285 This pass is always skipped on architectures that do not have
6286 instructions to support this. Enabled by default at @option{-O} and
6287 higher on architectures that support this.
6291 Perform dead code elimination (DCE) on RTL@.
6292 Enabled by default at @option{-O} and higher.
6296 Perform dead store elimination (DSE) on RTL@.
6297 Enabled by default at @option{-O} and higher.
6299 @item -fif-conversion
6300 @opindex fif-conversion
6301 Attempt to transform conditional jumps into branch-less equivalents. This
6302 include use of conditional moves, min, max, set flags and abs instructions, and
6303 some tricks doable by standard arithmetics. The use of conditional execution
6304 on chips where it is available is controlled by @code{if-conversion2}.
6306 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6308 @item -fif-conversion2
6309 @opindex fif-conversion2
6310 Use conditional execution (where available) to transform conditional jumps into
6311 branch-less equivalents.
6313 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6315 @item -fdelete-null-pointer-checks
6316 @opindex fdelete-null-pointer-checks
6317 Assume that programs cannot safely dereference null pointers, and that
6318 no code or data element resides there. This enables simple constant
6319 folding optimizations at all optimization levels. In addition, other
6320 optimization passes in GCC use this flag to control global dataflow
6321 analyses that eliminate useless checks for null pointers; these assume
6322 that if a pointer is checked after it has already been dereferenced,
6325 Note however that in some environments this assumption is not true.
6326 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6327 for programs which depend on that behavior.
6329 Some targets, especially embedded ones, disable this option at all levels.
6330 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6331 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6332 are enabled independently at different optimization levels.
6334 @item -fexpensive-optimizations
6335 @opindex fexpensive-optimizations
6336 Perform a number of minor optimizations that are relatively expensive.
6338 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6340 @item -foptimize-register-move
6342 @opindex foptimize-register-move
6344 Attempt to reassign register numbers in move instructions and as
6345 operands of other simple instructions in order to maximize the amount of
6346 register tying. This is especially helpful on machines with two-operand
6349 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6352 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6354 @item -fira-algorithm=@var{algorithm}
6355 Use specified coloring algorithm for the integrated register
6356 allocator. The @var{algorithm} argument should be @code{priority} or
6357 @code{CB}. The first algorithm specifies Chow's priority coloring,
6358 the second one specifies Chaitin-Briggs coloring. The second
6359 algorithm can be unimplemented for some architectures. If it is
6360 implemented, it is the default because Chaitin-Briggs coloring as a
6361 rule generates a better code.
6363 @item -fira-region=@var{region}
6364 Use specified regions for the integrated register allocator. The
6365 @var{region} argument should be one of @code{all}, @code{mixed}, or
6366 @code{one}. The first value means using all loops as register
6367 allocation regions, the second value which is the default means using
6368 all loops except for loops with small register pressure as the
6369 regions, and third one means using all function as a single region.
6370 The first value can give best result for machines with small size and
6371 irregular register set, the third one results in faster and generates
6372 decent code and the smallest size code, and the default value usually
6373 give the best results in most cases and for most architectures.
6375 @item -fira-coalesce
6376 @opindex fira-coalesce
6377 Do optimistic register coalescing. This option might be profitable for
6378 architectures with big regular register files.
6380 @item -fira-loop-pressure
6381 @opindex fira-loop-pressure
6382 Use IRA to evaluate register pressure in loops for decision to move
6383 loop invariants. Usage of this option usually results in generation
6384 of faster and smaller code on machines with big register files (>= 32
6385 registers) but it can slow compiler down.
6387 This option is enabled at level @option{-O3} for some targets.
6389 @item -fno-ira-share-save-slots
6390 @opindex fno-ira-share-save-slots
6391 Switch off sharing stack slots used for saving call used hard
6392 registers living through a call. Each hard register will get a
6393 separate stack slot and as a result function stack frame will be
6396 @item -fno-ira-share-spill-slots
6397 @opindex fno-ira-share-spill-slots
6398 Switch off sharing stack slots allocated for pseudo-registers. Each
6399 pseudo-register which did not get a hard register will get a separate
6400 stack slot and as a result function stack frame will be bigger.
6402 @item -fira-verbose=@var{n}
6403 @opindex fira-verbose
6404 Set up how verbose dump file for the integrated register allocator
6405 will be. Default value is 5. If the value is greater or equal to 10,
6406 the dump file will be stderr as if the value were @var{n} minus 10.
6408 @item -fdelayed-branch
6409 @opindex fdelayed-branch
6410 If supported for the target machine, attempt to reorder instructions
6411 to exploit instruction slots available after delayed branch
6414 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6416 @item -fschedule-insns
6417 @opindex fschedule-insns
6418 If supported for the target machine, attempt to reorder instructions to
6419 eliminate execution stalls due to required data being unavailable. This
6420 helps machines that have slow floating point or memory load instructions
6421 by allowing other instructions to be issued until the result of the load
6422 or floating point instruction is required.
6424 Enabled at levels @option{-O2}, @option{-O3}.
6426 @item -fschedule-insns2
6427 @opindex fschedule-insns2
6428 Similar to @option{-fschedule-insns}, but requests an additional pass of
6429 instruction scheduling after register allocation has been done. This is
6430 especially useful on machines with a relatively small number of
6431 registers and where memory load instructions take more than one cycle.
6433 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6435 @item -fno-sched-interblock
6436 @opindex fno-sched-interblock
6437 Don't schedule instructions across basic blocks. This is normally
6438 enabled by default when scheduling before register allocation, i.e.@:
6439 with @option{-fschedule-insns} or at @option{-O2} or higher.
6441 @item -fno-sched-spec
6442 @opindex fno-sched-spec
6443 Don't allow speculative motion of non-load instructions. This is normally
6444 enabled by default when scheduling before register allocation, i.e.@:
6445 with @option{-fschedule-insns} or at @option{-O2} or higher.
6447 @item -fsched-pressure
6448 @opindex fsched-pressure
6449 Enable register pressure sensitive insn scheduling before the register
6450 allocation. This only makes sense when scheduling before register
6451 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6452 @option{-O2} or higher. Usage of this option can improve the
6453 generated code and decrease its size by preventing register pressure
6454 increase above the number of available hard registers and as a
6455 consequence register spills in the register allocation.
6457 @item -fsched-spec-load
6458 @opindex fsched-spec-load
6459 Allow speculative motion of some load instructions. This only makes
6460 sense when scheduling before register allocation, i.e.@: with
6461 @option{-fschedule-insns} or at @option{-O2} or higher.
6463 @item -fsched-spec-load-dangerous
6464 @opindex fsched-spec-load-dangerous
6465 Allow speculative motion of more load instructions. This only makes
6466 sense when scheduling before register allocation, i.e.@: with
6467 @option{-fschedule-insns} or at @option{-O2} or higher.
6469 @item -fsched-stalled-insns
6470 @itemx -fsched-stalled-insns=@var{n}
6471 @opindex fsched-stalled-insns
6472 Define how many insns (if any) can be moved prematurely from the queue
6473 of stalled insns into the ready list, during the second scheduling pass.
6474 @option{-fno-sched-stalled-insns} means that no insns will be moved
6475 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6476 on how many queued insns can be moved prematurely.
6477 @option{-fsched-stalled-insns} without a value is equivalent to
6478 @option{-fsched-stalled-insns=1}.
6480 @item -fsched-stalled-insns-dep
6481 @itemx -fsched-stalled-insns-dep=@var{n}
6482 @opindex fsched-stalled-insns-dep
6483 Define how many insn groups (cycles) will be examined for a dependency
6484 on a stalled insn that is candidate for premature removal from the queue
6485 of stalled insns. This has an effect only during the second scheduling pass,
6486 and only if @option{-fsched-stalled-insns} is used.
6487 @option{-fno-sched-stalled-insns-dep} is equivalent to
6488 @option{-fsched-stalled-insns-dep=0}.
6489 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6490 @option{-fsched-stalled-insns-dep=1}.
6492 @item -fsched2-use-superblocks
6493 @opindex fsched2-use-superblocks
6494 When scheduling after register allocation, do use superblock scheduling
6495 algorithm. Superblock scheduling allows motion across basic block boundaries
6496 resulting on faster schedules. This option is experimental, as not all machine
6497 descriptions used by GCC model the CPU closely enough to avoid unreliable
6498 results from the algorithm.
6500 This only makes sense when scheduling after register allocation, i.e.@: with
6501 @option{-fschedule-insns2} or at @option{-O2} or higher.
6503 @item -fsched-group-heuristic
6504 @opindex fsched-group-heuristic
6505 Enable the group heuristic in the scheduler. This heuristic favors
6506 the instruction that belongs to a schedule group. This is enabled
6507 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6508 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6510 @item -fsched-critical-path-heuristic
6511 @opindex fsched-critical-path-heuristic
6512 Enable the critical-path heuristic in the scheduler. This heuristic favors
6513 instructions on the critical path. This is enabled by default when
6514 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6515 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6517 @item -fsched-spec-insn-heuristic
6518 @opindex fsched-spec-insn-heuristic
6519 Enable the speculative instruction heuristic in the scheduler. This
6520 heuristic favors speculative instructions with greater dependency weakness.
6521 This is enabled by default when scheduling is enabled, i.e.@:
6522 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6523 or at @option{-O2} or higher.
6525 @item -fsched-rank-heuristic
6526 @opindex fsched-rank-heuristic
6527 Enable the rank heuristic in the scheduler. This heuristic favors
6528 the instruction belonging to a basic block with greater size or frequency.
6529 This is enabled by default when scheduling is enabled, i.e.@:
6530 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6531 at @option{-O2} or higher.
6533 @item -fsched-last-insn-heuristic
6534 @opindex fsched-last-insn-heuristic
6535 Enable the last-instruction heuristic in the scheduler. This heuristic
6536 favors the instruction that is less dependent on the last instruction
6537 scheduled. This is enabled by default when scheduling is enabled,
6538 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6539 at @option{-O2} or higher.
6541 @item -fsched-dep-count-heuristic
6542 @opindex fsched-dep-count-heuristic
6543 Enable the dependent-count heuristic in the scheduler. This heuristic
6544 favors the instruction that has more instructions depending on it.
6545 This is enabled by default when scheduling is enabled, i.e.@:
6546 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6547 at @option{-O2} or higher.
6549 @item -freschedule-modulo-scheduled-loops
6550 @opindex freschedule-modulo-scheduled-loops
6551 The modulo scheduling comes before the traditional scheduling, if a loop
6552 was modulo scheduled we may want to prevent the later scheduling passes
6553 from changing its schedule, we use this option to control that.
6555 @item -fselective-scheduling
6556 @opindex fselective-scheduling
6557 Schedule instructions using selective scheduling algorithm. Selective
6558 scheduling runs instead of the first scheduler pass.
6560 @item -fselective-scheduling2
6561 @opindex fselective-scheduling2
6562 Schedule instructions using selective scheduling algorithm. Selective
6563 scheduling runs instead of the second scheduler pass.
6565 @item -fsel-sched-pipelining
6566 @opindex fsel-sched-pipelining
6567 Enable software pipelining of innermost loops during selective scheduling.
6568 This option has no effect until one of @option{-fselective-scheduling} or
6569 @option{-fselective-scheduling2} is turned on.
6571 @item -fsel-sched-pipelining-outer-loops
6572 @opindex fsel-sched-pipelining-outer-loops
6573 When pipelining loops during selective scheduling, also pipeline outer loops.
6574 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6576 @item -fcaller-saves
6577 @opindex fcaller-saves
6578 Enable values to be allocated in registers that will be clobbered by
6579 function calls, by emitting extra instructions to save and restore the
6580 registers around such calls. Such allocation is done only when it
6581 seems to result in better code than would otherwise be produced.
6583 This option is always enabled by default on certain machines, usually
6584 those which have no call-preserved registers to use instead.
6586 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6588 @item -fconserve-stack
6589 @opindex fconserve-stack
6590 Attempt to minimize stack usage. The compiler will attempt to use less
6591 stack space, even if that makes the program slower. This option
6592 implies setting the @option{large-stack-frame} parameter to 100
6593 and the @option{large-stack-frame-growth} parameter to 400.
6595 @item -ftree-reassoc
6596 @opindex ftree-reassoc
6597 Perform reassociation on trees. This flag is enabled by default
6598 at @option{-O} and higher.
6602 Perform partial redundancy elimination (PRE) on trees. This flag is
6603 enabled by default at @option{-O2} and @option{-O3}.
6605 @item -ftree-forwprop
6606 @opindex ftree-forwprop
6607 Perform forward propagation on trees. This flag is enabled by default
6608 at @option{-O} and higher.
6612 Perform full redundancy elimination (FRE) on trees. The difference
6613 between FRE and PRE is that FRE only considers expressions
6614 that are computed on all paths leading to the redundant computation.
6615 This analysis is faster than PRE, though it exposes fewer redundancies.
6616 This flag is enabled by default at @option{-O} and higher.
6618 @item -ftree-phiprop
6619 @opindex ftree-phiprop
6620 Perform hoisting of loads from conditional pointers on trees. This
6621 pass is enabled by default at @option{-O} and higher.
6623 @item -ftree-copy-prop
6624 @opindex ftree-copy-prop
6625 Perform copy propagation on trees. This pass eliminates unnecessary
6626 copy operations. This flag is enabled by default at @option{-O} and
6629 @item -fipa-pure-const
6630 @opindex fipa-pure-const
6631 Discover which functions are pure or constant.
6632 Enabled by default at @option{-O} and higher.
6634 @item -fipa-reference
6635 @opindex fipa-reference
6636 Discover which static variables do not escape cannot escape the
6638 Enabled by default at @option{-O} and higher.
6640 @item -fipa-struct-reorg
6641 @opindex fipa-struct-reorg
6642 Perform structure reorganization optimization, that change C-like structures
6643 layout in order to better utilize spatial locality. This transformation is
6644 affective for programs containing arrays of structures. Available in two
6645 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6646 or static (which uses built-in heuristics). It works only in whole program
6647 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6648 enabled. Structures considered @samp{cold} by this transformation are not
6649 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6651 With this flag, the program debug info reflects a new structure layout.
6655 Perform interprocedural pointer analysis and interprocedural modification
6656 and reference analysis. This option can cause excessive memory and
6657 compile-time usage on large compilation units. It is not enabled by
6658 default at any optimization level.
6661 @opindex fipa-profile
6662 Perform interprocedural profile propagation. The functions called only from
6663 cold functions are marked as cold. Also functions executed once (such as
6664 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
6665 functions and loop less parts of functions executed once are then optimized for
6667 Enabled by default at @option{-O} and higher.
6671 Perform interprocedural constant propagation.
6672 This optimization analyzes the program to determine when values passed
6673 to functions are constants and then optimizes accordingly.
6674 This optimization can substantially increase performance
6675 if the application has constants passed to functions.
6676 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6678 @item -fipa-cp-clone
6679 @opindex fipa-cp-clone
6680 Perform function cloning to make interprocedural constant propagation stronger.
6681 When enabled, interprocedural constant propagation will perform function cloning
6682 when externally visible function can be called with constant arguments.
6683 Because this optimization can create multiple copies of functions,
6684 it may significantly increase code size
6685 (see @option{--param ipcp-unit-growth=@var{value}}).
6686 This flag is enabled by default at @option{-O3}.
6688 @item -fipa-matrix-reorg
6689 @opindex fipa-matrix-reorg
6690 Perform matrix flattening and transposing.
6691 Matrix flattening tries to replace an @math{m}-dimensional matrix
6692 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6693 This reduces the level of indirection needed for accessing the elements
6694 of the matrix. The second optimization is matrix transposing that
6695 attempts to change the order of the matrix's dimensions in order to
6696 improve cache locality.
6697 Both optimizations need the @option{-fwhole-program} flag.
6698 Transposing is enabled only if profiling information is available.
6702 Perform forward store motion on trees. This flag is
6703 enabled by default at @option{-O} and higher.
6707 Perform sparse conditional constant propagation (CCP) on trees. This
6708 pass only operates on local scalar variables and is enabled by default
6709 at @option{-O} and higher.
6711 @item -ftree-switch-conversion
6712 Perform conversion of simple initializations in a switch to
6713 initializations from a scalar array. This flag is enabled by default
6714 at @option{-O2} and higher.
6718 Perform dead code elimination (DCE) on trees. This flag is enabled by
6719 default at @option{-O} and higher.
6721 @item -ftree-builtin-call-dce
6722 @opindex ftree-builtin-call-dce
6723 Perform conditional dead code elimination (DCE) for calls to builtin functions
6724 that may set @code{errno} but are otherwise side-effect free. This flag is
6725 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6728 @item -ftree-dominator-opts
6729 @opindex ftree-dominator-opts
6730 Perform a variety of simple scalar cleanups (constant/copy
6731 propagation, redundancy elimination, range propagation and expression
6732 simplification) based on a dominator tree traversal. This also
6733 performs jump threading (to reduce jumps to jumps). This flag is
6734 enabled by default at @option{-O} and higher.
6738 Perform dead store elimination (DSE) on trees. A dead store is a store into
6739 a memory location which will later be overwritten by another store without
6740 any intervening loads. In this case the earlier store can be deleted. This
6741 flag is enabled by default at @option{-O} and higher.
6745 Perform loop header copying on trees. This is beneficial since it increases
6746 effectiveness of code motion optimizations. It also saves one jump. This flag
6747 is enabled by default at @option{-O} and higher. It is not enabled
6748 for @option{-Os}, since it usually increases code size.
6750 @item -ftree-loop-optimize
6751 @opindex ftree-loop-optimize
6752 Perform loop optimizations on trees. This flag is enabled by default
6753 at @option{-O} and higher.
6755 @item -ftree-loop-linear
6756 @opindex ftree-loop-linear
6757 Perform linear loop transformations on tree. This flag can improve cache
6758 performance and allow further loop optimizations to take place.
6760 @item -floop-interchange
6761 Perform loop interchange transformations on loops. Interchanging two
6762 nested loops switches the inner and outer loops. For example, given a
6767 A(J, I) = A(J, I) * C
6771 loop interchange will transform the loop as if the user had written:
6775 A(J, I) = A(J, I) * C
6779 which can be beneficial when @code{N} is larger than the caches,
6780 because in Fortran, the elements of an array are stored in memory
6781 contiguously by column, and the original loop iterates over rows,
6782 potentially creating at each access a cache miss. This optimization
6783 applies to all the languages supported by GCC and is not limited to
6784 Fortran. To use this code transformation, GCC has to be configured
6785 with @option{--with-ppl} and @option{--with-cloog} to enable the
6786 Graphite loop transformation infrastructure.
6788 @item -floop-strip-mine
6789 Perform loop strip mining transformations on loops. Strip mining
6790 splits a loop into two nested loops. The outer loop has strides
6791 equal to the strip size and the inner loop has strides of the
6792 original loop within a strip. The strip length can be changed
6793 using the @option{loop-block-tile-size} parameter. For example,
6800 loop strip mining will transform the loop as if the user had written:
6803 DO I = II, min (II + 50, N)
6808 This optimization applies to all the languages supported by GCC and is
6809 not limited to Fortran. To use this code transformation, GCC has to
6810 be configured with @option{--with-ppl} and @option{--with-cloog} to
6811 enable the Graphite loop transformation infrastructure.
6814 Perform loop blocking transformations on loops. Blocking strip mines
6815 each loop in the loop nest such that the memory accesses of the
6816 element loops fit inside caches. The strip length can be changed
6817 using the @option{loop-block-tile-size} parameter. For example, given
6822 A(J, I) = B(I) + C(J)
6826 loop blocking will transform the loop as if the user had written:
6830 DO I = II, min (II + 50, N)
6831 DO J = JJ, min (JJ + 50, M)
6832 A(J, I) = B(I) + C(J)
6838 which can be beneficial when @code{M} is larger than the caches,
6839 because the innermost loop will iterate over a smaller amount of data
6840 that can be kept in the caches. This optimization applies to all the
6841 languages supported by GCC and is not limited to Fortran. To use this
6842 code transformation, GCC has to be configured with @option{--with-ppl}
6843 and @option{--with-cloog} to enable the Graphite loop transformation
6846 @item -fgraphite-identity
6847 @opindex fgraphite-identity
6848 Enable the identity transformation for graphite. For every SCoP we generate
6849 the polyhedral representation and transform it back to gimple. Using
6850 @option{-fgraphite-identity} we can check the costs or benefits of the
6851 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6852 are also performed by the code generator CLooG, like index splitting and
6853 dead code elimination in loops.
6855 @item -floop-parallelize-all
6856 Use the Graphite data dependence analysis to identify loops that can
6857 be parallelized. Parallelize all the loops that can be analyzed to
6858 not contain loop carried dependences without checking that it is
6859 profitable to parallelize the loops.
6861 @item -fcheck-data-deps
6862 @opindex fcheck-data-deps
6863 Compare the results of several data dependence analyzers. This option
6864 is used for debugging the data dependence analyzers.
6866 @item -ftree-loop-distribution
6867 Perform loop distribution. This flag can improve cache performance on
6868 big loop bodies and allow further loop optimizations, like
6869 parallelization or vectorization, to take place. For example, the loop
6886 @item -ftree-loop-im
6887 @opindex ftree-loop-im
6888 Perform loop invariant motion on trees. This pass moves only invariants that
6889 would be hard to handle at RTL level (function calls, operations that expand to
6890 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6891 operands of conditions that are invariant out of the loop, so that we can use
6892 just trivial invariantness analysis in loop unswitching. The pass also includes
6895 @item -ftree-loop-ivcanon
6896 @opindex ftree-loop-ivcanon
6897 Create a canonical counter for number of iterations in the loop for that
6898 determining number of iterations requires complicated analysis. Later
6899 optimizations then may determine the number easily. Useful especially
6900 in connection with unrolling.
6904 Perform induction variable optimizations (strength reduction, induction
6905 variable merging and induction variable elimination) on trees.
6907 @item -ftree-parallelize-loops=n
6908 @opindex ftree-parallelize-loops
6909 Parallelize loops, i.e., split their iteration space to run in n threads.
6910 This is only possible for loops whose iterations are independent
6911 and can be arbitrarily reordered. The optimization is only
6912 profitable on multiprocessor machines, for loops that are CPU-intensive,
6913 rather than constrained e.g.@: by memory bandwidth. This option
6914 implies @option{-pthread}, and thus is only supported on targets
6915 that have support for @option{-pthread}.
6919 Perform function-local points-to analysis on trees. This flag is
6920 enabled by default at @option{-O} and higher.
6924 Perform scalar replacement of aggregates. This pass replaces structure
6925 references with scalars to prevent committing structures to memory too
6926 early. This flag is enabled by default at @option{-O} and higher.
6928 @item -ftree-copyrename
6929 @opindex ftree-copyrename
6930 Perform copy renaming on trees. This pass attempts to rename compiler
6931 temporaries to other variables at copy locations, usually resulting in
6932 variable names which more closely resemble the original variables. This flag
6933 is enabled by default at @option{-O} and higher.
6937 Perform temporary expression replacement during the SSA->normal phase. Single
6938 use/single def temporaries are replaced at their use location with their
6939 defining expression. This results in non-GIMPLE code, but gives the expanders
6940 much more complex trees to work on resulting in better RTL generation. This is
6941 enabled by default at @option{-O} and higher.
6943 @item -ftree-vectorize
6944 @opindex ftree-vectorize
6945 Perform loop vectorization on trees. This flag is enabled by default at
6948 @item -ftree-slp-vectorize
6949 @opindex ftree-slp-vectorize
6950 Perform basic block vectorization on trees. This flag is enabled by default at
6951 @option{-O3} and when @option{-ftree-vectorize} is enabled.
6953 @item -ftree-vect-loop-version
6954 @opindex ftree-vect-loop-version
6955 Perform loop versioning when doing loop vectorization on trees. When a loop
6956 appears to be vectorizable except that data alignment or data dependence cannot
6957 be determined at compile time then vectorized and non-vectorized versions of
6958 the loop are generated along with runtime checks for alignment or dependence
6959 to control which version is executed. This option is enabled by default
6960 except at level @option{-Os} where it is disabled.
6962 @item -fvect-cost-model
6963 @opindex fvect-cost-model
6964 Enable cost model for vectorization.
6968 Perform Value Range Propagation on trees. This is similar to the
6969 constant propagation pass, but instead of values, ranges of values are
6970 propagated. This allows the optimizers to remove unnecessary range
6971 checks like array bound checks and null pointer checks. This is
6972 enabled by default at @option{-O2} and higher. Null pointer check
6973 elimination is only done if @option{-fdelete-null-pointer-checks} is
6978 Perform tail duplication to enlarge superblock size. This transformation
6979 simplifies the control flow of the function allowing other optimizations to do
6982 @item -funroll-loops
6983 @opindex funroll-loops
6984 Unroll loops whose number of iterations can be determined at compile
6985 time or upon entry to the loop. @option{-funroll-loops} implies
6986 @option{-frerun-cse-after-loop}. This option makes code larger,
6987 and may or may not make it run faster.
6989 @item -funroll-all-loops
6990 @opindex funroll-all-loops
6991 Unroll all loops, even if their number of iterations is uncertain when
6992 the loop is entered. This usually makes programs run more slowly.
6993 @option{-funroll-all-loops} implies the same options as
6994 @option{-funroll-loops},
6996 @item -fsplit-ivs-in-unroller
6997 @opindex fsplit-ivs-in-unroller
6998 Enables expressing of values of induction variables in later iterations
6999 of the unrolled loop using the value in the first iteration. This breaks
7000 long dependency chains, thus improving efficiency of the scheduling passes.
7002 Combination of @option{-fweb} and CSE is often sufficient to obtain the
7003 same effect. However in cases the loop body is more complicated than
7004 a single basic block, this is not reliable. It also does not work at all
7005 on some of the architectures due to restrictions in the CSE pass.
7007 This optimization is enabled by default.
7009 @item -fvariable-expansion-in-unroller
7010 @opindex fvariable-expansion-in-unroller
7011 With this option, the compiler will create multiple copies of some
7012 local variables when unrolling a loop which can result in superior code.
7014 @item -fpredictive-commoning
7015 @opindex fpredictive-commoning
7016 Perform predictive commoning optimization, i.e., reusing computations
7017 (especially memory loads and stores) performed in previous
7018 iterations of loops.
7020 This option is enabled at level @option{-O3}.
7022 @item -fprefetch-loop-arrays
7023 @opindex fprefetch-loop-arrays
7024 If supported by the target machine, generate instructions to prefetch
7025 memory to improve the performance of loops that access large arrays.
7027 This option may generate better or worse code; results are highly
7028 dependent on the structure of loops within the source code.
7030 Disabled at level @option{-Os}.
7033 @itemx -fno-peephole2
7034 @opindex fno-peephole
7035 @opindex fno-peephole2
7036 Disable any machine-specific peephole optimizations. The difference
7037 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7038 are implemented in the compiler; some targets use one, some use the
7039 other, a few use both.
7041 @option{-fpeephole} is enabled by default.
7042 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7044 @item -fno-guess-branch-probability
7045 @opindex fno-guess-branch-probability
7046 Do not guess branch probabilities using heuristics.
7048 GCC will use heuristics to guess branch probabilities if they are
7049 not provided by profiling feedback (@option{-fprofile-arcs}). These
7050 heuristics are based on the control flow graph. If some branch probabilities
7051 are specified by @samp{__builtin_expect}, then the heuristics will be
7052 used to guess branch probabilities for the rest of the control flow graph,
7053 taking the @samp{__builtin_expect} info into account. The interactions
7054 between the heuristics and @samp{__builtin_expect} can be complex, and in
7055 some cases, it may be useful to disable the heuristics so that the effects
7056 of @samp{__builtin_expect} are easier to understand.
7058 The default is @option{-fguess-branch-probability} at levels
7059 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7061 @item -freorder-blocks
7062 @opindex freorder-blocks
7063 Reorder basic blocks in the compiled function in order to reduce number of
7064 taken branches and improve code locality.
7066 Enabled at levels @option{-O2}, @option{-O3}.
7068 @item -freorder-blocks-and-partition
7069 @opindex freorder-blocks-and-partition
7070 In addition to reordering basic blocks in the compiled function, in order
7071 to reduce number of taken branches, partitions hot and cold basic blocks
7072 into separate sections of the assembly and .o files, to improve
7073 paging and cache locality performance.
7075 This optimization is automatically turned off in the presence of
7076 exception handling, for linkonce sections, for functions with a user-defined
7077 section attribute and on any architecture that does not support named
7080 @item -freorder-functions
7081 @opindex freorder-functions
7082 Reorder functions in the object file in order to
7083 improve code locality. This is implemented by using special
7084 subsections @code{.text.hot} for most frequently executed functions and
7085 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
7086 the linker so object file format must support named sections and linker must
7087 place them in a reasonable way.
7089 Also profile feedback must be available in to make this option effective. See
7090 @option{-fprofile-arcs} for details.
7092 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7094 @item -fstrict-aliasing
7095 @opindex fstrict-aliasing
7096 Allow the compiler to assume the strictest aliasing rules applicable to
7097 the language being compiled. For C (and C++), this activates
7098 optimizations based on the type of expressions. In particular, an
7099 object of one type is assumed never to reside at the same address as an
7100 object of a different type, unless the types are almost the same. For
7101 example, an @code{unsigned int} can alias an @code{int}, but not a
7102 @code{void*} or a @code{double}. A character type may alias any other
7105 @anchor{Type-punning}Pay special attention to code like this:
7118 The practice of reading from a different union member than the one most
7119 recently written to (called ``type-punning'') is common. Even with
7120 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7121 is accessed through the union type. So, the code above will work as
7122 expected. @xref{Structures unions enumerations and bit-fields
7123 implementation}. However, this code might not:
7134 Similarly, access by taking the address, casting the resulting pointer
7135 and dereferencing the result has undefined behavior, even if the cast
7136 uses a union type, e.g.:
7140 return ((union a_union *) &d)->i;
7144 The @option{-fstrict-aliasing} option is enabled at levels
7145 @option{-O2}, @option{-O3}, @option{-Os}.
7147 @item -fstrict-overflow
7148 @opindex fstrict-overflow
7149 Allow the compiler to assume strict signed overflow rules, depending
7150 on the language being compiled. For C (and C++) this means that
7151 overflow when doing arithmetic with signed numbers is undefined, which
7152 means that the compiler may assume that it will not happen. This
7153 permits various optimizations. For example, the compiler will assume
7154 that an expression like @code{i + 10 > i} will always be true for
7155 signed @code{i}. This assumption is only valid if signed overflow is
7156 undefined, as the expression is false if @code{i + 10} overflows when
7157 using twos complement arithmetic. When this option is in effect any
7158 attempt to determine whether an operation on signed numbers will
7159 overflow must be written carefully to not actually involve overflow.
7161 This option also allows the compiler to assume strict pointer
7162 semantics: given a pointer to an object, if adding an offset to that
7163 pointer does not produce a pointer to the same object, the addition is
7164 undefined. This permits the compiler to conclude that @code{p + u >
7165 p} is always true for a pointer @code{p} and unsigned integer
7166 @code{u}. This assumption is only valid because pointer wraparound is
7167 undefined, as the expression is false if @code{p + u} overflows using
7168 twos complement arithmetic.
7170 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7171 that integer signed overflow is fully defined: it wraps. When
7172 @option{-fwrapv} is used, there is no difference between
7173 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7174 integers. With @option{-fwrapv} certain types of overflow are
7175 permitted. For example, if the compiler gets an overflow when doing
7176 arithmetic on constants, the overflowed value can still be used with
7177 @option{-fwrapv}, but not otherwise.
7179 The @option{-fstrict-overflow} option is enabled at levels
7180 @option{-O2}, @option{-O3}, @option{-Os}.
7182 @item -falign-functions
7183 @itemx -falign-functions=@var{n}
7184 @opindex falign-functions
7185 Align the start of functions to the next power-of-two greater than
7186 @var{n}, skipping up to @var{n} bytes. For instance,
7187 @option{-falign-functions=32} aligns functions to the next 32-byte
7188 boundary, but @option{-falign-functions=24} would align to the next
7189 32-byte boundary only if this can be done by skipping 23 bytes or less.
7191 @option{-fno-align-functions} and @option{-falign-functions=1} are
7192 equivalent and mean that functions will not be aligned.
7194 Some assemblers only support this flag when @var{n} is a power of two;
7195 in that case, it is rounded up.
7197 If @var{n} is not specified or is zero, use a machine-dependent default.
7199 Enabled at levels @option{-O2}, @option{-O3}.
7201 @item -falign-labels
7202 @itemx -falign-labels=@var{n}
7203 @opindex falign-labels
7204 Align all branch targets to a power-of-two boundary, skipping up to
7205 @var{n} bytes like @option{-falign-functions}. This option can easily
7206 make code slower, because it must insert dummy operations for when the
7207 branch target is reached in the usual flow of the code.
7209 @option{-fno-align-labels} and @option{-falign-labels=1} are
7210 equivalent and mean that labels will not be aligned.
7212 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7213 are greater than this value, then their values are used instead.
7215 If @var{n} is not specified or is zero, use a machine-dependent default
7216 which is very likely to be @samp{1}, meaning no alignment.
7218 Enabled at levels @option{-O2}, @option{-O3}.
7221 @itemx -falign-loops=@var{n}
7222 @opindex falign-loops
7223 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7224 like @option{-falign-functions}. The hope is that the loop will be
7225 executed many times, which will make up for any execution of the dummy
7228 @option{-fno-align-loops} and @option{-falign-loops=1} are
7229 equivalent and mean that loops will not be aligned.
7231 If @var{n} is not specified or is zero, use a machine-dependent default.
7233 Enabled at levels @option{-O2}, @option{-O3}.
7236 @itemx -falign-jumps=@var{n}
7237 @opindex falign-jumps
7238 Align branch targets to a power-of-two boundary, for branch targets
7239 where the targets can only be reached by jumping, skipping up to @var{n}
7240 bytes like @option{-falign-functions}. In this case, no dummy operations
7243 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7244 equivalent and mean that loops will not be aligned.
7246 If @var{n} is not specified or is zero, use a machine-dependent default.
7248 Enabled at levels @option{-O2}, @option{-O3}.
7250 @item -funit-at-a-time
7251 @opindex funit-at-a-time
7252 This option is left for compatibility reasons. @option{-funit-at-a-time}
7253 has no effect, while @option{-fno-unit-at-a-time} implies
7254 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7258 @item -fno-toplevel-reorder
7259 @opindex fno-toplevel-reorder
7260 Do not reorder top-level functions, variables, and @code{asm}
7261 statements. Output them in the same order that they appear in the
7262 input file. When this option is used, unreferenced static variables
7263 will not be removed. This option is intended to support existing code
7264 which relies on a particular ordering. For new code, it is better to
7267 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7268 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7273 Constructs webs as commonly used for register allocation purposes and assign
7274 each web individual pseudo register. This allows the register allocation pass
7275 to operate on pseudos directly, but also strengthens several other optimization
7276 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7277 however, make debugging impossible, since variables will no longer stay in a
7280 Enabled by default with @option{-funroll-loops}.
7282 @item -fwhole-program
7283 @opindex fwhole-program
7284 Assume that the current compilation unit represents the whole program being
7285 compiled. All public functions and variables with the exception of @code{main}
7286 and those merged by attribute @code{externally_visible} become static functions
7287 and in effect are optimized more aggressively by interprocedural optimizers.
7288 While this option is equivalent to proper use of the @code{static} keyword for
7289 programs consisting of a single file, in combination with option
7290 @option{-combine}, @option{-flto} or @option{-fwhopr} this flag can be used to
7291 compile many smaller scale programs since the functions and variables become
7292 local for the whole combined compilation unit, not for the single source file
7295 This option implies @option{-fwhole-file} for Fortran programs.
7299 This option runs the standard link-time optimizer. When invoked
7300 with source code, it generates GIMPLE (one of GCC's internal
7301 representations) and writes it to special ELF sections in the object
7302 file. When the object files are linked together, all the function
7303 bodies are read from these ELF sections and instantiated as if they
7304 had been part of the same translation unit.
7306 To use the link-timer optimizer, @option{-flto} needs to be specified at
7307 compile time and during the final link. For example,
7310 gcc -c -O2 -flto foo.c
7311 gcc -c -O2 -flto bar.c
7312 gcc -o myprog -flto -O2 foo.o bar.o
7315 The first two invocations to GCC will save a bytecode representation
7316 of GIMPLE into special ELF sections inside @file{foo.o} and
7317 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7318 @file{foo.o} and @file{bar.o}, merge the two files into a single
7319 internal image, and compile the result as usual. Since both
7320 @file{foo.o} and @file{bar.o} are merged into a single image, this
7321 causes all the inter-procedural analyses and optimizations in GCC to
7322 work across the two files as if they were a single one. This means,
7323 for example, that the inliner will be able to inline functions in
7324 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7326 Another (simpler) way to enable link-time optimization is,
7329 gcc -o myprog -flto -O2 foo.c bar.c
7332 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7333 merge them together into a single GIMPLE representation and optimize
7334 them as usual to produce @file{myprog}.
7336 The only important thing to keep in mind is that to enable link-time
7337 optimizations the @option{-flto} flag needs to be passed to both the
7338 compile and the link commands.
7340 Note that when a file is compiled with @option{-flto}, the generated
7341 object file will be larger than a regular object file because it will
7342 contain GIMPLE bytecodes and the usual final code. This means that
7343 object files with LTO information can be linked as a normal object
7344 file. So, in the previous example, if the final link is done with
7347 gcc -o myprog foo.o bar.o
7350 The only difference will be that no inter-procedural optimizations
7351 will be applied to produce @file{myprog}. The two object files
7352 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7355 Additionally, the optimization flags used to compile individual files
7356 are not necessarily related to those used at link-time. For instance,
7359 gcc -c -O0 -flto foo.c
7360 gcc -c -O0 -flto bar.c
7361 gcc -o myprog -flto -O3 foo.o bar.o
7364 This will produce individual object files with unoptimized assembler
7365 code, but the resulting binary @file{myprog} will be optimized at
7366 @option{-O3}. Now, if the final binary is generated without
7367 @option{-flto}, then @file{myprog} will not be optimized.
7369 When producing the final binary with @option{-flto}, GCC will only
7370 apply link-time optimizations to those files that contain bytecode.
7371 Therefore, you can mix and match object files and libraries with
7372 GIMPLE bytecodes and final object code. GCC will automatically select
7373 which files to optimize in LTO mode and which files to link without
7376 There are some code generation flags that GCC will preserve when
7377 generating bytecodes, as they need to be used during the final link
7378 stage. Currently, the following options are saved into the GIMPLE
7379 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7380 @option{-m} target flags.
7382 At link time, these options are read-in and reapplied. Note that the
7383 current implementation makes no attempt at recognizing conflicting
7384 values for these options. If two or more files have a conflicting
7385 value (e.g., one file is compiled with @option{-fPIC} and another
7386 isn't), the compiler will simply use the last value read from the
7387 bytecode files. It is recommended, then, that all the files
7388 participating in the same link be compiled with the same options.
7390 Another feature of LTO is that it is possible to apply interprocedural
7391 optimizations on files written in different languages. This requires
7392 some support in the language front end. Currently, the C, C++ and
7393 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7394 something like this should work
7399 gfortran -c -flto baz.f90
7400 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7403 Notice that the final link is done with @command{g++} to get the C++
7404 runtime libraries and @option{-lgfortran} is added to get the Fortran
7405 runtime libraries. In general, when mixing languages in LTO mode, you
7406 should use the same link command used when mixing languages in a
7407 regular (non-LTO) compilation. This means that if your build process
7408 was mixing languages before, all you need to add is @option{-flto} to
7409 all the compile and link commands.
7411 If LTO encounters objects with C linkage declared with incompatible
7412 types in separate translation units to be linked together (undefined
7413 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7414 issued. The behavior is still undefined at runtime.
7416 If object files containing GIMPLE bytecode are stored in a library
7417 archive, say @file{libfoo.a}, it is possible to extract and use them
7418 in an LTO link if you are using @command{gold} as the linker (which,
7419 in turn requires GCC to be configured with @option{--enable-gold}).
7420 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7424 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7427 With the linker plugin enabled, @command{gold} will extract the needed
7428 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7429 to make them part of the aggregated GIMPLE image to be optimized.
7431 If you are not using @command{gold} and/or do not specify
7432 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7433 will be extracted and linked as usual, but they will not participate
7434 in the LTO optimization process.
7436 Link time optimizations do not require the presence of the whole
7437 program to operate. If the program does not require any symbols to
7438 be exported, it is possible to combine @option{-flto} and
7439 @option{-fwhopr} with @option{-fwhole-program} to allow the
7440 interprocedural optimizers to use more aggressive assumptions which
7441 may lead to improved optimization opportunities.
7443 Regarding portability: the current implementation of LTO makes no
7444 attempt at generating bytecode that can be ported between different
7445 types of hosts. The bytecode files are versioned and there is a
7446 strict version check, so bytecode files generated in one version of
7447 GCC will not work with an older/newer version of GCC.
7449 Link time optimization does not play well with generating debugging
7450 information. Combining @option{-flto} or @option{-fwhopr} with
7451 @option{-g} is experimental.
7453 This option is disabled by default.
7455 @item -fwhopr[=@var{n}]
7457 This option is identical in functionality to @option{-flto} but it
7458 differs in how the final link stage is executed. Instead of loading
7459 all the function bodies in memory, the callgraph is analyzed and
7460 optimization decisions are made (whole program analysis or WPA). Once
7461 optimization decisions are made, the callgraph is partitioned and the
7462 different sections are compiled separately (local transformations or
7463 LTRANS)@. This process allows optimizations on very large programs
7464 that otherwise would not fit in memory. This option enables
7465 @option{-fwpa} and @option{-fltrans} automatically.
7467 If you specify the optional @var{n} the link stage is executed in
7468 parallel using @var{n} parallel jobs by utilizing an installed
7469 @command{make} program. The environment variable @env{MAKE} may be
7470 used to override the program used.
7472 Disabled by default.
7476 This is an internal option used by GCC when compiling with
7477 @option{-fwhopr}. You should never need to use it.
7479 This option runs the link-time optimizer in the whole-program-analysis
7480 (WPA) mode, which reads in summary information from all inputs and
7481 performs a whole-program analysis based on summary information only.
7482 It generates object files for subsequent runs of the link-time
7483 optimizer where individual object files are optimized using both
7484 summary information from the WPA mode and the actual function bodies.
7485 It then drives the LTRANS phase.
7487 Disabled by default.
7491 This is an internal option used by GCC when compiling with
7492 @option{-fwhopr}. You should never need to use it.
7494 This option runs the link-time optimizer in the local-transformation (LTRANS)
7495 mode, which reads in output from a previous run of the LTO in WPA mode.
7496 In the LTRANS mode, LTO optimizes an object and produces the final assembly.
7498 Disabled by default.
7500 @item -fltrans-output-list=@var{file}
7501 @opindex fltrans-output-list
7502 This is an internal option used by GCC when compiling with
7503 @option{-fwhopr}. You should never need to use it.
7505 This option specifies a file to which the names of LTRANS output files are
7506 written. This option is only meaningful in conjunction with @option{-fwpa}.
7508 Disabled by default.
7510 @item -flto-compression-level=@var{n}
7511 This option specifies the level of compression used for intermediate
7512 language written to LTO object files, and is only meaningful in
7513 conjunction with LTO mode (@option{-fwhopr}, @option{-flto}). Valid
7514 values are 0 (no compression) to 9 (maximum compression). Values
7515 outside this range are clamped to either 0 or 9. If the option is not
7516 given, a default balanced compression setting is used.
7519 Prints a report with internal details on the workings of the link-time
7520 optimizer. The contents of this report vary from version to version,
7521 it is meant to be useful to GCC developers when processing object
7522 files in LTO mode (via @option{-fwhopr} or @option{-flto}).
7524 Disabled by default.
7526 @item -fuse-linker-plugin
7527 Enables the extraction of objects with GIMPLE bytecode information
7528 from library archives. This option relies on features available only
7529 in @command{gold}, so to use this you must configure GCC with
7530 @option{--enable-gold}. See @option{-flto} for a description on the
7531 effect of this flag and how to use it.
7533 Disabled by default.
7535 @item -fcprop-registers
7536 @opindex fcprop-registers
7537 After register allocation and post-register allocation instruction splitting,
7538 we perform a copy-propagation pass to try to reduce scheduling dependencies
7539 and occasionally eliminate the copy.
7541 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7543 @item -fprofile-correction
7544 @opindex fprofile-correction
7545 Profiles collected using an instrumented binary for multi-threaded programs may
7546 be inconsistent due to missed counter updates. When this option is specified,
7547 GCC will use heuristics to correct or smooth out such inconsistencies. By
7548 default, GCC will emit an error message when an inconsistent profile is detected.
7550 @item -fprofile-dir=@var{path}
7551 @opindex fprofile-dir
7553 Set the directory to search the profile data files in to @var{path}.
7554 This option affects only the profile data generated by
7555 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7556 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7557 and its related options.
7558 By default, GCC will use the current directory as @var{path}
7559 thus the profile data file will appear in the same directory as the object file.
7561 @item -fprofile-generate
7562 @itemx -fprofile-generate=@var{path}
7563 @opindex fprofile-generate
7565 Enable options usually used for instrumenting application to produce
7566 profile useful for later recompilation with profile feedback based
7567 optimization. You must use @option{-fprofile-generate} both when
7568 compiling and when linking your program.
7570 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7572 If @var{path} is specified, GCC will look at the @var{path} to find
7573 the profile feedback data files. See @option{-fprofile-dir}.
7576 @itemx -fprofile-use=@var{path}
7577 @opindex fprofile-use
7578 Enable profile feedback directed optimizations, and optimizations
7579 generally profitable only with profile feedback available.
7581 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7582 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7584 By default, GCC emits an error message if the feedback profiles do not
7585 match the source code. This error can be turned into a warning by using
7586 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7589 If @var{path} is specified, GCC will look at the @var{path} to find
7590 the profile feedback data files. See @option{-fprofile-dir}.
7593 The following options control compiler behavior regarding floating
7594 point arithmetic. These options trade off between speed and
7595 correctness. All must be specifically enabled.
7599 @opindex ffloat-store
7600 Do not store floating point variables in registers, and inhibit other
7601 options that might change whether a floating point value is taken from a
7604 @cindex floating point precision
7605 This option prevents undesirable excess precision on machines such as
7606 the 68000 where the floating registers (of the 68881) keep more
7607 precision than a @code{double} is supposed to have. Similarly for the
7608 x86 architecture. For most programs, the excess precision does only
7609 good, but a few programs rely on the precise definition of IEEE floating
7610 point. Use @option{-ffloat-store} for such programs, after modifying
7611 them to store all pertinent intermediate computations into variables.
7613 @item -fexcess-precision=@var{style}
7614 @opindex fexcess-precision
7615 This option allows further control over excess precision on machines
7616 where floating-point registers have more precision than the IEEE
7617 @code{float} and @code{double} types and the processor does not
7618 support operations rounding to those types. By default,
7619 @option{-fexcess-precision=fast} is in effect; this means that
7620 operations are carried out in the precision of the registers and that
7621 it is unpredictable when rounding to the types specified in the source
7622 code takes place. When compiling C, if
7623 @option{-fexcess-precision=standard} is specified then excess
7624 precision will follow the rules specified in ISO C99; in particular,
7625 both casts and assignments cause values to be rounded to their
7626 semantic types (whereas @option{-ffloat-store} only affects
7627 assignments). This option is enabled by default for C if a strict
7628 conformance option such as @option{-std=c99} is used.
7631 @option{-fexcess-precision=standard} is not implemented for languages
7632 other than C, and has no effect if
7633 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7634 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7635 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7636 semantics apply without excess precision, and in the latter, rounding
7641 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7642 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7643 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7645 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7647 This option is not turned on by any @option{-O} option since
7648 it can result in incorrect output for programs which depend on
7649 an exact implementation of IEEE or ISO rules/specifications for
7650 math functions. It may, however, yield faster code for programs
7651 that do not require the guarantees of these specifications.
7653 @item -fno-math-errno
7654 @opindex fno-math-errno
7655 Do not set ERRNO after calling math functions that are executed
7656 with a single instruction, e.g., sqrt. A program that relies on
7657 IEEE exceptions for math error handling may want to use this flag
7658 for speed while maintaining IEEE arithmetic compatibility.
7660 This option is not turned on by any @option{-O} option since
7661 it can result in incorrect output for programs which depend on
7662 an exact implementation of IEEE or ISO rules/specifications for
7663 math functions. It may, however, yield faster code for programs
7664 that do not require the guarantees of these specifications.
7666 The default is @option{-fmath-errno}.
7668 On Darwin systems, the math library never sets @code{errno}. There is
7669 therefore no reason for the compiler to consider the possibility that
7670 it might, and @option{-fno-math-errno} is the default.
7672 @item -funsafe-math-optimizations
7673 @opindex funsafe-math-optimizations
7675 Allow optimizations for floating-point arithmetic that (a) assume
7676 that arguments and results are valid and (b) may violate IEEE or
7677 ANSI standards. When used at link-time, it may include libraries
7678 or startup files that change the default FPU control word or other
7679 similar optimizations.
7681 This option is not turned on by any @option{-O} option since
7682 it can result in incorrect output for programs which depend on
7683 an exact implementation of IEEE or ISO rules/specifications for
7684 math functions. It may, however, yield faster code for programs
7685 that do not require the guarantees of these specifications.
7686 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7687 @option{-fassociative-math} and @option{-freciprocal-math}.
7689 The default is @option{-fno-unsafe-math-optimizations}.
7691 @item -fassociative-math
7692 @opindex fassociative-math
7694 Allow re-association of operands in series of floating-point operations.
7695 This violates the ISO C and C++ language standard by possibly changing
7696 computation result. NOTE: re-ordering may change the sign of zero as
7697 well as ignore NaNs and inhibit or create underflow or overflow (and
7698 thus cannot be used on a code which relies on rounding behavior like
7699 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7700 and thus may not be used when ordered comparisons are required.
7701 This option requires that both @option{-fno-signed-zeros} and
7702 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7703 much sense with @option{-frounding-math}. For Fortran the option
7704 is automatically enabled when both @option{-fno-signed-zeros} and
7705 @option{-fno-trapping-math} are in effect.
7707 The default is @option{-fno-associative-math}.
7709 @item -freciprocal-math
7710 @opindex freciprocal-math
7712 Allow the reciprocal of a value to be used instead of dividing by
7713 the value if this enables optimizations. For example @code{x / y}
7714 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7715 is subject to common subexpression elimination. Note that this loses
7716 precision and increases the number of flops operating on the value.
7718 The default is @option{-fno-reciprocal-math}.
7720 @item -ffinite-math-only
7721 @opindex ffinite-math-only
7722 Allow optimizations for floating-point arithmetic that assume
7723 that arguments and results are not NaNs or +-Infs.
7725 This option is not turned on by any @option{-O} option since
7726 it can result in incorrect output for programs which depend on
7727 an exact implementation of IEEE or ISO rules/specifications for
7728 math functions. It may, however, yield faster code for programs
7729 that do not require the guarantees of these specifications.
7731 The default is @option{-fno-finite-math-only}.
7733 @item -fno-signed-zeros
7734 @opindex fno-signed-zeros
7735 Allow optimizations for floating point arithmetic that ignore the
7736 signedness of zero. IEEE arithmetic specifies the behavior of
7737 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7738 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7739 This option implies that the sign of a zero result isn't significant.
7741 The default is @option{-fsigned-zeros}.
7743 @item -fno-trapping-math
7744 @opindex fno-trapping-math
7745 Compile code assuming that floating-point operations cannot generate
7746 user-visible traps. These traps include division by zero, overflow,
7747 underflow, inexact result and invalid operation. This option requires
7748 that @option{-fno-signaling-nans} be in effect. Setting this option may
7749 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7751 This option should never be turned on by any @option{-O} option since
7752 it can result in incorrect output for programs which depend on
7753 an exact implementation of IEEE or ISO rules/specifications for
7756 The default is @option{-ftrapping-math}.
7758 @item -frounding-math
7759 @opindex frounding-math
7760 Disable transformations and optimizations that assume default floating
7761 point rounding behavior. This is round-to-zero for all floating point
7762 to integer conversions, and round-to-nearest for all other arithmetic
7763 truncations. This option should be specified for programs that change
7764 the FP rounding mode dynamically, or that may be executed with a
7765 non-default rounding mode. This option disables constant folding of
7766 floating point expressions at compile-time (which may be affected by
7767 rounding mode) and arithmetic transformations that are unsafe in the
7768 presence of sign-dependent rounding modes.
7770 The default is @option{-fno-rounding-math}.
7772 This option is experimental and does not currently guarantee to
7773 disable all GCC optimizations that are affected by rounding mode.
7774 Future versions of GCC may provide finer control of this setting
7775 using C99's @code{FENV_ACCESS} pragma. This command line option
7776 will be used to specify the default state for @code{FENV_ACCESS}.
7778 @item -fsignaling-nans
7779 @opindex fsignaling-nans
7780 Compile code assuming that IEEE signaling NaNs may generate user-visible
7781 traps during floating-point operations. Setting this option disables
7782 optimizations that may change the number of exceptions visible with
7783 signaling NaNs. This option implies @option{-ftrapping-math}.
7785 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7788 The default is @option{-fno-signaling-nans}.
7790 This option is experimental and does not currently guarantee to
7791 disable all GCC optimizations that affect signaling NaN behavior.
7793 @item -fsingle-precision-constant
7794 @opindex fsingle-precision-constant
7795 Treat floating point constant as single precision constant instead of
7796 implicitly converting it to double precision constant.
7798 @item -fcx-limited-range
7799 @opindex fcx-limited-range
7800 When enabled, this option states that a range reduction step is not
7801 needed when performing complex division. Also, there is no checking
7802 whether the result of a complex multiplication or division is @code{NaN
7803 + I*NaN}, with an attempt to rescue the situation in that case. The
7804 default is @option{-fno-cx-limited-range}, but is enabled by
7805 @option{-ffast-math}.
7807 This option controls the default setting of the ISO C99
7808 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7811 @item -fcx-fortran-rules
7812 @opindex fcx-fortran-rules
7813 Complex multiplication and division follow Fortran rules. Range
7814 reduction is done as part of complex division, but there is no checking
7815 whether the result of a complex multiplication or division is @code{NaN
7816 + I*NaN}, with an attempt to rescue the situation in that case.
7818 The default is @option{-fno-cx-fortran-rules}.
7822 The following options control optimizations that may improve
7823 performance, but are not enabled by any @option{-O} options. This
7824 section includes experimental options that may produce broken code.
7827 @item -fbranch-probabilities
7828 @opindex fbranch-probabilities
7829 After running a program compiled with @option{-fprofile-arcs}
7830 (@pxref{Debugging Options,, Options for Debugging Your Program or
7831 @command{gcc}}), you can compile it a second time using
7832 @option{-fbranch-probabilities}, to improve optimizations based on
7833 the number of times each branch was taken. When the program
7834 compiled with @option{-fprofile-arcs} exits it saves arc execution
7835 counts to a file called @file{@var{sourcename}.gcda} for each source
7836 file. The information in this data file is very dependent on the
7837 structure of the generated code, so you must use the same source code
7838 and the same optimization options for both compilations.
7840 With @option{-fbranch-probabilities}, GCC puts a
7841 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7842 These can be used to improve optimization. Currently, they are only
7843 used in one place: in @file{reorg.c}, instead of guessing which path a
7844 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7845 exactly determine which path is taken more often.
7847 @item -fprofile-values
7848 @opindex fprofile-values
7849 If combined with @option{-fprofile-arcs}, it adds code so that some
7850 data about values of expressions in the program is gathered.
7852 With @option{-fbranch-probabilities}, it reads back the data gathered
7853 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7854 notes to instructions for their later usage in optimizations.
7856 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7860 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7861 a code to gather information about values of expressions.
7863 With @option{-fbranch-probabilities}, it reads back the data gathered
7864 and actually performs the optimizations based on them.
7865 Currently the optimizations include specialization of division operation
7866 using the knowledge about the value of the denominator.
7868 @item -frename-registers
7869 @opindex frename-registers
7870 Attempt to avoid false dependencies in scheduled code by making use
7871 of registers left over after register allocation. This optimization
7872 will most benefit processors with lots of registers. Depending on the
7873 debug information format adopted by the target, however, it can
7874 make debugging impossible, since variables will no longer stay in
7875 a ``home register''.
7877 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
7881 Perform tail duplication to enlarge superblock size. This transformation
7882 simplifies the control flow of the function allowing other optimizations to do
7885 Enabled with @option{-fprofile-use}.
7887 @item -funroll-loops
7888 @opindex funroll-loops
7889 Unroll loops whose number of iterations can be determined at compile time or
7890 upon entry to the loop. @option{-funroll-loops} implies
7891 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7892 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7893 small constant number of iterations). This option makes code larger, and may
7894 or may not make it run faster.
7896 Enabled with @option{-fprofile-use}.
7898 @item -funroll-all-loops
7899 @opindex funroll-all-loops
7900 Unroll all loops, even if their number of iterations is uncertain when
7901 the loop is entered. This usually makes programs run more slowly.
7902 @option{-funroll-all-loops} implies the same options as
7903 @option{-funroll-loops}.
7906 @opindex fpeel-loops
7907 Peels the loops for that there is enough information that they do not
7908 roll much (from profile feedback). It also turns on complete loop peeling
7909 (i.e.@: complete removal of loops with small constant number of iterations).
7911 Enabled with @option{-fprofile-use}.
7913 @item -fmove-loop-invariants
7914 @opindex fmove-loop-invariants
7915 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7916 at level @option{-O1}
7918 @item -funswitch-loops
7919 @opindex funswitch-loops
7920 Move branches with loop invariant conditions out of the loop, with duplicates
7921 of the loop on both branches (modified according to result of the condition).
7923 @item -ffunction-sections
7924 @itemx -fdata-sections
7925 @opindex ffunction-sections
7926 @opindex fdata-sections
7927 Place each function or data item into its own section in the output
7928 file if the target supports arbitrary sections. The name of the
7929 function or the name of the data item determines the section's name
7932 Use these options on systems where the linker can perform optimizations
7933 to improve locality of reference in the instruction space. Most systems
7934 using the ELF object format and SPARC processors running Solaris 2 have
7935 linkers with such optimizations. AIX may have these optimizations in
7938 Only use these options when there are significant benefits from doing
7939 so. When you specify these options, the assembler and linker will
7940 create larger object and executable files and will also be slower.
7941 You will not be able to use @code{gprof} on all systems if you
7942 specify this option and you may have problems with debugging if
7943 you specify both this option and @option{-g}.
7945 @item -fbranch-target-load-optimize
7946 @opindex fbranch-target-load-optimize
7947 Perform branch target register load optimization before prologue / epilogue
7949 The use of target registers can typically be exposed only during reload,
7950 thus hoisting loads out of loops and doing inter-block scheduling needs
7951 a separate optimization pass.
7953 @item -fbranch-target-load-optimize2
7954 @opindex fbranch-target-load-optimize2
7955 Perform branch target register load optimization after prologue / epilogue
7958 @item -fbtr-bb-exclusive
7959 @opindex fbtr-bb-exclusive
7960 When performing branch target register load optimization, don't reuse
7961 branch target registers in within any basic block.
7963 @item -fstack-protector
7964 @opindex fstack-protector
7965 Emit extra code to check for buffer overflows, such as stack smashing
7966 attacks. This is done by adding a guard variable to functions with
7967 vulnerable objects. This includes functions that call alloca, and
7968 functions with buffers larger than 8 bytes. The guards are initialized
7969 when a function is entered and then checked when the function exits.
7970 If a guard check fails, an error message is printed and the program exits.
7972 @item -fstack-protector-all
7973 @opindex fstack-protector-all
7974 Like @option{-fstack-protector} except that all functions are protected.
7976 @item -fsection-anchors
7977 @opindex fsection-anchors
7978 Try to reduce the number of symbolic address calculations by using
7979 shared ``anchor'' symbols to address nearby objects. This transformation
7980 can help to reduce the number of GOT entries and GOT accesses on some
7983 For example, the implementation of the following function @code{foo}:
7987 int foo (void) @{ return a + b + c; @}
7990 would usually calculate the addresses of all three variables, but if you
7991 compile it with @option{-fsection-anchors}, it will access the variables
7992 from a common anchor point instead. The effect is similar to the
7993 following pseudocode (which isn't valid C):
7998 register int *xr = &x;
7999 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
8003 Not all targets support this option.
8005 @item --param @var{name}=@var{value}
8007 In some places, GCC uses various constants to control the amount of
8008 optimization that is done. For example, GCC will not inline functions
8009 that contain more that a certain number of instructions. You can
8010 control some of these constants on the command-line using the
8011 @option{--param} option.
8013 The names of specific parameters, and the meaning of the values, are
8014 tied to the internals of the compiler, and are subject to change
8015 without notice in future releases.
8017 In each case, the @var{value} is an integer. The allowable choices for
8018 @var{name} are given in the following table:
8021 @item struct-reorg-cold-struct-ratio
8022 The threshold ratio (as a percentage) between a structure frequency
8023 and the frequency of the hottest structure in the program. This parameter
8024 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
8025 We say that if the ratio of a structure frequency, calculated by profiling,
8026 to the hottest structure frequency in the program is less than this
8027 parameter, then structure reorganization is not applied to this structure.
8030 @item predictable-branch-outcome
8031 When branch is predicted to be taken with probability lower than this threshold
8032 (in percent), then it is considered well predictable. The default is 10.
8034 @item max-crossjump-edges
8035 The maximum number of incoming edges to consider for crossjumping.
8036 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8037 the number of edges incoming to each block. Increasing values mean
8038 more aggressive optimization, making the compile time increase with
8039 probably small improvement in executable size.
8041 @item min-crossjump-insns
8042 The minimum number of instructions which must be matched at the end
8043 of two blocks before crossjumping will be performed on them. This
8044 value is ignored in the case where all instructions in the block being
8045 crossjumped from are matched. The default value is 5.
8047 @item max-grow-copy-bb-insns
8048 The maximum code size expansion factor when copying basic blocks
8049 instead of jumping. The expansion is relative to a jump instruction.
8050 The default value is 8.
8052 @item max-goto-duplication-insns
8053 The maximum number of instructions to duplicate to a block that jumps
8054 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
8055 passes, GCC factors computed gotos early in the compilation process,
8056 and unfactors them as late as possible. Only computed jumps at the
8057 end of a basic blocks with no more than max-goto-duplication-insns are
8058 unfactored. The default value is 8.
8060 @item max-delay-slot-insn-search
8061 The maximum number of instructions to consider when looking for an
8062 instruction to fill a delay slot. If more than this arbitrary number of
8063 instructions is searched, the time savings from filling the delay slot
8064 will be minimal so stop searching. Increasing values mean more
8065 aggressive optimization, making the compile time increase with probably
8066 small improvement in executable run time.
8068 @item max-delay-slot-live-search
8069 When trying to fill delay slots, the maximum number of instructions to
8070 consider when searching for a block with valid live register
8071 information. Increasing this arbitrarily chosen value means more
8072 aggressive optimization, increasing the compile time. This parameter
8073 should be removed when the delay slot code is rewritten to maintain the
8076 @item max-gcse-memory
8077 The approximate maximum amount of memory that will be allocated in
8078 order to perform the global common subexpression elimination
8079 optimization. If more memory than specified is required, the
8080 optimization will not be done.
8082 @item max-pending-list-length
8083 The maximum number of pending dependencies scheduling will allow
8084 before flushing the current state and starting over. Large functions
8085 with few branches or calls can create excessively large lists which
8086 needlessly consume memory and resources.
8088 @item max-inline-insns-single
8089 Several parameters control the tree inliner used in gcc.
8090 This number sets the maximum number of instructions (counted in GCC's
8091 internal representation) in a single function that the tree inliner
8092 will consider for inlining. This only affects functions declared
8093 inline and methods implemented in a class declaration (C++).
8094 The default value is 300.
8096 @item max-inline-insns-auto
8097 When you use @option{-finline-functions} (included in @option{-O3}),
8098 a lot of functions that would otherwise not be considered for inlining
8099 by the compiler will be investigated. To those functions, a different
8100 (more restrictive) limit compared to functions declared inline can
8102 The default value is 50.
8104 @item large-function-insns
8105 The limit specifying really large functions. For functions larger than this
8106 limit after inlining, inlining is constrained by
8107 @option{--param large-function-growth}. This parameter is useful primarily
8108 to avoid extreme compilation time caused by non-linear algorithms used by the
8110 The default value is 2700.
8112 @item large-function-growth
8113 Specifies maximal growth of large function caused by inlining in percents.
8114 The default value is 100 which limits large function growth to 2.0 times
8117 @item large-unit-insns
8118 The limit specifying large translation unit. Growth caused by inlining of
8119 units larger than this limit is limited by @option{--param inline-unit-growth}.
8120 For small units this might be too tight (consider unit consisting of function A
8121 that is inline and B that just calls A three time. If B is small relative to
8122 A, the growth of unit is 300\% and yet such inlining is very sane. For very
8123 large units consisting of small inlineable functions however the overall unit
8124 growth limit is needed to avoid exponential explosion of code size. Thus for
8125 smaller units, the size is increased to @option{--param large-unit-insns}
8126 before applying @option{--param inline-unit-growth}. The default is 10000
8128 @item inline-unit-growth
8129 Specifies maximal overall growth of the compilation unit caused by inlining.
8130 The default value is 30 which limits unit growth to 1.3 times the original
8133 @item ipcp-unit-growth
8134 Specifies maximal overall growth of the compilation unit caused by
8135 interprocedural constant propagation. The default value is 10 which limits
8136 unit growth to 1.1 times the original size.
8138 @item large-stack-frame
8139 The limit specifying large stack frames. While inlining the algorithm is trying
8140 to not grow past this limit too much. Default value is 256 bytes.
8142 @item large-stack-frame-growth
8143 Specifies maximal growth of large stack frames caused by inlining in percents.
8144 The default value is 1000 which limits large stack frame growth to 11 times
8147 @item max-inline-insns-recursive
8148 @itemx max-inline-insns-recursive-auto
8149 Specifies maximum number of instructions out-of-line copy of self recursive inline
8150 function can grow into by performing recursive inlining.
8152 For functions declared inline @option{--param max-inline-insns-recursive} is
8153 taken into account. For function not declared inline, recursive inlining
8154 happens only when @option{-finline-functions} (included in @option{-O3}) is
8155 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8156 default value is 450.
8158 @item max-inline-recursive-depth
8159 @itemx max-inline-recursive-depth-auto
8160 Specifies maximum recursion depth used by the recursive inlining.
8162 For functions declared inline @option{--param max-inline-recursive-depth} is
8163 taken into account. For function not declared inline, recursive inlining
8164 happens only when @option{-finline-functions} (included in @option{-O3}) is
8165 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8168 @item min-inline-recursive-probability
8169 Recursive inlining is profitable only for function having deep recursion
8170 in average and can hurt for function having little recursion depth by
8171 increasing the prologue size or complexity of function body to other
8174 When profile feedback is available (see @option{-fprofile-generate}) the actual
8175 recursion depth can be guessed from probability that function will recurse via
8176 given call expression. This parameter limits inlining only to call expression
8177 whose probability exceeds given threshold (in percents). The default value is
8180 @item early-inlining-insns
8181 Specify growth that early inliner can make. In effect it increases amount of
8182 inlining for code having large abstraction penalty. The default value is 8.
8184 @item max-early-inliner-iterations
8185 @itemx max-early-inliner-iterations
8186 Limit of iterations of early inliner. This basically bounds number of nested
8187 indirect calls early inliner can resolve. Deeper chains are still handled by
8190 @item min-vect-loop-bound
8191 The minimum number of iterations under which a loop will not get vectorized
8192 when @option{-ftree-vectorize} is used. The number of iterations after
8193 vectorization needs to be greater than the value specified by this option
8194 to allow vectorization. The default value is 0.
8196 @item max-unrolled-insns
8197 The maximum number of instructions that a loop should have if that loop
8198 is unrolled, and if the loop is unrolled, it determines how many times
8199 the loop code is unrolled.
8201 @item max-average-unrolled-insns
8202 The maximum number of instructions biased by probabilities of their execution
8203 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8204 it determines how many times the loop code is unrolled.
8206 @item max-unroll-times
8207 The maximum number of unrollings of a single loop.
8209 @item max-peeled-insns
8210 The maximum number of instructions that a loop should have if that loop
8211 is peeled, and if the loop is peeled, it determines how many times
8212 the loop code is peeled.
8214 @item max-peel-times
8215 The maximum number of peelings of a single loop.
8217 @item max-completely-peeled-insns
8218 The maximum number of insns of a completely peeled loop.
8220 @item max-completely-peel-times
8221 The maximum number of iterations of a loop to be suitable for complete peeling.
8223 @item max-completely-peel-loop-nest-depth
8224 The maximum depth of a loop nest suitable for complete peeling.
8226 @item max-unswitch-insns
8227 The maximum number of insns of an unswitched loop.
8229 @item max-unswitch-level
8230 The maximum number of branches unswitched in a single loop.
8233 The minimum cost of an expensive expression in the loop invariant motion.
8235 @item iv-consider-all-candidates-bound
8236 Bound on number of candidates for induction variables below that
8237 all candidates are considered for each use in induction variable
8238 optimizations. Only the most relevant candidates are considered
8239 if there are more candidates, to avoid quadratic time complexity.
8241 @item iv-max-considered-uses
8242 The induction variable optimizations give up on loops that contain more
8243 induction variable uses.
8245 @item iv-always-prune-cand-set-bound
8246 If number of candidates in the set is smaller than this value,
8247 we always try to remove unnecessary ivs from the set during its
8248 optimization when a new iv is added to the set.
8250 @item scev-max-expr-size
8251 Bound on size of expressions used in the scalar evolutions analyzer.
8252 Large expressions slow the analyzer.
8254 @item omega-max-vars
8255 The maximum number of variables in an Omega constraint system.
8256 The default value is 128.
8258 @item omega-max-geqs
8259 The maximum number of inequalities in an Omega constraint system.
8260 The default value is 256.
8263 The maximum number of equalities in an Omega constraint system.
8264 The default value is 128.
8266 @item omega-max-wild-cards
8267 The maximum number of wildcard variables that the Omega solver will
8268 be able to insert. The default value is 18.
8270 @item omega-hash-table-size
8271 The size of the hash table in the Omega solver. The default value is
8274 @item omega-max-keys
8275 The maximal number of keys used by the Omega solver. The default
8278 @item omega-eliminate-redundant-constraints
8279 When set to 1, use expensive methods to eliminate all redundant
8280 constraints. The default value is 0.
8282 @item vect-max-version-for-alignment-checks
8283 The maximum number of runtime checks that can be performed when
8284 doing loop versioning for alignment in the vectorizer. See option
8285 ftree-vect-loop-version for more information.
8287 @item vect-max-version-for-alias-checks
8288 The maximum number of runtime checks that can be performed when
8289 doing loop versioning for alias in the vectorizer. See option
8290 ftree-vect-loop-version for more information.
8292 @item max-iterations-to-track
8294 The maximum number of iterations of a loop the brute force algorithm
8295 for analysis of # of iterations of the loop tries to evaluate.
8297 @item hot-bb-count-fraction
8298 Select fraction of the maximal count of repetitions of basic block in program
8299 given basic block needs to have to be considered hot.
8301 @item hot-bb-frequency-fraction
8302 Select fraction of the maximal frequency of executions of basic block in
8303 function given basic block needs to have to be considered hot
8305 @item max-predicted-iterations
8306 The maximum number of loop iterations we predict statically. This is useful
8307 in cases where function contain single loop with known bound and other loop
8308 with unknown. We predict the known number of iterations correctly, while
8309 the unknown number of iterations average to roughly 10. This means that the
8310 loop without bounds would appear artificially cold relative to the other one.
8312 @item align-threshold
8314 Select fraction of the maximal frequency of executions of basic block in
8315 function given basic block will get aligned.
8317 @item align-loop-iterations
8319 A loop expected to iterate at lest the selected number of iterations will get
8322 @item tracer-dynamic-coverage
8323 @itemx tracer-dynamic-coverage-feedback
8325 This value is used to limit superblock formation once the given percentage of
8326 executed instructions is covered. This limits unnecessary code size
8329 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8330 feedback is available. The real profiles (as opposed to statically estimated
8331 ones) are much less balanced allowing the threshold to be larger value.
8333 @item tracer-max-code-growth
8334 Stop tail duplication once code growth has reached given percentage. This is
8335 rather hokey argument, as most of the duplicates will be eliminated later in
8336 cross jumping, so it may be set to much higher values than is the desired code
8339 @item tracer-min-branch-ratio
8341 Stop reverse growth when the reverse probability of best edge is less than this
8342 threshold (in percent).
8344 @item tracer-min-branch-ratio
8345 @itemx tracer-min-branch-ratio-feedback
8347 Stop forward growth if the best edge do have probability lower than this
8350 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8351 compilation for profile feedback and one for compilation without. The value
8352 for compilation with profile feedback needs to be more conservative (higher) in
8353 order to make tracer effective.
8355 @item max-cse-path-length
8357 Maximum number of basic blocks on path that cse considers. The default is 10.
8360 The maximum instructions CSE process before flushing. The default is 1000.
8362 @item ggc-min-expand
8364 GCC uses a garbage collector to manage its own memory allocation. This
8365 parameter specifies the minimum percentage by which the garbage
8366 collector's heap should be allowed to expand between collections.
8367 Tuning this may improve compilation speed; it has no effect on code
8370 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8371 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8372 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8373 GCC is not able to calculate RAM on a particular platform, the lower
8374 bound of 30% is used. Setting this parameter and
8375 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8376 every opportunity. This is extremely slow, but can be useful for
8379 @item ggc-min-heapsize
8381 Minimum size of the garbage collector's heap before it begins bothering
8382 to collect garbage. The first collection occurs after the heap expands
8383 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8384 tuning this may improve compilation speed, and has no effect on code
8387 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8388 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8389 with a lower bound of 4096 (four megabytes) and an upper bound of
8390 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8391 particular platform, the lower bound is used. Setting this parameter
8392 very large effectively disables garbage collection. Setting this
8393 parameter and @option{ggc-min-expand} to zero causes a full collection
8394 to occur at every opportunity.
8396 @item max-reload-search-insns
8397 The maximum number of instruction reload should look backward for equivalent
8398 register. Increasing values mean more aggressive optimization, making the
8399 compile time increase with probably slightly better performance. The default
8402 @item max-cselib-memory-locations
8403 The maximum number of memory locations cselib should take into account.
8404 Increasing values mean more aggressive optimization, making the compile time
8405 increase with probably slightly better performance. The default value is 500.
8407 @item reorder-blocks-duplicate
8408 @itemx reorder-blocks-duplicate-feedback
8410 Used by basic block reordering pass to decide whether to use unconditional
8411 branch or duplicate the code on its destination. Code is duplicated when its
8412 estimated size is smaller than this value multiplied by the estimated size of
8413 unconditional jump in the hot spots of the program.
8415 The @option{reorder-block-duplicate-feedback} is used only when profile
8416 feedback is available and may be set to higher values than
8417 @option{reorder-block-duplicate} since information about the hot spots is more
8420 @item max-sched-ready-insns
8421 The maximum number of instructions ready to be issued the scheduler should
8422 consider at any given time during the first scheduling pass. Increasing
8423 values mean more thorough searches, making the compilation time increase
8424 with probably little benefit. The default value is 100.
8426 @item max-sched-region-blocks
8427 The maximum number of blocks in a region to be considered for
8428 interblock scheduling. The default value is 10.
8430 @item max-pipeline-region-blocks
8431 The maximum number of blocks in a region to be considered for
8432 pipelining in the selective scheduler. The default value is 15.
8434 @item max-sched-region-insns
8435 The maximum number of insns in a region to be considered for
8436 interblock scheduling. The default value is 100.
8438 @item max-pipeline-region-insns
8439 The maximum number of insns in a region to be considered for
8440 pipelining in the selective scheduler. The default value is 200.
8443 The minimum probability (in percents) of reaching a source block
8444 for interblock speculative scheduling. The default value is 40.
8446 @item max-sched-extend-regions-iters
8447 The maximum number of iterations through CFG to extend regions.
8448 0 - disable region extension,
8449 N - do at most N iterations.
8450 The default value is 0.
8452 @item max-sched-insn-conflict-delay
8453 The maximum conflict delay for an insn to be considered for speculative motion.
8454 The default value is 3.
8456 @item sched-spec-prob-cutoff
8457 The minimal probability of speculation success (in percents), so that
8458 speculative insn will be scheduled.
8459 The default value is 40.
8461 @item sched-mem-true-dep-cost
8462 Minimal distance (in CPU cycles) between store and load targeting same
8463 memory locations. The default value is 1.
8465 @item selsched-max-lookahead
8466 The maximum size of the lookahead window of selective scheduling. It is a
8467 depth of search for available instructions.
8468 The default value is 50.
8470 @item selsched-max-sched-times
8471 The maximum number of times that an instruction will be scheduled during
8472 selective scheduling. This is the limit on the number of iterations
8473 through which the instruction may be pipelined. The default value is 2.
8475 @item selsched-max-insns-to-rename
8476 The maximum number of best instructions in the ready list that are considered
8477 for renaming in the selective scheduler. The default value is 2.
8479 @item max-last-value-rtl
8480 The maximum size measured as number of RTLs that can be recorded in an expression
8481 in combiner for a pseudo register as last known value of that register. The default
8484 @item integer-share-limit
8485 Small integer constants can use a shared data structure, reducing the
8486 compiler's memory usage and increasing its speed. This sets the maximum
8487 value of a shared integer constant. The default value is 256.
8489 @item min-virtual-mappings
8490 Specifies the minimum number of virtual mappings in the incremental
8491 SSA updater that should be registered to trigger the virtual mappings
8492 heuristic defined by virtual-mappings-ratio. The default value is
8495 @item virtual-mappings-ratio
8496 If the number of virtual mappings is virtual-mappings-ratio bigger
8497 than the number of virtual symbols to be updated, then the incremental
8498 SSA updater switches to a full update for those symbols. The default
8501 @item ssp-buffer-size
8502 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8503 protection when @option{-fstack-protection} is used.
8505 @item max-jump-thread-duplication-stmts
8506 Maximum number of statements allowed in a block that needs to be
8507 duplicated when threading jumps.
8509 @item max-fields-for-field-sensitive
8510 Maximum number of fields in a structure we will treat in
8511 a field sensitive manner during pointer analysis. The default is zero
8512 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8514 @item prefetch-latency
8515 Estimate on average number of instructions that are executed before
8516 prefetch finishes. The distance we prefetch ahead is proportional
8517 to this constant. Increasing this number may also lead to less
8518 streams being prefetched (see @option{simultaneous-prefetches}).
8520 @item simultaneous-prefetches
8521 Maximum number of prefetches that can run at the same time.
8523 @item l1-cache-line-size
8524 The size of cache line in L1 cache, in bytes.
8527 The size of L1 cache, in kilobytes.
8530 The size of L2 cache, in kilobytes.
8532 @item min-insn-to-prefetch-ratio
8533 The minimum ratio between the number of instructions and the
8534 number of prefetches to enable prefetching in a loop.
8536 @item prefetch-min-insn-to-mem-ratio
8537 The minimum ratio between the number of instructions and the
8538 number of memory references to enable prefetching in a loop.
8540 @item use-canonical-types
8541 Whether the compiler should use the ``canonical'' type system. By
8542 default, this should always be 1, which uses a more efficient internal
8543 mechanism for comparing types in C++ and Objective-C++. However, if
8544 bugs in the canonical type system are causing compilation failures,
8545 set this value to 0 to disable canonical types.
8547 @item switch-conversion-max-branch-ratio
8548 Switch initialization conversion will refuse to create arrays that are
8549 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8550 branches in the switch.
8552 @item max-partial-antic-length
8553 Maximum length of the partial antic set computed during the tree
8554 partial redundancy elimination optimization (@option{-ftree-pre}) when
8555 optimizing at @option{-O3} and above. For some sorts of source code
8556 the enhanced partial redundancy elimination optimization can run away,
8557 consuming all of the memory available on the host machine. This
8558 parameter sets a limit on the length of the sets that are computed,
8559 which prevents the runaway behavior. Setting a value of 0 for
8560 this parameter will allow an unlimited set length.
8562 @item sccvn-max-scc-size
8563 Maximum size of a strongly connected component (SCC) during SCCVN
8564 processing. If this limit is hit, SCCVN processing for the whole
8565 function will not be done and optimizations depending on it will
8566 be disabled. The default maximum SCC size is 10000.
8568 @item ira-max-loops-num
8569 IRA uses a regional register allocation by default. If a function
8570 contains loops more than number given by the parameter, only at most
8571 given number of the most frequently executed loops will form regions
8572 for the regional register allocation. The default value of the
8575 @item ira-max-conflict-table-size
8576 Although IRA uses a sophisticated algorithm of compression conflict
8577 table, the table can be still big for huge functions. If the conflict
8578 table for a function could be more than size in MB given by the
8579 parameter, the conflict table is not built and faster, simpler, and
8580 lower quality register allocation algorithm will be used. The
8581 algorithm do not use pseudo-register conflicts. The default value of
8582 the parameter is 2000.
8584 @item ira-loop-reserved-regs
8585 IRA can be used to evaluate more accurate register pressure in loops
8586 for decision to move loop invariants (see @option{-O3}). The number
8587 of available registers reserved for some other purposes is described
8588 by this parameter. The default value of the parameter is 2 which is
8589 minimal number of registers needed for execution of typical
8590 instruction. This value is the best found from numerous experiments.
8592 @item loop-invariant-max-bbs-in-loop
8593 Loop invariant motion can be very expensive, both in compile time and
8594 in amount of needed compile time memory, with very large loops. Loops
8595 with more basic blocks than this parameter won't have loop invariant
8596 motion optimization performed on them. The default value of the
8597 parameter is 1000 for -O1 and 10000 for -O2 and above.
8599 @item max-vartrack-size
8600 Sets a maximum number of hash table slots to use during variable
8601 tracking dataflow analysis of any function. If this limit is exceeded
8602 with variable tracking at assignments enabled, analysis for that
8603 function is retried without it, after removing all debug insns from
8604 the function. If the limit is exceeded even without debug insns, var
8605 tracking analysis is completely disabled for the function. Setting
8606 the parameter to zero makes it unlimited.
8608 @item min-nondebug-insn-uid
8609 Use uids starting at this parameter for nondebug insns. The range below
8610 the parameter is reserved exclusively for debug insns created by
8611 @option{-fvar-tracking-assignments}, but debug insns may get
8612 (non-overlapping) uids above it if the reserved range is exhausted.
8614 @item ipa-sra-ptr-growth-factor
8615 IPA-SRA will replace a pointer to an aggregate with one or more new
8616 parameters only when their cumulative size is less or equal to
8617 @option{ipa-sra-ptr-growth-factor} times the size of the original
8620 @item graphite-max-nb-scop-params
8621 To avoid exponential effects in the Graphite loop transforms, the
8622 number of parameters in a Static Control Part (SCoP) is bounded. The
8623 default value is 10 parameters. A variable whose value is unknown at
8624 compile time and defined outside a SCoP is a parameter of the SCoP.
8626 @item graphite-max-bbs-per-function
8627 To avoid exponential effects in the detection of SCoPs, the size of
8628 the functions analyzed by Graphite is bounded. The default value is
8631 @item loop-block-tile-size
8632 Loop blocking or strip mining transforms, enabled with
8633 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
8634 loop in the loop nest by a given number of iterations. The strip
8635 length can be changed using the @option{loop-block-tile-size}
8636 parameter. The default value is 51 iterations.
8641 @node Preprocessor Options
8642 @section Options Controlling the Preprocessor
8643 @cindex preprocessor options
8644 @cindex options, preprocessor
8646 These options control the C preprocessor, which is run on each C source
8647 file before actual compilation.
8649 If you use the @option{-E} option, nothing is done except preprocessing.
8650 Some of these options make sense only together with @option{-E} because
8651 they cause the preprocessor output to be unsuitable for actual
8655 @item -Wp,@var{option}
8657 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8658 and pass @var{option} directly through to the preprocessor. If
8659 @var{option} contains commas, it is split into multiple options at the
8660 commas. However, many options are modified, translated or interpreted
8661 by the compiler driver before being passed to the preprocessor, and
8662 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8663 interface is undocumented and subject to change, so whenever possible
8664 you should avoid using @option{-Wp} and let the driver handle the
8667 @item -Xpreprocessor @var{option}
8668 @opindex Xpreprocessor
8669 Pass @var{option} as an option to the preprocessor. You can use this to
8670 supply system-specific preprocessor options which GCC does not know how to
8673 If you want to pass an option that takes an argument, you must use
8674 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8677 @include cppopts.texi
8679 @node Assembler Options
8680 @section Passing Options to the Assembler
8682 @c prevent bad page break with this line
8683 You can pass options to the assembler.
8686 @item -Wa,@var{option}
8688 Pass @var{option} as an option to the assembler. If @var{option}
8689 contains commas, it is split into multiple options at the commas.
8691 @item -Xassembler @var{option}
8693 Pass @var{option} as an option to the assembler. You can use this to
8694 supply system-specific assembler options which GCC does not know how to
8697 If you want to pass an option that takes an argument, you must use
8698 @option{-Xassembler} twice, once for the option and once for the argument.
8703 @section Options for Linking
8704 @cindex link options
8705 @cindex options, linking
8707 These options come into play when the compiler links object files into
8708 an executable output file. They are meaningless if the compiler is
8709 not doing a link step.
8713 @item @var{object-file-name}
8714 A file name that does not end in a special recognized suffix is
8715 considered to name an object file or library. (Object files are
8716 distinguished from libraries by the linker according to the file
8717 contents.) If linking is done, these object files are used as input
8726 If any of these options is used, then the linker is not run, and
8727 object file names should not be used as arguments. @xref{Overall
8731 @item -l@var{library}
8732 @itemx -l @var{library}
8734 Search the library named @var{library} when linking. (The second
8735 alternative with the library as a separate argument is only for
8736 POSIX compliance and is not recommended.)
8738 It makes a difference where in the command you write this option; the
8739 linker searches and processes libraries and object files in the order they
8740 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8741 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8742 to functions in @samp{z}, those functions may not be loaded.
8744 The linker searches a standard list of directories for the library,
8745 which is actually a file named @file{lib@var{library}.a}. The linker
8746 then uses this file as if it had been specified precisely by name.
8748 The directories searched include several standard system directories
8749 plus any that you specify with @option{-L}.
8751 Normally the files found this way are library files---archive files
8752 whose members are object files. The linker handles an archive file by
8753 scanning through it for members which define symbols that have so far
8754 been referenced but not defined. But if the file that is found is an
8755 ordinary object file, it is linked in the usual fashion. The only
8756 difference between using an @option{-l} option and specifying a file name
8757 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8758 and searches several directories.
8762 You need this special case of the @option{-l} option in order to
8763 link an Objective-C or Objective-C++ program.
8766 @opindex nostartfiles
8767 Do not use the standard system startup files when linking.
8768 The standard system libraries are used normally, unless @option{-nostdlib}
8769 or @option{-nodefaultlibs} is used.
8771 @item -nodefaultlibs
8772 @opindex nodefaultlibs
8773 Do not use the standard system libraries when linking.
8774 Only the libraries you specify will be passed to the linker, options
8775 specifying linkage of the system libraries, such as @code{-static-libgcc}
8776 or @code{-shared-libgcc}, will be ignored.
8777 The standard startup files are used normally, unless @option{-nostartfiles}
8778 is used. The compiler may generate calls to @code{memcmp},
8779 @code{memset}, @code{memcpy} and @code{memmove}.
8780 These entries are usually resolved by entries in
8781 libc. These entry points should be supplied through some other
8782 mechanism when this option is specified.
8786 Do not use the standard system startup files or libraries when linking.
8787 No startup files and only the libraries you specify will be passed to
8788 the linker, options specifying linkage of the system libraries, such as
8789 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8790 The compiler may generate calls to @code{memcmp}, @code{memset},
8791 @code{memcpy} and @code{memmove}.
8792 These entries are usually resolved by entries in
8793 libc. These entry points should be supplied through some other
8794 mechanism when this option is specified.
8796 @cindex @option{-lgcc}, use with @option{-nostdlib}
8797 @cindex @option{-nostdlib} and unresolved references
8798 @cindex unresolved references and @option{-nostdlib}
8799 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8800 @cindex @option{-nodefaultlibs} and unresolved references
8801 @cindex unresolved references and @option{-nodefaultlibs}
8802 One of the standard libraries bypassed by @option{-nostdlib} and
8803 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8804 that GCC uses to overcome shortcomings of particular machines, or special
8805 needs for some languages.
8806 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8807 Collection (GCC) Internals},
8808 for more discussion of @file{libgcc.a}.)
8809 In most cases, you need @file{libgcc.a} even when you want to avoid
8810 other standard libraries. In other words, when you specify @option{-nostdlib}
8811 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8812 This ensures that you have no unresolved references to internal GCC
8813 library subroutines. (For example, @samp{__main}, used to ensure C++
8814 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8815 GNU Compiler Collection (GCC) Internals}.)
8819 Produce a position independent executable on targets which support it.
8820 For predictable results, you must also specify the same set of options
8821 that were used to generate code (@option{-fpie}, @option{-fPIE},
8822 or model suboptions) when you specify this option.
8826 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8827 that support it. This instructs the linker to add all symbols, not
8828 only used ones, to the dynamic symbol table. This option is needed
8829 for some uses of @code{dlopen} or to allow obtaining backtraces
8830 from within a program.
8834 Remove all symbol table and relocation information from the executable.
8838 On systems that support dynamic linking, this prevents linking with the shared
8839 libraries. On other systems, this option has no effect.
8843 Produce a shared object which can then be linked with other objects to
8844 form an executable. Not all systems support this option. For predictable
8845 results, you must also specify the same set of options that were used to
8846 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8847 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8848 needs to build supplementary stub code for constructors to work. On
8849 multi-libbed systems, @samp{gcc -shared} must select the correct support
8850 libraries to link against. Failing to supply the correct flags may lead
8851 to subtle defects. Supplying them in cases where they are not necessary
8854 @item -shared-libgcc
8855 @itemx -static-libgcc
8856 @opindex shared-libgcc
8857 @opindex static-libgcc
8858 On systems that provide @file{libgcc} as a shared library, these options
8859 force the use of either the shared or static version respectively.
8860 If no shared version of @file{libgcc} was built when the compiler was
8861 configured, these options have no effect.
8863 There are several situations in which an application should use the
8864 shared @file{libgcc} instead of the static version. The most common
8865 of these is when the application wishes to throw and catch exceptions
8866 across different shared libraries. In that case, each of the libraries
8867 as well as the application itself should use the shared @file{libgcc}.
8869 Therefore, the G++ and GCJ drivers automatically add
8870 @option{-shared-libgcc} whenever you build a shared library or a main
8871 executable, because C++ and Java programs typically use exceptions, so
8872 this is the right thing to do.
8874 If, instead, you use the GCC driver to create shared libraries, you may
8875 find that they will not always be linked with the shared @file{libgcc}.
8876 If GCC finds, at its configuration time, that you have a non-GNU linker
8877 or a GNU linker that does not support option @option{--eh-frame-hdr},
8878 it will link the shared version of @file{libgcc} into shared libraries
8879 by default. Otherwise, it will take advantage of the linker and optimize
8880 away the linking with the shared version of @file{libgcc}, linking with
8881 the static version of libgcc by default. This allows exceptions to
8882 propagate through such shared libraries, without incurring relocation
8883 costs at library load time.
8885 However, if a library or main executable is supposed to throw or catch
8886 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8887 for the languages used in the program, or using the option
8888 @option{-shared-libgcc}, such that it is linked with the shared
8891 @item -static-libstdc++
8892 When the @command{g++} program is used to link a C++ program, it will
8893 normally automatically link against @option{libstdc++}. If
8894 @file{libstdc++} is available as a shared library, and the
8895 @option{-static} option is not used, then this will link against the
8896 shared version of @file{libstdc++}. That is normally fine. However, it
8897 is sometimes useful to freeze the version of @file{libstdc++} used by
8898 the program without going all the way to a fully static link. The
8899 @option{-static-libstdc++} option directs the @command{g++} driver to
8900 link @file{libstdc++} statically, without necessarily linking other
8901 libraries statically.
8905 Bind references to global symbols when building a shared object. Warn
8906 about any unresolved references (unless overridden by the link editor
8907 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8910 @item -T @var{script}
8912 @cindex linker script
8913 Use @var{script} as the linker script. This option is supported by most
8914 systems using the GNU linker. On some targets, such as bare-board
8915 targets without an operating system, the @option{-T} option may be required
8916 when linking to avoid references to undefined symbols.
8918 @item -Xlinker @var{option}
8920 Pass @var{option} as an option to the linker. You can use this to
8921 supply system-specific linker options which GCC does not know how to
8924 If you want to pass an option that takes a separate argument, you must use
8925 @option{-Xlinker} twice, once for the option and once for the argument.
8926 For example, to pass @option{-assert definitions}, you must write
8927 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8928 @option{-Xlinker "-assert definitions"}, because this passes the entire
8929 string as a single argument, which is not what the linker expects.
8931 When using the GNU linker, it is usually more convenient to pass
8932 arguments to linker options using the @option{@var{option}=@var{value}}
8933 syntax than as separate arguments. For example, you can specify
8934 @samp{-Xlinker -Map=output.map} rather than
8935 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8936 this syntax for command-line options.
8938 @item -Wl,@var{option}
8940 Pass @var{option} as an option to the linker. If @var{option} contains
8941 commas, it is split into multiple options at the commas. You can use this
8942 syntax to pass an argument to the option.
8943 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8944 linker. When using the GNU linker, you can also get the same effect with
8945 @samp{-Wl,-Map=output.map}.
8947 @item -u @var{symbol}
8949 Pretend the symbol @var{symbol} is undefined, to force linking of
8950 library modules to define it. You can use @option{-u} multiple times with
8951 different symbols to force loading of additional library modules.
8954 @node Directory Options
8955 @section Options for Directory Search
8956 @cindex directory options
8957 @cindex options, directory search
8960 These options specify directories to search for header files, for
8961 libraries and for parts of the compiler:
8966 Add the directory @var{dir} to the head of the list of directories to be
8967 searched for header files. This can be used to override a system header
8968 file, substituting your own version, since these directories are
8969 searched before the system header file directories. However, you should
8970 not use this option to add directories that contain vendor-supplied
8971 system header files (use @option{-isystem} for that). If you use more than
8972 one @option{-I} option, the directories are scanned in left-to-right
8973 order; the standard system directories come after.
8975 If a standard system include directory, or a directory specified with
8976 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8977 option will be ignored. The directory will still be searched but as a
8978 system directory at its normal position in the system include chain.
8979 This is to ensure that GCC's procedure to fix buggy system headers and
8980 the ordering for the include_next directive are not inadvertently changed.
8981 If you really need to change the search order for system directories,
8982 use the @option{-nostdinc} and/or @option{-isystem} options.
8984 @item -iplugindir=@var{dir}
8985 Set the directory to search for plugins which are passed
8986 by @option{-fplugin=@var{name}} instead of
8987 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
8988 to be used by the user, but only passed by the driver.
8990 @item -iquote@var{dir}
8992 Add the directory @var{dir} to the head of the list of directories to
8993 be searched for header files only for the case of @samp{#include
8994 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8995 otherwise just like @option{-I}.
8999 Add directory @var{dir} to the list of directories to be searched
9002 @item -B@var{prefix}
9004 This option specifies where to find the executables, libraries,
9005 include files, and data files of the compiler itself.
9007 The compiler driver program runs one or more of the subprograms
9008 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
9009 @var{prefix} as a prefix for each program it tries to run, both with and
9010 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
9012 For each subprogram to be run, the compiler driver first tries the
9013 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
9014 was not specified, the driver tries two standard prefixes, which are
9015 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
9016 those results in a file name that is found, the unmodified program
9017 name is searched for using the directories specified in your
9018 @env{PATH} environment variable.
9020 The compiler will check to see if the path provided by the @option{-B}
9021 refers to a directory, and if necessary it will add a directory
9022 separator character at the end of the path.
9024 @option{-B} prefixes that effectively specify directory names also apply
9025 to libraries in the linker, because the compiler translates these
9026 options into @option{-L} options for the linker. They also apply to
9027 includes files in the preprocessor, because the compiler translates these
9028 options into @option{-isystem} options for the preprocessor. In this case,
9029 the compiler appends @samp{include} to the prefix.
9031 The run-time support file @file{libgcc.a} can also be searched for using
9032 the @option{-B} prefix, if needed. If it is not found there, the two
9033 standard prefixes above are tried, and that is all. The file is left
9034 out of the link if it is not found by those means.
9036 Another way to specify a prefix much like the @option{-B} prefix is to use
9037 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
9040 As a special kludge, if the path provided by @option{-B} is
9041 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
9042 9, then it will be replaced by @file{[dir/]include}. This is to help
9043 with boot-strapping the compiler.
9045 @item -specs=@var{file}
9047 Process @var{file} after the compiler reads in the standard @file{specs}
9048 file, in order to override the defaults that the @file{gcc} driver
9049 program uses when determining what switches to pass to @file{cc1},
9050 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
9051 @option{-specs=@var{file}} can be specified on the command line, and they
9052 are processed in order, from left to right.
9054 @item --sysroot=@var{dir}
9056 Use @var{dir} as the logical root directory for headers and libraries.
9057 For example, if the compiler would normally search for headers in
9058 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
9059 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
9061 If you use both this option and the @option{-isysroot} option, then
9062 the @option{--sysroot} option will apply to libraries, but the
9063 @option{-isysroot} option will apply to header files.
9065 The GNU linker (beginning with version 2.16) has the necessary support
9066 for this option. If your linker does not support this option, the
9067 header file aspect of @option{--sysroot} will still work, but the
9068 library aspect will not.
9072 This option has been deprecated. Please use @option{-iquote} instead for
9073 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
9074 Any directories you specify with @option{-I} options before the @option{-I-}
9075 option are searched only for the case of @samp{#include "@var{file}"};
9076 they are not searched for @samp{#include <@var{file}>}.
9078 If additional directories are specified with @option{-I} options after
9079 the @option{-I-}, these directories are searched for all @samp{#include}
9080 directives. (Ordinarily @emph{all} @option{-I} directories are used
9083 In addition, the @option{-I-} option inhibits the use of the current
9084 directory (where the current input file came from) as the first search
9085 directory for @samp{#include "@var{file}"}. There is no way to
9086 override this effect of @option{-I-}. With @option{-I.} you can specify
9087 searching the directory which was current when the compiler was
9088 invoked. That is not exactly the same as what the preprocessor does
9089 by default, but it is often satisfactory.
9091 @option{-I-} does not inhibit the use of the standard system directories
9092 for header files. Thus, @option{-I-} and @option{-nostdinc} are
9099 @section Specifying subprocesses and the switches to pass to them
9102 @command{gcc} is a driver program. It performs its job by invoking a
9103 sequence of other programs to do the work of compiling, assembling and
9104 linking. GCC interprets its command-line parameters and uses these to
9105 deduce which programs it should invoke, and which command-line options
9106 it ought to place on their command lines. This behavior is controlled
9107 by @dfn{spec strings}. In most cases there is one spec string for each
9108 program that GCC can invoke, but a few programs have multiple spec
9109 strings to control their behavior. The spec strings built into GCC can
9110 be overridden by using the @option{-specs=} command-line switch to specify
9113 @dfn{Spec files} are plaintext files that are used to construct spec
9114 strings. They consist of a sequence of directives separated by blank
9115 lines. The type of directive is determined by the first non-whitespace
9116 character on the line and it can be one of the following:
9119 @item %@var{command}
9120 Issues a @var{command} to the spec file processor. The commands that can
9124 @item %include <@var{file}>
9126 Search for @var{file} and insert its text at the current point in the
9129 @item %include_noerr <@var{file}>
9130 @cindex %include_noerr
9131 Just like @samp{%include}, but do not generate an error message if the include
9132 file cannot be found.
9134 @item %rename @var{old_name} @var{new_name}
9136 Rename the spec string @var{old_name} to @var{new_name}.
9140 @item *[@var{spec_name}]:
9141 This tells the compiler to create, override or delete the named spec
9142 string. All lines after this directive up to the next directive or
9143 blank line are considered to be the text for the spec string. If this
9144 results in an empty string then the spec will be deleted. (Or, if the
9145 spec did not exist, then nothing will happened.) Otherwise, if the spec
9146 does not currently exist a new spec will be created. If the spec does
9147 exist then its contents will be overridden by the text of this
9148 directive, unless the first character of that text is the @samp{+}
9149 character, in which case the text will be appended to the spec.
9151 @item [@var{suffix}]:
9152 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
9153 and up to the next directive or blank line are considered to make up the
9154 spec string for the indicated suffix. When the compiler encounters an
9155 input file with the named suffix, it will processes the spec string in
9156 order to work out how to compile that file. For example:
9163 This says that any input file whose name ends in @samp{.ZZ} should be
9164 passed to the program @samp{z-compile}, which should be invoked with the
9165 command-line switch @option{-input} and with the result of performing the
9166 @samp{%i} substitution. (See below.)
9168 As an alternative to providing a spec string, the text that follows a
9169 suffix directive can be one of the following:
9172 @item @@@var{language}
9173 This says that the suffix is an alias for a known @var{language}. This is
9174 similar to using the @option{-x} command-line switch to GCC to specify a
9175 language explicitly. For example:
9182 Says that .ZZ files are, in fact, C++ source files.
9185 This causes an error messages saying:
9188 @var{name} compiler not installed on this system.
9192 GCC already has an extensive list of suffixes built into it.
9193 This directive will add an entry to the end of the list of suffixes, but
9194 since the list is searched from the end backwards, it is effectively
9195 possible to override earlier entries using this technique.
9199 GCC has the following spec strings built into it. Spec files can
9200 override these strings or create their own. Note that individual
9201 targets can also add their own spec strings to this list.
9204 asm Options to pass to the assembler
9205 asm_final Options to pass to the assembler post-processor
9206 cpp Options to pass to the C preprocessor
9207 cc1 Options to pass to the C compiler
9208 cc1plus Options to pass to the C++ compiler
9209 endfile Object files to include at the end of the link
9210 link Options to pass to the linker
9211 lib Libraries to include on the command line to the linker
9212 libgcc Decides which GCC support library to pass to the linker
9213 linker Sets the name of the linker
9214 predefines Defines to be passed to the C preprocessor
9215 signed_char Defines to pass to CPP to say whether @code{char} is signed
9217 startfile Object files to include at the start of the link
9220 Here is a small example of a spec file:
9226 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9229 This example renames the spec called @samp{lib} to @samp{old_lib} and
9230 then overrides the previous definition of @samp{lib} with a new one.
9231 The new definition adds in some extra command-line options before
9232 including the text of the old definition.
9234 @dfn{Spec strings} are a list of command-line options to be passed to their
9235 corresponding program. In addition, the spec strings can contain
9236 @samp{%}-prefixed sequences to substitute variable text or to
9237 conditionally insert text into the command line. Using these constructs
9238 it is possible to generate quite complex command lines.
9240 Here is a table of all defined @samp{%}-sequences for spec
9241 strings. Note that spaces are not generated automatically around the
9242 results of expanding these sequences. Therefore you can concatenate them
9243 together or combine them with constant text in a single argument.
9247 Substitute one @samp{%} into the program name or argument.
9250 Substitute the name of the input file being processed.
9253 Substitute the basename of the input file being processed.
9254 This is the substring up to (and not including) the last period
9255 and not including the directory.
9258 This is the same as @samp{%b}, but include the file suffix (text after
9262 Marks the argument containing or following the @samp{%d} as a
9263 temporary file name, so that that file will be deleted if GCC exits
9264 successfully. Unlike @samp{%g}, this contributes no text to the
9267 @item %g@var{suffix}
9268 Substitute a file name that has suffix @var{suffix} and is chosen
9269 once per compilation, and mark the argument in the same way as
9270 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9271 name is now chosen in a way that is hard to predict even when previously
9272 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9273 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9274 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9275 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9276 was simply substituted with a file name chosen once per compilation,
9277 without regard to any appended suffix (which was therefore treated
9278 just like ordinary text), making such attacks more likely to succeed.
9280 @item %u@var{suffix}
9281 Like @samp{%g}, but generates a new temporary file name even if
9282 @samp{%u@var{suffix}} was already seen.
9284 @item %U@var{suffix}
9285 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9286 new one if there is no such last file name. In the absence of any
9287 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9288 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9289 would involve the generation of two distinct file names, one
9290 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9291 simply substituted with a file name chosen for the previous @samp{%u},
9292 without regard to any appended suffix.
9294 @item %j@var{suffix}
9295 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9296 writable, and if save-temps is off; otherwise, substitute the name
9297 of a temporary file, just like @samp{%u}. This temporary file is not
9298 meant for communication between processes, but rather as a junk
9301 @item %|@var{suffix}
9302 @itemx %m@var{suffix}
9303 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9304 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9305 all. These are the two most common ways to instruct a program that it
9306 should read from standard input or write to standard output. If you
9307 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9308 construct: see for example @file{f/lang-specs.h}.
9310 @item %.@var{SUFFIX}
9311 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9312 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9313 terminated by the next space or %.
9316 Marks the argument containing or following the @samp{%w} as the
9317 designated output file of this compilation. This puts the argument
9318 into the sequence of arguments that @samp{%o} will substitute later.
9321 Substitutes the names of all the output files, with spaces
9322 automatically placed around them. You should write spaces
9323 around the @samp{%o} as well or the results are undefined.
9324 @samp{%o} is for use in the specs for running the linker.
9325 Input files whose names have no recognized suffix are not compiled
9326 at all, but they are included among the output files, so they will
9330 Substitutes the suffix for object files. Note that this is
9331 handled specially when it immediately follows @samp{%g, %u, or %U},
9332 because of the need for those to form complete file names. The
9333 handling is such that @samp{%O} is treated exactly as if it had already
9334 been substituted, except that @samp{%g, %u, and %U} do not currently
9335 support additional @var{suffix} characters following @samp{%O} as they would
9336 following, for example, @samp{.o}.
9339 Substitutes the standard macro predefinitions for the
9340 current target machine. Use this when running @code{cpp}.
9343 Like @samp{%p}, but puts @samp{__} before and after the name of each
9344 predefined macro, except for macros that start with @samp{__} or with
9345 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9349 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9350 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9351 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9352 and @option{-imultilib} as necessary.
9355 Current argument is the name of a library or startup file of some sort.
9356 Search for that file in a standard list of directories and substitute
9357 the full name found. The current working directory is included in the
9358 list of directories scanned.
9361 Current argument is the name of a linker script. Search for that file
9362 in the current list of directories to scan for libraries. If the file
9363 is located insert a @option{--script} option into the command line
9364 followed by the full path name found. If the file is not found then
9365 generate an error message. Note: the current working directory is not
9369 Print @var{str} as an error message. @var{str} is terminated by a newline.
9370 Use this when inconsistent options are detected.
9373 Substitute the contents of spec string @var{name} at this point.
9376 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9378 @item %x@{@var{option}@}
9379 Accumulate an option for @samp{%X}.
9382 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9386 Output the accumulated assembler options specified by @option{-Wa}.
9389 Output the accumulated preprocessor options specified by @option{-Wp}.
9392 Process the @code{asm} spec. This is used to compute the
9393 switches to be passed to the assembler.
9396 Process the @code{asm_final} spec. This is a spec string for
9397 passing switches to an assembler post-processor, if such a program is
9401 Process the @code{link} spec. This is the spec for computing the
9402 command line passed to the linker. Typically it will make use of the
9403 @samp{%L %G %S %D and %E} sequences.
9406 Dump out a @option{-L} option for each directory that GCC believes might
9407 contain startup files. If the target supports multilibs then the
9408 current multilib directory will be prepended to each of these paths.
9411 Process the @code{lib} spec. This is a spec string for deciding which
9412 libraries should be included on the command line to the linker.
9415 Process the @code{libgcc} spec. This is a spec string for deciding
9416 which GCC support library should be included on the command line to the linker.
9419 Process the @code{startfile} spec. This is a spec for deciding which
9420 object files should be the first ones passed to the linker. Typically
9421 this might be a file named @file{crt0.o}.
9424 Process the @code{endfile} spec. This is a spec string that specifies
9425 the last object files that will be passed to the linker.
9428 Process the @code{cpp} spec. This is used to construct the arguments
9429 to be passed to the C preprocessor.
9432 Process the @code{cc1} spec. This is used to construct the options to be
9433 passed to the actual C compiler (@samp{cc1}).
9436 Process the @code{cc1plus} spec. This is used to construct the options to be
9437 passed to the actual C++ compiler (@samp{cc1plus}).
9440 Substitute the variable part of a matched option. See below.
9441 Note that each comma in the substituted string is replaced by
9445 Remove all occurrences of @code{-S} from the command line. Note---this
9446 command is position dependent. @samp{%} commands in the spec string
9447 before this one will see @code{-S}, @samp{%} commands in the spec string
9448 after this one will not.
9450 @item %:@var{function}(@var{args})
9451 Call the named function @var{function}, passing it @var{args}.
9452 @var{args} is first processed as a nested spec string, then split
9453 into an argument vector in the usual fashion. The function returns
9454 a string which is processed as if it had appeared literally as part
9455 of the current spec.
9457 The following built-in spec functions are provided:
9461 The @code{getenv} spec function takes two arguments: an environment
9462 variable name and a string. If the environment variable is not
9463 defined, a fatal error is issued. Otherwise, the return value is the
9464 value of the environment variable concatenated with the string. For
9465 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9468 %:getenv(TOPDIR /include)
9471 expands to @file{/path/to/top/include}.
9473 @item @code{if-exists}
9474 The @code{if-exists} spec function takes one argument, an absolute
9475 pathname to a file. If the file exists, @code{if-exists} returns the
9476 pathname. Here is a small example of its usage:
9480 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9483 @item @code{if-exists-else}
9484 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9485 spec function, except that it takes two arguments. The first argument is
9486 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9487 returns the pathname. If it does not exist, it returns the second argument.
9488 This way, @code{if-exists-else} can be used to select one file or another,
9489 based on the existence of the first. Here is a small example of its usage:
9493 crt0%O%s %:if-exists(crti%O%s) \
9494 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9497 @item @code{replace-outfile}
9498 The @code{replace-outfile} spec function takes two arguments. It looks for the
9499 first argument in the outfiles array and replaces it with the second argument. Here
9500 is a small example of its usage:
9503 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9506 @item @code{print-asm-header}
9507 The @code{print-asm-header} function takes no arguments and simply
9508 prints a banner like:
9514 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9517 It is used to separate compiler options from assembler options
9518 in the @option{--target-help} output.
9522 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9523 If that switch was not specified, this substitutes nothing. Note that
9524 the leading dash is omitted when specifying this option, and it is
9525 automatically inserted if the substitution is performed. Thus the spec
9526 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9527 and would output the command line option @option{-foo}.
9529 @item %W@{@code{S}@}
9530 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9533 @item %@{@code{S}*@}
9534 Substitutes all the switches specified to GCC whose names start
9535 with @code{-S}, but which also take an argument. This is used for
9536 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9537 GCC considers @option{-o foo} as being
9538 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9539 text, including the space. Thus two arguments would be generated.
9541 @item %@{@code{S}*&@code{T}*@}
9542 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9543 (the order of @code{S} and @code{T} in the spec is not significant).
9544 There can be any number of ampersand-separated variables; for each the
9545 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9547 @item %@{@code{S}:@code{X}@}
9548 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9550 @item %@{!@code{S}:@code{X}@}
9551 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9553 @item %@{@code{S}*:@code{X}@}
9554 Substitutes @code{X} if one or more switches whose names start with
9555 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9556 once, no matter how many such switches appeared. However, if @code{%*}
9557 appears somewhere in @code{X}, then @code{X} will be substituted once
9558 for each matching switch, with the @code{%*} replaced by the part of
9559 that switch that matched the @code{*}.
9561 @item %@{.@code{S}:@code{X}@}
9562 Substitutes @code{X}, if processing a file with suffix @code{S}.
9564 @item %@{!.@code{S}:@code{X}@}
9565 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9567 @item %@{,@code{S}:@code{X}@}
9568 Substitutes @code{X}, if processing a file for language @code{S}.
9570 @item %@{!,@code{S}:@code{X}@}
9571 Substitutes @code{X}, if not processing a file for language @code{S}.
9573 @item %@{@code{S}|@code{P}:@code{X}@}
9574 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9575 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9576 @code{*} sequences as well, although they have a stronger binding than
9577 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9578 alternatives must be starred, and only the first matching alternative
9581 For example, a spec string like this:
9584 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9587 will output the following command-line options from the following input
9588 command-line options:
9593 -d fred.c -foo -baz -boggle
9594 -d jim.d -bar -baz -boggle
9597 @item %@{S:X; T:Y; :D@}
9599 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9600 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9601 be as many clauses as you need. This may be combined with @code{.},
9602 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9607 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9608 construct may contain other nested @samp{%} constructs or spaces, or
9609 even newlines. They are processed as usual, as described above.
9610 Trailing white space in @code{X} is ignored. White space may also
9611 appear anywhere on the left side of the colon in these constructs,
9612 except between @code{.} or @code{*} and the corresponding word.
9614 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9615 handled specifically in these constructs. If another value of
9616 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9617 @option{-W} switch is found later in the command line, the earlier
9618 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9619 just one letter, which passes all matching options.
9621 The character @samp{|} at the beginning of the predicate text is used to
9622 indicate that a command should be piped to the following command, but
9623 only if @option{-pipe} is specified.
9625 It is built into GCC which switches take arguments and which do not.
9626 (You might think it would be useful to generalize this to allow each
9627 compiler's spec to say which switches take arguments. But this cannot
9628 be done in a consistent fashion. GCC cannot even decide which input
9629 files have been specified without knowing which switches take arguments,
9630 and it must know which input files to compile in order to tell which
9633 GCC also knows implicitly that arguments starting in @option{-l} are to be
9634 treated as compiler output files, and passed to the linker in their
9635 proper position among the other output files.
9637 @c man begin OPTIONS
9639 @node Target Options
9640 @section Specifying Target Machine and Compiler Version
9641 @cindex target options
9642 @cindex cross compiling
9643 @cindex specifying machine version
9644 @cindex specifying compiler version and target machine
9645 @cindex compiler version, specifying
9646 @cindex target machine, specifying
9648 The usual way to run GCC is to run the executable called @file{gcc}, or
9649 @file{<machine>-gcc} when cross-compiling, or
9650 @file{<machine>-gcc-<version>} to run a version other than the one that
9653 @node Submodel Options
9654 @section Hardware Models and Configurations
9655 @cindex submodel options
9656 @cindex specifying hardware config
9657 @cindex hardware models and configurations, specifying
9658 @cindex machine dependent options
9660 Each target machine types can have its own
9661 special options, starting with @samp{-m}, to choose among various
9662 hardware models or configurations---for example, 68010 vs 68020,
9663 floating coprocessor or none. A single installed version of the
9664 compiler can compile for any model or configuration, according to the
9667 Some configurations of the compiler also support additional special
9668 options, usually for compatibility with other compilers on the same
9671 @c This list is ordered alphanumerically by subsection name.
9672 @c It should be the same order and spelling as these options are listed
9673 @c in Machine Dependent Options
9679 * Blackfin Options::
9683 * DEC Alpha Options::
9684 * DEC Alpha/VMS Options::
9687 * GNU/Linux Options::
9690 * i386 and x86-64 Options::
9691 * i386 and x86-64 Windows Options::
9693 * IA-64/VMS Options::
9705 * picoChip Options::
9707 * RS/6000 and PowerPC Options::
9709 * S/390 and zSeries Options::
9714 * System V Options::
9719 * Xstormy16 Options::
9725 @subsection ARC Options
9728 These options are defined for ARC implementations:
9733 Compile code for little endian mode. This is the default.
9737 Compile code for big endian mode.
9740 @opindex mmangle-cpu
9741 Prepend the name of the cpu to all public symbol names.
9742 In multiple-processor systems, there are many ARC variants with different
9743 instruction and register set characteristics. This flag prevents code
9744 compiled for one cpu to be linked with code compiled for another.
9745 No facility exists for handling variants that are ``almost identical''.
9746 This is an all or nothing option.
9748 @item -mcpu=@var{cpu}
9750 Compile code for ARC variant @var{cpu}.
9751 Which variants are supported depend on the configuration.
9752 All variants support @option{-mcpu=base}, this is the default.
9754 @item -mtext=@var{text-section}
9755 @itemx -mdata=@var{data-section}
9756 @itemx -mrodata=@var{readonly-data-section}
9760 Put functions, data, and readonly data in @var{text-section},
9761 @var{data-section}, and @var{readonly-data-section} respectively
9762 by default. This can be overridden with the @code{section} attribute.
9763 @xref{Variable Attributes}.
9768 @subsection ARM Options
9771 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9775 @item -mabi=@var{name}
9777 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9778 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9781 @opindex mapcs-frame
9782 Generate a stack frame that is compliant with the ARM Procedure Call
9783 Standard for all functions, even if this is not strictly necessary for
9784 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9785 with this option will cause the stack frames not to be generated for
9786 leaf functions. The default is @option{-mno-apcs-frame}.
9790 This is a synonym for @option{-mapcs-frame}.
9793 @c not currently implemented
9794 @item -mapcs-stack-check
9795 @opindex mapcs-stack-check
9796 Generate code to check the amount of stack space available upon entry to
9797 every function (that actually uses some stack space). If there is
9798 insufficient space available then either the function
9799 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9800 called, depending upon the amount of stack space required. The run time
9801 system is required to provide these functions. The default is
9802 @option{-mno-apcs-stack-check}, since this produces smaller code.
9804 @c not currently implemented
9806 @opindex mapcs-float
9807 Pass floating point arguments using the float point registers. This is
9808 one of the variants of the APCS@. This option is recommended if the
9809 target hardware has a floating point unit or if a lot of floating point
9810 arithmetic is going to be performed by the code. The default is
9811 @option{-mno-apcs-float}, since integer only code is slightly increased in
9812 size if @option{-mapcs-float} is used.
9814 @c not currently implemented
9815 @item -mapcs-reentrant
9816 @opindex mapcs-reentrant
9817 Generate reentrant, position independent code. The default is
9818 @option{-mno-apcs-reentrant}.
9821 @item -mthumb-interwork
9822 @opindex mthumb-interwork
9823 Generate code which supports calling between the ARM and Thumb
9824 instruction sets. Without this option the two instruction sets cannot
9825 be reliably used inside one program. The default is
9826 @option{-mno-thumb-interwork}, since slightly larger code is generated
9827 when @option{-mthumb-interwork} is specified.
9829 @item -mno-sched-prolog
9830 @opindex mno-sched-prolog
9831 Prevent the reordering of instructions in the function prolog, or the
9832 merging of those instruction with the instructions in the function's
9833 body. This means that all functions will start with a recognizable set
9834 of instructions (or in fact one of a choice from a small set of
9835 different function prologues), and this information can be used to
9836 locate the start if functions inside an executable piece of code. The
9837 default is @option{-msched-prolog}.
9839 @item -mfloat-abi=@var{name}
9841 Specifies which floating-point ABI to use. Permissible values
9842 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9844 Specifying @samp{soft} causes GCC to generate output containing
9845 library calls for floating-point operations.
9846 @samp{softfp} allows the generation of code using hardware floating-point
9847 instructions, but still uses the soft-float calling conventions.
9848 @samp{hard} allows generation of floating-point instructions
9849 and uses FPU-specific calling conventions.
9851 The default depends on the specific target configuration. Note that
9852 the hard-float and soft-float ABIs are not link-compatible; you must
9853 compile your entire program with the same ABI, and link with a
9854 compatible set of libraries.
9857 @opindex mhard-float
9858 Equivalent to @option{-mfloat-abi=hard}.
9861 @opindex msoft-float
9862 Equivalent to @option{-mfloat-abi=soft}.
9864 @item -mlittle-endian
9865 @opindex mlittle-endian
9866 Generate code for a processor running in little-endian mode. This is
9867 the default for all standard configurations.
9870 @opindex mbig-endian
9871 Generate code for a processor running in big-endian mode; the default is
9872 to compile code for a little-endian processor.
9874 @item -mwords-little-endian
9875 @opindex mwords-little-endian
9876 This option only applies when generating code for big-endian processors.
9877 Generate code for a little-endian word order but a big-endian byte
9878 order. That is, a byte order of the form @samp{32107654}. Note: this
9879 option should only be used if you require compatibility with code for
9880 big-endian ARM processors generated by versions of the compiler prior to
9883 @item -mcpu=@var{name}
9885 This specifies the name of the target ARM processor. GCC uses this name
9886 to determine what kind of instructions it can emit when generating
9887 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9888 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9889 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9890 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9891 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9893 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9894 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9895 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9896 @samp{strongarm1110},
9897 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9898 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9899 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9900 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9901 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9902 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9903 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9904 @samp{cortex-a5}, @samp{cortex-a8}, @samp{cortex-a9},
9905 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9908 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9910 @item -mtune=@var{name}
9912 This option is very similar to the @option{-mcpu=} option, except that
9913 instead of specifying the actual target processor type, and hence
9914 restricting which instructions can be used, it specifies that GCC should
9915 tune the performance of the code as if the target were of the type
9916 specified in this option, but still choosing the instructions that it
9917 will generate based on the cpu specified by a @option{-mcpu=} option.
9918 For some ARM implementations better performance can be obtained by using
9921 @item -march=@var{name}
9923 This specifies the name of the target ARM architecture. GCC uses this
9924 name to determine what kind of instructions it can emit when generating
9925 assembly code. This option can be used in conjunction with or instead
9926 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9927 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9928 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9929 @samp{armv6}, @samp{armv6j},
9930 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9931 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9932 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9934 @item -mfpu=@var{name}
9935 @itemx -mfpe=@var{number}
9936 @itemx -mfp=@var{number}
9940 This specifies what floating point hardware (or hardware emulation) is
9941 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9942 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
9943 @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
9944 @samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
9945 @samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
9946 @option{-mfp} and @option{-mfpe} are synonyms for
9947 @option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
9950 If @option{-msoft-float} is specified this specifies the format of
9951 floating point values.
9953 @item -mfp16-format=@var{name}
9954 @opindex mfp16-format
9955 Specify the format of the @code{__fp16} half-precision floating-point type.
9956 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
9957 the default is @samp{none}, in which case the @code{__fp16} type is not
9958 defined. @xref{Half-Precision}, for more information.
9960 @item -mstructure-size-boundary=@var{n}
9961 @opindex mstructure-size-boundary
9962 The size of all structures and unions will be rounded up to a multiple
9963 of the number of bits set by this option. Permissible values are 8, 32
9964 and 64. The default value varies for different toolchains. For the COFF
9965 targeted toolchain the default value is 8. A value of 64 is only allowed
9966 if the underlying ABI supports it.
9968 Specifying the larger number can produce faster, more efficient code, but
9969 can also increase the size of the program. Different values are potentially
9970 incompatible. Code compiled with one value cannot necessarily expect to
9971 work with code or libraries compiled with another value, if they exchange
9972 information using structures or unions.
9974 @item -mabort-on-noreturn
9975 @opindex mabort-on-noreturn
9976 Generate a call to the function @code{abort} at the end of a
9977 @code{noreturn} function. It will be executed if the function tries to
9981 @itemx -mno-long-calls
9982 @opindex mlong-calls
9983 @opindex mno-long-calls
9984 Tells the compiler to perform function calls by first loading the
9985 address of the function into a register and then performing a subroutine
9986 call on this register. This switch is needed if the target function
9987 will lie outside of the 64 megabyte addressing range of the offset based
9988 version of subroutine call instruction.
9990 Even if this switch is enabled, not all function calls will be turned
9991 into long calls. The heuristic is that static functions, functions
9992 which have the @samp{short-call} attribute, functions that are inside
9993 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9994 definitions have already been compiled within the current compilation
9995 unit, will not be turned into long calls. The exception to this rule is
9996 that weak function definitions, functions with the @samp{long-call}
9997 attribute or the @samp{section} attribute, and functions that are within
9998 the scope of a @samp{#pragma long_calls} directive, will always be
9999 turned into long calls.
10001 This feature is not enabled by default. Specifying
10002 @option{-mno-long-calls} will restore the default behavior, as will
10003 placing the function calls within the scope of a @samp{#pragma
10004 long_calls_off} directive. Note these switches have no effect on how
10005 the compiler generates code to handle function calls via function
10008 @item -msingle-pic-base
10009 @opindex msingle-pic-base
10010 Treat the register used for PIC addressing as read-only, rather than
10011 loading it in the prologue for each function. The run-time system is
10012 responsible for initializing this register with an appropriate value
10013 before execution begins.
10015 @item -mpic-register=@var{reg}
10016 @opindex mpic-register
10017 Specify the register to be used for PIC addressing. The default is R10
10018 unless stack-checking is enabled, when R9 is used.
10020 @item -mcirrus-fix-invalid-insns
10021 @opindex mcirrus-fix-invalid-insns
10022 @opindex mno-cirrus-fix-invalid-insns
10023 Insert NOPs into the instruction stream to in order to work around
10024 problems with invalid Maverick instruction combinations. This option
10025 is only valid if the @option{-mcpu=ep9312} option has been used to
10026 enable generation of instructions for the Cirrus Maverick floating
10027 point co-processor. This option is not enabled by default, since the
10028 problem is only present in older Maverick implementations. The default
10029 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
10032 @item -mpoke-function-name
10033 @opindex mpoke-function-name
10034 Write the name of each function into the text section, directly
10035 preceding the function prologue. The generated code is similar to this:
10039 .ascii "arm_poke_function_name", 0
10042 .word 0xff000000 + (t1 - t0)
10043 arm_poke_function_name
10045 stmfd sp!, @{fp, ip, lr, pc@}
10049 When performing a stack backtrace, code can inspect the value of
10050 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
10051 location @code{pc - 12} and the top 8 bits are set, then we know that
10052 there is a function name embedded immediately preceding this location
10053 and has length @code{((pc[-3]) & 0xff000000)}.
10057 Generate code for the Thumb instruction set. The default is to
10058 use the 32-bit ARM instruction set.
10059 This option automatically enables either 16-bit Thumb-1 or
10060 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
10061 and @option{-march=@var{name}} options. This option is not passed to the
10062 assembler. If you want to force assembler files to be interpreted as Thumb code,
10063 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
10064 option directly to the assembler by prefixing it with @option{-Wa}.
10067 @opindex mtpcs-frame
10068 Generate a stack frame that is compliant with the Thumb Procedure Call
10069 Standard for all non-leaf functions. (A leaf function is one that does
10070 not call any other functions.) The default is @option{-mno-tpcs-frame}.
10072 @item -mtpcs-leaf-frame
10073 @opindex mtpcs-leaf-frame
10074 Generate a stack frame that is compliant with the Thumb Procedure Call
10075 Standard for all leaf functions. (A leaf function is one that does
10076 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
10078 @item -mcallee-super-interworking
10079 @opindex mcallee-super-interworking
10080 Gives all externally visible functions in the file being compiled an ARM
10081 instruction set header which switches to Thumb mode before executing the
10082 rest of the function. This allows these functions to be called from
10083 non-interworking code. This option is not valid in AAPCS configurations
10084 because interworking is enabled by default.
10086 @item -mcaller-super-interworking
10087 @opindex mcaller-super-interworking
10088 Allows calls via function pointers (including virtual functions) to
10089 execute correctly regardless of whether the target code has been
10090 compiled for interworking or not. There is a small overhead in the cost
10091 of executing a function pointer if this option is enabled. This option
10092 is not valid in AAPCS configurations because interworking is enabled
10095 @item -mtp=@var{name}
10097 Specify the access model for the thread local storage pointer. The valid
10098 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10099 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
10100 (supported in the arm6k architecture), and @option{auto}, which uses the
10101 best available method for the selected processor. The default setting is
10104 @item -mword-relocations
10105 @opindex mword-relocations
10106 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
10107 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10108 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10111 @item -mfix-cortex-m3-ldrd
10112 @opindex mfix-cortex-m3-ldrd
10113 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
10114 with overlapping destination and base registers are used. This option avoids
10115 generating these instructions. This option is enabled by default when
10116 @option{-mcpu=cortex-m3} is specified.
10121 @subsection AVR Options
10122 @cindex AVR Options
10124 These options are defined for AVR implementations:
10127 @item -mmcu=@var{mcu}
10129 Specify ATMEL AVR instruction set or MCU type.
10131 Instruction set avr1 is for the minimal AVR core, not supported by the C
10132 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10133 attiny11, attiny12, attiny15, attiny28).
10135 Instruction set avr2 (default) is for the classic AVR core with up to
10136 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10137 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10138 at90c8534, at90s8535).
10140 Instruction set avr3 is for the classic AVR core with up to 128K program
10141 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10143 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10144 memory space (MCU types: atmega8, atmega83, atmega85).
10146 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10147 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10148 atmega64, atmega128, at43usb355, at94k).
10150 @item -mno-interrupts
10151 @opindex mno-interrupts
10152 Generated code is not compatible with hardware interrupts.
10153 Code size will be smaller.
10155 @item -mcall-prologues
10156 @opindex mcall-prologues
10157 Functions prologues/epilogues expanded as call to appropriate
10158 subroutines. Code size will be smaller.
10161 @opindex mtiny-stack
10162 Change only the low 8 bits of the stack pointer.
10166 Assume int to be 8 bit integer. This affects the sizes of all types: A
10167 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10168 and long long will be 4 bytes. Please note that this option does not
10169 comply to the C standards, but it will provide you with smaller code
10173 @node Blackfin Options
10174 @subsection Blackfin Options
10175 @cindex Blackfin Options
10178 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10180 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10181 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10182 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10183 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10184 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10185 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10186 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10188 The optional @var{sirevision} specifies the silicon revision of the target
10189 Blackfin processor. Any workarounds available for the targeted silicon revision
10190 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10191 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10192 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10193 hexadecimal digits representing the major and minor numbers in the silicon
10194 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10195 is not defined. If @var{sirevision} is @samp{any}, the
10196 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10197 If this optional @var{sirevision} is not used, GCC assumes the latest known
10198 silicon revision of the targeted Blackfin processor.
10200 Support for @samp{bf561} is incomplete. For @samp{bf561},
10201 Only the processor macro is defined.
10202 Without this option, @samp{bf532} is used as the processor by default.
10203 The corresponding predefined processor macros for @var{cpu} is to
10204 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10205 provided by libgloss to be linked in if @option{-msim} is not given.
10209 Specifies that the program will be run on the simulator. This causes
10210 the simulator BSP provided by libgloss to be linked in. This option
10211 has effect only for @samp{bfin-elf} toolchain.
10212 Certain other options, such as @option{-mid-shared-library} and
10213 @option{-mfdpic}, imply @option{-msim}.
10215 @item -momit-leaf-frame-pointer
10216 @opindex momit-leaf-frame-pointer
10217 Don't keep the frame pointer in a register for leaf functions. This
10218 avoids the instructions to save, set up and restore frame pointers and
10219 makes an extra register available in leaf functions. The option
10220 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10221 which might make debugging harder.
10223 @item -mspecld-anomaly
10224 @opindex mspecld-anomaly
10225 When enabled, the compiler will ensure that the generated code does not
10226 contain speculative loads after jump instructions. If this option is used,
10227 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10229 @item -mno-specld-anomaly
10230 @opindex mno-specld-anomaly
10231 Don't generate extra code to prevent speculative loads from occurring.
10233 @item -mcsync-anomaly
10234 @opindex mcsync-anomaly
10235 When enabled, the compiler will ensure that the generated code does not
10236 contain CSYNC or SSYNC instructions too soon after conditional branches.
10237 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10239 @item -mno-csync-anomaly
10240 @opindex mno-csync-anomaly
10241 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10242 occurring too soon after a conditional branch.
10246 When enabled, the compiler is free to take advantage of the knowledge that
10247 the entire program fits into the low 64k of memory.
10250 @opindex mno-low-64k
10251 Assume that the program is arbitrarily large. This is the default.
10253 @item -mstack-check-l1
10254 @opindex mstack-check-l1
10255 Do stack checking using information placed into L1 scratchpad memory by the
10258 @item -mid-shared-library
10259 @opindex mid-shared-library
10260 Generate code that supports shared libraries via the library ID method.
10261 This allows for execute in place and shared libraries in an environment
10262 without virtual memory management. This option implies @option{-fPIC}.
10263 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10265 @item -mno-id-shared-library
10266 @opindex mno-id-shared-library
10267 Generate code that doesn't assume ID based shared libraries are being used.
10268 This is the default.
10270 @item -mleaf-id-shared-library
10271 @opindex mleaf-id-shared-library
10272 Generate code that supports shared libraries via the library ID method,
10273 but assumes that this library or executable won't link against any other
10274 ID shared libraries. That allows the compiler to use faster code for jumps
10277 @item -mno-leaf-id-shared-library
10278 @opindex mno-leaf-id-shared-library
10279 Do not assume that the code being compiled won't link against any ID shared
10280 libraries. Slower code will be generated for jump and call insns.
10282 @item -mshared-library-id=n
10283 @opindex mshared-library-id
10284 Specified the identification number of the ID based shared library being
10285 compiled. Specifying a value of 0 will generate more compact code, specifying
10286 other values will force the allocation of that number to the current
10287 library but is no more space or time efficient than omitting this option.
10291 Generate code that allows the data segment to be located in a different
10292 area of memory from the text segment. This allows for execute in place in
10293 an environment without virtual memory management by eliminating relocations
10294 against the text section.
10296 @item -mno-sep-data
10297 @opindex mno-sep-data
10298 Generate code that assumes that the data segment follows the text segment.
10299 This is the default.
10302 @itemx -mno-long-calls
10303 @opindex mlong-calls
10304 @opindex mno-long-calls
10305 Tells the compiler to perform function calls by first loading the
10306 address of the function into a register and then performing a subroutine
10307 call on this register. This switch is needed if the target function
10308 will lie outside of the 24 bit addressing range of the offset based
10309 version of subroutine call instruction.
10311 This feature is not enabled by default. Specifying
10312 @option{-mno-long-calls} will restore the default behavior. Note these
10313 switches have no effect on how the compiler generates code to handle
10314 function calls via function pointers.
10318 Link with the fast floating-point library. This library relaxes some of
10319 the IEEE floating-point standard's rules for checking inputs against
10320 Not-a-Number (NAN), in the interest of performance.
10323 @opindex minline-plt
10324 Enable inlining of PLT entries in function calls to functions that are
10325 not known to bind locally. It has no effect without @option{-mfdpic}.
10328 @opindex mmulticore
10329 Build standalone application for multicore Blackfin processor. Proper
10330 start files and link scripts will be used to support multicore.
10331 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10332 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10333 @option{-mcorea} or @option{-mcoreb}. If it's used without
10334 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10335 programming model is used. In this model, the main function of Core B
10336 should be named as coreb_main. If it's used with @option{-mcorea} or
10337 @option{-mcoreb}, one application per core programming model is used.
10338 If this option is not used, single core application programming
10343 Build standalone application for Core A of BF561 when using
10344 one application per core programming model. Proper start files
10345 and link scripts will be used to support Core A. This option
10346 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10350 Build standalone application for Core B of BF561 when using
10351 one application per core programming model. Proper start files
10352 and link scripts will be used to support Core B. This option
10353 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10354 should be used instead of main. It must be used with
10355 @option{-mmulticore}.
10359 Build standalone application for SDRAM. Proper start files and
10360 link scripts will be used to put the application into SDRAM.
10361 Loader should initialize SDRAM before loading the application
10362 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10366 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10367 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10368 are enabled; for standalone applications the default is off.
10372 @subsection CRIS Options
10373 @cindex CRIS Options
10375 These options are defined specifically for the CRIS ports.
10378 @item -march=@var{architecture-type}
10379 @itemx -mcpu=@var{architecture-type}
10382 Generate code for the specified architecture. The choices for
10383 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10384 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10385 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10388 @item -mtune=@var{architecture-type}
10390 Tune to @var{architecture-type} everything applicable about the generated
10391 code, except for the ABI and the set of available instructions. The
10392 choices for @var{architecture-type} are the same as for
10393 @option{-march=@var{architecture-type}}.
10395 @item -mmax-stack-frame=@var{n}
10396 @opindex mmax-stack-frame
10397 Warn when the stack frame of a function exceeds @var{n} bytes.
10403 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10404 @option{-march=v3} and @option{-march=v8} respectively.
10406 @item -mmul-bug-workaround
10407 @itemx -mno-mul-bug-workaround
10408 @opindex mmul-bug-workaround
10409 @opindex mno-mul-bug-workaround
10410 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10411 models where it applies. This option is active by default.
10415 Enable CRIS-specific verbose debug-related information in the assembly
10416 code. This option also has the effect to turn off the @samp{#NO_APP}
10417 formatted-code indicator to the assembler at the beginning of the
10422 Do not use condition-code results from previous instruction; always emit
10423 compare and test instructions before use of condition codes.
10425 @item -mno-side-effects
10426 @opindex mno-side-effects
10427 Do not emit instructions with side-effects in addressing modes other than
10430 @item -mstack-align
10431 @itemx -mno-stack-align
10432 @itemx -mdata-align
10433 @itemx -mno-data-align
10434 @itemx -mconst-align
10435 @itemx -mno-const-align
10436 @opindex mstack-align
10437 @opindex mno-stack-align
10438 @opindex mdata-align
10439 @opindex mno-data-align
10440 @opindex mconst-align
10441 @opindex mno-const-align
10442 These options (no-options) arranges (eliminate arrangements) for the
10443 stack-frame, individual data and constants to be aligned for the maximum
10444 single data access size for the chosen CPU model. The default is to
10445 arrange for 32-bit alignment. ABI details such as structure layout are
10446 not affected by these options.
10454 Similar to the stack- data- and const-align options above, these options
10455 arrange for stack-frame, writable data and constants to all be 32-bit,
10456 16-bit or 8-bit aligned. The default is 32-bit alignment.
10458 @item -mno-prologue-epilogue
10459 @itemx -mprologue-epilogue
10460 @opindex mno-prologue-epilogue
10461 @opindex mprologue-epilogue
10462 With @option{-mno-prologue-epilogue}, the normal function prologue and
10463 epilogue that sets up the stack-frame are omitted and no return
10464 instructions or return sequences are generated in the code. Use this
10465 option only together with visual inspection of the compiled code: no
10466 warnings or errors are generated when call-saved registers must be saved,
10467 or storage for local variable needs to be allocated.
10471 @opindex mno-gotplt
10473 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10474 instruction sequences that load addresses for functions from the PLT part
10475 of the GOT rather than (traditional on other architectures) calls to the
10476 PLT@. The default is @option{-mgotplt}.
10480 Legacy no-op option only recognized with the cris-axis-elf and
10481 cris-axis-linux-gnu targets.
10485 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10489 This option, recognized for the cris-axis-elf arranges
10490 to link with input-output functions from a simulator library. Code,
10491 initialized data and zero-initialized data are allocated consecutively.
10495 Like @option{-sim}, but pass linker options to locate initialized data at
10496 0x40000000 and zero-initialized data at 0x80000000.
10500 @subsection CRX Options
10501 @cindex CRX Options
10503 These options are defined specifically for the CRX ports.
10509 Enable the use of multiply-accumulate instructions. Disabled by default.
10512 @opindex mpush-args
10513 Push instructions will be used to pass outgoing arguments when functions
10514 are called. Enabled by default.
10517 @node Darwin Options
10518 @subsection Darwin Options
10519 @cindex Darwin options
10521 These options are defined for all architectures running the Darwin operating
10524 FSF GCC on Darwin does not create ``fat'' object files; it will create
10525 an object file for the single architecture that it was built to
10526 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10527 @option{-arch} options are used; it does so by running the compiler or
10528 linker multiple times and joining the results together with
10531 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10532 @samp{i686}) is determined by the flags that specify the ISA
10533 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10534 @option{-force_cpusubtype_ALL} option can be used to override this.
10536 The Darwin tools vary in their behavior when presented with an ISA
10537 mismatch. The assembler, @file{as}, will only permit instructions to
10538 be used that are valid for the subtype of the file it is generating,
10539 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10540 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10541 and print an error if asked to create a shared library with a less
10542 restrictive subtype than its input files (for instance, trying to put
10543 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10544 for executables, @file{ld}, will quietly give the executable the most
10545 restrictive subtype of any of its input files.
10550 Add the framework directory @var{dir} to the head of the list of
10551 directories to be searched for header files. These directories are
10552 interleaved with those specified by @option{-I} options and are
10553 scanned in a left-to-right order.
10555 A framework directory is a directory with frameworks in it. A
10556 framework is a directory with a @samp{"Headers"} and/or
10557 @samp{"PrivateHeaders"} directory contained directly in it that ends
10558 in @samp{".framework"}. The name of a framework is the name of this
10559 directory excluding the @samp{".framework"}. Headers associated with
10560 the framework are found in one of those two directories, with
10561 @samp{"Headers"} being searched first. A subframework is a framework
10562 directory that is in a framework's @samp{"Frameworks"} directory.
10563 Includes of subframework headers can only appear in a header of a
10564 framework that contains the subframework, or in a sibling subframework
10565 header. Two subframeworks are siblings if they occur in the same
10566 framework. A subframework should not have the same name as a
10567 framework, a warning will be issued if this is violated. Currently a
10568 subframework cannot have subframeworks, in the future, the mechanism
10569 may be extended to support this. The standard frameworks can be found
10570 in @samp{"/System/Library/Frameworks"} and
10571 @samp{"/Library/Frameworks"}. An example include looks like
10572 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10573 the name of the framework and header.h is found in the
10574 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10576 @item -iframework@var{dir}
10577 @opindex iframework
10578 Like @option{-F} except the directory is a treated as a system
10579 directory. The main difference between this @option{-iframework} and
10580 @option{-F} is that with @option{-iframework} the compiler does not
10581 warn about constructs contained within header files found via
10582 @var{dir}. This option is valid only for the C family of languages.
10586 Emit debugging information for symbols that are used. For STABS
10587 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10588 This is by default ON@.
10592 Emit debugging information for all symbols and types.
10594 @item -mmacosx-version-min=@var{version}
10595 The earliest version of MacOS X that this executable will run on
10596 is @var{version}. Typical values of @var{version} include @code{10.1},
10597 @code{10.2}, and @code{10.3.9}.
10599 If the compiler was built to use the system's headers by default,
10600 then the default for this option is the system version on which the
10601 compiler is running, otherwise the default is to make choices which
10602 are compatible with as many systems and code bases as possible.
10606 Enable kernel development mode. The @option{-mkernel} option sets
10607 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10608 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10609 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10610 applicable. This mode also sets @option{-mno-altivec},
10611 @option{-msoft-float}, @option{-fno-builtin} and
10612 @option{-mlong-branch} for PowerPC targets.
10614 @item -mone-byte-bool
10615 @opindex mone-byte-bool
10616 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10617 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10618 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10619 option has no effect on x86.
10621 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10622 to generate code that is not binary compatible with code generated
10623 without that switch. Using this switch may require recompiling all
10624 other modules in a program, including system libraries. Use this
10625 switch to conform to a non-default data model.
10627 @item -mfix-and-continue
10628 @itemx -ffix-and-continue
10629 @itemx -findirect-data
10630 @opindex mfix-and-continue
10631 @opindex ffix-and-continue
10632 @opindex findirect-data
10633 Generate code suitable for fast turn around development. Needed to
10634 enable gdb to dynamically load @code{.o} files into already running
10635 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10636 are provided for backwards compatibility.
10640 Loads all members of static archive libraries.
10641 See man ld(1) for more information.
10643 @item -arch_errors_fatal
10644 @opindex arch_errors_fatal
10645 Cause the errors having to do with files that have the wrong architecture
10648 @item -bind_at_load
10649 @opindex bind_at_load
10650 Causes the output file to be marked such that the dynamic linker will
10651 bind all undefined references when the file is loaded or launched.
10655 Produce a Mach-o bundle format file.
10656 See man ld(1) for more information.
10658 @item -bundle_loader @var{executable}
10659 @opindex bundle_loader
10660 This option specifies the @var{executable} that will be loading the build
10661 output file being linked. See man ld(1) for more information.
10664 @opindex dynamiclib
10665 When passed this option, GCC will produce a dynamic library instead of
10666 an executable when linking, using the Darwin @file{libtool} command.
10668 @item -force_cpusubtype_ALL
10669 @opindex force_cpusubtype_ALL
10670 This causes GCC's output file to have the @var{ALL} subtype, instead of
10671 one controlled by the @option{-mcpu} or @option{-march} option.
10673 @item -allowable_client @var{client_name}
10674 @itemx -client_name
10675 @itemx -compatibility_version
10676 @itemx -current_version
10678 @itemx -dependency-file
10680 @itemx -dylinker_install_name
10682 @itemx -exported_symbols_list
10684 @itemx -flat_namespace
10685 @itemx -force_flat_namespace
10686 @itemx -headerpad_max_install_names
10689 @itemx -install_name
10690 @itemx -keep_private_externs
10691 @itemx -multi_module
10692 @itemx -multiply_defined
10693 @itemx -multiply_defined_unused
10695 @itemx -no_dead_strip_inits_and_terms
10696 @itemx -nofixprebinding
10697 @itemx -nomultidefs
10699 @itemx -noseglinkedit
10700 @itemx -pagezero_size
10702 @itemx -prebind_all_twolevel_modules
10703 @itemx -private_bundle
10704 @itemx -read_only_relocs
10706 @itemx -sectobjectsymbols
10710 @itemx -sectobjectsymbols
10713 @itemx -segs_read_only_addr
10714 @itemx -segs_read_write_addr
10715 @itemx -seg_addr_table
10716 @itemx -seg_addr_table_filename
10717 @itemx -seglinkedit
10719 @itemx -segs_read_only_addr
10720 @itemx -segs_read_write_addr
10721 @itemx -single_module
10723 @itemx -sub_library
10724 @itemx -sub_umbrella
10725 @itemx -twolevel_namespace
10728 @itemx -unexported_symbols_list
10729 @itemx -weak_reference_mismatches
10730 @itemx -whatsloaded
10731 @opindex allowable_client
10732 @opindex client_name
10733 @opindex compatibility_version
10734 @opindex current_version
10735 @opindex dead_strip
10736 @opindex dependency-file
10737 @opindex dylib_file
10738 @opindex dylinker_install_name
10740 @opindex exported_symbols_list
10742 @opindex flat_namespace
10743 @opindex force_flat_namespace
10744 @opindex headerpad_max_install_names
10745 @opindex image_base
10747 @opindex install_name
10748 @opindex keep_private_externs
10749 @opindex multi_module
10750 @opindex multiply_defined
10751 @opindex multiply_defined_unused
10752 @opindex noall_load
10753 @opindex no_dead_strip_inits_and_terms
10754 @opindex nofixprebinding
10755 @opindex nomultidefs
10757 @opindex noseglinkedit
10758 @opindex pagezero_size
10760 @opindex prebind_all_twolevel_modules
10761 @opindex private_bundle
10762 @opindex read_only_relocs
10764 @opindex sectobjectsymbols
10767 @opindex sectcreate
10768 @opindex sectobjectsymbols
10771 @opindex segs_read_only_addr
10772 @opindex segs_read_write_addr
10773 @opindex seg_addr_table
10774 @opindex seg_addr_table_filename
10775 @opindex seglinkedit
10777 @opindex segs_read_only_addr
10778 @opindex segs_read_write_addr
10779 @opindex single_module
10781 @opindex sub_library
10782 @opindex sub_umbrella
10783 @opindex twolevel_namespace
10786 @opindex unexported_symbols_list
10787 @opindex weak_reference_mismatches
10788 @opindex whatsloaded
10789 These options are passed to the Darwin linker. The Darwin linker man page
10790 describes them in detail.
10793 @node DEC Alpha Options
10794 @subsection DEC Alpha Options
10796 These @samp{-m} options are defined for the DEC Alpha implementations:
10799 @item -mno-soft-float
10800 @itemx -msoft-float
10801 @opindex mno-soft-float
10802 @opindex msoft-float
10803 Use (do not use) the hardware floating-point instructions for
10804 floating-point operations. When @option{-msoft-float} is specified,
10805 functions in @file{libgcc.a} will be used to perform floating-point
10806 operations. Unless they are replaced by routines that emulate the
10807 floating-point operations, or compiled in such a way as to call such
10808 emulations routines, these routines will issue floating-point
10809 operations. If you are compiling for an Alpha without floating-point
10810 operations, you must ensure that the library is built so as not to call
10813 Note that Alpha implementations without floating-point operations are
10814 required to have floating-point registers.
10817 @itemx -mno-fp-regs
10819 @opindex mno-fp-regs
10820 Generate code that uses (does not use) the floating-point register set.
10821 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10822 register set is not used, floating point operands are passed in integer
10823 registers as if they were integers and floating-point results are passed
10824 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10825 so any function with a floating-point argument or return value called by code
10826 compiled with @option{-mno-fp-regs} must also be compiled with that
10829 A typical use of this option is building a kernel that does not use,
10830 and hence need not save and restore, any floating-point registers.
10834 The Alpha architecture implements floating-point hardware optimized for
10835 maximum performance. It is mostly compliant with the IEEE floating
10836 point standard. However, for full compliance, software assistance is
10837 required. This option generates code fully IEEE compliant code
10838 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10839 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10840 defined during compilation. The resulting code is less efficient but is
10841 able to correctly support denormalized numbers and exceptional IEEE
10842 values such as not-a-number and plus/minus infinity. Other Alpha
10843 compilers call this option @option{-ieee_with_no_inexact}.
10845 @item -mieee-with-inexact
10846 @opindex mieee-with-inexact
10847 This is like @option{-mieee} except the generated code also maintains
10848 the IEEE @var{inexact-flag}. Turning on this option causes the
10849 generated code to implement fully-compliant IEEE math. In addition to
10850 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10851 macro. On some Alpha implementations the resulting code may execute
10852 significantly slower than the code generated by default. Since there is
10853 very little code that depends on the @var{inexact-flag}, you should
10854 normally not specify this option. Other Alpha compilers call this
10855 option @option{-ieee_with_inexact}.
10857 @item -mfp-trap-mode=@var{trap-mode}
10858 @opindex mfp-trap-mode
10859 This option controls what floating-point related traps are enabled.
10860 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10861 The trap mode can be set to one of four values:
10865 This is the default (normal) setting. The only traps that are enabled
10866 are the ones that cannot be disabled in software (e.g., division by zero
10870 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10874 Like @samp{u}, but the instructions are marked to be safe for software
10875 completion (see Alpha architecture manual for details).
10878 Like @samp{su}, but inexact traps are enabled as well.
10881 @item -mfp-rounding-mode=@var{rounding-mode}
10882 @opindex mfp-rounding-mode
10883 Selects the IEEE rounding mode. Other Alpha compilers call this option
10884 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10889 Normal IEEE rounding mode. Floating point numbers are rounded towards
10890 the nearest machine number or towards the even machine number in case
10894 Round towards minus infinity.
10897 Chopped rounding mode. Floating point numbers are rounded towards zero.
10900 Dynamic rounding mode. A field in the floating point control register
10901 (@var{fpcr}, see Alpha architecture reference manual) controls the
10902 rounding mode in effect. The C library initializes this register for
10903 rounding towards plus infinity. Thus, unless your program modifies the
10904 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10907 @item -mtrap-precision=@var{trap-precision}
10908 @opindex mtrap-precision
10909 In the Alpha architecture, floating point traps are imprecise. This
10910 means without software assistance it is impossible to recover from a
10911 floating trap and program execution normally needs to be terminated.
10912 GCC can generate code that can assist operating system trap handlers
10913 in determining the exact location that caused a floating point trap.
10914 Depending on the requirements of an application, different levels of
10915 precisions can be selected:
10919 Program precision. This option is the default and means a trap handler
10920 can only identify which program caused a floating point exception.
10923 Function precision. The trap handler can determine the function that
10924 caused a floating point exception.
10927 Instruction precision. The trap handler can determine the exact
10928 instruction that caused a floating point exception.
10931 Other Alpha compilers provide the equivalent options called
10932 @option{-scope_safe} and @option{-resumption_safe}.
10934 @item -mieee-conformant
10935 @opindex mieee-conformant
10936 This option marks the generated code as IEEE conformant. You must not
10937 use this option unless you also specify @option{-mtrap-precision=i} and either
10938 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10939 is to emit the line @samp{.eflag 48} in the function prologue of the
10940 generated assembly file. Under DEC Unix, this has the effect that
10941 IEEE-conformant math library routines will be linked in.
10943 @item -mbuild-constants
10944 @opindex mbuild-constants
10945 Normally GCC examines a 32- or 64-bit integer constant to
10946 see if it can construct it from smaller constants in two or three
10947 instructions. If it cannot, it will output the constant as a literal and
10948 generate code to load it from the data segment at runtime.
10950 Use this option to require GCC to construct @emph{all} integer constants
10951 using code, even if it takes more instructions (the maximum is six).
10953 You would typically use this option to build a shared library dynamic
10954 loader. Itself a shared library, it must relocate itself in memory
10955 before it can find the variables and constants in its own data segment.
10961 Select whether to generate code to be assembled by the vendor-supplied
10962 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10980 Indicate whether GCC should generate code to use the optional BWX,
10981 CIX, FIX and MAX instruction sets. The default is to use the instruction
10982 sets supported by the CPU type specified via @option{-mcpu=} option or that
10983 of the CPU on which GCC was built if none was specified.
10986 @itemx -mfloat-ieee
10987 @opindex mfloat-vax
10988 @opindex mfloat-ieee
10989 Generate code that uses (does not use) VAX F and G floating point
10990 arithmetic instead of IEEE single and double precision.
10992 @item -mexplicit-relocs
10993 @itemx -mno-explicit-relocs
10994 @opindex mexplicit-relocs
10995 @opindex mno-explicit-relocs
10996 Older Alpha assemblers provided no way to generate symbol relocations
10997 except via assembler macros. Use of these macros does not allow
10998 optimal instruction scheduling. GNU binutils as of version 2.12
10999 supports a new syntax that allows the compiler to explicitly mark
11000 which relocations should apply to which instructions. This option
11001 is mostly useful for debugging, as GCC detects the capabilities of
11002 the assembler when it is built and sets the default accordingly.
11005 @itemx -mlarge-data
11006 @opindex msmall-data
11007 @opindex mlarge-data
11008 When @option{-mexplicit-relocs} is in effect, static data is
11009 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
11010 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
11011 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
11012 16-bit relocations off of the @code{$gp} register. This limits the
11013 size of the small data area to 64KB, but allows the variables to be
11014 directly accessed via a single instruction.
11016 The default is @option{-mlarge-data}. With this option the data area
11017 is limited to just below 2GB@. Programs that require more than 2GB of
11018 data must use @code{malloc} or @code{mmap} to allocate the data in the
11019 heap instead of in the program's data segment.
11021 When generating code for shared libraries, @option{-fpic} implies
11022 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
11025 @itemx -mlarge-text
11026 @opindex msmall-text
11027 @opindex mlarge-text
11028 When @option{-msmall-text} is used, the compiler assumes that the
11029 code of the entire program (or shared library) fits in 4MB, and is
11030 thus reachable with a branch instruction. When @option{-msmall-data}
11031 is used, the compiler can assume that all local symbols share the
11032 same @code{$gp} value, and thus reduce the number of instructions
11033 required for a function call from 4 to 1.
11035 The default is @option{-mlarge-text}.
11037 @item -mcpu=@var{cpu_type}
11039 Set the instruction set and instruction scheduling parameters for
11040 machine type @var{cpu_type}. You can specify either the @samp{EV}
11041 style name or the corresponding chip number. GCC supports scheduling
11042 parameters for the EV4, EV5 and EV6 family of processors and will
11043 choose the default values for the instruction set from the processor
11044 you specify. If you do not specify a processor type, GCC will default
11045 to the processor on which the compiler was built.
11047 Supported values for @var{cpu_type} are
11053 Schedules as an EV4 and has no instruction set extensions.
11057 Schedules as an EV5 and has no instruction set extensions.
11061 Schedules as an EV5 and supports the BWX extension.
11066 Schedules as an EV5 and supports the BWX and MAX extensions.
11070 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
11074 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
11077 Native Linux/GNU toolchains also support the value @samp{native},
11078 which selects the best architecture option for the host processor.
11079 @option{-mcpu=native} has no effect if GCC does not recognize
11082 @item -mtune=@var{cpu_type}
11084 Set only the instruction scheduling parameters for machine type
11085 @var{cpu_type}. The instruction set is not changed.
11087 Native Linux/GNU toolchains also support the value @samp{native},
11088 which selects the best architecture option for the host processor.
11089 @option{-mtune=native} has no effect if GCC does not recognize
11092 @item -mmemory-latency=@var{time}
11093 @opindex mmemory-latency
11094 Sets the latency the scheduler should assume for typical memory
11095 references as seen by the application. This number is highly
11096 dependent on the memory access patterns used by the application
11097 and the size of the external cache on the machine.
11099 Valid options for @var{time} are
11103 A decimal number representing clock cycles.
11109 The compiler contains estimates of the number of clock cycles for
11110 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
11111 (also called Dcache, Scache, and Bcache), as well as to main memory.
11112 Note that L3 is only valid for EV5.
11117 @node DEC Alpha/VMS Options
11118 @subsection DEC Alpha/VMS Options
11120 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
11123 @item -mvms-return-codes
11124 @opindex mvms-return-codes
11125 Return VMS condition codes from main. The default is to return POSIX
11126 style condition (e.g.@: error) codes.
11128 @item -mdebug-main=@var{prefix}
11129 @opindex mdebug-main=@var{prefix}
11130 Flag the first routine whose name starts with @var{prefix} as the main
11131 routine for the debugger.
11135 Default to 64bit memory allocation routines.
11139 @subsection FR30 Options
11140 @cindex FR30 Options
11142 These options are defined specifically for the FR30 port.
11146 @item -msmall-model
11147 @opindex msmall-model
11148 Use the small address space model. This can produce smaller code, but
11149 it does assume that all symbolic values and addresses will fit into a
11154 Assume that run-time support has been provided and so there is no need
11155 to include the simulator library (@file{libsim.a}) on the linker
11161 @subsection FRV Options
11162 @cindex FRV Options
11168 Only use the first 32 general purpose registers.
11173 Use all 64 general purpose registers.
11178 Use only the first 32 floating point registers.
11183 Use all 64 floating point registers
11186 @opindex mhard-float
11188 Use hardware instructions for floating point operations.
11191 @opindex msoft-float
11193 Use library routines for floating point operations.
11198 Dynamically allocate condition code registers.
11203 Do not try to dynamically allocate condition code registers, only
11204 use @code{icc0} and @code{fcc0}.
11209 Change ABI to use double word insns.
11214 Do not use double word instructions.
11219 Use floating point double instructions.
11222 @opindex mno-double
11224 Do not use floating point double instructions.
11229 Use media instructions.
11234 Do not use media instructions.
11239 Use multiply and add/subtract instructions.
11242 @opindex mno-muladd
11244 Do not use multiply and add/subtract instructions.
11249 Select the FDPIC ABI, that uses function descriptors to represent
11250 pointers to functions. Without any PIC/PIE-related options, it
11251 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11252 assumes GOT entries and small data are within a 12-bit range from the
11253 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11254 are computed with 32 bits.
11255 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11258 @opindex minline-plt
11260 Enable inlining of PLT entries in function calls to functions that are
11261 not known to bind locally. It has no effect without @option{-mfdpic}.
11262 It's enabled by default if optimizing for speed and compiling for
11263 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11264 optimization option such as @option{-O3} or above is present in the
11270 Assume a large TLS segment when generating thread-local code.
11275 Do not assume a large TLS segment when generating thread-local code.
11280 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11281 that is known to be in read-only sections. It's enabled by default,
11282 except for @option{-fpic} or @option{-fpie}: even though it may help
11283 make the global offset table smaller, it trades 1 instruction for 4.
11284 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11285 one of which may be shared by multiple symbols, and it avoids the need
11286 for a GOT entry for the referenced symbol, so it's more likely to be a
11287 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11289 @item -multilib-library-pic
11290 @opindex multilib-library-pic
11292 Link with the (library, not FD) pic libraries. It's implied by
11293 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11294 @option{-fpic} without @option{-mfdpic}. You should never have to use
11298 @opindex mlinked-fp
11300 Follow the EABI requirement of always creating a frame pointer whenever
11301 a stack frame is allocated. This option is enabled by default and can
11302 be disabled with @option{-mno-linked-fp}.
11305 @opindex mlong-calls
11307 Use indirect addressing to call functions outside the current
11308 compilation unit. This allows the functions to be placed anywhere
11309 within the 32-bit address space.
11311 @item -malign-labels
11312 @opindex malign-labels
11314 Try to align labels to an 8-byte boundary by inserting nops into the
11315 previous packet. This option only has an effect when VLIW packing
11316 is enabled. It doesn't create new packets; it merely adds nops to
11319 @item -mlibrary-pic
11320 @opindex mlibrary-pic
11322 Generate position-independent EABI code.
11327 Use only the first four media accumulator registers.
11332 Use all eight media accumulator registers.
11337 Pack VLIW instructions.
11342 Do not pack VLIW instructions.
11345 @opindex mno-eflags
11347 Do not mark ABI switches in e_flags.
11350 @opindex mcond-move
11352 Enable the use of conditional-move instructions (default).
11354 This switch is mainly for debugging the compiler and will likely be removed
11355 in a future version.
11357 @item -mno-cond-move
11358 @opindex mno-cond-move
11360 Disable the use of conditional-move instructions.
11362 This switch is mainly for debugging the compiler and will likely be removed
11363 in a future version.
11368 Enable the use of conditional set instructions (default).
11370 This switch is mainly for debugging the compiler and will likely be removed
11371 in a future version.
11376 Disable the use of conditional set instructions.
11378 This switch is mainly for debugging the compiler and will likely be removed
11379 in a future version.
11382 @opindex mcond-exec
11384 Enable the use of conditional execution (default).
11386 This switch is mainly for debugging the compiler and will likely be removed
11387 in a future version.
11389 @item -mno-cond-exec
11390 @opindex mno-cond-exec
11392 Disable the use of conditional execution.
11394 This switch is mainly for debugging the compiler and will likely be removed
11395 in a future version.
11397 @item -mvliw-branch
11398 @opindex mvliw-branch
11400 Run a pass to pack branches into VLIW instructions (default).
11402 This switch is mainly for debugging the compiler and will likely be removed
11403 in a future version.
11405 @item -mno-vliw-branch
11406 @opindex mno-vliw-branch
11408 Do not run a pass to pack branches into VLIW instructions.
11410 This switch is mainly for debugging the compiler and will likely be removed
11411 in a future version.
11413 @item -mmulti-cond-exec
11414 @opindex mmulti-cond-exec
11416 Enable optimization of @code{&&} and @code{||} in conditional execution
11419 This switch is mainly for debugging the compiler and will likely be removed
11420 in a future version.
11422 @item -mno-multi-cond-exec
11423 @opindex mno-multi-cond-exec
11425 Disable optimization of @code{&&} and @code{||} in conditional execution.
11427 This switch is mainly for debugging the compiler and will likely be removed
11428 in a future version.
11430 @item -mnested-cond-exec
11431 @opindex mnested-cond-exec
11433 Enable nested conditional execution optimizations (default).
11435 This switch is mainly for debugging the compiler and will likely be removed
11436 in a future version.
11438 @item -mno-nested-cond-exec
11439 @opindex mno-nested-cond-exec
11441 Disable nested conditional execution optimizations.
11443 This switch is mainly for debugging the compiler and will likely be removed
11444 in a future version.
11446 @item -moptimize-membar
11447 @opindex moptimize-membar
11449 This switch removes redundant @code{membar} instructions from the
11450 compiler generated code. It is enabled by default.
11452 @item -mno-optimize-membar
11453 @opindex mno-optimize-membar
11455 This switch disables the automatic removal of redundant @code{membar}
11456 instructions from the generated code.
11458 @item -mtomcat-stats
11459 @opindex mtomcat-stats
11461 Cause gas to print out tomcat statistics.
11463 @item -mcpu=@var{cpu}
11466 Select the processor type for which to generate code. Possible values are
11467 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11468 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11472 @node GNU/Linux Options
11473 @subsection GNU/Linux Options
11475 These @samp{-m} options are defined for GNU/Linux targets:
11480 Use the GNU C library instead of uClibc. This is the default except
11481 on @samp{*-*-linux-*uclibc*} targets.
11485 Use uClibc instead of the GNU C library. This is the default on
11486 @samp{*-*-linux-*uclibc*} targets.
11489 @node H8/300 Options
11490 @subsection H8/300 Options
11492 These @samp{-m} options are defined for the H8/300 implementations:
11497 Shorten some address references at link time, when possible; uses the
11498 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11499 ld, Using ld}, for a fuller description.
11503 Generate code for the H8/300H@.
11507 Generate code for the H8S@.
11511 Generate code for the H8S and H8/300H in the normal mode. This switch
11512 must be used either with @option{-mh} or @option{-ms}.
11516 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11520 Make @code{int} data 32 bits by default.
11523 @opindex malign-300
11524 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11525 The default for the H8/300H and H8S is to align longs and floats on 4
11527 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11528 This option has no effect on the H8/300.
11532 @subsection HPPA Options
11533 @cindex HPPA Options
11535 These @samp{-m} options are defined for the HPPA family of computers:
11538 @item -march=@var{architecture-type}
11540 Generate code for the specified architecture. The choices for
11541 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11542 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11543 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11544 architecture option for your machine. Code compiled for lower numbered
11545 architectures will run on higher numbered architectures, but not the
11548 @item -mpa-risc-1-0
11549 @itemx -mpa-risc-1-1
11550 @itemx -mpa-risc-2-0
11551 @opindex mpa-risc-1-0
11552 @opindex mpa-risc-1-1
11553 @opindex mpa-risc-2-0
11554 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11557 @opindex mbig-switch
11558 Generate code suitable for big switch tables. Use this option only if
11559 the assembler/linker complain about out of range branches within a switch
11562 @item -mjump-in-delay
11563 @opindex mjump-in-delay
11564 Fill delay slots of function calls with unconditional jump instructions
11565 by modifying the return pointer for the function call to be the target
11566 of the conditional jump.
11568 @item -mdisable-fpregs
11569 @opindex mdisable-fpregs
11570 Prevent floating point registers from being used in any manner. This is
11571 necessary for compiling kernels which perform lazy context switching of
11572 floating point registers. If you use this option and attempt to perform
11573 floating point operations, the compiler will abort.
11575 @item -mdisable-indexing
11576 @opindex mdisable-indexing
11577 Prevent the compiler from using indexing address modes. This avoids some
11578 rather obscure problems when compiling MIG generated code under MACH@.
11580 @item -mno-space-regs
11581 @opindex mno-space-regs
11582 Generate code that assumes the target has no space registers. This allows
11583 GCC to generate faster indirect calls and use unscaled index address modes.
11585 Such code is suitable for level 0 PA systems and kernels.
11587 @item -mfast-indirect-calls
11588 @opindex mfast-indirect-calls
11589 Generate code that assumes calls never cross space boundaries. This
11590 allows GCC to emit code which performs faster indirect calls.
11592 This option will not work in the presence of shared libraries or nested
11595 @item -mfixed-range=@var{register-range}
11596 @opindex mfixed-range
11597 Generate code treating the given register range as fixed registers.
11598 A fixed register is one that the register allocator can not use. This is
11599 useful when compiling kernel code. A register range is specified as
11600 two registers separated by a dash. Multiple register ranges can be
11601 specified separated by a comma.
11603 @item -mlong-load-store
11604 @opindex mlong-load-store
11605 Generate 3-instruction load and store sequences as sometimes required by
11606 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11609 @item -mportable-runtime
11610 @opindex mportable-runtime
11611 Use the portable calling conventions proposed by HP for ELF systems.
11615 Enable the use of assembler directives only GAS understands.
11617 @item -mschedule=@var{cpu-type}
11619 Schedule code according to the constraints for the machine type
11620 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11621 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11622 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11623 proper scheduling option for your machine. The default scheduling is
11627 @opindex mlinker-opt
11628 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11629 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11630 linkers in which they give bogus error messages when linking some programs.
11633 @opindex msoft-float
11634 Generate output containing library calls for floating point.
11635 @strong{Warning:} the requisite libraries are not available for all HPPA
11636 targets. Normally the facilities of the machine's usual C compiler are
11637 used, but this cannot be done directly in cross-compilation. You must make
11638 your own arrangements to provide suitable library functions for
11641 @option{-msoft-float} changes the calling convention in the output file;
11642 therefore, it is only useful if you compile @emph{all} of a program with
11643 this option. In particular, you need to compile @file{libgcc.a}, the
11644 library that comes with GCC, with @option{-msoft-float} in order for
11649 Generate the predefine, @code{_SIO}, for server IO@. The default is
11650 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11651 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11652 options are available under HP-UX and HI-UX@.
11656 Use GNU ld specific options. This passes @option{-shared} to ld when
11657 building a shared library. It is the default when GCC is configured,
11658 explicitly or implicitly, with the GNU linker. This option does not
11659 have any affect on which ld is called, it only changes what parameters
11660 are passed to that ld. The ld that is called is determined by the
11661 @option{--with-ld} configure option, GCC's program search path, and
11662 finally by the user's @env{PATH}. The linker used by GCC can be printed
11663 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11664 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11668 Use HP ld specific options. This passes @option{-b} to ld when building
11669 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11670 links. It is the default when GCC is configured, explicitly or
11671 implicitly, with the HP linker. This option does not have any affect on
11672 which ld is called, it only changes what parameters are passed to that
11673 ld. The ld that is called is determined by the @option{--with-ld}
11674 configure option, GCC's program search path, and finally by the user's
11675 @env{PATH}. The linker used by GCC can be printed using @samp{which
11676 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11677 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11680 @opindex mno-long-calls
11681 Generate code that uses long call sequences. This ensures that a call
11682 is always able to reach linker generated stubs. The default is to generate
11683 long calls only when the distance from the call site to the beginning
11684 of the function or translation unit, as the case may be, exceeds a
11685 predefined limit set by the branch type being used. The limits for
11686 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11687 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11690 Distances are measured from the beginning of functions when using the
11691 @option{-ffunction-sections} option, or when using the @option{-mgas}
11692 and @option{-mno-portable-runtime} options together under HP-UX with
11695 It is normally not desirable to use this option as it will degrade
11696 performance. However, it may be useful in large applications,
11697 particularly when partial linking is used to build the application.
11699 The types of long calls used depends on the capabilities of the
11700 assembler and linker, and the type of code being generated. The
11701 impact on systems that support long absolute calls, and long pic
11702 symbol-difference or pc-relative calls should be relatively small.
11703 However, an indirect call is used on 32-bit ELF systems in pic code
11704 and it is quite long.
11706 @item -munix=@var{unix-std}
11708 Generate compiler predefines and select a startfile for the specified
11709 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11710 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11711 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11712 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11713 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11716 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11717 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11718 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11719 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11720 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11721 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11723 It is @emph{important} to note that this option changes the interfaces
11724 for various library routines. It also affects the operational behavior
11725 of the C library. Thus, @emph{extreme} care is needed in using this
11728 Library code that is intended to operate with more than one UNIX
11729 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11730 as appropriate. Most GNU software doesn't provide this capability.
11734 Suppress the generation of link options to search libdld.sl when the
11735 @option{-static} option is specified on HP-UX 10 and later.
11739 The HP-UX implementation of setlocale in libc has a dependency on
11740 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11741 when the @option{-static} option is specified, special link options
11742 are needed to resolve this dependency.
11744 On HP-UX 10 and later, the GCC driver adds the necessary options to
11745 link with libdld.sl when the @option{-static} option is specified.
11746 This causes the resulting binary to be dynamic. On the 64-bit port,
11747 the linkers generate dynamic binaries by default in any case. The
11748 @option{-nolibdld} option can be used to prevent the GCC driver from
11749 adding these link options.
11753 Add support for multithreading with the @dfn{dce thread} library
11754 under HP-UX@. This option sets flags for both the preprocessor and
11758 @node i386 and x86-64 Options
11759 @subsection Intel 386 and AMD x86-64 Options
11760 @cindex i386 Options
11761 @cindex x86-64 Options
11762 @cindex Intel 386 Options
11763 @cindex AMD x86-64 Options
11765 These @samp{-m} options are defined for the i386 and x86-64 family of
11769 @item -mtune=@var{cpu-type}
11771 Tune to @var{cpu-type} everything applicable about the generated code, except
11772 for the ABI and the set of available instructions. The choices for
11773 @var{cpu-type} are:
11776 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11777 If you know the CPU on which your code will run, then you should use
11778 the corresponding @option{-mtune} option instead of
11779 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11780 of your application will have, then you should use this option.
11782 As new processors are deployed in the marketplace, the behavior of this
11783 option will change. Therefore, if you upgrade to a newer version of
11784 GCC, the code generated option will change to reflect the processors
11785 that were most common when that version of GCC was released.
11787 There is no @option{-march=generic} option because @option{-march}
11788 indicates the instruction set the compiler can use, and there is no
11789 generic instruction set applicable to all processors. In contrast,
11790 @option{-mtune} indicates the processor (or, in this case, collection of
11791 processors) for which the code is optimized.
11793 This selects the CPU to tune for at compilation time by determining
11794 the processor type of the compiling machine. Using @option{-mtune=native}
11795 will produce code optimized for the local machine under the constraints
11796 of the selected instruction set. Using @option{-march=native} will
11797 enable all instruction subsets supported by the local machine (hence
11798 the result might not run on different machines).
11800 Original Intel's i386 CPU@.
11802 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11803 @item i586, pentium
11804 Intel Pentium CPU with no MMX support.
11806 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11808 Intel PentiumPro CPU@.
11810 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11811 instruction set will be used, so the code will run on all i686 family chips.
11813 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11814 @item pentium3, pentium3m
11815 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11818 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11819 support. Used by Centrino notebooks.
11820 @item pentium4, pentium4m
11821 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11823 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11826 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11827 SSE2 and SSE3 instruction set support.
11829 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11830 instruction set support.
11832 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11833 instruction set support.
11835 AMD K6 CPU with MMX instruction set support.
11837 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
11838 @item athlon, athlon-tbird
11839 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
11841 @item athlon-4, athlon-xp, athlon-mp
11842 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
11843 instruction set support.
11844 @item k8, opteron, athlon64, athlon-fx
11845 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11846 MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
11847 @item k8-sse3, opteron-sse3, athlon64-sse3
11848 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11849 @item amdfam10, barcelona
11850 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11851 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
11852 instruction set extensions.)
11854 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11857 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
11858 instruction set support.
11860 Via C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
11861 implemented for this chip.)
11863 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11864 implemented for this chip.)
11866 Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
11869 While picking a specific @var{cpu-type} will schedule things appropriately
11870 for that particular chip, the compiler will not generate any code that
11871 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11874 @item -march=@var{cpu-type}
11876 Generate instructions for the machine type @var{cpu-type}. The choices
11877 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11878 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11880 @item -mcpu=@var{cpu-type}
11882 A deprecated synonym for @option{-mtune}.
11884 @item -mfpmath=@var{unit}
11886 Generate floating point arithmetics for selected unit @var{unit}. The choices
11887 for @var{unit} are:
11891 Use the standard 387 floating point coprocessor present majority of chips and
11892 emulated otherwise. Code compiled with this option will run almost everywhere.
11893 The temporary results are computed in 80bit precision instead of precision
11894 specified by the type resulting in slightly different results compared to most
11895 of other chips. See @option{-ffloat-store} for more detailed description.
11897 This is the default choice for i386 compiler.
11900 Use scalar floating point instructions present in the SSE instruction set.
11901 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11902 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11903 instruction set supports only single precision arithmetics, thus the double and
11904 extended precision arithmetics is still done using 387. Later version, present
11905 only in Pentium4 and the future AMD x86-64 chips supports double precision
11908 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11909 or @option{-msse2} switches to enable SSE extensions and make this option
11910 effective. For the x86-64 compiler, these extensions are enabled by default.
11912 The resulting code should be considerably faster in the majority of cases and avoid
11913 the numerical instability problems of 387 code, but may break some existing
11914 code that expects temporaries to be 80bit.
11916 This is the default choice for the x86-64 compiler.
11921 Attempt to utilize both instruction sets at once. This effectively double the
11922 amount of available registers and on chips with separate execution units for
11923 387 and SSE the execution resources too. Use this option with care, as it is
11924 still experimental, because the GCC register allocator does not model separate
11925 functional units well resulting in instable performance.
11928 @item -masm=@var{dialect}
11929 @opindex masm=@var{dialect}
11930 Output asm instructions using selected @var{dialect}. Supported
11931 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11932 not support @samp{intel}.
11935 @itemx -mno-ieee-fp
11937 @opindex mno-ieee-fp
11938 Control whether or not the compiler uses IEEE floating point
11939 comparisons. These handle correctly the case where the result of a
11940 comparison is unordered.
11943 @opindex msoft-float
11944 Generate output containing library calls for floating point.
11945 @strong{Warning:} the requisite libraries are not part of GCC@.
11946 Normally the facilities of the machine's usual C compiler are used, but
11947 this can't be done directly in cross-compilation. You must make your
11948 own arrangements to provide suitable library functions for
11951 On machines where a function returns floating point results in the 80387
11952 register stack, some floating point opcodes may be emitted even if
11953 @option{-msoft-float} is used.
11955 @item -mno-fp-ret-in-387
11956 @opindex mno-fp-ret-in-387
11957 Do not use the FPU registers for return values of functions.
11959 The usual calling convention has functions return values of types
11960 @code{float} and @code{double} in an FPU register, even if there
11961 is no FPU@. The idea is that the operating system should emulate
11964 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11965 in ordinary CPU registers instead.
11967 @item -mno-fancy-math-387
11968 @opindex mno-fancy-math-387
11969 Some 387 emulators do not support the @code{sin}, @code{cos} and
11970 @code{sqrt} instructions for the 387. Specify this option to avoid
11971 generating those instructions. This option is the default on FreeBSD,
11972 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11973 indicates that the target cpu will always have an FPU and so the
11974 instruction will not need emulation. As of revision 2.6.1, these
11975 instructions are not generated unless you also use the
11976 @option{-funsafe-math-optimizations} switch.
11978 @item -malign-double
11979 @itemx -mno-align-double
11980 @opindex malign-double
11981 @opindex mno-align-double
11982 Control whether GCC aligns @code{double}, @code{long double}, and
11983 @code{long long} variables on a two word boundary or a one word
11984 boundary. Aligning @code{double} variables on a two word boundary will
11985 produce code that runs somewhat faster on a @samp{Pentium} at the
11986 expense of more memory.
11988 On x86-64, @option{-malign-double} is enabled by default.
11990 @strong{Warning:} if you use the @option{-malign-double} switch,
11991 structures containing the above types will be aligned differently than
11992 the published application binary interface specifications for the 386
11993 and will not be binary compatible with structures in code compiled
11994 without that switch.
11996 @item -m96bit-long-double
11997 @itemx -m128bit-long-double
11998 @opindex m96bit-long-double
11999 @opindex m128bit-long-double
12000 These switches control the size of @code{long double} type. The i386
12001 application binary interface specifies the size to be 96 bits,
12002 so @option{-m96bit-long-double} is the default in 32 bit mode.
12004 Modern architectures (Pentium and newer) would prefer @code{long double}
12005 to be aligned to an 8 or 16 byte boundary. In arrays or structures
12006 conforming to the ABI, this would not be possible. So specifying a
12007 @option{-m128bit-long-double} will align @code{long double}
12008 to a 16 byte boundary by padding the @code{long double} with an additional
12011 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
12012 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
12014 Notice that neither of these options enable any extra precision over the x87
12015 standard of 80 bits for a @code{long double}.
12017 @strong{Warning:} if you override the default value for your target ABI, the
12018 structures and arrays containing @code{long double} variables will change
12019 their size as well as function calling convention for function taking
12020 @code{long double} will be modified. Hence they will not be binary
12021 compatible with arrays or structures in code compiled without that switch.
12023 @item -mlarge-data-threshold=@var{number}
12024 @opindex mlarge-data-threshold=@var{number}
12025 When @option{-mcmodel=medium} is specified, the data greater than
12026 @var{threshold} are placed in large data section. This value must be the
12027 same across all object linked into the binary and defaults to 65535.
12031 Use a different function-calling convention, in which functions that
12032 take a fixed number of arguments return with the @code{ret} @var{num}
12033 instruction, which pops their arguments while returning. This saves one
12034 instruction in the caller since there is no need to pop the arguments
12037 You can specify that an individual function is called with this calling
12038 sequence with the function attribute @samp{stdcall}. You can also
12039 override the @option{-mrtd} option by using the function attribute
12040 @samp{cdecl}. @xref{Function Attributes}.
12042 @strong{Warning:} this calling convention is incompatible with the one
12043 normally used on Unix, so you cannot use it if you need to call
12044 libraries compiled with the Unix compiler.
12046 Also, you must provide function prototypes for all functions that
12047 take variable numbers of arguments (including @code{printf});
12048 otherwise incorrect code will be generated for calls to those
12051 In addition, seriously incorrect code will result if you call a
12052 function with too many arguments. (Normally, extra arguments are
12053 harmlessly ignored.)
12055 @item -mregparm=@var{num}
12057 Control how many registers are used to pass integer arguments. By
12058 default, no registers are used to pass arguments, and at most 3
12059 registers can be used. You can control this behavior for a specific
12060 function by using the function attribute @samp{regparm}.
12061 @xref{Function Attributes}.
12063 @strong{Warning:} if you use this switch, and
12064 @var{num} is nonzero, then you must build all modules with the same
12065 value, including any libraries. This includes the system libraries and
12069 @opindex msseregparm
12070 Use SSE register passing conventions for float and double arguments
12071 and return values. You can control this behavior for a specific
12072 function by using the function attribute @samp{sseregparm}.
12073 @xref{Function Attributes}.
12075 @strong{Warning:} if you use this switch then you must build all
12076 modules with the same value, including any libraries. This includes
12077 the system libraries and startup modules.
12086 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
12087 is specified, the significands of results of floating-point operations are
12088 rounded to 24 bits (single precision); @option{-mpc64} rounds the
12089 significands of results of floating-point operations to 53 bits (double
12090 precision) and @option{-mpc80} rounds the significands of results of
12091 floating-point operations to 64 bits (extended double precision), which is
12092 the default. When this option is used, floating-point operations in higher
12093 precisions are not available to the programmer without setting the FPU
12094 control word explicitly.
12096 Setting the rounding of floating-point operations to less than the default
12097 80 bits can speed some programs by 2% or more. Note that some mathematical
12098 libraries assume that extended precision (80 bit) floating-point operations
12099 are enabled by default; routines in such libraries could suffer significant
12100 loss of accuracy, typically through so-called "catastrophic cancellation",
12101 when this option is used to set the precision to less than extended precision.
12103 @item -mstackrealign
12104 @opindex mstackrealign
12105 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
12106 option will generate an alternate prologue and epilogue that realigns the
12107 runtime stack if necessary. This supports mixing legacy codes that keep
12108 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
12109 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
12110 applicable to individual functions.
12112 @item -mpreferred-stack-boundary=@var{num}
12113 @opindex mpreferred-stack-boundary
12114 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
12115 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
12116 the default is 4 (16 bytes or 128 bits).
12118 @item -mincoming-stack-boundary=@var{num}
12119 @opindex mincoming-stack-boundary
12120 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
12121 boundary. If @option{-mincoming-stack-boundary} is not specified,
12122 the one specified by @option{-mpreferred-stack-boundary} will be used.
12124 On Pentium and PentiumPro, @code{double} and @code{long double} values
12125 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
12126 suffer significant run time performance penalties. On Pentium III, the
12127 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
12128 properly if it is not 16 byte aligned.
12130 To ensure proper alignment of this values on the stack, the stack boundary
12131 must be as aligned as that required by any value stored on the stack.
12132 Further, every function must be generated such that it keeps the stack
12133 aligned. Thus calling a function compiled with a higher preferred
12134 stack boundary from a function compiled with a lower preferred stack
12135 boundary will most likely misalign the stack. It is recommended that
12136 libraries that use callbacks always use the default setting.
12138 This extra alignment does consume extra stack space, and generally
12139 increases code size. Code that is sensitive to stack space usage, such
12140 as embedded systems and operating system kernels, may want to reduce the
12141 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12185 These switches enable or disable the use of instructions in the MMX,
12186 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, FMA4, XOP,
12187 LWP, ABM or 3DNow!@: extended instruction sets.
12188 These extensions are also available as built-in functions: see
12189 @ref{X86 Built-in Functions}, for details of the functions enabled and
12190 disabled by these switches.
12192 To have SSE/SSE2 instructions generated automatically from floating-point
12193 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12195 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12196 generates new AVX instructions or AVX equivalence for all SSEx instructions
12199 These options will enable GCC to use these extended instructions in
12200 generated code, even without @option{-mfpmath=sse}. Applications which
12201 perform runtime CPU detection must compile separate files for each
12202 supported architecture, using the appropriate flags. In particular,
12203 the file containing the CPU detection code should be compiled without
12207 @itemx -mno-fused-madd
12208 @opindex mfused-madd
12209 @opindex mno-fused-madd
12210 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12211 instructions. The default is to use these instructions.
12215 This option instructs GCC to emit a @code{cld} instruction in the prologue
12216 of functions that use string instructions. String instructions depend on
12217 the DF flag to select between autoincrement or autodecrement mode. While the
12218 ABI specifies the DF flag to be cleared on function entry, some operating
12219 systems violate this specification by not clearing the DF flag in their
12220 exception dispatchers. The exception handler can be invoked with the DF flag
12221 set which leads to wrong direction mode, when string instructions are used.
12222 This option can be enabled by default on 32-bit x86 targets by configuring
12223 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12224 instructions can be suppressed with the @option{-mno-cld} compiler option
12229 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12230 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12231 data types. This is useful for high resolution counters that could be updated
12232 by multiple processors (or cores). This instruction is generated as part of
12233 atomic built-in functions: see @ref{Atomic Builtins} for details.
12237 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12238 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12239 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12240 SAHF are load and store instructions, respectively, for certain status flags.
12241 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12242 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12246 This option will enable GCC to use movbe instruction to implement
12247 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12251 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12252 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12253 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12257 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12258 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12259 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12260 variants) for single precision floating point arguments. These instructions
12261 are generated only when @option{-funsafe-math-optimizations} is enabled
12262 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12263 Note that while the throughput of the sequence is higher than the throughput
12264 of the non-reciprocal instruction, the precision of the sequence can be
12265 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12267 Note that GCC implements 1.0f/sqrtf(x) in terms of RSQRTSS (or RSQRTPS)
12268 already with @option{-ffast-math} (or the above option combination), and
12269 doesn't need @option{-mrecip}.
12271 @item -mveclibabi=@var{type}
12272 @opindex mveclibabi
12273 Specifies the ABI type to use for vectorizing intrinsics using an
12274 external library. Supported types are @code{svml} for the Intel short
12275 vector math library and @code{acml} for the AMD math core library style
12276 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12277 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12278 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12279 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12280 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12281 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12282 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12283 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12284 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12285 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12286 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12287 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12288 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12289 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12290 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12291 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12292 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12293 compatible library will have to be specified at link time.
12295 @item -mabi=@var{name}
12297 Generate code for the specified calling convention. Permissible values
12298 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12299 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12300 ABI when targeting Windows. On all other systems, the default is the
12301 SYSV ABI. You can control this behavior for a specific function by
12302 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12303 @xref{Function Attributes}.
12306 @itemx -mno-push-args
12307 @opindex mpush-args
12308 @opindex mno-push-args
12309 Use PUSH operations to store outgoing parameters. This method is shorter
12310 and usually equally fast as method using SUB/MOV operations and is enabled
12311 by default. In some cases disabling it may improve performance because of
12312 improved scheduling and reduced dependencies.
12314 @item -maccumulate-outgoing-args
12315 @opindex maccumulate-outgoing-args
12316 If enabled, the maximum amount of space required for outgoing arguments will be
12317 computed in the function prologue. This is faster on most modern CPUs
12318 because of reduced dependencies, improved scheduling and reduced stack usage
12319 when preferred stack boundary is not equal to 2. The drawback is a notable
12320 increase in code size. This switch implies @option{-mno-push-args}.
12324 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12325 on thread-safe exception handling must compile and link all code with the
12326 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12327 @option{-D_MT}; when linking, it links in a special thread helper library
12328 @option{-lmingwthrd} which cleans up per thread exception handling data.
12330 @item -mno-align-stringops
12331 @opindex mno-align-stringops
12332 Do not align destination of inlined string operations. This switch reduces
12333 code size and improves performance in case the destination is already aligned,
12334 but GCC doesn't know about it.
12336 @item -minline-all-stringops
12337 @opindex minline-all-stringops
12338 By default GCC inlines string operations only when destination is known to be
12339 aligned at least to 4 byte boundary. This enables more inlining, increase code
12340 size, but may improve performance of code that depends on fast memcpy, strlen
12341 and memset for short lengths.
12343 @item -minline-stringops-dynamically
12344 @opindex minline-stringops-dynamically
12345 For string operation of unknown size, inline runtime checks so for small
12346 blocks inline code is used, while for large blocks library call is used.
12348 @item -mstringop-strategy=@var{alg}
12349 @opindex mstringop-strategy=@var{alg}
12350 Overwrite internal decision heuristic about particular algorithm to inline
12351 string operation with. The allowed values are @code{rep_byte},
12352 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12353 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12354 expanding inline loop, @code{libcall} for always expanding library call.
12356 @item -momit-leaf-frame-pointer
12357 @opindex momit-leaf-frame-pointer
12358 Don't keep the frame pointer in a register for leaf functions. This
12359 avoids the instructions to save, set up and restore frame pointers and
12360 makes an extra register available in leaf functions. The option
12361 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12362 which might make debugging harder.
12364 @item -mtls-direct-seg-refs
12365 @itemx -mno-tls-direct-seg-refs
12366 @opindex mtls-direct-seg-refs
12367 Controls whether TLS variables may be accessed with offsets from the
12368 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12369 or whether the thread base pointer must be added. Whether or not this
12370 is legal depends on the operating system, and whether it maps the
12371 segment to cover the entire TLS area.
12373 For systems that use GNU libc, the default is on.
12376 @itemx -mno-sse2avx
12378 Specify that the assembler should encode SSE instructions with VEX
12379 prefix. The option @option{-mavx} turns this on by default.
12382 These @samp{-m} switches are supported in addition to the above
12383 on AMD x86-64 processors in 64-bit environments.
12390 Generate code for a 32-bit or 64-bit environment.
12391 The 32-bit environment sets int, long and pointer to 32 bits and
12392 generates code that runs on any i386 system.
12393 The 64-bit environment sets int to 32 bits and long and pointer
12394 to 64 bits and generates code for AMD's x86-64 architecture. For
12395 darwin only the -m64 option turns off the @option{-fno-pic} and
12396 @option{-mdynamic-no-pic} options.
12398 @item -mno-red-zone
12399 @opindex mno-red-zone
12400 Do not use a so called red zone for x86-64 code. The red zone is mandated
12401 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12402 stack pointer that will not be modified by signal or interrupt handlers
12403 and therefore can be used for temporary data without adjusting the stack
12404 pointer. The flag @option{-mno-red-zone} disables this red zone.
12406 @item -mcmodel=small
12407 @opindex mcmodel=small
12408 Generate code for the small code model: the program and its symbols must
12409 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12410 Programs can be statically or dynamically linked. This is the default
12413 @item -mcmodel=kernel
12414 @opindex mcmodel=kernel
12415 Generate code for the kernel code model. The kernel runs in the
12416 negative 2 GB of the address space.
12417 This model has to be used for Linux kernel code.
12419 @item -mcmodel=medium
12420 @opindex mcmodel=medium
12421 Generate code for the medium model: The program is linked in the lower 2
12422 GB of the address space. Small symbols are also placed there. Symbols
12423 with sizes larger than @option{-mlarge-data-threshold} are put into
12424 large data or bss sections and can be located above 2GB. Programs can
12425 be statically or dynamically linked.
12427 @item -mcmodel=large
12428 @opindex mcmodel=large
12429 Generate code for the large model: This model makes no assumptions
12430 about addresses and sizes of sections.
12433 @node IA-64 Options
12434 @subsection IA-64 Options
12435 @cindex IA-64 Options
12437 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12441 @opindex mbig-endian
12442 Generate code for a big endian target. This is the default for HP-UX@.
12444 @item -mlittle-endian
12445 @opindex mlittle-endian
12446 Generate code for a little endian target. This is the default for AIX5
12452 @opindex mno-gnu-as
12453 Generate (or don't) code for the GNU assembler. This is the default.
12454 @c Also, this is the default if the configure option @option{--with-gnu-as}
12460 @opindex mno-gnu-ld
12461 Generate (or don't) code for the GNU linker. This is the default.
12462 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12467 Generate code that does not use a global pointer register. The result
12468 is not position independent code, and violates the IA-64 ABI@.
12470 @item -mvolatile-asm-stop
12471 @itemx -mno-volatile-asm-stop
12472 @opindex mvolatile-asm-stop
12473 @opindex mno-volatile-asm-stop
12474 Generate (or don't) a stop bit immediately before and after volatile asm
12477 @item -mregister-names
12478 @itemx -mno-register-names
12479 @opindex mregister-names
12480 @opindex mno-register-names
12481 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12482 the stacked registers. This may make assembler output more readable.
12488 Disable (or enable) optimizations that use the small data section. This may
12489 be useful for working around optimizer bugs.
12491 @item -mconstant-gp
12492 @opindex mconstant-gp
12493 Generate code that uses a single constant global pointer value. This is
12494 useful when compiling kernel code.
12498 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12499 This is useful when compiling firmware code.
12501 @item -minline-float-divide-min-latency
12502 @opindex minline-float-divide-min-latency
12503 Generate code for inline divides of floating point values
12504 using the minimum latency algorithm.
12506 @item -minline-float-divide-max-throughput
12507 @opindex minline-float-divide-max-throughput
12508 Generate code for inline divides of floating point values
12509 using the maximum throughput algorithm.
12511 @item -mno-inline-float-divide
12512 @opindex mno-inline-float-divide
12513 Do not generate inline code for divides of floating point values.
12515 @item -minline-int-divide-min-latency
12516 @opindex minline-int-divide-min-latency
12517 Generate code for inline divides of integer values
12518 using the minimum latency algorithm.
12520 @item -minline-int-divide-max-throughput
12521 @opindex minline-int-divide-max-throughput
12522 Generate code for inline divides of integer values
12523 using the maximum throughput algorithm.
12525 @item -mno-inline-int-divide
12526 @opindex mno-inline-int-divide
12527 Do not generate inline code for divides of integer values.
12529 @item -minline-sqrt-min-latency
12530 @opindex minline-sqrt-min-latency
12531 Generate code for inline square roots
12532 using the minimum latency algorithm.
12534 @item -minline-sqrt-max-throughput
12535 @opindex minline-sqrt-max-throughput
12536 Generate code for inline square roots
12537 using the maximum throughput algorithm.
12539 @item -mno-inline-sqrt
12540 @opindex mno-inline-sqrt
12541 Do not generate inline code for sqrt.
12544 @itemx -mno-fused-madd
12545 @opindex mfused-madd
12546 @opindex mno-fused-madd
12547 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12548 instructions. The default is to use these instructions.
12550 @item -mno-dwarf2-asm
12551 @itemx -mdwarf2-asm
12552 @opindex mno-dwarf2-asm
12553 @opindex mdwarf2-asm
12554 Don't (or do) generate assembler code for the DWARF2 line number debugging
12555 info. This may be useful when not using the GNU assembler.
12557 @item -mearly-stop-bits
12558 @itemx -mno-early-stop-bits
12559 @opindex mearly-stop-bits
12560 @opindex mno-early-stop-bits
12561 Allow stop bits to be placed earlier than immediately preceding the
12562 instruction that triggered the stop bit. This can improve instruction
12563 scheduling, but does not always do so.
12565 @item -mfixed-range=@var{register-range}
12566 @opindex mfixed-range
12567 Generate code treating the given register range as fixed registers.
12568 A fixed register is one that the register allocator can not use. This is
12569 useful when compiling kernel code. A register range is specified as
12570 two registers separated by a dash. Multiple register ranges can be
12571 specified separated by a comma.
12573 @item -mtls-size=@var{tls-size}
12575 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12578 @item -mtune=@var{cpu-type}
12580 Tune the instruction scheduling for a particular CPU, Valid values are
12581 itanium, itanium1, merced, itanium2, and mckinley.
12587 Generate code for a 32-bit or 64-bit environment.
12588 The 32-bit environment sets int, long and pointer to 32 bits.
12589 The 64-bit environment sets int to 32 bits and long and pointer
12590 to 64 bits. These are HP-UX specific flags.
12592 @item -mno-sched-br-data-spec
12593 @itemx -msched-br-data-spec
12594 @opindex mno-sched-br-data-spec
12595 @opindex msched-br-data-spec
12596 (Dis/En)able data speculative scheduling before reload.
12597 This will result in generation of the ld.a instructions and
12598 the corresponding check instructions (ld.c / chk.a).
12599 The default is 'disable'.
12601 @item -msched-ar-data-spec
12602 @itemx -mno-sched-ar-data-spec
12603 @opindex msched-ar-data-spec
12604 @opindex mno-sched-ar-data-spec
12605 (En/Dis)able data speculative scheduling after reload.
12606 This will result in generation of the ld.a instructions and
12607 the corresponding check instructions (ld.c / chk.a).
12608 The default is 'enable'.
12610 @item -mno-sched-control-spec
12611 @itemx -msched-control-spec
12612 @opindex mno-sched-control-spec
12613 @opindex msched-control-spec
12614 (Dis/En)able control speculative scheduling. This feature is
12615 available only during region scheduling (i.e.@: before reload).
12616 This will result in generation of the ld.s instructions and
12617 the corresponding check instructions chk.s .
12618 The default is 'disable'.
12620 @item -msched-br-in-data-spec
12621 @itemx -mno-sched-br-in-data-spec
12622 @opindex msched-br-in-data-spec
12623 @opindex mno-sched-br-in-data-spec
12624 (En/Dis)able speculative scheduling of the instructions that
12625 are dependent on the data speculative loads before reload.
12626 This is effective only with @option{-msched-br-data-spec} enabled.
12627 The default is 'enable'.
12629 @item -msched-ar-in-data-spec
12630 @itemx -mno-sched-ar-in-data-spec
12631 @opindex msched-ar-in-data-spec
12632 @opindex mno-sched-ar-in-data-spec
12633 (En/Dis)able speculative scheduling of the instructions that
12634 are dependent on the data speculative loads after reload.
12635 This is effective only with @option{-msched-ar-data-spec} enabled.
12636 The default is 'enable'.
12638 @item -msched-in-control-spec
12639 @itemx -mno-sched-in-control-spec
12640 @opindex msched-in-control-spec
12641 @opindex mno-sched-in-control-spec
12642 (En/Dis)able speculative scheduling of the instructions that
12643 are dependent on the control speculative loads.
12644 This is effective only with @option{-msched-control-spec} enabled.
12645 The default is 'enable'.
12647 @item -mno-sched-prefer-non-data-spec-insns
12648 @itemx -msched-prefer-non-data-spec-insns
12649 @opindex mno-sched-prefer-non-data-spec-insns
12650 @opindex msched-prefer-non-data-spec-insns
12651 If enabled, data speculative instructions will be chosen for schedule
12652 only if there are no other choices at the moment. This will make
12653 the use of the data speculation much more conservative.
12654 The default is 'disable'.
12656 @item -mno-sched-prefer-non-control-spec-insns
12657 @itemx -msched-prefer-non-control-spec-insns
12658 @opindex mno-sched-prefer-non-control-spec-insns
12659 @opindex msched-prefer-non-control-spec-insns
12660 If enabled, control speculative instructions will be chosen for schedule
12661 only if there are no other choices at the moment. This will make
12662 the use of the control speculation much more conservative.
12663 The default is 'disable'.
12665 @item -mno-sched-count-spec-in-critical-path
12666 @itemx -msched-count-spec-in-critical-path
12667 @opindex mno-sched-count-spec-in-critical-path
12668 @opindex msched-count-spec-in-critical-path
12669 If enabled, speculative dependencies will be considered during
12670 computation of the instructions priorities. This will make the use of the
12671 speculation a bit more conservative.
12672 The default is 'disable'.
12674 @item -msched-spec-ldc
12675 @opindex msched-spec-ldc
12676 Use a simple data speculation check. This option is on by default.
12678 @item -msched-control-spec-ldc
12679 @opindex msched-spec-ldc
12680 Use a simple check for control speculation. This option is on by default.
12682 @item -msched-stop-bits-after-every-cycle
12683 @opindex msched-stop-bits-after-every-cycle
12684 Place a stop bit after every cycle when scheduling. This option is on
12687 @item -msched-fp-mem-deps-zero-cost
12688 @opindex msched-fp-mem-deps-zero-cost
12689 Assume that floating-point stores and loads are not likely to cause a conflict
12690 when placed into the same instruction group. This option is disabled by
12693 @item -msel-sched-dont-check-control-spec
12694 @opindex msel-sched-dont-check-control-spec
12695 Generate checks for control speculation in selective scheduling.
12696 This flag is disabled by default.
12698 @item -msched-max-memory-insns=@var{max-insns}
12699 @opindex msched-max-memory-insns
12700 Limit on the number of memory insns per instruction group, giving lower
12701 priority to subsequent memory insns attempting to schedule in the same
12702 instruction group. Frequently useful to prevent cache bank conflicts.
12703 The default value is 1.
12705 @item -msched-max-memory-insns-hard-limit
12706 @opindex msched-max-memory-insns-hard-limit
12707 Disallow more than `msched-max-memory-insns' in instruction group.
12708 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12709 when limit is reached but may still schedule memory operations.
12713 @node IA-64/VMS Options
12714 @subsection IA-64/VMS Options
12716 These @samp{-m} options are defined for the IA-64/VMS implementations:
12719 @item -mvms-return-codes
12720 @opindex mvms-return-codes
12721 Return VMS condition codes from main. The default is to return POSIX
12722 style condition (e.g.@ error) codes.
12724 @item -mdebug-main=@var{prefix}
12725 @opindex mdebug-main=@var{prefix}
12726 Flag the first routine whose name starts with @var{prefix} as the main
12727 routine for the debugger.
12731 Default to 64bit memory allocation routines.
12735 @subsection LM32 Options
12736 @cindex LM32 options
12738 These @option{-m} options are defined for the Lattice Mico32 architecture:
12741 @item -mbarrel-shift-enabled
12742 @opindex mbarrel-shift-enabled
12743 Enable barrel-shift instructions.
12745 @item -mdivide-enabled
12746 @opindex mdivide-enabled
12747 Enable divide and modulus instructions.
12749 @item -mmultiply-enabled
12750 @opindex multiply-enabled
12751 Enable multiply instructions.
12753 @item -msign-extend-enabled
12754 @opindex msign-extend-enabled
12755 Enable sign extend instructions.
12757 @item -muser-enabled
12758 @opindex muser-enabled
12759 Enable user-defined instructions.
12764 @subsection M32C Options
12765 @cindex M32C options
12768 @item -mcpu=@var{name}
12770 Select the CPU for which code is generated. @var{name} may be one of
12771 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12772 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12773 the M32C/80 series.
12777 Specifies that the program will be run on the simulator. This causes
12778 an alternate runtime library to be linked in which supports, for
12779 example, file I/O@. You must not use this option when generating
12780 programs that will run on real hardware; you must provide your own
12781 runtime library for whatever I/O functions are needed.
12783 @item -memregs=@var{number}
12785 Specifies the number of memory-based pseudo-registers GCC will use
12786 during code generation. These pseudo-registers will be used like real
12787 registers, so there is a tradeoff between GCC's ability to fit the
12788 code into available registers, and the performance penalty of using
12789 memory instead of registers. Note that all modules in a program must
12790 be compiled with the same value for this option. Because of that, you
12791 must not use this option with the default runtime libraries gcc
12796 @node M32R/D Options
12797 @subsection M32R/D Options
12798 @cindex M32R/D options
12800 These @option{-m} options are defined for Renesas M32R/D architectures:
12805 Generate code for the M32R/2@.
12809 Generate code for the M32R/X@.
12813 Generate code for the M32R@. This is the default.
12815 @item -mmodel=small
12816 @opindex mmodel=small
12817 Assume all objects live in the lower 16MB of memory (so that their addresses
12818 can be loaded with the @code{ld24} instruction), and assume all subroutines
12819 are reachable with the @code{bl} instruction.
12820 This is the default.
12822 The addressability of a particular object can be set with the
12823 @code{model} attribute.
12825 @item -mmodel=medium
12826 @opindex mmodel=medium
12827 Assume objects may be anywhere in the 32-bit address space (the compiler
12828 will generate @code{seth/add3} instructions to load their addresses), and
12829 assume all subroutines are reachable with the @code{bl} instruction.
12831 @item -mmodel=large
12832 @opindex mmodel=large
12833 Assume objects may be anywhere in the 32-bit address space (the compiler
12834 will generate @code{seth/add3} instructions to load their addresses), and
12835 assume subroutines may not be reachable with the @code{bl} instruction
12836 (the compiler will generate the much slower @code{seth/add3/jl}
12837 instruction sequence).
12840 @opindex msdata=none
12841 Disable use of the small data area. Variables will be put into
12842 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12843 @code{section} attribute has been specified).
12844 This is the default.
12846 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12847 Objects may be explicitly put in the small data area with the
12848 @code{section} attribute using one of these sections.
12850 @item -msdata=sdata
12851 @opindex msdata=sdata
12852 Put small global and static data in the small data area, but do not
12853 generate special code to reference them.
12856 @opindex msdata=use
12857 Put small global and static data in the small data area, and generate
12858 special instructions to reference them.
12862 @cindex smaller data references
12863 Put global and static objects less than or equal to @var{num} bytes
12864 into the small data or bss sections instead of the normal data or bss
12865 sections. The default value of @var{num} is 8.
12866 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12867 for this option to have any effect.
12869 All modules should be compiled with the same @option{-G @var{num}} value.
12870 Compiling with different values of @var{num} may or may not work; if it
12871 doesn't the linker will give an error message---incorrect code will not be
12876 Makes the M32R specific code in the compiler display some statistics
12877 that might help in debugging programs.
12879 @item -malign-loops
12880 @opindex malign-loops
12881 Align all loops to a 32-byte boundary.
12883 @item -mno-align-loops
12884 @opindex mno-align-loops
12885 Do not enforce a 32-byte alignment for loops. This is the default.
12887 @item -missue-rate=@var{number}
12888 @opindex missue-rate=@var{number}
12889 Issue @var{number} instructions per cycle. @var{number} can only be 1
12892 @item -mbranch-cost=@var{number}
12893 @opindex mbranch-cost=@var{number}
12894 @var{number} can only be 1 or 2. If it is 1 then branches will be
12895 preferred over conditional code, if it is 2, then the opposite will
12898 @item -mflush-trap=@var{number}
12899 @opindex mflush-trap=@var{number}
12900 Specifies the trap number to use to flush the cache. The default is
12901 12. Valid numbers are between 0 and 15 inclusive.
12903 @item -mno-flush-trap
12904 @opindex mno-flush-trap
12905 Specifies that the cache cannot be flushed by using a trap.
12907 @item -mflush-func=@var{name}
12908 @opindex mflush-func=@var{name}
12909 Specifies the name of the operating system function to call to flush
12910 the cache. The default is @emph{_flush_cache}, but a function call
12911 will only be used if a trap is not available.
12913 @item -mno-flush-func
12914 @opindex mno-flush-func
12915 Indicates that there is no OS function for flushing the cache.
12919 @node M680x0 Options
12920 @subsection M680x0 Options
12921 @cindex M680x0 options
12923 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
12924 The default settings depend on which architecture was selected when
12925 the compiler was configured; the defaults for the most common choices
12929 @item -march=@var{arch}
12931 Generate code for a specific M680x0 or ColdFire instruction set
12932 architecture. Permissible values of @var{arch} for M680x0
12933 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
12934 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
12935 architectures are selected according to Freescale's ISA classification
12936 and the permissible values are: @samp{isaa}, @samp{isaaplus},
12937 @samp{isab} and @samp{isac}.
12939 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
12940 code for a ColdFire target. The @var{arch} in this macro is one of the
12941 @option{-march} arguments given above.
12943 When used together, @option{-march} and @option{-mtune} select code
12944 that runs on a family of similar processors but that is optimized
12945 for a particular microarchitecture.
12947 @item -mcpu=@var{cpu}
12949 Generate code for a specific M680x0 or ColdFire processor.
12950 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12951 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12952 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12953 below, which also classifies the CPUs into families:
12955 @multitable @columnfractions 0.20 0.80
12956 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12957 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
12958 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12959 @item @samp{5206e} @tab @samp{5206e}
12960 @item @samp{5208} @tab @samp{5207} @samp{5208}
12961 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12962 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12963 @item @samp{5216} @tab @samp{5214} @samp{5216}
12964 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12965 @item @samp{5225} @tab @samp{5224} @samp{5225}
12966 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
12967 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12968 @item @samp{5249} @tab @samp{5249}
12969 @item @samp{5250} @tab @samp{5250}
12970 @item @samp{5271} @tab @samp{5270} @samp{5271}
12971 @item @samp{5272} @tab @samp{5272}
12972 @item @samp{5275} @tab @samp{5274} @samp{5275}
12973 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12974 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
12975 @item @samp{5307} @tab @samp{5307}
12976 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12977 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12978 @item @samp{5407} @tab @samp{5407}
12979 @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}
12982 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12983 @var{arch} is compatible with @var{cpu}. Other combinations of
12984 @option{-mcpu} and @option{-march} are rejected.
12986 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12987 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12988 where the value of @var{family} is given by the table above.
12990 @item -mtune=@var{tune}
12992 Tune the code for a particular microarchitecture, within the
12993 constraints set by @option{-march} and @option{-mcpu}.
12994 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12995 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12996 and @samp{cpu32}. The ColdFire microarchitectures
12997 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12999 You can also use @option{-mtune=68020-40} for code that needs
13000 to run relatively well on 68020, 68030 and 68040 targets.
13001 @option{-mtune=68020-60} is similar but includes 68060 targets
13002 as well. These two options select the same tuning decisions as
13003 @option{-m68020-40} and @option{-m68020-60} respectively.
13005 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
13006 when tuning for 680x0 architecture @var{arch}. It also defines
13007 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
13008 option is used. If gcc is tuning for a range of architectures,
13009 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
13010 it defines the macros for every architecture in the range.
13012 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
13013 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
13014 of the arguments given above.
13020 Generate output for a 68000. This is the default
13021 when the compiler is configured for 68000-based systems.
13022 It is equivalent to @option{-march=68000}.
13024 Use this option for microcontrollers with a 68000 or EC000 core,
13025 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
13029 Generate output for a 68010. This is the default
13030 when the compiler is configured for 68010-based systems.
13031 It is equivalent to @option{-march=68010}.
13037 Generate output for a 68020. This is the default
13038 when the compiler is configured for 68020-based systems.
13039 It is equivalent to @option{-march=68020}.
13043 Generate output for a 68030. This is the default when the compiler is
13044 configured for 68030-based systems. It is equivalent to
13045 @option{-march=68030}.
13049 Generate output for a 68040. This is the default when the compiler is
13050 configured for 68040-based systems. It is equivalent to
13051 @option{-march=68040}.
13053 This option inhibits the use of 68881/68882 instructions that have to be
13054 emulated by software on the 68040. Use this option if your 68040 does not
13055 have code to emulate those instructions.
13059 Generate output for a 68060. This is the default when the compiler is
13060 configured for 68060-based systems. It is equivalent to
13061 @option{-march=68060}.
13063 This option inhibits the use of 68020 and 68881/68882 instructions that
13064 have to be emulated by software on the 68060. Use this option if your 68060
13065 does not have code to emulate those instructions.
13069 Generate output for a CPU32. This is the default
13070 when the compiler is configured for CPU32-based systems.
13071 It is equivalent to @option{-march=cpu32}.
13073 Use this option for microcontrollers with a
13074 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
13075 68336, 68340, 68341, 68349 and 68360.
13079 Generate output for a 520X ColdFire CPU@. This is the default
13080 when the compiler is configured for 520X-based systems.
13081 It is equivalent to @option{-mcpu=5206}, and is now deprecated
13082 in favor of that option.
13084 Use this option for microcontroller with a 5200 core, including
13085 the MCF5202, MCF5203, MCF5204 and MCF5206.
13089 Generate output for a 5206e ColdFire CPU@. The option is now
13090 deprecated in favor of the equivalent @option{-mcpu=5206e}.
13094 Generate output for a member of the ColdFire 528X family.
13095 The option is now deprecated in favor of the equivalent
13096 @option{-mcpu=528x}.
13100 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
13101 in favor of the equivalent @option{-mcpu=5307}.
13105 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
13106 in favor of the equivalent @option{-mcpu=5407}.
13110 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
13111 This includes use of hardware floating point instructions.
13112 The option is equivalent to @option{-mcpu=547x}, and is now
13113 deprecated in favor of that option.
13117 Generate output for a 68040, without using any of the new instructions.
13118 This results in code which can run relatively efficiently on either a
13119 68020/68881 or a 68030 or a 68040. The generated code does use the
13120 68881 instructions that are emulated on the 68040.
13122 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
13126 Generate output for a 68060, without using any of the new instructions.
13127 This results in code which can run relatively efficiently on either a
13128 68020/68881 or a 68030 or a 68040. The generated code does use the
13129 68881 instructions that are emulated on the 68060.
13131 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
13135 @opindex mhard-float
13137 Generate floating-point instructions. This is the default for 68020
13138 and above, and for ColdFire devices that have an FPU@. It defines the
13139 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
13140 on ColdFire targets.
13143 @opindex msoft-float
13144 Do not generate floating-point instructions; use library calls instead.
13145 This is the default for 68000, 68010, and 68832 targets. It is also
13146 the default for ColdFire devices that have no FPU.
13152 Generate (do not generate) ColdFire hardware divide and remainder
13153 instructions. If @option{-march} is used without @option{-mcpu},
13154 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
13155 architectures. Otherwise, the default is taken from the target CPU
13156 (either the default CPU, or the one specified by @option{-mcpu}). For
13157 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
13158 @option{-mcpu=5206e}.
13160 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
13164 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13165 Additionally, parameters passed on the stack are also aligned to a
13166 16-bit boundary even on targets whose API mandates promotion to 32-bit.
13170 Do not consider type @code{int} to be 16 bits wide. This is the default.
13173 @itemx -mno-bitfield
13174 @opindex mnobitfield
13175 @opindex mno-bitfield
13176 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
13177 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
13181 Do use the bit-field instructions. The @option{-m68020} option implies
13182 @option{-mbitfield}. This is the default if you use a configuration
13183 designed for a 68020.
13187 Use a different function-calling convention, in which functions
13188 that take a fixed number of arguments return with the @code{rtd}
13189 instruction, which pops their arguments while returning. This
13190 saves one instruction in the caller since there is no need to pop
13191 the arguments there.
13193 This calling convention is incompatible with the one normally
13194 used on Unix, so you cannot use it if you need to call libraries
13195 compiled with the Unix compiler.
13197 Also, you must provide function prototypes for all functions that
13198 take variable numbers of arguments (including @code{printf});
13199 otherwise incorrect code will be generated for calls to those
13202 In addition, seriously incorrect code will result if you call a
13203 function with too many arguments. (Normally, extra arguments are
13204 harmlessly ignored.)
13206 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13207 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13211 Do not use the calling conventions selected by @option{-mrtd}.
13212 This is the default.
13215 @itemx -mno-align-int
13216 @opindex malign-int
13217 @opindex mno-align-int
13218 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13219 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13220 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13221 Aligning variables on 32-bit boundaries produces code that runs somewhat
13222 faster on processors with 32-bit busses at the expense of more memory.
13224 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13225 align structures containing the above types differently than
13226 most published application binary interface specifications for the m68k.
13230 Use the pc-relative addressing mode of the 68000 directly, instead of
13231 using a global offset table. At present, this option implies @option{-fpic},
13232 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13233 not presently supported with @option{-mpcrel}, though this could be supported for
13234 68020 and higher processors.
13236 @item -mno-strict-align
13237 @itemx -mstrict-align
13238 @opindex mno-strict-align
13239 @opindex mstrict-align
13240 Do not (do) assume that unaligned memory references will be handled by
13244 Generate code that allows the data segment to be located in a different
13245 area of memory from the text segment. This allows for execute in place in
13246 an environment without virtual memory management. This option implies
13249 @item -mno-sep-data
13250 Generate code that assumes that the data segment follows the text segment.
13251 This is the default.
13253 @item -mid-shared-library
13254 Generate code that supports shared libraries via the library ID method.
13255 This allows for execute in place and shared libraries in an environment
13256 without virtual memory management. This option implies @option{-fPIC}.
13258 @item -mno-id-shared-library
13259 Generate code that doesn't assume ID based shared libraries are being used.
13260 This is the default.
13262 @item -mshared-library-id=n
13263 Specified the identification number of the ID based shared library being
13264 compiled. Specifying a value of 0 will generate more compact code, specifying
13265 other values will force the allocation of that number to the current
13266 library but is no more space or time efficient than omitting this option.
13272 When generating position-independent code for ColdFire, generate code
13273 that works if the GOT has more than 8192 entries. This code is
13274 larger and slower than code generated without this option. On M680x0
13275 processors, this option is not needed; @option{-fPIC} suffices.
13277 GCC normally uses a single instruction to load values from the GOT@.
13278 While this is relatively efficient, it only works if the GOT
13279 is smaller than about 64k. Anything larger causes the linker
13280 to report an error such as:
13282 @cindex relocation truncated to fit (ColdFire)
13284 relocation truncated to fit: R_68K_GOT16O foobar
13287 If this happens, you should recompile your code with @option{-mxgot}.
13288 It should then work with very large GOTs. However, code generated with
13289 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13290 the value of a global symbol.
13292 Note that some linkers, including newer versions of the GNU linker,
13293 can create multiple GOTs and sort GOT entries. If you have such a linker,
13294 you should only need to use @option{-mxgot} when compiling a single
13295 object file that accesses more than 8192 GOT entries. Very few do.
13297 These options have no effect unless GCC is generating
13298 position-independent code.
13302 @node M68hc1x Options
13303 @subsection M68hc1x Options
13304 @cindex M68hc1x options
13306 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13307 microcontrollers. The default values for these options depends on
13308 which style of microcontroller was selected when the compiler was configured;
13309 the defaults for the most common choices are given below.
13316 Generate output for a 68HC11. This is the default
13317 when the compiler is configured for 68HC11-based systems.
13323 Generate output for a 68HC12. This is the default
13324 when the compiler is configured for 68HC12-based systems.
13330 Generate output for a 68HCS12.
13332 @item -mauto-incdec
13333 @opindex mauto-incdec
13334 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13341 Enable the use of 68HC12 min and max instructions.
13344 @itemx -mno-long-calls
13345 @opindex mlong-calls
13346 @opindex mno-long-calls
13347 Treat all calls as being far away (near). If calls are assumed to be
13348 far away, the compiler will use the @code{call} instruction to
13349 call a function and the @code{rtc} instruction for returning.
13353 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13355 @item -msoft-reg-count=@var{count}
13356 @opindex msoft-reg-count
13357 Specify the number of pseudo-soft registers which are used for the
13358 code generation. The maximum number is 32. Using more pseudo-soft
13359 register may or may not result in better code depending on the program.
13360 The default is 4 for 68HC11 and 2 for 68HC12.
13364 @node MCore Options
13365 @subsection MCore Options
13366 @cindex MCore options
13368 These are the @samp{-m} options defined for the Motorola M*Core
13374 @itemx -mno-hardlit
13376 @opindex mno-hardlit
13377 Inline constants into the code stream if it can be done in two
13378 instructions or less.
13384 Use the divide instruction. (Enabled by default).
13386 @item -mrelax-immediate
13387 @itemx -mno-relax-immediate
13388 @opindex mrelax-immediate
13389 @opindex mno-relax-immediate
13390 Allow arbitrary sized immediates in bit operations.
13392 @item -mwide-bitfields
13393 @itemx -mno-wide-bitfields
13394 @opindex mwide-bitfields
13395 @opindex mno-wide-bitfields
13396 Always treat bit-fields as int-sized.
13398 @item -m4byte-functions
13399 @itemx -mno-4byte-functions
13400 @opindex m4byte-functions
13401 @opindex mno-4byte-functions
13402 Force all functions to be aligned to a four byte boundary.
13404 @item -mcallgraph-data
13405 @itemx -mno-callgraph-data
13406 @opindex mcallgraph-data
13407 @opindex mno-callgraph-data
13408 Emit callgraph information.
13411 @itemx -mno-slow-bytes
13412 @opindex mslow-bytes
13413 @opindex mno-slow-bytes
13414 Prefer word access when reading byte quantities.
13416 @item -mlittle-endian
13417 @itemx -mbig-endian
13418 @opindex mlittle-endian
13419 @opindex mbig-endian
13420 Generate code for a little endian target.
13426 Generate code for the 210 processor.
13430 Assume that run-time support has been provided and so omit the
13431 simulator library (@file{libsim.a)} from the linker command line.
13433 @item -mstack-increment=@var{size}
13434 @opindex mstack-increment
13435 Set the maximum amount for a single stack increment operation. Large
13436 values can increase the speed of programs which contain functions
13437 that need a large amount of stack space, but they can also trigger a
13438 segmentation fault if the stack is extended too much. The default
13444 @subsection MeP Options
13445 @cindex MeP options
13451 Enables the @code{abs} instruction, which is the absolute difference
13452 between two registers.
13456 Enables all the optional instructions - average, multiply, divide, bit
13457 operations, leading zero, absolute difference, min/max, clip, and
13463 Enables the @code{ave} instruction, which computes the average of two
13466 @item -mbased=@var{n}
13468 Variables of size @var{n} bytes or smaller will be placed in the
13469 @code{.based} section by default. Based variables use the @code{$tp}
13470 register as a base register, and there is a 128 byte limit to the
13471 @code{.based} section.
13475 Enables the bit operation instructions - bit test (@code{btstm}), set
13476 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13477 test-and-set (@code{tas}).
13479 @item -mc=@var{name}
13481 Selects which section constant data will be placed in. @var{name} may
13482 be @code{tiny}, @code{near}, or @code{far}.
13486 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13487 useful unless you also provide @code{-mminmax}.
13489 @item -mconfig=@var{name}
13491 Selects one of the build-in core configurations. Each MeP chip has
13492 one or more modules in it; each module has a core CPU and a variety of
13493 coprocessors, optional instructions, and peripherals. The
13494 @code{MeP-Integrator} tool, not part of GCC, provides these
13495 configurations through this option; using this option is the same as
13496 using all the corresponding command line options. The default
13497 configuration is @code{default}.
13501 Enables the coprocessor instructions. By default, this is a 32-bit
13502 coprocessor. Note that the coprocessor is normally enabled via the
13503 @code{-mconfig=} option.
13507 Enables the 32-bit coprocessor's instructions.
13511 Enables the 64-bit coprocessor's instructions.
13515 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13519 Causes constant variables to be placed in the @code{.near} section.
13523 Enables the @code{div} and @code{divu} instructions.
13527 Generate big-endian code.
13531 Generate little-endian code.
13533 @item -mio-volatile
13534 @opindex mio-volatile
13535 Tells the compiler that any variable marked with the @code{io}
13536 attribute is to be considered volatile.
13540 Causes variables to be assigned to the @code{.far} section by default.
13544 Enables the @code{leadz} (leading zero) instruction.
13548 Causes variables to be assigned to the @code{.near} section by default.
13552 Enables the @code{min} and @code{max} instructions.
13556 Enables the multiplication and multiply-accumulate instructions.
13560 Disables all the optional instructions enabled by @code{-mall-opts}.
13564 Enables the @code{repeat} and @code{erepeat} instructions, used for
13565 low-overhead looping.
13569 Causes all variables to default to the @code{.tiny} section. Note
13570 that there is a 65536 byte limit to this section. Accesses to these
13571 variables use the @code{%gp} base register.
13575 Enables the saturation instructions. Note that the compiler does not
13576 currently generate these itself, but this option is included for
13577 compatibility with other tools, like @code{as}.
13581 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13585 Link the simulator runtime libraries.
13589 Link the simulator runtime libraries, excluding built-in support
13590 for reset and exception vectors and tables.
13594 Causes all functions to default to the @code{.far} section. Without
13595 this option, functions default to the @code{.near} section.
13597 @item -mtiny=@var{n}
13599 Variables that are @var{n} bytes or smaller will be allocated to the
13600 @code{.tiny} section. These variables use the @code{$gp} base
13601 register. The default for this option is 4, but note that there's a
13602 65536 byte limit to the @code{.tiny} section.
13607 @subsection MIPS Options
13608 @cindex MIPS options
13614 Generate big-endian code.
13618 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13621 @item -march=@var{arch}
13623 Generate code that will run on @var{arch}, which can be the name of a
13624 generic MIPS ISA, or the name of a particular processor.
13626 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13627 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13628 The processor names are:
13629 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13630 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13631 @samp{5kc}, @samp{5kf},
13633 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13634 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13635 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13636 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13637 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13638 @samp{loongson2e}, @samp{loongson2f},
13642 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13643 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13644 @samp{rm7000}, @samp{rm9000},
13645 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13648 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13649 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13651 The special value @samp{from-abi} selects the
13652 most compatible architecture for the selected ABI (that is,
13653 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13655 Native Linux/GNU toolchains also support the value @samp{native},
13656 which selects the best architecture option for the host processor.
13657 @option{-march=native} has no effect if GCC does not recognize
13660 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13661 (for example, @samp{-march=r2k}). Prefixes are optional, and
13662 @samp{vr} may be written @samp{r}.
13664 Names of the form @samp{@var{n}f2_1} refer to processors with
13665 FPUs clocked at half the rate of the core, names of the form
13666 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13667 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13668 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13669 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13670 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13671 accepted as synonyms for @samp{@var{n}f1_1}.
13673 GCC defines two macros based on the value of this option. The first
13674 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13675 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13676 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13677 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13678 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13680 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13681 above. In other words, it will have the full prefix and will not
13682 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13683 the macro names the resolved architecture (either @samp{"mips1"} or
13684 @samp{"mips3"}). It names the default architecture when no
13685 @option{-march} option is given.
13687 @item -mtune=@var{arch}
13689 Optimize for @var{arch}. Among other things, this option controls
13690 the way instructions are scheduled, and the perceived cost of arithmetic
13691 operations. The list of @var{arch} values is the same as for
13694 When this option is not used, GCC will optimize for the processor
13695 specified by @option{-march}. By using @option{-march} and
13696 @option{-mtune} together, it is possible to generate code that will
13697 run on a family of processors, but optimize the code for one
13698 particular member of that family.
13700 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13701 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13702 @samp{-march} ones described above.
13706 Equivalent to @samp{-march=mips1}.
13710 Equivalent to @samp{-march=mips2}.
13714 Equivalent to @samp{-march=mips3}.
13718 Equivalent to @samp{-march=mips4}.
13722 Equivalent to @samp{-march=mips32}.
13726 Equivalent to @samp{-march=mips32r2}.
13730 Equivalent to @samp{-march=mips64}.
13734 Equivalent to @samp{-march=mips64r2}.
13739 @opindex mno-mips16
13740 Generate (do not generate) MIPS16 code. If GCC is targetting a
13741 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13743 MIPS16 code generation can also be controlled on a per-function basis
13744 by means of @code{mips16} and @code{nomips16} attributes.
13745 @xref{Function Attributes}, for more information.
13747 @item -mflip-mips16
13748 @opindex mflip-mips16
13749 Generate MIPS16 code on alternating functions. This option is provided
13750 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13751 not intended for ordinary use in compiling user code.
13753 @item -minterlink-mips16
13754 @itemx -mno-interlink-mips16
13755 @opindex minterlink-mips16
13756 @opindex mno-interlink-mips16
13757 Require (do not require) that non-MIPS16 code be link-compatible with
13760 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13761 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13762 therefore disables direct jumps unless GCC knows that the target of the
13763 jump is not MIPS16.
13775 Generate code for the given ABI@.
13777 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13778 generates 64-bit code when you select a 64-bit architecture, but you
13779 can use @option{-mgp32} to get 32-bit code instead.
13781 For information about the O64 ABI, see
13782 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13784 GCC supports a variant of the o32 ABI in which floating-point registers
13785 are 64 rather than 32 bits wide. You can select this combination with
13786 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13787 and @samp{mfhc1} instructions and is therefore only supported for
13788 MIPS32R2 processors.
13790 The register assignments for arguments and return values remain the
13791 same, but each scalar value is passed in a single 64-bit register
13792 rather than a pair of 32-bit registers. For example, scalar
13793 floating-point values are returned in @samp{$f0} only, not a
13794 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13795 remains the same, but all 64 bits are saved.
13798 @itemx -mno-abicalls
13800 @opindex mno-abicalls
13801 Generate (do not generate) code that is suitable for SVR4-style
13802 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13807 Generate (do not generate) code that is fully position-independent,
13808 and that can therefore be linked into shared libraries. This option
13809 only affects @option{-mabicalls}.
13811 All @option{-mabicalls} code has traditionally been position-independent,
13812 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13813 as an extension, the GNU toolchain allows executables to use absolute
13814 accesses for locally-binding symbols. It can also use shorter GP
13815 initialization sequences and generate direct calls to locally-defined
13816 functions. This mode is selected by @option{-mno-shared}.
13818 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13819 objects that can only be linked by the GNU linker. However, the option
13820 does not affect the ABI of the final executable; it only affects the ABI
13821 of relocatable objects. Using @option{-mno-shared} will generally make
13822 executables both smaller and quicker.
13824 @option{-mshared} is the default.
13830 Assume (do not assume) that the static and dynamic linkers
13831 support PLTs and copy relocations. This option only affects
13832 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13833 has no effect without @samp{-msym32}.
13835 You can make @option{-mplt} the default by configuring
13836 GCC with @option{--with-mips-plt}. The default is
13837 @option{-mno-plt} otherwise.
13843 Lift (do not lift) the usual restrictions on the size of the global
13846 GCC normally uses a single instruction to load values from the GOT@.
13847 While this is relatively efficient, it will only work if the GOT
13848 is smaller than about 64k. Anything larger will cause the linker
13849 to report an error such as:
13851 @cindex relocation truncated to fit (MIPS)
13853 relocation truncated to fit: R_MIPS_GOT16 foobar
13856 If this happens, you should recompile your code with @option{-mxgot}.
13857 It should then work with very large GOTs, although it will also be
13858 less efficient, since it will take three instructions to fetch the
13859 value of a global symbol.
13861 Note that some linkers can create multiple GOTs. If you have such a
13862 linker, you should only need to use @option{-mxgot} when a single object
13863 file accesses more than 64k's worth of GOT entries. Very few do.
13865 These options have no effect unless GCC is generating position
13870 Assume that general-purpose registers are 32 bits wide.
13874 Assume that general-purpose registers are 64 bits wide.
13878 Assume that floating-point registers are 32 bits wide.
13882 Assume that floating-point registers are 64 bits wide.
13885 @opindex mhard-float
13886 Use floating-point coprocessor instructions.
13889 @opindex msoft-float
13890 Do not use floating-point coprocessor instructions. Implement
13891 floating-point calculations using library calls instead.
13893 @item -msingle-float
13894 @opindex msingle-float
13895 Assume that the floating-point coprocessor only supports single-precision
13898 @item -mdouble-float
13899 @opindex mdouble-float
13900 Assume that the floating-point coprocessor supports double-precision
13901 operations. This is the default.
13907 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13908 implement atomic memory built-in functions. When neither option is
13909 specified, GCC will use the instructions if the target architecture
13912 @option{-mllsc} is useful if the runtime environment can emulate the
13913 instructions and @option{-mno-llsc} can be useful when compiling for
13914 nonstandard ISAs. You can make either option the default by
13915 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
13916 respectively. @option{--with-llsc} is the default for some
13917 configurations; see the installation documentation for details.
13923 Use (do not use) revision 1 of the MIPS DSP ASE@.
13924 @xref{MIPS DSP Built-in Functions}. This option defines the
13925 preprocessor macro @samp{__mips_dsp}. It also defines
13926 @samp{__mips_dsp_rev} to 1.
13932 Use (do not use) revision 2 of the MIPS DSP ASE@.
13933 @xref{MIPS DSP Built-in Functions}. This option defines the
13934 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
13935 It also defines @samp{__mips_dsp_rev} to 2.
13938 @itemx -mno-smartmips
13939 @opindex msmartmips
13940 @opindex mno-smartmips
13941 Use (do not use) the MIPS SmartMIPS ASE.
13943 @item -mpaired-single
13944 @itemx -mno-paired-single
13945 @opindex mpaired-single
13946 @opindex mno-paired-single
13947 Use (do not use) paired-single floating-point instructions.
13948 @xref{MIPS Paired-Single Support}. This option requires
13949 hardware floating-point support to be enabled.
13955 Use (do not use) MIPS Digital Media Extension instructions.
13956 This option can only be used when generating 64-bit code and requires
13957 hardware floating-point support to be enabled.
13962 @opindex mno-mips3d
13963 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
13964 The option @option{-mips3d} implies @option{-mpaired-single}.
13970 Use (do not use) MT Multithreading instructions.
13974 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
13975 an explanation of the default and the way that the pointer size is
13980 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
13982 The default size of @code{int}s, @code{long}s and pointers depends on
13983 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
13984 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
13985 32-bit @code{long}s. Pointers are the same size as @code{long}s,
13986 or the same size as integer registers, whichever is smaller.
13992 Assume (do not assume) that all symbols have 32-bit values, regardless
13993 of the selected ABI@. This option is useful in combination with
13994 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
13995 to generate shorter and faster references to symbolic addresses.
13999 Put definitions of externally-visible data in a small data section
14000 if that data is no bigger than @var{num} bytes. GCC can then access
14001 the data more efficiently; see @option{-mgpopt} for details.
14003 The default @option{-G} option depends on the configuration.
14005 @item -mlocal-sdata
14006 @itemx -mno-local-sdata
14007 @opindex mlocal-sdata
14008 @opindex mno-local-sdata
14009 Extend (do not extend) the @option{-G} behavior to local data too,
14010 such as to static variables in C@. @option{-mlocal-sdata} is the
14011 default for all configurations.
14013 If the linker complains that an application is using too much small data,
14014 you might want to try rebuilding the less performance-critical parts with
14015 @option{-mno-local-sdata}. You might also want to build large
14016 libraries with @option{-mno-local-sdata}, so that the libraries leave
14017 more room for the main program.
14019 @item -mextern-sdata
14020 @itemx -mno-extern-sdata
14021 @opindex mextern-sdata
14022 @opindex mno-extern-sdata
14023 Assume (do not assume) that externally-defined data will be in
14024 a small data section if that data is within the @option{-G} limit.
14025 @option{-mextern-sdata} is the default for all configurations.
14027 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
14028 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
14029 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
14030 is placed in a small data section. If @var{Var} is defined by another
14031 module, you must either compile that module with a high-enough
14032 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
14033 definition. If @var{Var} is common, you must link the application
14034 with a high-enough @option{-G} setting.
14036 The easiest way of satisfying these restrictions is to compile
14037 and link every module with the same @option{-G} option. However,
14038 you may wish to build a library that supports several different
14039 small data limits. You can do this by compiling the library with
14040 the highest supported @option{-G} setting and additionally using
14041 @option{-mno-extern-sdata} to stop the library from making assumptions
14042 about externally-defined data.
14048 Use (do not use) GP-relative accesses for symbols that are known to be
14049 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
14050 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
14053 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
14054 might not hold the value of @code{_gp}. For example, if the code is
14055 part of a library that might be used in a boot monitor, programs that
14056 call boot monitor routines will pass an unknown value in @code{$gp}.
14057 (In such situations, the boot monitor itself would usually be compiled
14058 with @option{-G0}.)
14060 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
14061 @option{-mno-extern-sdata}.
14063 @item -membedded-data
14064 @itemx -mno-embedded-data
14065 @opindex membedded-data
14066 @opindex mno-embedded-data
14067 Allocate variables to the read-only data section first if possible, then
14068 next in the small data section if possible, otherwise in data. This gives
14069 slightly slower code than the default, but reduces the amount of RAM required
14070 when executing, and thus may be preferred for some embedded systems.
14072 @item -muninit-const-in-rodata
14073 @itemx -mno-uninit-const-in-rodata
14074 @opindex muninit-const-in-rodata
14075 @opindex mno-uninit-const-in-rodata
14076 Put uninitialized @code{const} variables in the read-only data section.
14077 This option is only meaningful in conjunction with @option{-membedded-data}.
14079 @item -mcode-readable=@var{setting}
14080 @opindex mcode-readable
14081 Specify whether GCC may generate code that reads from executable sections.
14082 There are three possible settings:
14085 @item -mcode-readable=yes
14086 Instructions may freely access executable sections. This is the
14089 @item -mcode-readable=pcrel
14090 MIPS16 PC-relative load instructions can access executable sections,
14091 but other instructions must not do so. This option is useful on 4KSc
14092 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
14093 It is also useful on processors that can be configured to have a dual
14094 instruction/data SRAM interface and that, like the M4K, automatically
14095 redirect PC-relative loads to the instruction RAM.
14097 @item -mcode-readable=no
14098 Instructions must not access executable sections. This option can be
14099 useful on targets that are configured to have a dual instruction/data
14100 SRAM interface but that (unlike the M4K) do not automatically redirect
14101 PC-relative loads to the instruction RAM.
14104 @item -msplit-addresses
14105 @itemx -mno-split-addresses
14106 @opindex msplit-addresses
14107 @opindex mno-split-addresses
14108 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
14109 relocation operators. This option has been superseded by
14110 @option{-mexplicit-relocs} but is retained for backwards compatibility.
14112 @item -mexplicit-relocs
14113 @itemx -mno-explicit-relocs
14114 @opindex mexplicit-relocs
14115 @opindex mno-explicit-relocs
14116 Use (do not use) assembler relocation operators when dealing with symbolic
14117 addresses. The alternative, selected by @option{-mno-explicit-relocs},
14118 is to use assembler macros instead.
14120 @option{-mexplicit-relocs} is the default if GCC was configured
14121 to use an assembler that supports relocation operators.
14123 @item -mcheck-zero-division
14124 @itemx -mno-check-zero-division
14125 @opindex mcheck-zero-division
14126 @opindex mno-check-zero-division
14127 Trap (do not trap) on integer division by zero.
14129 The default is @option{-mcheck-zero-division}.
14131 @item -mdivide-traps
14132 @itemx -mdivide-breaks
14133 @opindex mdivide-traps
14134 @opindex mdivide-breaks
14135 MIPS systems check for division by zero by generating either a
14136 conditional trap or a break instruction. Using traps results in
14137 smaller code, but is only supported on MIPS II and later. Also, some
14138 versions of the Linux kernel have a bug that prevents trap from
14139 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
14140 allow conditional traps on architectures that support them and
14141 @option{-mdivide-breaks} to force the use of breaks.
14143 The default is usually @option{-mdivide-traps}, but this can be
14144 overridden at configure time using @option{--with-divide=breaks}.
14145 Divide-by-zero checks can be completely disabled using
14146 @option{-mno-check-zero-division}.
14151 @opindex mno-memcpy
14152 Force (do not force) the use of @code{memcpy()} for non-trivial block
14153 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
14154 most constant-sized copies.
14157 @itemx -mno-long-calls
14158 @opindex mlong-calls
14159 @opindex mno-long-calls
14160 Disable (do not disable) use of the @code{jal} instruction. Calling
14161 functions using @code{jal} is more efficient but requires the caller
14162 and callee to be in the same 256 megabyte segment.
14164 This option has no effect on abicalls code. The default is
14165 @option{-mno-long-calls}.
14171 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
14172 instructions, as provided by the R4650 ISA@.
14175 @itemx -mno-fused-madd
14176 @opindex mfused-madd
14177 @opindex mno-fused-madd
14178 Enable (disable) use of the floating point multiply-accumulate
14179 instructions, when they are available. The default is
14180 @option{-mfused-madd}.
14182 When multiply-accumulate instructions are used, the intermediate
14183 product is calculated to infinite precision and is not subject to
14184 the FCSR Flush to Zero bit. This may be undesirable in some
14189 Tell the MIPS assembler to not run its preprocessor over user
14190 assembler files (with a @samp{.s} suffix) when assembling them.
14193 @itemx -mno-fix-r4000
14194 @opindex mfix-r4000
14195 @opindex mno-fix-r4000
14196 Work around certain R4000 CPU errata:
14199 A double-word or a variable shift may give an incorrect result if executed
14200 immediately after starting an integer division.
14202 A double-word or a variable shift may give an incorrect result if executed
14203 while an integer multiplication is in progress.
14205 An integer division may give an incorrect result if started in a delay slot
14206 of a taken branch or a jump.
14210 @itemx -mno-fix-r4400
14211 @opindex mfix-r4400
14212 @opindex mno-fix-r4400
14213 Work around certain R4400 CPU errata:
14216 A double-word or a variable shift may give an incorrect result if executed
14217 immediately after starting an integer division.
14221 @itemx -mno-fix-r10000
14222 @opindex mfix-r10000
14223 @opindex mno-fix-r10000
14224 Work around certain R10000 errata:
14227 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14228 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14231 This option can only be used if the target architecture supports
14232 branch-likely instructions. @option{-mfix-r10000} is the default when
14233 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14237 @itemx -mno-fix-vr4120
14238 @opindex mfix-vr4120
14239 Work around certain VR4120 errata:
14242 @code{dmultu} does not always produce the correct result.
14244 @code{div} and @code{ddiv} do not always produce the correct result if one
14245 of the operands is negative.
14247 The workarounds for the division errata rely on special functions in
14248 @file{libgcc.a}. At present, these functions are only provided by
14249 the @code{mips64vr*-elf} configurations.
14251 Other VR4120 errata require a nop to be inserted between certain pairs of
14252 instructions. These errata are handled by the assembler, not by GCC itself.
14255 @opindex mfix-vr4130
14256 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14257 workarounds are implemented by the assembler rather than by GCC,
14258 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14259 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14260 instructions are available instead.
14263 @itemx -mno-fix-sb1
14265 Work around certain SB-1 CPU core errata.
14266 (This flag currently works around the SB-1 revision 2
14267 ``F1'' and ``F2'' floating point errata.)
14269 @item -mr10k-cache-barrier=@var{setting}
14270 @opindex mr10k-cache-barrier
14271 Specify whether GCC should insert cache barriers to avoid the
14272 side-effects of speculation on R10K processors.
14274 In common with many processors, the R10K tries to predict the outcome
14275 of a conditional branch and speculatively executes instructions from
14276 the ``taken'' branch. It later aborts these instructions if the
14277 predicted outcome was wrong. However, on the R10K, even aborted
14278 instructions can have side effects.
14280 This problem only affects kernel stores and, depending on the system,
14281 kernel loads. As an example, a speculatively-executed store may load
14282 the target memory into cache and mark the cache line as dirty, even if
14283 the store itself is later aborted. If a DMA operation writes to the
14284 same area of memory before the ``dirty'' line is flushed, the cached
14285 data will overwrite the DMA-ed data. See the R10K processor manual
14286 for a full description, including other potential problems.
14288 One workaround is to insert cache barrier instructions before every memory
14289 access that might be speculatively executed and that might have side
14290 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14291 controls GCC's implementation of this workaround. It assumes that
14292 aborted accesses to any byte in the following regions will not have
14297 the memory occupied by the current function's stack frame;
14300 the memory occupied by an incoming stack argument;
14303 the memory occupied by an object with a link-time-constant address.
14306 It is the kernel's responsibility to ensure that speculative
14307 accesses to these regions are indeed safe.
14309 If the input program contains a function declaration such as:
14315 then the implementation of @code{foo} must allow @code{j foo} and
14316 @code{jal foo} to be executed speculatively. GCC honors this
14317 restriction for functions it compiles itself. It expects non-GCC
14318 functions (such as hand-written assembly code) to do the same.
14320 The option has three forms:
14323 @item -mr10k-cache-barrier=load-store
14324 Insert a cache barrier before a load or store that might be
14325 speculatively executed and that might have side effects even
14328 @item -mr10k-cache-barrier=store
14329 Insert a cache barrier before a store that might be speculatively
14330 executed and that might have side effects even if aborted.
14332 @item -mr10k-cache-barrier=none
14333 Disable the insertion of cache barriers. This is the default setting.
14336 @item -mflush-func=@var{func}
14337 @itemx -mno-flush-func
14338 @opindex mflush-func
14339 Specifies the function to call to flush the I and D caches, or to not
14340 call any such function. If called, the function must take the same
14341 arguments as the common @code{_flush_func()}, that is, the address of the
14342 memory range for which the cache is being flushed, the size of the
14343 memory range, and the number 3 (to flush both caches). The default
14344 depends on the target GCC was configured for, but commonly is either
14345 @samp{_flush_func} or @samp{__cpu_flush}.
14347 @item mbranch-cost=@var{num}
14348 @opindex mbranch-cost
14349 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14350 This cost is only a heuristic and is not guaranteed to produce
14351 consistent results across releases. A zero cost redundantly selects
14352 the default, which is based on the @option{-mtune} setting.
14354 @item -mbranch-likely
14355 @itemx -mno-branch-likely
14356 @opindex mbranch-likely
14357 @opindex mno-branch-likely
14358 Enable or disable use of Branch Likely instructions, regardless of the
14359 default for the selected architecture. By default, Branch Likely
14360 instructions may be generated if they are supported by the selected
14361 architecture. An exception is for the MIPS32 and MIPS64 architectures
14362 and processors which implement those architectures; for those, Branch
14363 Likely instructions will not be generated by default because the MIPS32
14364 and MIPS64 architectures specifically deprecate their use.
14366 @item -mfp-exceptions
14367 @itemx -mno-fp-exceptions
14368 @opindex mfp-exceptions
14369 Specifies whether FP exceptions are enabled. This affects how we schedule
14370 FP instructions for some processors. The default is that FP exceptions are
14373 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14374 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14377 @item -mvr4130-align
14378 @itemx -mno-vr4130-align
14379 @opindex mvr4130-align
14380 The VR4130 pipeline is two-way superscalar, but can only issue two
14381 instructions together if the first one is 8-byte aligned. When this
14382 option is enabled, GCC will align pairs of instructions that it
14383 thinks should execute in parallel.
14385 This option only has an effect when optimizing for the VR4130.
14386 It normally makes code faster, but at the expense of making it bigger.
14387 It is enabled by default at optimization level @option{-O3}.
14392 Enable (disable) generation of @code{synci} instructions on
14393 architectures that support it. The @code{synci} instructions (if
14394 enabled) will be generated when @code{__builtin___clear_cache()} is
14397 This option defaults to @code{-mno-synci}, but the default can be
14398 overridden by configuring with @code{--with-synci}.
14400 When compiling code for single processor systems, it is generally safe
14401 to use @code{synci}. However, on many multi-core (SMP) systems, it
14402 will not invalidate the instruction caches on all cores and may lead
14403 to undefined behavior.
14405 @item -mrelax-pic-calls
14406 @itemx -mno-relax-pic-calls
14407 @opindex mrelax-pic-calls
14408 Try to turn PIC calls that are normally dispatched via register
14409 @code{$25} into direct calls. This is only possible if the linker can
14410 resolve the destination at link-time and if the destination is within
14411 range for a direct call.
14413 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14414 an assembler and a linker that supports the @code{.reloc} assembly
14415 directive and @code{-mexplicit-relocs} is in effect. With
14416 @code{-mno-explicit-relocs}, this optimization can be performed by the
14417 assembler and the linker alone without help from the compiler.
14419 @item -mmcount-ra-address
14420 @itemx -mno-mcount-ra-address
14421 @opindex mmcount-ra-address
14422 @opindex mno-mcount-ra-address
14423 Emit (do not emit) code that allows @code{_mcount} to modify the
14424 calling function's return address. When enabled, this option extends
14425 the usual @code{_mcount} interface with a new @var{ra-address}
14426 parameter, which has type @code{intptr_t *} and is passed in register
14427 @code{$12}. @code{_mcount} can then modify the return address by
14428 doing both of the following:
14431 Returning the new address in register @code{$31}.
14433 Storing the new address in @code{*@var{ra-address}},
14434 if @var{ra-address} is nonnull.
14437 The default is @option{-mno-mcount-ra-address}.
14442 @subsection MMIX Options
14443 @cindex MMIX Options
14445 These options are defined for the MMIX:
14449 @itemx -mno-libfuncs
14451 @opindex mno-libfuncs
14452 Specify that intrinsic library functions are being compiled, passing all
14453 values in registers, no matter the size.
14456 @itemx -mno-epsilon
14458 @opindex mno-epsilon
14459 Generate floating-point comparison instructions that compare with respect
14460 to the @code{rE} epsilon register.
14462 @item -mabi=mmixware
14464 @opindex mabi=mmixware
14466 Generate code that passes function parameters and return values that (in
14467 the called function) are seen as registers @code{$0} and up, as opposed to
14468 the GNU ABI which uses global registers @code{$231} and up.
14470 @item -mzero-extend
14471 @itemx -mno-zero-extend
14472 @opindex mzero-extend
14473 @opindex mno-zero-extend
14474 When reading data from memory in sizes shorter than 64 bits, use (do not
14475 use) zero-extending load instructions by default, rather than
14476 sign-extending ones.
14479 @itemx -mno-knuthdiv
14481 @opindex mno-knuthdiv
14482 Make the result of a division yielding a remainder have the same sign as
14483 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14484 remainder follows the sign of the dividend. Both methods are
14485 arithmetically valid, the latter being almost exclusively used.
14487 @item -mtoplevel-symbols
14488 @itemx -mno-toplevel-symbols
14489 @opindex mtoplevel-symbols
14490 @opindex mno-toplevel-symbols
14491 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14492 code can be used with the @code{PREFIX} assembly directive.
14496 Generate an executable in the ELF format, rather than the default
14497 @samp{mmo} format used by the @command{mmix} simulator.
14499 @item -mbranch-predict
14500 @itemx -mno-branch-predict
14501 @opindex mbranch-predict
14502 @opindex mno-branch-predict
14503 Use (do not use) the probable-branch instructions, when static branch
14504 prediction indicates a probable branch.
14506 @item -mbase-addresses
14507 @itemx -mno-base-addresses
14508 @opindex mbase-addresses
14509 @opindex mno-base-addresses
14510 Generate (do not generate) code that uses @emph{base addresses}. Using a
14511 base address automatically generates a request (handled by the assembler
14512 and the linker) for a constant to be set up in a global register. The
14513 register is used for one or more base address requests within the range 0
14514 to 255 from the value held in the register. The generally leads to short
14515 and fast code, but the number of different data items that can be
14516 addressed is limited. This means that a program that uses lots of static
14517 data may require @option{-mno-base-addresses}.
14519 @item -msingle-exit
14520 @itemx -mno-single-exit
14521 @opindex msingle-exit
14522 @opindex mno-single-exit
14523 Force (do not force) generated code to have a single exit point in each
14527 @node MN10300 Options
14528 @subsection MN10300 Options
14529 @cindex MN10300 options
14531 These @option{-m} options are defined for Matsushita MN10300 architectures:
14536 Generate code to avoid bugs in the multiply instructions for the MN10300
14537 processors. This is the default.
14539 @item -mno-mult-bug
14540 @opindex mno-mult-bug
14541 Do not generate code to avoid bugs in the multiply instructions for the
14542 MN10300 processors.
14546 Generate code which uses features specific to the AM33 processor.
14550 Do not generate code which uses features specific to the AM33 processor. This
14553 @item -mreturn-pointer-on-d0
14554 @opindex mreturn-pointer-on-d0
14555 When generating a function which returns a pointer, return the pointer
14556 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14557 only in a0, and attempts to call such functions without a prototype
14558 would result in errors. Note that this option is on by default; use
14559 @option{-mno-return-pointer-on-d0} to disable it.
14563 Do not link in the C run-time initialization object file.
14567 Indicate to the linker that it should perform a relaxation optimization pass
14568 to shorten branches, calls and absolute memory addresses. This option only
14569 has an effect when used on the command line for the final link step.
14571 This option makes symbolic debugging impossible.
14574 @node PDP-11 Options
14575 @subsection PDP-11 Options
14576 @cindex PDP-11 Options
14578 These options are defined for the PDP-11:
14583 Use hardware FPP floating point. This is the default. (FIS floating
14584 point on the PDP-11/40 is not supported.)
14587 @opindex msoft-float
14588 Do not use hardware floating point.
14592 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14596 Return floating-point results in memory. This is the default.
14600 Generate code for a PDP-11/40.
14604 Generate code for a PDP-11/45. This is the default.
14608 Generate code for a PDP-11/10.
14610 @item -mbcopy-builtin
14611 @opindex mbcopy-builtin
14612 Use inline @code{movmemhi} patterns for copying memory. This is the
14617 Do not use inline @code{movmemhi} patterns for copying memory.
14623 Use 16-bit @code{int}. This is the default.
14629 Use 32-bit @code{int}.
14632 @itemx -mno-float32
14634 @opindex mno-float32
14635 Use 64-bit @code{float}. This is the default.
14638 @itemx -mno-float64
14640 @opindex mno-float64
14641 Use 32-bit @code{float}.
14645 Use @code{abshi2} pattern. This is the default.
14649 Do not use @code{abshi2} pattern.
14651 @item -mbranch-expensive
14652 @opindex mbranch-expensive
14653 Pretend that branches are expensive. This is for experimenting with
14654 code generation only.
14656 @item -mbranch-cheap
14657 @opindex mbranch-cheap
14658 Do not pretend that branches are expensive. This is the default.
14662 Generate code for a system with split I&D@.
14666 Generate code for a system without split I&D@. This is the default.
14670 Use Unix assembler syntax. This is the default when configured for
14671 @samp{pdp11-*-bsd}.
14675 Use DEC assembler syntax. This is the default when configured for any
14676 PDP-11 target other than @samp{pdp11-*-bsd}.
14679 @node picoChip Options
14680 @subsection picoChip Options
14681 @cindex picoChip options
14683 These @samp{-m} options are defined for picoChip implementations:
14687 @item -mae=@var{ae_type}
14689 Set the instruction set, register set, and instruction scheduling
14690 parameters for array element type @var{ae_type}. Supported values
14691 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14693 @option{-mae=ANY} selects a completely generic AE type. Code
14694 generated with this option will run on any of the other AE types. The
14695 code will not be as efficient as it would be if compiled for a specific
14696 AE type, and some types of operation (e.g., multiplication) will not
14697 work properly on all types of AE.
14699 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14700 for compiled code, and is the default.
14702 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14703 option may suffer from poor performance of byte (char) manipulation,
14704 since the DSP AE does not provide hardware support for byte load/stores.
14706 @item -msymbol-as-address
14707 Enable the compiler to directly use a symbol name as an address in a
14708 load/store instruction, without first loading it into a
14709 register. Typically, the use of this option will generate larger
14710 programs, which run faster than when the option isn't used. However, the
14711 results vary from program to program, so it is left as a user option,
14712 rather than being permanently enabled.
14714 @item -mno-inefficient-warnings
14715 Disables warnings about the generation of inefficient code. These
14716 warnings can be generated, for example, when compiling code which
14717 performs byte-level memory operations on the MAC AE type. The MAC AE has
14718 no hardware support for byte-level memory operations, so all byte
14719 load/stores must be synthesized from word load/store operations. This is
14720 inefficient and a warning will be generated indicating to the programmer
14721 that they should rewrite the code to avoid byte operations, or to target
14722 an AE type which has the necessary hardware support. This option enables
14723 the warning to be turned off.
14727 @node PowerPC Options
14728 @subsection PowerPC Options
14729 @cindex PowerPC options
14731 These are listed under @xref{RS/6000 and PowerPC Options}.
14733 @node RS/6000 and PowerPC Options
14734 @subsection IBM RS/6000 and PowerPC Options
14735 @cindex RS/6000 and PowerPC Options
14736 @cindex IBM RS/6000 and PowerPC Options
14738 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14745 @itemx -mno-powerpc
14746 @itemx -mpowerpc-gpopt
14747 @itemx -mno-powerpc-gpopt
14748 @itemx -mpowerpc-gfxopt
14749 @itemx -mno-powerpc-gfxopt
14751 @itemx -mno-powerpc64
14755 @itemx -mno-popcntb
14757 @itemx -mno-popcntd
14765 @itemx -mno-hard-dfp
14769 @opindex mno-power2
14771 @opindex mno-powerpc
14772 @opindex mpowerpc-gpopt
14773 @opindex mno-powerpc-gpopt
14774 @opindex mpowerpc-gfxopt
14775 @opindex mno-powerpc-gfxopt
14776 @opindex mpowerpc64
14777 @opindex mno-powerpc64
14781 @opindex mno-popcntb
14783 @opindex mno-popcntd
14789 @opindex mno-mfpgpr
14791 @opindex mno-hard-dfp
14792 GCC supports two related instruction set architectures for the
14793 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14794 instructions supported by the @samp{rios} chip set used in the original
14795 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14796 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14797 the IBM 4xx, 6xx, and follow-on microprocessors.
14799 Neither architecture is a subset of the other. However there is a
14800 large common subset of instructions supported by both. An MQ
14801 register is included in processors supporting the POWER architecture.
14803 You use these options to specify which instructions are available on the
14804 processor you are using. The default value of these options is
14805 determined when configuring GCC@. Specifying the
14806 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14807 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14808 rather than the options listed above.
14810 The @option{-mpower} option allows GCC to generate instructions that
14811 are found only in the POWER architecture and to use the MQ register.
14812 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14813 to generate instructions that are present in the POWER2 architecture but
14814 not the original POWER architecture.
14816 The @option{-mpowerpc} option allows GCC to generate instructions that
14817 are found only in the 32-bit subset of the PowerPC architecture.
14818 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14819 GCC to use the optional PowerPC architecture instructions in the
14820 General Purpose group, including floating-point square root. Specifying
14821 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14822 use the optional PowerPC architecture instructions in the Graphics
14823 group, including floating-point select.
14825 The @option{-mmfcrf} option allows GCC to generate the move from
14826 condition register field instruction implemented on the POWER4
14827 processor and other processors that support the PowerPC V2.01
14829 The @option{-mpopcntb} option allows GCC to generate the popcount and
14830 double precision FP reciprocal estimate instruction implemented on the
14831 POWER5 processor and other processors that support the PowerPC V2.02
14833 The @option{-mpopcntd} option allows GCC to generate the popcount
14834 instruction implemented on the POWER7 processor and other processors
14835 that support the PowerPC V2.06 architecture.
14836 The @option{-mfprnd} option allows GCC to generate the FP round to
14837 integer instructions implemented on the POWER5+ processor and other
14838 processors that support the PowerPC V2.03 architecture.
14839 The @option{-mcmpb} option allows GCC to generate the compare bytes
14840 instruction implemented on the POWER6 processor and other processors
14841 that support the PowerPC V2.05 architecture.
14842 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14843 general purpose register instructions implemented on the POWER6X
14844 processor and other processors that support the extended PowerPC V2.05
14846 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14847 point instructions implemented on some POWER processors.
14849 The @option{-mpowerpc64} option allows GCC to generate the additional
14850 64-bit instructions that are found in the full PowerPC64 architecture
14851 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14852 @option{-mno-powerpc64}.
14854 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14855 will use only the instructions in the common subset of both
14856 architectures plus some special AIX common-mode calls, and will not use
14857 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14858 permits GCC to use any instruction from either architecture and to
14859 allow use of the MQ register; specify this for the Motorola MPC601.
14861 @item -mnew-mnemonics
14862 @itemx -mold-mnemonics
14863 @opindex mnew-mnemonics
14864 @opindex mold-mnemonics
14865 Select which mnemonics to use in the generated assembler code. With
14866 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14867 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14868 assembler mnemonics defined for the POWER architecture. Instructions
14869 defined in only one architecture have only one mnemonic; GCC uses that
14870 mnemonic irrespective of which of these options is specified.
14872 GCC defaults to the mnemonics appropriate for the architecture in
14873 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14874 value of these option. Unless you are building a cross-compiler, you
14875 should normally not specify either @option{-mnew-mnemonics} or
14876 @option{-mold-mnemonics}, but should instead accept the default.
14878 @item -mcpu=@var{cpu_type}
14880 Set architecture type, register usage, choice of mnemonics, and
14881 instruction scheduling parameters for machine type @var{cpu_type}.
14882 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14883 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14884 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
14885 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
14886 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14887 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
14888 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
14889 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
14890 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
14891 @samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14892 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14894 @option{-mcpu=common} selects a completely generic processor. Code
14895 generated under this option will run on any POWER or PowerPC processor.
14896 GCC will use only the instructions in the common subset of both
14897 architectures, and will not use the MQ register. GCC assumes a generic
14898 processor model for scheduling purposes.
14900 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14901 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14902 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14903 types, with an appropriate, generic processor model assumed for
14904 scheduling purposes.
14906 The other options specify a specific processor. Code generated under
14907 those options will run best on that processor, and may not run at all on
14910 The @option{-mcpu} options automatically enable or disable the
14913 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
14914 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
14915 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
14916 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
14918 The particular options set for any particular CPU will vary between
14919 compiler versions, depending on what setting seems to produce optimal
14920 code for that CPU; it doesn't necessarily reflect the actual hardware's
14921 capabilities. If you wish to set an individual option to a particular
14922 value, you may specify it after the @option{-mcpu} option, like
14923 @samp{-mcpu=970 -mno-altivec}.
14925 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
14926 not enabled or disabled by the @option{-mcpu} option at present because
14927 AIX does not have full support for these options. You may still
14928 enable or disable them individually if you're sure it'll work in your
14931 @item -mtune=@var{cpu_type}
14933 Set the instruction scheduling parameters for machine type
14934 @var{cpu_type}, but do not set the architecture type, register usage, or
14935 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
14936 values for @var{cpu_type} are used for @option{-mtune} as for
14937 @option{-mcpu}. If both are specified, the code generated will use the
14938 architecture, registers, and mnemonics set by @option{-mcpu}, but the
14939 scheduling parameters set by @option{-mtune}.
14945 Generate code to compute division as reciprocal estimate and iterative
14946 refinement, creating opportunities for increased throughput. This
14947 feature requires: optional PowerPC Graphics instruction set for single
14948 precision and FRE instruction for double precision, assuming divides
14949 cannot generate user-visible traps, and the domain values not include
14950 Infinities, denormals or zero denominator.
14953 @itemx -mno-altivec
14955 @opindex mno-altivec
14956 Generate code that uses (does not use) AltiVec instructions, and also
14957 enable the use of built-in functions that allow more direct access to
14958 the AltiVec instruction set. You may also need to set
14959 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
14965 @opindex mno-vrsave
14966 Generate VRSAVE instructions when generating AltiVec code.
14968 @item -mgen-cell-microcode
14969 @opindex mgen-cell-microcode
14970 Generate Cell microcode instructions
14972 @item -mwarn-cell-microcode
14973 @opindex mwarn-cell-microcode
14974 Warning when a Cell microcode instruction is going to emitted. An example
14975 of a Cell microcode instruction is a variable shift.
14978 @opindex msecure-plt
14979 Generate code that allows ld and ld.so to build executables and shared
14980 libraries with non-exec .plt and .got sections. This is a PowerPC
14981 32-bit SYSV ABI option.
14985 Generate code that uses a BSS .plt section that ld.so fills in, and
14986 requires .plt and .got sections that are both writable and executable.
14987 This is a PowerPC 32-bit SYSV ABI option.
14993 This switch enables or disables the generation of ISEL instructions.
14995 @item -misel=@var{yes/no}
14996 This switch has been deprecated. Use @option{-misel} and
14997 @option{-mno-isel} instead.
15003 This switch enables or disables the generation of SPE simd
15009 @opindex mno-paired
15010 This switch enables or disables the generation of PAIRED simd
15013 @item -mspe=@var{yes/no}
15014 This option has been deprecated. Use @option{-mspe} and
15015 @option{-mno-spe} instead.
15021 Generate code that uses (does not use) vector/scalar (VSX)
15022 instructions, and also enable the use of built-in functions that allow
15023 more direct access to the VSX instruction set.
15025 @item -mfloat-gprs=@var{yes/single/double/no}
15026 @itemx -mfloat-gprs
15027 @opindex mfloat-gprs
15028 This switch enables or disables the generation of floating point
15029 operations on the general purpose registers for architectures that
15032 The argument @var{yes} or @var{single} enables the use of
15033 single-precision floating point operations.
15035 The argument @var{double} enables the use of single and
15036 double-precision floating point operations.
15038 The argument @var{no} disables floating point operations on the
15039 general purpose registers.
15041 This option is currently only available on the MPC854x.
15047 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
15048 targets (including GNU/Linux). The 32-bit environment sets int, long
15049 and pointer to 32 bits and generates code that runs on any PowerPC
15050 variant. The 64-bit environment sets int to 32 bits and long and
15051 pointer to 64 bits, and generates code for PowerPC64, as for
15052 @option{-mpowerpc64}.
15055 @itemx -mno-fp-in-toc
15056 @itemx -mno-sum-in-toc
15057 @itemx -mminimal-toc
15059 @opindex mno-fp-in-toc
15060 @opindex mno-sum-in-toc
15061 @opindex mminimal-toc
15062 Modify generation of the TOC (Table Of Contents), which is created for
15063 every executable file. The @option{-mfull-toc} option is selected by
15064 default. In that case, GCC will allocate at least one TOC entry for
15065 each unique non-automatic variable reference in your program. GCC
15066 will also place floating-point constants in the TOC@. However, only
15067 16,384 entries are available in the TOC@.
15069 If you receive a linker error message that saying you have overflowed
15070 the available TOC space, you can reduce the amount of TOC space used
15071 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
15072 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
15073 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
15074 generate code to calculate the sum of an address and a constant at
15075 run-time instead of putting that sum into the TOC@. You may specify one
15076 or both of these options. Each causes GCC to produce very slightly
15077 slower and larger code at the expense of conserving TOC space.
15079 If you still run out of space in the TOC even when you specify both of
15080 these options, specify @option{-mminimal-toc} instead. This option causes
15081 GCC to make only one TOC entry for every file. When you specify this
15082 option, GCC will produce code that is slower and larger but which
15083 uses extremely little TOC space. You may wish to use this option
15084 only on files that contain less frequently executed code.
15090 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
15091 @code{long} type, and the infrastructure needed to support them.
15092 Specifying @option{-maix64} implies @option{-mpowerpc64} and
15093 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
15094 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
15097 @itemx -mno-xl-compat
15098 @opindex mxl-compat
15099 @opindex mno-xl-compat
15100 Produce code that conforms more closely to IBM XL compiler semantics
15101 when using AIX-compatible ABI@. Pass floating-point arguments to
15102 prototyped functions beyond the register save area (RSA) on the stack
15103 in addition to argument FPRs. Do not assume that most significant
15104 double in 128-bit long double value is properly rounded when comparing
15105 values and converting to double. Use XL symbol names for long double
15108 The AIX calling convention was extended but not initially documented to
15109 handle an obscure K&R C case of calling a function that takes the
15110 address of its arguments with fewer arguments than declared. IBM XL
15111 compilers access floating point arguments which do not fit in the
15112 RSA from the stack when a subroutine is compiled without
15113 optimization. Because always storing floating-point arguments on the
15114 stack is inefficient and rarely needed, this option is not enabled by
15115 default and only is necessary when calling subroutines compiled by IBM
15116 XL compilers without optimization.
15120 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
15121 application written to use message passing with special startup code to
15122 enable the application to run. The system must have PE installed in the
15123 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
15124 must be overridden with the @option{-specs=} option to specify the
15125 appropriate directory location. The Parallel Environment does not
15126 support threads, so the @option{-mpe} option and the @option{-pthread}
15127 option are incompatible.
15129 @item -malign-natural
15130 @itemx -malign-power
15131 @opindex malign-natural
15132 @opindex malign-power
15133 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
15134 @option{-malign-natural} overrides the ABI-defined alignment of larger
15135 types, such as floating-point doubles, on their natural size-based boundary.
15136 The option @option{-malign-power} instructs GCC to follow the ABI-specified
15137 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
15139 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
15143 @itemx -mhard-float
15144 @opindex msoft-float
15145 @opindex mhard-float
15146 Generate code that does not use (uses) the floating-point register set.
15147 Software floating point emulation is provided if you use the
15148 @option{-msoft-float} option, and pass the option to GCC when linking.
15150 @item -msingle-float
15151 @itemx -mdouble-float
15152 @opindex msingle-float
15153 @opindex mdouble-float
15154 Generate code for single or double-precision floating point operations.
15155 @option{-mdouble-float} implies @option{-msingle-float}.
15158 @opindex msimple-fpu
15159 Do not generate sqrt and div instructions for hardware floating point unit.
15163 Specify type of floating point unit. Valid values are @var{sp_lite}
15164 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
15165 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
15166 and @var{dp_full} (equivalent to -mdouble-float).
15169 @opindex mxilinx-fpu
15170 Perform optimizations for floating point unit on Xilinx PPC 405/440.
15173 @itemx -mno-multiple
15175 @opindex mno-multiple
15176 Generate code that uses (does not use) the load multiple word
15177 instructions and the store multiple word instructions. These
15178 instructions are generated by default on POWER systems, and not
15179 generated on PowerPC systems. Do not use @option{-mmultiple} on little
15180 endian PowerPC systems, since those instructions do not work when the
15181 processor is in little endian mode. The exceptions are PPC740 and
15182 PPC750 which permit the instructions usage in little endian mode.
15187 @opindex mno-string
15188 Generate code that uses (does not use) the load string instructions
15189 and the store string word instructions to save multiple registers and
15190 do small block moves. These instructions are generated by default on
15191 POWER systems, and not generated on PowerPC systems. Do not use
15192 @option{-mstring} on little endian PowerPC systems, since those
15193 instructions do not work when the processor is in little endian mode.
15194 The exceptions are PPC740 and PPC750 which permit the instructions
15195 usage in little endian mode.
15200 @opindex mno-update
15201 Generate code that uses (does not use) the load or store instructions
15202 that update the base register to the address of the calculated memory
15203 location. These instructions are generated by default. If you use
15204 @option{-mno-update}, there is a small window between the time that the
15205 stack pointer is updated and the address of the previous frame is
15206 stored, which means code that walks the stack frame across interrupts or
15207 signals may get corrupted data.
15209 @item -mavoid-indexed-addresses
15210 @itemx -mno-avoid-indexed-addresses
15211 @opindex mavoid-indexed-addresses
15212 @opindex mno-avoid-indexed-addresses
15213 Generate code that tries to avoid (not avoid) the use of indexed load
15214 or store instructions. These instructions can incur a performance
15215 penalty on Power6 processors in certain situations, such as when
15216 stepping through large arrays that cross a 16M boundary. This option
15217 is enabled by default when targetting Power6 and disabled otherwise.
15220 @itemx -mno-fused-madd
15221 @opindex mfused-madd
15222 @opindex mno-fused-madd
15223 Generate code that uses (does not use) the floating point multiply and
15224 accumulate instructions. These instructions are generated by default if
15225 hardware floating is used.
15231 Generate code that uses (does not use) the half-word multiply and
15232 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15233 These instructions are generated by default when targetting those
15240 Generate code that uses (does not use) the string-search @samp{dlmzb}
15241 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15242 generated by default when targetting those processors.
15244 @item -mno-bit-align
15246 @opindex mno-bit-align
15247 @opindex mbit-align
15248 On System V.4 and embedded PowerPC systems do not (do) force structures
15249 and unions that contain bit-fields to be aligned to the base type of the
15252 For example, by default a structure containing nothing but 8
15253 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15254 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15255 the structure would be aligned to a 1 byte boundary and be one byte in
15258 @item -mno-strict-align
15259 @itemx -mstrict-align
15260 @opindex mno-strict-align
15261 @opindex mstrict-align
15262 On System V.4 and embedded PowerPC systems do not (do) assume that
15263 unaligned memory references will be handled by the system.
15265 @item -mrelocatable
15266 @itemx -mno-relocatable
15267 @opindex mrelocatable
15268 @opindex mno-relocatable
15269 On embedded PowerPC systems generate code that allows (does not allow)
15270 the program to be relocated to a different address at runtime. If you
15271 use @option{-mrelocatable} on any module, all objects linked together must
15272 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
15274 @item -mrelocatable-lib
15275 @itemx -mno-relocatable-lib
15276 @opindex mrelocatable-lib
15277 @opindex mno-relocatable-lib
15278 On embedded PowerPC systems generate code that allows (does not allow)
15279 the program to be relocated to a different address at runtime. Modules
15280 compiled with @option{-mrelocatable-lib} can be linked with either modules
15281 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
15282 with modules compiled with the @option{-mrelocatable} options.
15288 On System V.4 and embedded PowerPC systems do not (do) assume that
15289 register 2 contains a pointer to a global area pointing to the addresses
15290 used in the program.
15293 @itemx -mlittle-endian
15295 @opindex mlittle-endian
15296 On System V.4 and embedded PowerPC systems compile code for the
15297 processor in little endian mode. The @option{-mlittle-endian} option is
15298 the same as @option{-mlittle}.
15301 @itemx -mbig-endian
15303 @opindex mbig-endian
15304 On System V.4 and embedded PowerPC systems compile code for the
15305 processor in big endian mode. The @option{-mbig-endian} option is
15306 the same as @option{-mbig}.
15308 @item -mdynamic-no-pic
15309 @opindex mdynamic-no-pic
15310 On Darwin and Mac OS X systems, compile code so that it is not
15311 relocatable, but that its external references are relocatable. The
15312 resulting code is suitable for applications, but not shared
15315 @item -mprioritize-restricted-insns=@var{priority}
15316 @opindex mprioritize-restricted-insns
15317 This option controls the priority that is assigned to
15318 dispatch-slot restricted instructions during the second scheduling
15319 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15320 @var{no/highest/second-highest} priority to dispatch slot restricted
15323 @item -msched-costly-dep=@var{dependence_type}
15324 @opindex msched-costly-dep
15325 This option controls which dependences are considered costly
15326 by the target during instruction scheduling. The argument
15327 @var{dependence_type} takes one of the following values:
15328 @var{no}: no dependence is costly,
15329 @var{all}: all dependences are costly,
15330 @var{true_store_to_load}: a true dependence from store to load is costly,
15331 @var{store_to_load}: any dependence from store to load is costly,
15332 @var{number}: any dependence which latency >= @var{number} is costly.
15334 @item -minsert-sched-nops=@var{scheme}
15335 @opindex minsert-sched-nops
15336 This option controls which nop insertion scheme will be used during
15337 the second scheduling pass. The argument @var{scheme} takes one of the
15339 @var{no}: Don't insert nops.
15340 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15341 according to the scheduler's grouping.
15342 @var{regroup_exact}: Insert nops to force costly dependent insns into
15343 separate groups. Insert exactly as many nops as needed to force an insn
15344 to a new group, according to the estimated processor grouping.
15345 @var{number}: Insert nops to force costly dependent insns into
15346 separate groups. Insert @var{number} nops to force an insn to a new group.
15349 @opindex mcall-sysv
15350 On System V.4 and embedded PowerPC systems compile code using calling
15351 conventions that adheres to the March 1995 draft of the System V
15352 Application Binary Interface, PowerPC processor supplement. This is the
15353 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15355 @item -mcall-sysv-eabi
15357 @opindex mcall-sysv-eabi
15358 @opindex mcall-eabi
15359 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15361 @item -mcall-sysv-noeabi
15362 @opindex mcall-sysv-noeabi
15363 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15365 @item -mcall-aixdesc
15367 On System V.4 and embedded PowerPC systems compile code for the AIX
15371 @opindex mcall-linux
15372 On System V.4 and embedded PowerPC systems compile code for the
15373 Linux-based GNU system.
15377 On System V.4 and embedded PowerPC systems compile code for the
15378 Hurd-based GNU system.
15380 @item -mcall-freebsd
15381 @opindex mcall-freebsd
15382 On System V.4 and embedded PowerPC systems compile code for the
15383 FreeBSD operating system.
15385 @item -mcall-netbsd
15386 @opindex mcall-netbsd
15387 On System V.4 and embedded PowerPC systems compile code for the
15388 NetBSD operating system.
15390 @item -mcall-openbsd
15391 @opindex mcall-netbsd
15392 On System V.4 and embedded PowerPC systems compile code for the
15393 OpenBSD operating system.
15395 @item -maix-struct-return
15396 @opindex maix-struct-return
15397 Return all structures in memory (as specified by the AIX ABI)@.
15399 @item -msvr4-struct-return
15400 @opindex msvr4-struct-return
15401 Return structures smaller than 8 bytes in registers (as specified by the
15404 @item -mabi=@var{abi-type}
15406 Extend the current ABI with a particular extension, or remove such extension.
15407 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15408 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15412 Extend the current ABI with SPE ABI extensions. This does not change
15413 the default ABI, instead it adds the SPE ABI extensions to the current
15417 @opindex mabi=no-spe
15418 Disable Booke SPE ABI extensions for the current ABI@.
15420 @item -mabi=ibmlongdouble
15421 @opindex mabi=ibmlongdouble
15422 Change the current ABI to use IBM extended precision long double.
15423 This is a PowerPC 32-bit SYSV ABI option.
15425 @item -mabi=ieeelongdouble
15426 @opindex mabi=ieeelongdouble
15427 Change the current ABI to use IEEE extended precision long double.
15428 This is a PowerPC 32-bit Linux ABI option.
15431 @itemx -mno-prototype
15432 @opindex mprototype
15433 @opindex mno-prototype
15434 On System V.4 and embedded PowerPC systems assume that all calls to
15435 variable argument functions are properly prototyped. Otherwise, the
15436 compiler must insert an instruction before every non prototyped call to
15437 set or clear bit 6 of the condition code register (@var{CR}) to
15438 indicate whether floating point values were passed in the floating point
15439 registers in case the function takes a variable arguments. With
15440 @option{-mprototype}, only calls to prototyped variable argument functions
15441 will set or clear the bit.
15445 On embedded PowerPC systems, assume that the startup module is called
15446 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15447 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15452 On embedded PowerPC systems, assume that the startup module is called
15453 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15458 On embedded PowerPC systems, assume that the startup module is called
15459 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15462 @item -myellowknife
15463 @opindex myellowknife
15464 On embedded PowerPC systems, assume that the startup module is called
15465 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15470 On System V.4 and embedded PowerPC systems, specify that you are
15471 compiling for a VxWorks system.
15475 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15476 header to indicate that @samp{eabi} extended relocations are used.
15482 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15483 Embedded Applications Binary Interface (eabi) which is a set of
15484 modifications to the System V.4 specifications. Selecting @option{-meabi}
15485 means that the stack is aligned to an 8 byte boundary, a function
15486 @code{__eabi} is called to from @code{main} to set up the eabi
15487 environment, and the @option{-msdata} option can use both @code{r2} and
15488 @code{r13} to point to two separate small data areas. Selecting
15489 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15490 do not call an initialization function from @code{main}, and the
15491 @option{-msdata} option will only use @code{r13} to point to a single
15492 small data area. The @option{-meabi} option is on by default if you
15493 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15496 @opindex msdata=eabi
15497 On System V.4 and embedded PowerPC systems, put small initialized
15498 @code{const} global and static data in the @samp{.sdata2} section, which
15499 is pointed to by register @code{r2}. Put small initialized
15500 non-@code{const} global and static data in the @samp{.sdata} section,
15501 which is pointed to by register @code{r13}. Put small uninitialized
15502 global and static data in the @samp{.sbss} section, which is adjacent to
15503 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15504 incompatible with the @option{-mrelocatable} option. The
15505 @option{-msdata=eabi} option also sets the @option{-memb} option.
15508 @opindex msdata=sysv
15509 On System V.4 and embedded PowerPC systems, put small global and static
15510 data in the @samp{.sdata} section, which is pointed to by register
15511 @code{r13}. Put small uninitialized global and static data in the
15512 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15513 The @option{-msdata=sysv} option is incompatible with the
15514 @option{-mrelocatable} option.
15516 @item -msdata=default
15518 @opindex msdata=default
15520 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15521 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15522 same as @option{-msdata=sysv}.
15525 @opindex msdata=data
15526 On System V.4 and embedded PowerPC systems, put small global
15527 data in the @samp{.sdata} section. Put small uninitialized global
15528 data in the @samp{.sbss} section. Do not use register @code{r13}
15529 to address small data however. This is the default behavior unless
15530 other @option{-msdata} options are used.
15534 @opindex msdata=none
15536 On embedded PowerPC systems, put all initialized global and static data
15537 in the @samp{.data} section, and all uninitialized data in the
15538 @samp{.bss} section.
15542 @cindex smaller data references (PowerPC)
15543 @cindex .sdata/.sdata2 references (PowerPC)
15544 On embedded PowerPC systems, put global and static items less than or
15545 equal to @var{num} bytes into the small data or bss sections instead of
15546 the normal data or bss section. By default, @var{num} is 8. The
15547 @option{-G @var{num}} switch is also passed to the linker.
15548 All modules should be compiled with the same @option{-G @var{num}} value.
15551 @itemx -mno-regnames
15553 @opindex mno-regnames
15554 On System V.4 and embedded PowerPC systems do (do not) emit register
15555 names in the assembly language output using symbolic forms.
15558 @itemx -mno-longcall
15560 @opindex mno-longcall
15561 By default assume that all calls are far away so that a longer more
15562 expensive calling sequence is required. This is required for calls
15563 further than 32 megabytes (33,554,432 bytes) from the current location.
15564 A short call will be generated if the compiler knows
15565 the call cannot be that far away. This setting can be overridden by
15566 the @code{shortcall} function attribute, or by @code{#pragma
15569 Some linkers are capable of detecting out-of-range calls and generating
15570 glue code on the fly. On these systems, long calls are unnecessary and
15571 generate slower code. As of this writing, the AIX linker can do this,
15572 as can the GNU linker for PowerPC/64. It is planned to add this feature
15573 to the GNU linker for 32-bit PowerPC systems as well.
15575 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15576 callee, L42'', plus a ``branch island'' (glue code). The two target
15577 addresses represent the callee and the ``branch island''. The
15578 Darwin/PPC linker will prefer the first address and generate a ``bl
15579 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15580 otherwise, the linker will generate ``bl L42'' to call the ``branch
15581 island''. The ``branch island'' is appended to the body of the
15582 calling function; it computes the full 32-bit address of the callee
15585 On Mach-O (Darwin) systems, this option directs the compiler emit to
15586 the glue for every direct call, and the Darwin linker decides whether
15587 to use or discard it.
15589 In the future, we may cause GCC to ignore all longcall specifications
15590 when the linker is known to generate glue.
15592 @item -mtls-markers
15593 @itemx -mno-tls-markers
15594 @opindex mtls-markers
15595 @opindex mno-tls-markers
15596 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15597 specifying the function argument. The relocation allows ld to
15598 reliably associate function call with argument setup instructions for
15599 TLS optimization, which in turn allows gcc to better schedule the
15604 Adds support for multithreading with the @dfn{pthreads} library.
15605 This option sets flags for both the preprocessor and linker.
15610 @subsection RX Options
15613 These command line options are defined for RX targets:
15616 @item -m64bit-doubles
15617 @itemx -m32bit-doubles
15618 @opindex m64bit-doubles
15619 @opindex m32bit-doubles
15620 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
15621 or 32-bits (@option{-m32bit-doubles}) in size. The default is
15622 @option{-m32bit-doubles}. @emph{Note} RX floating point hardware only
15623 works on 32-bit values, which is why the default is
15624 @option{-m32bit-doubles}.
15630 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
15631 floating point hardware. The default is enabled for the @var{RX600}
15632 series and disabled for the @var{RX200} series.
15634 Floating point instructions will only be generated for 32-bit floating
15635 point values however, so if the @option{-m64bit-doubles} option is in
15636 use then the FPU hardware will not be used for doubles.
15638 @emph{Note} If the @option{-fpu} option is enabled then
15639 @option{-funsafe-math-optimizations} is also enabled automatically.
15640 This is because the RX FPU instructions are themselves unsafe.
15642 @item -mcpu=@var{name}
15643 @itemx -patch=@var{name}
15646 Selects the type of RX CPU to be targeted. Currently three types are
15647 supported, the generic @var{RX600} and @var{RX200} series hardware and
15648 the specific @var{RX610} cpu. The default is @var{RX600}.
15650 The only difference between @var{RX600} and @var{RX610} is that the
15651 @var{RX610} does not support the @code{MVTIPL} instruction.
15653 The @var{RX200} series does not have a hardware floating point unit
15654 and so @option{-nofpu} is enabled by default when this type is
15657 @item -mbig-endian-data
15658 @itemx -mlittle-endian-data
15659 @opindex mbig-endian-data
15660 @opindex mlittle-endian-data
15661 Store data (but not code) in the big-endian format. The default is
15662 @option{-mlittle-endian-data}, ie to store data in the little endian
15665 @item -msmall-data-limit=@var{N}
15666 @opindex msmall-data-limit
15667 Specifies the maximum size in bytes of global and static variables
15668 which can be placed into the small data area. Using the small data
15669 area can lead to smaller and faster code, but the size of area is
15670 limited and it is up to the programmer to ensure that the area does
15671 not overflow. Also when the small data area is used one of the RX's
15672 registers (@code{r13}) is reserved for use pointing to this area, so
15673 it is no longer available for use by the compiler. This could result
15674 in slower and/or larger code if variables which once could have been
15675 held in @code{r13} are now pushed onto the stack.
15677 Note, common variables (variables which have not been initialised) and
15678 constants are not placed into the small data area as they are assigned
15679 to other sections in the output executable.
15681 The default value is zero, which disables this feature. Note, this
15682 feature is not enabled by default with higher optimization levels
15683 (@option{-O2} etc) because of the potentially detrimental effects of
15684 reserving register @code{r13}. It is up to the programmer to
15685 experiment and discover whether this feature is of benefit to their
15692 Use the simulator runtime. The default is to use the libgloss board
15695 @item -mas100-syntax
15696 @itemx -mno-as100-syntax
15697 @opindex mas100-syntax
15698 @opindex mno-as100-syntax
15699 When generating assembler output use a syntax that is compatible with
15700 Renesas's AS100 assembler. This syntax can also be handled by the GAS
15701 assembler but it has some restrictions so generating it is not the
15704 @item -mmax-constant-size=@var{N}
15705 @opindex mmax-constant-size
15706 Specifies the maximum size, in bytes, of a constant that can be used as
15707 an operand in a RX instruction. Although the RX instruction set does
15708 allow constants of up to 4 bytes in length to be used in instructions,
15709 a longer value equates to a longer instruction. Thus in some
15710 circumstances it can be beneficial to restrict the size of constants
15711 that are used in instructions. Constants that are too big are instead
15712 placed into a constant pool and referenced via register indirection.
15714 The value @var{N} can be between 0 and 4. A value of 0 (the default)
15715 or 4 means that constants of any size are allowed.
15719 Enable linker relaxation. Linker relaxation is a process whereby the
15720 linker will attempt to reduce the size of a program by finding shorter
15721 versions of various instructions. Disabled by default.
15723 @item -mint-register=@var{N}
15724 @opindex mint-register
15725 Specify the number of registers to reserve for fast interrupt handler
15726 functions. The value @var{N} can be between 0 and 4. A value of 1
15727 means that register @code{r13} will be reserved for the exclusive use
15728 of fast interrupt handlers. A value of 2 reserves @code{r13} and
15729 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
15730 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
15731 A value of 0, the default, does not reserve any registers.
15733 @item -msave-acc-in-interrupts
15734 @opindex msave-acc-in-interrupts
15735 Specifies that interrupt handler functions should preserve the
15736 accumulator register. This is only necessary if normal code might use
15737 the accumulator register, for example because it performs 64-bit
15738 multiplications. The default is to ignore the accumulator as this
15739 makes the interrupt handlers faster.
15743 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
15744 has special significance to the RX port when used with the
15745 @code{interrupt} function attribute. This attribute indicates a
15746 function intended to process fast interrupts. GCC will will ensure
15747 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
15748 and/or @code{r13} and only provided that the normal use of the
15749 corresponding registers have been restricted via the
15750 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
15753 @node S/390 and zSeries Options
15754 @subsection S/390 and zSeries Options
15755 @cindex S/390 and zSeries Options
15757 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15761 @itemx -msoft-float
15762 @opindex mhard-float
15763 @opindex msoft-float
15764 Use (do not use) the hardware floating-point instructions and registers
15765 for floating-point operations. When @option{-msoft-float} is specified,
15766 functions in @file{libgcc.a} will be used to perform floating-point
15767 operations. When @option{-mhard-float} is specified, the compiler
15768 generates IEEE floating-point instructions. This is the default.
15771 @itemx -mno-hard-dfp
15773 @opindex mno-hard-dfp
15774 Use (do not use) the hardware decimal-floating-point instructions for
15775 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15776 specified, functions in @file{libgcc.a} will be used to perform
15777 decimal-floating-point operations. When @option{-mhard-dfp} is
15778 specified, the compiler generates decimal-floating-point hardware
15779 instructions. This is the default for @option{-march=z9-ec} or higher.
15781 @item -mlong-double-64
15782 @itemx -mlong-double-128
15783 @opindex mlong-double-64
15784 @opindex mlong-double-128
15785 These switches control the size of @code{long double} type. A size
15786 of 64bit makes the @code{long double} type equivalent to the @code{double}
15787 type. This is the default.
15790 @itemx -mno-backchain
15791 @opindex mbackchain
15792 @opindex mno-backchain
15793 Store (do not store) the address of the caller's frame as backchain pointer
15794 into the callee's stack frame.
15795 A backchain may be needed to allow debugging using tools that do not understand
15796 DWARF-2 call frame information.
15797 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15798 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15799 the backchain is placed into the topmost word of the 96/160 byte register
15802 In general, code compiled with @option{-mbackchain} is call-compatible with
15803 code compiled with @option{-mmo-backchain}; however, use of the backchain
15804 for debugging purposes usually requires that the whole binary is built with
15805 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15806 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15807 to build a linux kernel use @option{-msoft-float}.
15809 The default is to not maintain the backchain.
15811 @item -mpacked-stack
15812 @itemx -mno-packed-stack
15813 @opindex mpacked-stack
15814 @opindex mno-packed-stack
15815 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15816 specified, the compiler uses the all fields of the 96/160 byte register save
15817 area only for their default purpose; unused fields still take up stack space.
15818 When @option{-mpacked-stack} is specified, register save slots are densely
15819 packed at the top of the register save area; unused space is reused for other
15820 purposes, allowing for more efficient use of the available stack space.
15821 However, when @option{-mbackchain} is also in effect, the topmost word of
15822 the save area is always used to store the backchain, and the return address
15823 register is always saved two words below the backchain.
15825 As long as the stack frame backchain is not used, code generated with
15826 @option{-mpacked-stack} is call-compatible with code generated with
15827 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15828 S/390 or zSeries generated code that uses the stack frame backchain at run
15829 time, not just for debugging purposes. Such code is not call-compatible
15830 with code compiled with @option{-mpacked-stack}. Also, note that the
15831 combination of @option{-mbackchain},
15832 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15833 to build a linux kernel use @option{-msoft-float}.
15835 The default is to not use the packed stack layout.
15838 @itemx -mno-small-exec
15839 @opindex msmall-exec
15840 @opindex mno-small-exec
15841 Generate (or do not generate) code using the @code{bras} instruction
15842 to do subroutine calls.
15843 This only works reliably if the total executable size does not
15844 exceed 64k. The default is to use the @code{basr} instruction instead,
15845 which does not have this limitation.
15851 When @option{-m31} is specified, generate code compliant to the
15852 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15853 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15854 particular to generate 64-bit instructions. For the @samp{s390}
15855 targets, the default is @option{-m31}, while the @samp{s390x}
15856 targets default to @option{-m64}.
15862 When @option{-mzarch} is specified, generate code using the
15863 instructions available on z/Architecture.
15864 When @option{-mesa} is specified, generate code using the
15865 instructions available on ESA/390. Note that @option{-mesa} is
15866 not possible with @option{-m64}.
15867 When generating code compliant to the GNU/Linux for S/390 ABI,
15868 the default is @option{-mesa}. When generating code compliant
15869 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
15875 Generate (or do not generate) code using the @code{mvcle} instruction
15876 to perform block moves. When @option{-mno-mvcle} is specified,
15877 use a @code{mvc} loop instead. This is the default unless optimizing for
15884 Print (or do not print) additional debug information when compiling.
15885 The default is to not print debug information.
15887 @item -march=@var{cpu-type}
15889 Generate code that will run on @var{cpu-type}, which is the name of a system
15890 representing a certain processor type. Possible values for
15891 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
15892 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
15893 When generating code using the instructions available on z/Architecture,
15894 the default is @option{-march=z900}. Otherwise, the default is
15895 @option{-march=g5}.
15897 @item -mtune=@var{cpu-type}
15899 Tune to @var{cpu-type} everything applicable about the generated code,
15900 except for the ABI and the set of available instructions.
15901 The list of @var{cpu-type} values is the same as for @option{-march}.
15902 The default is the value used for @option{-march}.
15905 @itemx -mno-tpf-trace
15906 @opindex mtpf-trace
15907 @opindex mno-tpf-trace
15908 Generate code that adds (does not add) in TPF OS specific branches to trace
15909 routines in the operating system. This option is off by default, even
15910 when compiling for the TPF OS@.
15913 @itemx -mno-fused-madd
15914 @opindex mfused-madd
15915 @opindex mno-fused-madd
15916 Generate code that uses (does not use) the floating point multiply and
15917 accumulate instructions. These instructions are generated by default if
15918 hardware floating point is used.
15920 @item -mwarn-framesize=@var{framesize}
15921 @opindex mwarn-framesize
15922 Emit a warning if the current function exceeds the given frame size. Because
15923 this is a compile time check it doesn't need to be a real problem when the program
15924 runs. It is intended to identify functions which most probably cause
15925 a stack overflow. It is useful to be used in an environment with limited stack
15926 size e.g.@: the linux kernel.
15928 @item -mwarn-dynamicstack
15929 @opindex mwarn-dynamicstack
15930 Emit a warning if the function calls alloca or uses dynamically
15931 sized arrays. This is generally a bad idea with a limited stack size.
15933 @item -mstack-guard=@var{stack-guard}
15934 @itemx -mstack-size=@var{stack-size}
15935 @opindex mstack-guard
15936 @opindex mstack-size
15937 If these options are provided the s390 back end emits additional instructions in
15938 the function prologue which trigger a trap if the stack size is @var{stack-guard}
15939 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
15940 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
15941 the frame size of the compiled function is chosen.
15942 These options are intended to be used to help debugging stack overflow problems.
15943 The additionally emitted code causes only little overhead and hence can also be
15944 used in production like systems without greater performance degradation. The given
15945 values have to be exact powers of 2 and @var{stack-size} has to be greater than
15946 @var{stack-guard} without exceeding 64k.
15947 In order to be efficient the extra code makes the assumption that the stack starts
15948 at an address aligned to the value given by @var{stack-size}.
15949 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
15952 @node Score Options
15953 @subsection Score Options
15954 @cindex Score Options
15956 These options are defined for Score implementations:
15961 Compile code for big endian mode. This is the default.
15965 Compile code for little endian mode.
15969 Disable generate bcnz instruction.
15973 Enable generate unaligned load and store instruction.
15977 Enable the use of multiply-accumulate instructions. Disabled by default.
15981 Specify the SCORE5 as the target architecture.
15985 Specify the SCORE5U of the target architecture.
15989 Specify the SCORE7 as the target architecture. This is the default.
15993 Specify the SCORE7D as the target architecture.
15997 @subsection SH Options
15999 These @samp{-m} options are defined for the SH implementations:
16004 Generate code for the SH1.
16008 Generate code for the SH2.
16011 Generate code for the SH2e.
16015 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
16016 that the floating-point unit is not used.
16018 @item -m2a-single-only
16019 @opindex m2a-single-only
16020 Generate code for the SH2a-FPU, in such a way that no double-precision
16021 floating point operations are used.
16024 @opindex m2a-single
16025 Generate code for the SH2a-FPU assuming the floating-point unit is in
16026 single-precision mode by default.
16030 Generate code for the SH2a-FPU assuming the floating-point unit is in
16031 double-precision mode by default.
16035 Generate code for the SH3.
16039 Generate code for the SH3e.
16043 Generate code for the SH4 without a floating-point unit.
16045 @item -m4-single-only
16046 @opindex m4-single-only
16047 Generate code for the SH4 with a floating-point unit that only
16048 supports single-precision arithmetic.
16052 Generate code for the SH4 assuming the floating-point unit is in
16053 single-precision mode by default.
16057 Generate code for the SH4.
16061 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
16062 floating-point unit is not used.
16064 @item -m4a-single-only
16065 @opindex m4a-single-only
16066 Generate code for the SH4a, in such a way that no double-precision
16067 floating point operations are used.
16070 @opindex m4a-single
16071 Generate code for the SH4a assuming the floating-point unit is in
16072 single-precision mode by default.
16076 Generate code for the SH4a.
16080 Same as @option{-m4a-nofpu}, except that it implicitly passes
16081 @option{-dsp} to the assembler. GCC doesn't generate any DSP
16082 instructions at the moment.
16086 Compile code for the processor in big endian mode.
16090 Compile code for the processor in little endian mode.
16094 Align doubles at 64-bit boundaries. Note that this changes the calling
16095 conventions, and thus some functions from the standard C library will
16096 not work unless you recompile it first with @option{-mdalign}.
16100 Shorten some address references at link time, when possible; uses the
16101 linker option @option{-relax}.
16105 Use 32-bit offsets in @code{switch} tables. The default is to use
16110 Enable the use of bit manipulation instructions on SH2A.
16114 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
16115 alignment constraints.
16119 Comply with the calling conventions defined by Renesas.
16123 Comply with the calling conventions defined by Renesas.
16127 Comply with the calling conventions defined for GCC before the Renesas
16128 conventions were available. This option is the default for all
16129 targets of the SH toolchain except for @samp{sh-symbianelf}.
16132 @opindex mnomacsave
16133 Mark the @code{MAC} register as call-clobbered, even if
16134 @option{-mhitachi} is given.
16138 Increase IEEE-compliance of floating-point code.
16139 At the moment, this is equivalent to @option{-fno-finite-math-only}.
16140 When generating 16 bit SH opcodes, getting IEEE-conforming results for
16141 comparisons of NANs / infinities incurs extra overhead in every
16142 floating point comparison, therefore the default is set to
16143 @option{-ffinite-math-only}.
16145 @item -minline-ic_invalidate
16146 @opindex minline-ic_invalidate
16147 Inline code to invalidate instruction cache entries after setting up
16148 nested function trampolines.
16149 This option has no effect if -musermode is in effect and the selected
16150 code generation option (e.g. -m4) does not allow the use of the icbi
16152 If the selected code generation option does not allow the use of the icbi
16153 instruction, and -musermode is not in effect, the inlined code will
16154 manipulate the instruction cache address array directly with an associative
16155 write. This not only requires privileged mode, but it will also
16156 fail if the cache line had been mapped via the TLB and has become unmapped.
16160 Dump instruction size and location in the assembly code.
16163 @opindex mpadstruct
16164 This option is deprecated. It pads structures to multiple of 4 bytes,
16165 which is incompatible with the SH ABI@.
16169 Optimize for space instead of speed. Implied by @option{-Os}.
16172 @opindex mprefergot
16173 When generating position-independent code, emit function calls using
16174 the Global Offset Table instead of the Procedure Linkage Table.
16178 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
16179 if the inlined code would not work in user mode.
16180 This is the default when the target is @code{sh-*-linux*}.
16182 @item -multcost=@var{number}
16183 @opindex multcost=@var{number}
16184 Set the cost to assume for a multiply insn.
16186 @item -mdiv=@var{strategy}
16187 @opindex mdiv=@var{strategy}
16188 Set the division strategy to use for SHmedia code. @var{strategy} must be
16189 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
16190 inv:call2, inv:fp .
16191 "fp" performs the operation in floating point. This has a very high latency,
16192 but needs only a few instructions, so it might be a good choice if
16193 your code has enough easily exploitable ILP to allow the compiler to
16194 schedule the floating point instructions together with other instructions.
16195 Division by zero causes a floating point exception.
16196 "inv" uses integer operations to calculate the inverse of the divisor,
16197 and then multiplies the dividend with the inverse. This strategy allows
16198 cse and hoisting of the inverse calculation. Division by zero calculates
16199 an unspecified result, but does not trap.
16200 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
16201 have been found, or if the entire operation has been hoisted to the same
16202 place, the last stages of the inverse calculation are intertwined with the
16203 final multiply to reduce the overall latency, at the expense of using a few
16204 more instructions, and thus offering fewer scheduling opportunities with
16206 "call" calls a library function that usually implements the inv:minlat
16208 This gives high code density for m5-*media-nofpu compilations.
16209 "call2" uses a different entry point of the same library function, where it
16210 assumes that a pointer to a lookup table has already been set up, which
16211 exposes the pointer load to cse / code hoisting optimizations.
16212 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
16213 code generation, but if the code stays unoptimized, revert to the "call",
16214 "call2", or "fp" strategies, respectively. Note that the
16215 potentially-trapping side effect of division by zero is carried by a
16216 separate instruction, so it is possible that all the integer instructions
16217 are hoisted out, but the marker for the side effect stays where it is.
16218 A recombination to fp operations or a call is not possible in that case.
16219 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
16220 that the inverse calculation was nor separated from the multiply, they speed
16221 up division where the dividend fits into 20 bits (plus sign where applicable),
16222 by inserting a test to skip a number of operations in this case; this test
16223 slows down the case of larger dividends. inv20u assumes the case of a such
16224 a small dividend to be unlikely, and inv20l assumes it to be likely.
16226 @item -maccumulate-outgoing-args
16227 @opindex maccumulate-outgoing-args
16228 Reserve space once for outgoing arguments in the function prologue rather
16229 than around each call. Generally beneficial for performance and size. Also
16230 needed for unwinding to avoid changing the stack frame around conditional code.
16232 @item -mdivsi3_libfunc=@var{name}
16233 @opindex mdivsi3_libfunc=@var{name}
16234 Set the name of the library function used for 32 bit signed division to
16235 @var{name}. This only affect the name used in the call and inv:call
16236 division strategies, and the compiler will still expect the same
16237 sets of input/output/clobbered registers as if this option was not present.
16239 @item -mfixed-range=@var{register-range}
16240 @opindex mfixed-range
16241 Generate code treating the given register range as fixed registers.
16242 A fixed register is one that the register allocator can not use. This is
16243 useful when compiling kernel code. A register range is specified as
16244 two registers separated by a dash. Multiple register ranges can be
16245 specified separated by a comma.
16247 @item -madjust-unroll
16248 @opindex madjust-unroll
16249 Throttle unrolling to avoid thrashing target registers.
16250 This option only has an effect if the gcc code base supports the
16251 TARGET_ADJUST_UNROLL_MAX target hook.
16253 @item -mindexed-addressing
16254 @opindex mindexed-addressing
16255 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16256 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16257 semantics for the indexed addressing mode. The architecture allows the
16258 implementation of processors with 64 bit MMU, which the OS could use to
16259 get 32 bit addressing, but since no current hardware implementation supports
16260 this or any other way to make the indexed addressing mode safe to use in
16261 the 32 bit ABI, the default is -mno-indexed-addressing.
16263 @item -mgettrcost=@var{number}
16264 @opindex mgettrcost=@var{number}
16265 Set the cost assumed for the gettr instruction to @var{number}.
16266 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16270 Assume pt* instructions won't trap. This will generally generate better
16271 scheduled code, but is unsafe on current hardware. The current architecture
16272 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16273 This has the unintentional effect of making it unsafe to schedule ptabs /
16274 ptrel before a branch, or hoist it out of a loop. For example,
16275 __do_global_ctors, a part of libgcc that runs constructors at program
16276 startup, calls functions in a list which is delimited by @minus{}1. With the
16277 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16278 That means that all the constructors will be run a bit quicker, but when
16279 the loop comes to the end of the list, the program crashes because ptabs
16280 loads @minus{}1 into a target register. Since this option is unsafe for any
16281 hardware implementing the current architecture specification, the default
16282 is -mno-pt-fixed. Unless the user specifies a specific cost with
16283 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16284 this deters register allocation using target registers for storing
16287 @item -minvalid-symbols
16288 @opindex minvalid-symbols
16289 Assume symbols might be invalid. Ordinary function symbols generated by
16290 the compiler will always be valid to load with movi/shori/ptabs or
16291 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16292 to generate symbols that will cause ptabs / ptrel to trap.
16293 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16294 It will then prevent cross-basic-block cse, hoisting and most scheduling
16295 of symbol loads. The default is @option{-mno-invalid-symbols}.
16298 @node SPARC Options
16299 @subsection SPARC Options
16300 @cindex SPARC options
16302 These @samp{-m} options are supported on the SPARC:
16305 @item -mno-app-regs
16307 @opindex mno-app-regs
16309 Specify @option{-mapp-regs} to generate output using the global registers
16310 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16313 To be fully SVR4 ABI compliant at the cost of some performance loss,
16314 specify @option{-mno-app-regs}. You should compile libraries and system
16315 software with this option.
16318 @itemx -mhard-float
16320 @opindex mhard-float
16321 Generate output containing floating point instructions. This is the
16325 @itemx -msoft-float
16327 @opindex msoft-float
16328 Generate output containing library calls for floating point.
16329 @strong{Warning:} the requisite libraries are not available for all SPARC
16330 targets. Normally the facilities of the machine's usual C compiler are
16331 used, but this cannot be done directly in cross-compilation. You must make
16332 your own arrangements to provide suitable library functions for
16333 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16334 @samp{sparclite-*-*} do provide software floating point support.
16336 @option{-msoft-float} changes the calling convention in the output file;
16337 therefore, it is only useful if you compile @emph{all} of a program with
16338 this option. In particular, you need to compile @file{libgcc.a}, the
16339 library that comes with GCC, with @option{-msoft-float} in order for
16342 @item -mhard-quad-float
16343 @opindex mhard-quad-float
16344 Generate output containing quad-word (long double) floating point
16347 @item -msoft-quad-float
16348 @opindex msoft-quad-float
16349 Generate output containing library calls for quad-word (long double)
16350 floating point instructions. The functions called are those specified
16351 in the SPARC ABI@. This is the default.
16353 As of this writing, there are no SPARC implementations that have hardware
16354 support for the quad-word floating point instructions. They all invoke
16355 a trap handler for one of these instructions, and then the trap handler
16356 emulates the effect of the instruction. Because of the trap handler overhead,
16357 this is much slower than calling the ABI library routines. Thus the
16358 @option{-msoft-quad-float} option is the default.
16360 @item -mno-unaligned-doubles
16361 @itemx -munaligned-doubles
16362 @opindex mno-unaligned-doubles
16363 @opindex munaligned-doubles
16364 Assume that doubles have 8 byte alignment. This is the default.
16366 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16367 alignment only if they are contained in another type, or if they have an
16368 absolute address. Otherwise, it assumes they have 4 byte alignment.
16369 Specifying this option avoids some rare compatibility problems with code
16370 generated by other compilers. It is not the default because it results
16371 in a performance loss, especially for floating point code.
16373 @item -mno-faster-structs
16374 @itemx -mfaster-structs
16375 @opindex mno-faster-structs
16376 @opindex mfaster-structs
16377 With @option{-mfaster-structs}, the compiler assumes that structures
16378 should have 8 byte alignment. This enables the use of pairs of
16379 @code{ldd} and @code{std} instructions for copies in structure
16380 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16381 However, the use of this changed alignment directly violates the SPARC
16382 ABI@. Thus, it's intended only for use on targets where the developer
16383 acknowledges that their resulting code will not be directly in line with
16384 the rules of the ABI@.
16386 @item -mimpure-text
16387 @opindex mimpure-text
16388 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16389 the compiler to not pass @option{-z text} to the linker when linking a
16390 shared object. Using this option, you can link position-dependent
16391 code into a shared object.
16393 @option{-mimpure-text} suppresses the ``relocations remain against
16394 allocatable but non-writable sections'' linker error message.
16395 However, the necessary relocations will trigger copy-on-write, and the
16396 shared object is not actually shared across processes. Instead of
16397 using @option{-mimpure-text}, you should compile all source code with
16398 @option{-fpic} or @option{-fPIC}.
16400 This option is only available on SunOS and Solaris.
16402 @item -mcpu=@var{cpu_type}
16404 Set the instruction set, register set, and instruction scheduling parameters
16405 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16406 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16407 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16408 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16409 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16411 Default instruction scheduling parameters are used for values that select
16412 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16413 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16415 Here is a list of each supported architecture and their supported
16420 v8: supersparc, hypersparc
16421 sparclite: f930, f934, sparclite86x
16423 v9: ultrasparc, ultrasparc3, niagara, niagara2
16426 By default (unless configured otherwise), GCC generates code for the V7
16427 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16428 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16429 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16430 SPARCStation 1, 2, IPX etc.
16432 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16433 architecture. The only difference from V7 code is that the compiler emits
16434 the integer multiply and integer divide instructions which exist in SPARC-V8
16435 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16436 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16439 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16440 the SPARC architecture. This adds the integer multiply, integer divide step
16441 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16442 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16443 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16444 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16445 MB86934 chip, which is the more recent SPARClite with FPU@.
16447 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16448 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16449 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16450 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16451 optimizes it for the TEMIC SPARClet chip.
16453 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16454 architecture. This adds 64-bit integer and floating-point move instructions,
16455 3 additional floating-point condition code registers and conditional move
16456 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16457 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16458 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
16459 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
16460 @option{-mcpu=niagara}, the compiler additionally optimizes it for
16461 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
16462 additionally optimizes it for Sun UltraSPARC T2 chips.
16464 @item -mtune=@var{cpu_type}
16466 Set the instruction scheduling parameters for machine type
16467 @var{cpu_type}, but do not set the instruction set or register set that the
16468 option @option{-mcpu=@var{cpu_type}} would.
16470 The same values for @option{-mcpu=@var{cpu_type}} can be used for
16471 @option{-mtune=@var{cpu_type}}, but the only useful values are those
16472 that select a particular cpu implementation. Those are @samp{cypress},
16473 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
16474 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
16475 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
16480 @opindex mno-v8plus
16481 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
16482 difference from the V8 ABI is that the global and out registers are
16483 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
16484 mode for all SPARC-V9 processors.
16490 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
16491 Visual Instruction Set extensions. The default is @option{-mno-vis}.
16494 These @samp{-m} options are supported in addition to the above
16495 on SPARC-V9 processors in 64-bit environments:
16498 @item -mlittle-endian
16499 @opindex mlittle-endian
16500 Generate code for a processor running in little-endian mode. It is only
16501 available for a few configurations and most notably not on Solaris and Linux.
16507 Generate code for a 32-bit or 64-bit environment.
16508 The 32-bit environment sets int, long and pointer to 32 bits.
16509 The 64-bit environment sets int to 32 bits and long and pointer
16512 @item -mcmodel=medlow
16513 @opindex mcmodel=medlow
16514 Generate code for the Medium/Low code model: 64-bit addresses, programs
16515 must be linked in the low 32 bits of memory. Programs can be statically
16516 or dynamically linked.
16518 @item -mcmodel=medmid
16519 @opindex mcmodel=medmid
16520 Generate code for the Medium/Middle code model: 64-bit addresses, programs
16521 must be linked in the low 44 bits of memory, the text and data segments must
16522 be less than 2GB in size and the data segment must be located within 2GB of
16525 @item -mcmodel=medany
16526 @opindex mcmodel=medany
16527 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
16528 may be linked anywhere in memory, the text and data segments must be less
16529 than 2GB in size and the data segment must be located within 2GB of the
16532 @item -mcmodel=embmedany
16533 @opindex mcmodel=embmedany
16534 Generate code for the Medium/Anywhere code model for embedded systems:
16535 64-bit addresses, the text and data segments must be less than 2GB in
16536 size, both starting anywhere in memory (determined at link time). The
16537 global register %g4 points to the base of the data segment. Programs
16538 are statically linked and PIC is not supported.
16541 @itemx -mno-stack-bias
16542 @opindex mstack-bias
16543 @opindex mno-stack-bias
16544 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
16545 frame pointer if present, are offset by @minus{}2047 which must be added back
16546 when making stack frame references. This is the default in 64-bit mode.
16547 Otherwise, assume no such offset is present.
16550 These switches are supported in addition to the above on Solaris:
16555 Add support for multithreading using the Solaris threads library. This
16556 option sets flags for both the preprocessor and linker. This option does
16557 not affect the thread safety of object code produced by the compiler or
16558 that of libraries supplied with it.
16562 Add support for multithreading using the POSIX threads library. This
16563 option sets flags for both the preprocessor and linker. This option does
16564 not affect the thread safety of object code produced by the compiler or
16565 that of libraries supplied with it.
16569 This is a synonym for @option{-pthreads}.
16573 @subsection SPU Options
16574 @cindex SPU options
16576 These @samp{-m} options are supported on the SPU:
16580 @itemx -merror-reloc
16581 @opindex mwarn-reloc
16582 @opindex merror-reloc
16584 The loader for SPU does not handle dynamic relocations. By default, GCC
16585 will give an error when it generates code that requires a dynamic
16586 relocation. @option{-mno-error-reloc} disables the error,
16587 @option{-mwarn-reloc} will generate a warning instead.
16590 @itemx -munsafe-dma
16592 @opindex munsafe-dma
16594 Instructions which initiate or test completion of DMA must not be
16595 reordered with respect to loads and stores of the memory which is being
16596 accessed. Users typically address this problem using the volatile
16597 keyword, but that can lead to inefficient code in places where the
16598 memory is known to not change. Rather than mark the memory as volatile
16599 we treat the DMA instructions as potentially effecting all memory. With
16600 @option{-munsafe-dma} users must use the volatile keyword to protect
16603 @item -mbranch-hints
16604 @opindex mbranch-hints
16606 By default, GCC will generate a branch hint instruction to avoid
16607 pipeline stalls for always taken or probably taken branches. A hint
16608 will not be generated closer than 8 instructions away from its branch.
16609 There is little reason to disable them, except for debugging purposes,
16610 or to make an object a little bit smaller.
16614 @opindex msmall-mem
16615 @opindex mlarge-mem
16617 By default, GCC generates code assuming that addresses are never larger
16618 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
16619 a full 32 bit address.
16624 By default, GCC links against startup code that assumes the SPU-style
16625 main function interface (which has an unconventional parameter list).
16626 With @option{-mstdmain}, GCC will link your program against startup
16627 code that assumes a C99-style interface to @code{main}, including a
16628 local copy of @code{argv} strings.
16630 @item -mfixed-range=@var{register-range}
16631 @opindex mfixed-range
16632 Generate code treating the given register range as fixed registers.
16633 A fixed register is one that the register allocator can not use. This is
16634 useful when compiling kernel code. A register range is specified as
16635 two registers separated by a dash. Multiple register ranges can be
16636 specified separated by a comma.
16642 Compile code assuming that pointers to the PPU address space accessed
16643 via the @code{__ea} named address space qualifier are either 32 or 64
16644 bits wide. The default is 32 bits. As this is an ABI changing option,
16645 all object code in an executable must be compiled with the same setting.
16647 @item -maddress-space-conversion
16648 @itemx -mno-address-space-conversion
16649 @opindex maddress-space-conversion
16650 @opindex mno-address-space-conversion
16651 Allow/disallow treating the @code{__ea} address space as superset
16652 of the generic address space. This enables explicit type casts
16653 between @code{__ea} and generic pointer as well as implicit
16654 conversions of generic pointers to @code{__ea} pointers. The
16655 default is to allow address space pointer conversions.
16657 @item -mcache-size=@var{cache-size}
16658 @opindex mcache-size
16659 This option controls the version of libgcc that the compiler links to an
16660 executable and selects a software-managed cache for accessing variables
16661 in the @code{__ea} address space with a particular cache size. Possible
16662 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
16663 and @samp{128}. The default cache size is 64KB.
16665 @item -matomic-updates
16666 @itemx -mno-atomic-updates
16667 @opindex matomic-updates
16668 @opindex mno-atomic-updates
16669 This option controls the version of libgcc that the compiler links to an
16670 executable and selects whether atomic updates to the software-managed
16671 cache of PPU-side variables are used. If you use atomic updates, changes
16672 to a PPU variable from SPU code using the @code{__ea} named address space
16673 qualifier will not interfere with changes to other PPU variables residing
16674 in the same cache line from PPU code. If you do not use atomic updates,
16675 such interference may occur; however, writing back cache lines will be
16676 more efficient. The default behavior is to use atomic updates.
16679 @itemx -mdual-nops=@var{n}
16680 @opindex mdual-nops
16681 By default, GCC will insert nops to increase dual issue when it expects
16682 it to increase performance. @var{n} can be a value from 0 to 10. A
16683 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
16684 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
16686 @item -mhint-max-nops=@var{n}
16687 @opindex mhint-max-nops
16688 Maximum number of nops to insert for a branch hint. A branch hint must
16689 be at least 8 instructions away from the branch it is effecting. GCC
16690 will insert up to @var{n} nops to enforce this, otherwise it will not
16691 generate the branch hint.
16693 @item -mhint-max-distance=@var{n}
16694 @opindex mhint-max-distance
16695 The encoding of the branch hint instruction limits the hint to be within
16696 256 instructions of the branch it is effecting. By default, GCC makes
16697 sure it is within 125.
16700 @opindex msafe-hints
16701 Work around a hardware bug which causes the SPU to stall indefinitely.
16702 By default, GCC will insert the @code{hbrp} instruction to make sure
16703 this stall won't happen.
16707 @node System V Options
16708 @subsection Options for System V
16710 These additional options are available on System V Release 4 for
16711 compatibility with other compilers on those systems:
16716 Create a shared object.
16717 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16721 Identify the versions of each tool used by the compiler, in a
16722 @code{.ident} assembler directive in the output.
16726 Refrain from adding @code{.ident} directives to the output file (this is
16729 @item -YP,@var{dirs}
16731 Search the directories @var{dirs}, and no others, for libraries
16732 specified with @option{-l}.
16734 @item -Ym,@var{dir}
16736 Look in the directory @var{dir} to find the M4 preprocessor.
16737 The assembler uses this option.
16738 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16739 @c the generic assembler that comes with Solaris takes just -Ym.
16743 @subsection V850 Options
16744 @cindex V850 Options
16746 These @samp{-m} options are defined for V850 implementations:
16750 @itemx -mno-long-calls
16751 @opindex mlong-calls
16752 @opindex mno-long-calls
16753 Treat all calls as being far away (near). If calls are assumed to be
16754 far away, the compiler will always load the functions address up into a
16755 register, and call indirect through the pointer.
16761 Do not optimize (do optimize) basic blocks that use the same index
16762 pointer 4 or more times to copy pointer into the @code{ep} register, and
16763 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16764 option is on by default if you optimize.
16766 @item -mno-prolog-function
16767 @itemx -mprolog-function
16768 @opindex mno-prolog-function
16769 @opindex mprolog-function
16770 Do not use (do use) external functions to save and restore registers
16771 at the prologue and epilogue of a function. The external functions
16772 are slower, but use less code space if more than one function saves
16773 the same number of registers. The @option{-mprolog-function} option
16774 is on by default if you optimize.
16778 Try to make the code as small as possible. At present, this just turns
16779 on the @option{-mep} and @option{-mprolog-function} options.
16781 @item -mtda=@var{n}
16783 Put static or global variables whose size is @var{n} bytes or less into
16784 the tiny data area that register @code{ep} points to. The tiny data
16785 area can hold up to 256 bytes in total (128 bytes for byte references).
16787 @item -msda=@var{n}
16789 Put static or global variables whose size is @var{n} bytes or less into
16790 the small data area that register @code{gp} points to. The small data
16791 area can hold up to 64 kilobytes.
16793 @item -mzda=@var{n}
16795 Put static or global variables whose size is @var{n} bytes or less into
16796 the first 32 kilobytes of memory.
16800 Specify that the target processor is the V850.
16803 @opindex mbig-switch
16804 Generate code suitable for big switch tables. Use this option only if
16805 the assembler/linker complain about out of range branches within a switch
16810 This option will cause r2 and r5 to be used in the code generated by
16811 the compiler. This setting is the default.
16813 @item -mno-app-regs
16814 @opindex mno-app-regs
16815 This option will cause r2 and r5 to be treated as fixed registers.
16819 Specify that the target processor is the V850E1. The preprocessor
16820 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16821 this option is used.
16825 Specify that the target processor is the V850E@. The preprocessor
16826 constant @samp{__v850e__} will be defined if this option is used.
16828 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16829 are defined then a default target processor will be chosen and the
16830 relevant @samp{__v850*__} preprocessor constant will be defined.
16832 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16833 defined, regardless of which processor variant is the target.
16835 @item -mdisable-callt
16836 @opindex mdisable-callt
16837 This option will suppress generation of the CALLT instruction for the
16838 v850e and v850e1 flavors of the v850 architecture. The default is
16839 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16844 @subsection VAX Options
16845 @cindex VAX options
16847 These @samp{-m} options are defined for the VAX:
16852 Do not output certain jump instructions (@code{aobleq} and so on)
16853 that the Unix assembler for the VAX cannot handle across long
16858 Do output those jump instructions, on the assumption that you
16859 will assemble with the GNU assembler.
16863 Output code for g-format floating point numbers instead of d-format.
16866 @node VxWorks Options
16867 @subsection VxWorks Options
16868 @cindex VxWorks Options
16870 The options in this section are defined for all VxWorks targets.
16871 Options specific to the target hardware are listed with the other
16872 options for that target.
16877 GCC can generate code for both VxWorks kernels and real time processes
16878 (RTPs). This option switches from the former to the latter. It also
16879 defines the preprocessor macro @code{__RTP__}.
16882 @opindex non-static
16883 Link an RTP executable against shared libraries rather than static
16884 libraries. The options @option{-static} and @option{-shared} can
16885 also be used for RTPs (@pxref{Link Options}); @option{-static}
16892 These options are passed down to the linker. They are defined for
16893 compatibility with Diab.
16896 @opindex Xbind-lazy
16897 Enable lazy binding of function calls. This option is equivalent to
16898 @option{-Wl,-z,now} and is defined for compatibility with Diab.
16902 Disable lazy binding of function calls. This option is the default and
16903 is defined for compatibility with Diab.
16906 @node x86-64 Options
16907 @subsection x86-64 Options
16908 @cindex x86-64 options
16910 These are listed under @xref{i386 and x86-64 Options}.
16912 @node i386 and x86-64 Windows Options
16913 @subsection i386 and x86-64 Windows Options
16914 @cindex i386 and x86-64 Windows Options
16916 These additional options are available for Windows targets:
16921 This option is available for Cygwin and MinGW targets. It
16922 specifies that a console application is to be generated, by
16923 instructing the linker to set the PE header subsystem type
16924 required for console applications.
16925 This is the default behavior for Cygwin and MinGW targets.
16929 This option is available for Cygwin targets. It specifies that
16930 the Cygwin internal interface is to be used for predefined
16931 preprocessor macros, C runtime libraries and related linker
16932 paths and options. For Cygwin targets this is the default behavior.
16933 This option is deprecated and will be removed in a future release.
16936 @opindex mno-cygwin
16937 This option is available for Cygwin targets. It specifies that
16938 the MinGW internal interface is to be used instead of Cygwin's, by
16939 setting MinGW-related predefined macros and linker paths and default
16941 This option is deprecated and will be removed in a future release.
16945 This option is available for Cygwin and MinGW targets. It
16946 specifies that a DLL - a dynamic link library - is to be
16947 generated, enabling the selection of the required runtime
16948 startup object and entry point.
16950 @item -mnop-fun-dllimport
16951 @opindex mnop-fun-dllimport
16952 This option is available for Cygwin and MinGW targets. It
16953 specifies that the dllimport attribute should be ignored.
16957 This option is available for MinGW targets. It specifies
16958 that MinGW-specific thread support is to be used.
16962 This option is available for mingw-w64 targets. It specifies
16963 that the UNICODE macro is getting pre-defined and that the
16964 unicode capable runtime startup code is chosen.
16968 This option is available for Cygwin and MinGW targets. It
16969 specifies that the typical Windows pre-defined macros are to
16970 be set in the pre-processor, but does not influence the choice
16971 of runtime library/startup code.
16975 This option is available for Cygwin and MinGW targets. It
16976 specifies that a GUI application is to be generated by
16977 instructing the linker to set the PE header subsystem type
16980 @item -fno-set-stack-executable
16981 @opindex fno-set-stack-executable
16982 This option is available for MinGW targets. It specifies that
16983 the executable flag for stack used by nested functions isn't
16984 set. This is necessary for binaries running in kernel mode of
16985 Windows, as there the user32 API, which is used to set executable
16986 privileges, isn't available.
16988 @item -mpe-aligned-commons
16989 @opindex mpe-aligned-commons
16990 This option is available for Cygwin and MinGW targets. It
16991 specifies that the GNU extension to the PE file format that
16992 permits the correct alignment of COMMON variables should be
16993 used when generating code. It will be enabled by default if
16994 GCC detects that the target assembler found during configuration
16995 supports the feature.
16998 See also under @ref{i386 and x86-64 Options} for standard options.
17000 @node Xstormy16 Options
17001 @subsection Xstormy16 Options
17002 @cindex Xstormy16 Options
17004 These options are defined for Xstormy16:
17009 Choose startup files and linker script suitable for the simulator.
17012 @node Xtensa Options
17013 @subsection Xtensa Options
17014 @cindex Xtensa Options
17016 These options are supported for Xtensa targets:
17020 @itemx -mno-const16
17022 @opindex mno-const16
17023 Enable or disable use of @code{CONST16} instructions for loading
17024 constant values. The @code{CONST16} instruction is currently not a
17025 standard option from Tensilica. When enabled, @code{CONST16}
17026 instructions are always used in place of the standard @code{L32R}
17027 instructions. The use of @code{CONST16} is enabled by default only if
17028 the @code{L32R} instruction is not available.
17031 @itemx -mno-fused-madd
17032 @opindex mfused-madd
17033 @opindex mno-fused-madd
17034 Enable or disable use of fused multiply/add and multiply/subtract
17035 instructions in the floating-point option. This has no effect if the
17036 floating-point option is not also enabled. Disabling fused multiply/add
17037 and multiply/subtract instructions forces the compiler to use separate
17038 instructions for the multiply and add/subtract operations. This may be
17039 desirable in some cases where strict IEEE 754-compliant results are
17040 required: the fused multiply add/subtract instructions do not round the
17041 intermediate result, thereby producing results with @emph{more} bits of
17042 precision than specified by the IEEE standard. Disabling fused multiply
17043 add/subtract instructions also ensures that the program output is not
17044 sensitive to the compiler's ability to combine multiply and add/subtract
17047 @item -mserialize-volatile
17048 @itemx -mno-serialize-volatile
17049 @opindex mserialize-volatile
17050 @opindex mno-serialize-volatile
17051 When this option is enabled, GCC inserts @code{MEMW} instructions before
17052 @code{volatile} memory references to guarantee sequential consistency.
17053 The default is @option{-mserialize-volatile}. Use
17054 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
17056 @item -mtext-section-literals
17057 @itemx -mno-text-section-literals
17058 @opindex mtext-section-literals
17059 @opindex mno-text-section-literals
17060 Control the treatment of literal pools. The default is
17061 @option{-mno-text-section-literals}, which places literals in a separate
17062 section in the output file. This allows the literal pool to be placed
17063 in a data RAM/ROM, and it also allows the linker to combine literal
17064 pools from separate object files to remove redundant literals and
17065 improve code size. With @option{-mtext-section-literals}, the literals
17066 are interspersed in the text section in order to keep them as close as
17067 possible to their references. This may be necessary for large assembly
17070 @item -mtarget-align
17071 @itemx -mno-target-align
17072 @opindex mtarget-align
17073 @opindex mno-target-align
17074 When this option is enabled, GCC instructs the assembler to
17075 automatically align instructions to reduce branch penalties at the
17076 expense of some code density. The assembler attempts to widen density
17077 instructions to align branch targets and the instructions following call
17078 instructions. If there are not enough preceding safe density
17079 instructions to align a target, no widening will be performed. The
17080 default is @option{-mtarget-align}. These options do not affect the
17081 treatment of auto-aligned instructions like @code{LOOP}, which the
17082 assembler will always align, either by widening density instructions or
17083 by inserting no-op instructions.
17086 @itemx -mno-longcalls
17087 @opindex mlongcalls
17088 @opindex mno-longcalls
17089 When this option is enabled, GCC instructs the assembler to translate
17090 direct calls to indirect calls unless it can determine that the target
17091 of a direct call is in the range allowed by the call instruction. This
17092 translation typically occurs for calls to functions in other source
17093 files. Specifically, the assembler translates a direct @code{CALL}
17094 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
17095 The default is @option{-mno-longcalls}. This option should be used in
17096 programs where the call target can potentially be out of range. This
17097 option is implemented in the assembler, not the compiler, so the
17098 assembly code generated by GCC will still show direct call
17099 instructions---look at the disassembled object code to see the actual
17100 instructions. Note that the assembler will use an indirect call for
17101 every cross-file call, not just those that really will be out of range.
17104 @node zSeries Options
17105 @subsection zSeries Options
17106 @cindex zSeries options
17108 These are listed under @xref{S/390 and zSeries Options}.
17110 @node Code Gen Options
17111 @section Options for Code Generation Conventions
17112 @cindex code generation conventions
17113 @cindex options, code generation
17114 @cindex run-time options
17116 These machine-independent options control the interface conventions
17117 used in code generation.
17119 Most of them have both positive and negative forms; the negative form
17120 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
17121 one of the forms is listed---the one which is not the default. You
17122 can figure out the other form by either removing @samp{no-} or adding
17126 @item -fbounds-check
17127 @opindex fbounds-check
17128 For front-ends that support it, generate additional code to check that
17129 indices used to access arrays are within the declared range. This is
17130 currently only supported by the Java and Fortran front-ends, where
17131 this option defaults to true and false respectively.
17135 This option generates traps for signed overflow on addition, subtraction,
17136 multiplication operations.
17140 This option instructs the compiler to assume that signed arithmetic
17141 overflow of addition, subtraction and multiplication wraps around
17142 using twos-complement representation. This flag enables some optimizations
17143 and disables others. This option is enabled by default for the Java
17144 front-end, as required by the Java language specification.
17147 @opindex fexceptions
17148 Enable exception handling. Generates extra code needed to propagate
17149 exceptions. For some targets, this implies GCC will generate frame
17150 unwind information for all functions, which can produce significant data
17151 size overhead, although it does not affect execution. If you do not
17152 specify this option, GCC will enable it by default for languages like
17153 C++ which normally require exception handling, and disable it for
17154 languages like C that do not normally require it. However, you may need
17155 to enable this option when compiling C code that needs to interoperate
17156 properly with exception handlers written in C++. You may also wish to
17157 disable this option if you are compiling older C++ programs that don't
17158 use exception handling.
17160 @item -fnon-call-exceptions
17161 @opindex fnon-call-exceptions
17162 Generate code that allows trapping instructions to throw exceptions.
17163 Note that this requires platform-specific runtime support that does
17164 not exist everywhere. Moreover, it only allows @emph{trapping}
17165 instructions to throw exceptions, i.e.@: memory references or floating
17166 point instructions. It does not allow exceptions to be thrown from
17167 arbitrary signal handlers such as @code{SIGALRM}.
17169 @item -funwind-tables
17170 @opindex funwind-tables
17171 Similar to @option{-fexceptions}, except that it will just generate any needed
17172 static data, but will not affect the generated code in any other way.
17173 You will normally not enable this option; instead, a language processor
17174 that needs this handling would enable it on your behalf.
17176 @item -fasynchronous-unwind-tables
17177 @opindex fasynchronous-unwind-tables
17178 Generate unwind table in dwarf2 format, if supported by target machine. The
17179 table is exact at each instruction boundary, so it can be used for stack
17180 unwinding from asynchronous events (such as debugger or garbage collector).
17182 @item -fpcc-struct-return
17183 @opindex fpcc-struct-return
17184 Return ``short'' @code{struct} and @code{union} values in memory like
17185 longer ones, rather than in registers. This convention is less
17186 efficient, but it has the advantage of allowing intercallability between
17187 GCC-compiled files and files compiled with other compilers, particularly
17188 the Portable C Compiler (pcc).
17190 The precise convention for returning structures in memory depends
17191 on the target configuration macros.
17193 Short structures and unions are those whose size and alignment match
17194 that of some integer type.
17196 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
17197 switch is not binary compatible with code compiled with the
17198 @option{-freg-struct-return} switch.
17199 Use it to conform to a non-default application binary interface.
17201 @item -freg-struct-return
17202 @opindex freg-struct-return
17203 Return @code{struct} and @code{union} values in registers when possible.
17204 This is more efficient for small structures than
17205 @option{-fpcc-struct-return}.
17207 If you specify neither @option{-fpcc-struct-return} nor
17208 @option{-freg-struct-return}, GCC defaults to whichever convention is
17209 standard for the target. If there is no standard convention, GCC
17210 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
17211 the principal compiler. In those cases, we can choose the standard, and
17212 we chose the more efficient register return alternative.
17214 @strong{Warning:} code compiled with the @option{-freg-struct-return}
17215 switch is not binary compatible with code compiled with the
17216 @option{-fpcc-struct-return} switch.
17217 Use it to conform to a non-default application binary interface.
17219 @item -fshort-enums
17220 @opindex fshort-enums
17221 Allocate to an @code{enum} type only as many bytes as it needs for the
17222 declared range of possible values. Specifically, the @code{enum} type
17223 will be equivalent to the smallest integer type which has enough room.
17225 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
17226 code that is not binary compatible with code generated without that switch.
17227 Use it to conform to a non-default application binary interface.
17229 @item -fshort-double
17230 @opindex fshort-double
17231 Use the same size for @code{double} as for @code{float}.
17233 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
17234 code that is not binary compatible with code generated without that switch.
17235 Use it to conform to a non-default application binary interface.
17237 @item -fshort-wchar
17238 @opindex fshort-wchar
17239 Override the underlying type for @samp{wchar_t} to be @samp{short
17240 unsigned int} instead of the default for the target. This option is
17241 useful for building programs to run under WINE@.
17243 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
17244 code that is not binary compatible with code generated without that switch.
17245 Use it to conform to a non-default application binary interface.
17248 @opindex fno-common
17249 In C code, controls the placement of uninitialized global variables.
17250 Unix C compilers have traditionally permitted multiple definitions of
17251 such variables in different compilation units by placing the variables
17253 This is the behavior specified by @option{-fcommon}, and is the default
17254 for GCC on most targets.
17255 On the other hand, this behavior is not required by ISO C, and on some
17256 targets may carry a speed or code size penalty on variable references.
17257 The @option{-fno-common} option specifies that the compiler should place
17258 uninitialized global variables in the data section of the object file,
17259 rather than generating them as common blocks.
17260 This has the effect that if the same variable is declared
17261 (without @code{extern}) in two different compilations,
17262 you will get a multiple-definition error when you link them.
17263 In this case, you must compile with @option{-fcommon} instead.
17264 Compiling with @option{-fno-common} is useful on targets for which
17265 it provides better performance, or if you wish to verify that the
17266 program will work on other systems which always treat uninitialized
17267 variable declarations this way.
17271 Ignore the @samp{#ident} directive.
17273 @item -finhibit-size-directive
17274 @opindex finhibit-size-directive
17275 Don't output a @code{.size} assembler directive, or anything else that
17276 would cause trouble if the function is split in the middle, and the
17277 two halves are placed at locations far apart in memory. This option is
17278 used when compiling @file{crtstuff.c}; you should not need to use it
17281 @item -fverbose-asm
17282 @opindex fverbose-asm
17283 Put extra commentary information in the generated assembly code to
17284 make it more readable. This option is generally only of use to those
17285 who actually need to read the generated assembly code (perhaps while
17286 debugging the compiler itself).
17288 @option{-fno-verbose-asm}, the default, causes the
17289 extra information to be omitted and is useful when comparing two assembler
17292 @item -frecord-gcc-switches
17293 @opindex frecord-gcc-switches
17294 This switch causes the command line that was used to invoke the
17295 compiler to be recorded into the object file that is being created.
17296 This switch is only implemented on some targets and the exact format
17297 of the recording is target and binary file format dependent, but it
17298 usually takes the form of a section containing ASCII text. This
17299 switch is related to the @option{-fverbose-asm} switch, but that
17300 switch only records information in the assembler output file as
17301 comments, so it never reaches the object file.
17305 @cindex global offset table
17307 Generate position-independent code (PIC) suitable for use in a shared
17308 library, if supported for the target machine. Such code accesses all
17309 constant addresses through a global offset table (GOT)@. The dynamic
17310 loader resolves the GOT entries when the program starts (the dynamic
17311 loader is not part of GCC; it is part of the operating system). If
17312 the GOT size for the linked executable exceeds a machine-specific
17313 maximum size, you get an error message from the linker indicating that
17314 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17315 instead. (These maximums are 8k on the SPARC and 32k
17316 on the m68k and RS/6000. The 386 has no such limit.)
17318 Position-independent code requires special support, and therefore works
17319 only on certain machines. For the 386, GCC supports PIC for System V
17320 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17321 position-independent.
17323 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17328 If supported for the target machine, emit position-independent code,
17329 suitable for dynamic linking and avoiding any limit on the size of the
17330 global offset table. This option makes a difference on the m68k,
17331 PowerPC and SPARC@.
17333 Position-independent code requires special support, and therefore works
17334 only on certain machines.
17336 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17343 These options are similar to @option{-fpic} and @option{-fPIC}, but
17344 generated position independent code can be only linked into executables.
17345 Usually these options are used when @option{-pie} GCC option will be
17346 used during linking.
17348 @option{-fpie} and @option{-fPIE} both define the macros
17349 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17350 for @option{-fpie} and 2 for @option{-fPIE}.
17352 @item -fno-jump-tables
17353 @opindex fno-jump-tables
17354 Do not use jump tables for switch statements even where it would be
17355 more efficient than other code generation strategies. This option is
17356 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17357 building code which forms part of a dynamic linker and cannot
17358 reference the address of a jump table. On some targets, jump tables
17359 do not require a GOT and this option is not needed.
17361 @item -ffixed-@var{reg}
17363 Treat the register named @var{reg} as a fixed register; generated code
17364 should never refer to it (except perhaps as a stack pointer, frame
17365 pointer or in some other fixed role).
17367 @var{reg} must be the name of a register. The register names accepted
17368 are machine-specific and are defined in the @code{REGISTER_NAMES}
17369 macro in the machine description macro file.
17371 This flag does not have a negative form, because it specifies a
17374 @item -fcall-used-@var{reg}
17375 @opindex fcall-used
17376 Treat the register named @var{reg} as an allocable register that is
17377 clobbered by function calls. It may be allocated for temporaries or
17378 variables that do not live across a call. Functions compiled this way
17379 will not save and restore the register @var{reg}.
17381 It is an error to used this flag with the frame pointer or stack pointer.
17382 Use of this flag for other registers that have fixed pervasive roles in
17383 the machine's execution model will produce disastrous results.
17385 This flag does not have a negative form, because it specifies a
17388 @item -fcall-saved-@var{reg}
17389 @opindex fcall-saved
17390 Treat the register named @var{reg} as an allocable register saved by
17391 functions. It may be allocated even for temporaries or variables that
17392 live across a call. Functions compiled this way will save and restore
17393 the register @var{reg} if they use it.
17395 It is an error to used this flag with the frame pointer or stack pointer.
17396 Use of this flag for other registers that have fixed pervasive roles in
17397 the machine's execution model will produce disastrous results.
17399 A different sort of disaster will result from the use of this flag for
17400 a register in which function values may be returned.
17402 This flag does not have a negative form, because it specifies a
17405 @item -fpack-struct[=@var{n}]
17406 @opindex fpack-struct
17407 Without a value specified, pack all structure members together without
17408 holes. When a value is specified (which must be a small power of two), pack
17409 structure members according to this value, representing the maximum
17410 alignment (that is, objects with default alignment requirements larger than
17411 this will be output potentially unaligned at the next fitting location.
17413 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17414 code that is not binary compatible with code generated without that switch.
17415 Additionally, it makes the code suboptimal.
17416 Use it to conform to a non-default application binary interface.
17418 @item -finstrument-functions
17419 @opindex finstrument-functions
17420 Generate instrumentation calls for entry and exit to functions. Just
17421 after function entry and just before function exit, the following
17422 profiling functions will be called with the address of the current
17423 function and its call site. (On some platforms,
17424 @code{__builtin_return_address} does not work beyond the current
17425 function, so the call site information may not be available to the
17426 profiling functions otherwise.)
17429 void __cyg_profile_func_enter (void *this_fn,
17431 void __cyg_profile_func_exit (void *this_fn,
17435 The first argument is the address of the start of the current function,
17436 which may be looked up exactly in the symbol table.
17438 This instrumentation is also done for functions expanded inline in other
17439 functions. The profiling calls will indicate where, conceptually, the
17440 inline function is entered and exited. This means that addressable
17441 versions of such functions must be available. If all your uses of a
17442 function are expanded inline, this may mean an additional expansion of
17443 code size. If you use @samp{extern inline} in your C code, an
17444 addressable version of such functions must be provided. (This is
17445 normally the case anyways, but if you get lucky and the optimizer always
17446 expands the functions inline, you might have gotten away without
17447 providing static copies.)
17449 A function may be given the attribute @code{no_instrument_function}, in
17450 which case this instrumentation will not be done. This can be used, for
17451 example, for the profiling functions listed above, high-priority
17452 interrupt routines, and any functions from which the profiling functions
17453 cannot safely be called (perhaps signal handlers, if the profiling
17454 routines generate output or allocate memory).
17456 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17457 @opindex finstrument-functions-exclude-file-list
17459 Set the list of functions that are excluded from instrumentation (see
17460 the description of @code{-finstrument-functions}). If the file that
17461 contains a function definition matches with one of @var{file}, then
17462 that function is not instrumented. The match is done on substrings:
17463 if the @var{file} parameter is a substring of the file name, it is
17464 considered to be a match.
17467 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
17468 will exclude any inline function defined in files whose pathnames
17469 contain @code{/bits/stl} or @code{include/sys}.
17471 If, for some reason, you want to include letter @code{','} in one of
17472 @var{sym}, write @code{'\,'}. For example,
17473 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17474 (note the single quote surrounding the option).
17476 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
17477 @opindex finstrument-functions-exclude-function-list
17479 This is similar to @code{-finstrument-functions-exclude-file-list},
17480 but this option sets the list of function names to be excluded from
17481 instrumentation. The function name to be matched is its user-visible
17482 name, such as @code{vector<int> blah(const vector<int> &)}, not the
17483 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
17484 match is done on substrings: if the @var{sym} parameter is a substring
17485 of the function name, it is considered to be a match. For C99 and C++
17486 extended identifiers, the function name must be given in UTF-8, not
17487 using universal character names.
17489 @item -fstack-check
17490 @opindex fstack-check
17491 Generate code to verify that you do not go beyond the boundary of the
17492 stack. You should specify this flag if you are running in an
17493 environment with multiple threads, but only rarely need to specify it in
17494 a single-threaded environment since stack overflow is automatically
17495 detected on nearly all systems if there is only one stack.
17497 Note that this switch does not actually cause checking to be done; the
17498 operating system or the language runtime must do that. The switch causes
17499 generation of code to ensure that they see the stack being extended.
17501 You can additionally specify a string parameter: @code{no} means no
17502 checking, @code{generic} means force the use of old-style checking,
17503 @code{specific} means use the best checking method and is equivalent
17504 to bare @option{-fstack-check}.
17506 Old-style checking is a generic mechanism that requires no specific
17507 target support in the compiler but comes with the following drawbacks:
17511 Modified allocation strategy for large objects: they will always be
17512 allocated dynamically if their size exceeds a fixed threshold.
17515 Fixed limit on the size of the static frame of functions: when it is
17516 topped by a particular function, stack checking is not reliable and
17517 a warning is issued by the compiler.
17520 Inefficiency: because of both the modified allocation strategy and the
17521 generic implementation, the performances of the code are hampered.
17524 Note that old-style stack checking is also the fallback method for
17525 @code{specific} if no target support has been added in the compiler.
17527 @item -fstack-limit-register=@var{reg}
17528 @itemx -fstack-limit-symbol=@var{sym}
17529 @itemx -fno-stack-limit
17530 @opindex fstack-limit-register
17531 @opindex fstack-limit-symbol
17532 @opindex fno-stack-limit
17533 Generate code to ensure that the stack does not grow beyond a certain value,
17534 either the value of a register or the address of a symbol. If the stack
17535 would grow beyond the value, a signal is raised. For most targets,
17536 the signal is raised before the stack overruns the boundary, so
17537 it is possible to catch the signal without taking special precautions.
17539 For instance, if the stack starts at absolute address @samp{0x80000000}
17540 and grows downwards, you can use the flags
17541 @option{-fstack-limit-symbol=__stack_limit} and
17542 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
17543 of 128KB@. Note that this may only work with the GNU linker.
17545 @item -fleading-underscore
17546 @opindex fleading-underscore
17547 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
17548 change the way C symbols are represented in the object file. One use
17549 is to help link with legacy assembly code.
17551 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
17552 generate code that is not binary compatible with code generated without that
17553 switch. Use it to conform to a non-default application binary interface.
17554 Not all targets provide complete support for this switch.
17556 @item -ftls-model=@var{model}
17557 @opindex ftls-model
17558 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
17559 The @var{model} argument should be one of @code{global-dynamic},
17560 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
17562 The default without @option{-fpic} is @code{initial-exec}; with
17563 @option{-fpic} the default is @code{global-dynamic}.
17565 @item -fvisibility=@var{default|internal|hidden|protected}
17566 @opindex fvisibility
17567 Set the default ELF image symbol visibility to the specified option---all
17568 symbols will be marked with this unless overridden within the code.
17569 Using this feature can very substantially improve linking and
17570 load times of shared object libraries, produce more optimized
17571 code, provide near-perfect API export and prevent symbol clashes.
17572 It is @strong{strongly} recommended that you use this in any shared objects
17575 Despite the nomenclature, @code{default} always means public ie;
17576 available to be linked against from outside the shared object.
17577 @code{protected} and @code{internal} are pretty useless in real-world
17578 usage so the only other commonly used option will be @code{hidden}.
17579 The default if @option{-fvisibility} isn't specified is
17580 @code{default}, i.e., make every
17581 symbol public---this causes the same behavior as previous versions of
17584 A good explanation of the benefits offered by ensuring ELF
17585 symbols have the correct visibility is given by ``How To Write
17586 Shared Libraries'' by Ulrich Drepper (which can be found at
17587 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
17588 solution made possible by this option to marking things hidden when
17589 the default is public is to make the default hidden and mark things
17590 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
17591 and @code{__attribute__ ((visibility("default")))} instead of
17592 @code{__declspec(dllexport)} you get almost identical semantics with
17593 identical syntax. This is a great boon to those working with
17594 cross-platform projects.
17596 For those adding visibility support to existing code, you may find
17597 @samp{#pragma GCC visibility} of use. This works by you enclosing
17598 the declarations you wish to set visibility for with (for example)
17599 @samp{#pragma GCC visibility push(hidden)} and
17600 @samp{#pragma GCC visibility pop}.
17601 Bear in mind that symbol visibility should be viewed @strong{as
17602 part of the API interface contract} and thus all new code should
17603 always specify visibility when it is not the default ie; declarations
17604 only for use within the local DSO should @strong{always} be marked explicitly
17605 as hidden as so to avoid PLT indirection overheads---making this
17606 abundantly clear also aids readability and self-documentation of the code.
17607 Note that due to ISO C++ specification requirements, operator new and
17608 operator delete must always be of default visibility.
17610 Be aware that headers from outside your project, in particular system
17611 headers and headers from any other library you use, may not be
17612 expecting to be compiled with visibility other than the default. You
17613 may need to explicitly say @samp{#pragma GCC visibility push(default)}
17614 before including any such headers.
17616 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
17617 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
17618 no modifications. However, this means that calls to @samp{extern}
17619 functions with no explicit visibility will use the PLT, so it is more
17620 effective to use @samp{__attribute ((visibility))} and/or
17621 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
17622 declarations should be treated as hidden.
17624 Note that @samp{-fvisibility} does affect C++ vague linkage
17625 entities. This means that, for instance, an exception class that will
17626 be thrown between DSOs must be explicitly marked with default
17627 visibility so that the @samp{type_info} nodes will be unified between
17630 An overview of these techniques, their benefits and how to use them
17631 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
17637 @node Environment Variables
17638 @section Environment Variables Affecting GCC
17639 @cindex environment variables
17641 @c man begin ENVIRONMENT
17642 This section describes several environment variables that affect how GCC
17643 operates. Some of them work by specifying directories or prefixes to use
17644 when searching for various kinds of files. Some are used to specify other
17645 aspects of the compilation environment.
17647 Note that you can also specify places to search using options such as
17648 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
17649 take precedence over places specified using environment variables, which
17650 in turn take precedence over those specified by the configuration of GCC@.
17651 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
17652 GNU Compiler Collection (GCC) Internals}.
17657 @c @itemx LC_COLLATE
17659 @c @itemx LC_MONETARY
17660 @c @itemx LC_NUMERIC
17665 @c @findex LC_COLLATE
17666 @findex LC_MESSAGES
17667 @c @findex LC_MONETARY
17668 @c @findex LC_NUMERIC
17672 These environment variables control the way that GCC uses
17673 localization information that allow GCC to work with different
17674 national conventions. GCC inspects the locale categories
17675 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
17676 so. These locale categories can be set to any value supported by your
17677 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
17678 Kingdom encoded in UTF-8.
17680 The @env{LC_CTYPE} environment variable specifies character
17681 classification. GCC uses it to determine the character boundaries in
17682 a string; this is needed for some multibyte encodings that contain quote
17683 and escape characters that would otherwise be interpreted as a string
17686 The @env{LC_MESSAGES} environment variable specifies the language to
17687 use in diagnostic messages.
17689 If the @env{LC_ALL} environment variable is set, it overrides the value
17690 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17691 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17692 environment variable. If none of these variables are set, GCC
17693 defaults to traditional C English behavior.
17697 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17698 files. GCC uses temporary files to hold the output of one stage of
17699 compilation which is to be used as input to the next stage: for example,
17700 the output of the preprocessor, which is the input to the compiler
17703 @item GCC_EXEC_PREFIX
17704 @findex GCC_EXEC_PREFIX
17705 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17706 names of the subprograms executed by the compiler. No slash is added
17707 when this prefix is combined with the name of a subprogram, but you can
17708 specify a prefix that ends with a slash if you wish.
17710 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17711 an appropriate prefix to use based on the pathname it was invoked with.
17713 If GCC cannot find the subprogram using the specified prefix, it
17714 tries looking in the usual places for the subprogram.
17716 The default value of @env{GCC_EXEC_PREFIX} is
17717 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17718 the installed compiler. In many cases @var{prefix} is the value
17719 of @code{prefix} when you ran the @file{configure} script.
17721 Other prefixes specified with @option{-B} take precedence over this prefix.
17723 This prefix is also used for finding files such as @file{crt0.o} that are
17726 In addition, the prefix is used in an unusual way in finding the
17727 directories to search for header files. For each of the standard
17728 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17729 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17730 replacing that beginning with the specified prefix to produce an
17731 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17732 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17733 These alternate directories are searched first; the standard directories
17734 come next. If a standard directory begins with the configured
17735 @var{prefix} then the value of @var{prefix} is replaced by
17736 @env{GCC_EXEC_PREFIX} when looking for header files.
17738 @item COMPILER_PATH
17739 @findex COMPILER_PATH
17740 The value of @env{COMPILER_PATH} is a colon-separated list of
17741 directories, much like @env{PATH}. GCC tries the directories thus
17742 specified when searching for subprograms, if it can't find the
17743 subprograms using @env{GCC_EXEC_PREFIX}.
17746 @findex LIBRARY_PATH
17747 The value of @env{LIBRARY_PATH} is a colon-separated list of
17748 directories, much like @env{PATH}. When configured as a native compiler,
17749 GCC tries the directories thus specified when searching for special
17750 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17751 using GCC also uses these directories when searching for ordinary
17752 libraries for the @option{-l} option (but directories specified with
17753 @option{-L} come first).
17757 @cindex locale definition
17758 This variable is used to pass locale information to the compiler. One way in
17759 which this information is used is to determine the character set to be used
17760 when character literals, string literals and comments are parsed in C and C++.
17761 When the compiler is configured to allow multibyte characters,
17762 the following values for @env{LANG} are recognized:
17766 Recognize JIS characters.
17768 Recognize SJIS characters.
17770 Recognize EUCJP characters.
17773 If @env{LANG} is not defined, or if it has some other value, then the
17774 compiler will use mblen and mbtowc as defined by the default locale to
17775 recognize and translate multibyte characters.
17779 Some additional environments variables affect the behavior of the
17782 @include cppenv.texi
17786 @node Precompiled Headers
17787 @section Using Precompiled Headers
17788 @cindex precompiled headers
17789 @cindex speed of compilation
17791 Often large projects have many header files that are included in every
17792 source file. The time the compiler takes to process these header files
17793 over and over again can account for nearly all of the time required to
17794 build the project. To make builds faster, GCC allows users to
17795 `precompile' a header file; then, if builds can use the precompiled
17796 header file they will be much faster.
17798 To create a precompiled header file, simply compile it as you would any
17799 other file, if necessary using the @option{-x} option to make the driver
17800 treat it as a C or C++ header file. You will probably want to use a
17801 tool like @command{make} to keep the precompiled header up-to-date when
17802 the headers it contains change.
17804 A precompiled header file will be searched for when @code{#include} is
17805 seen in the compilation. As it searches for the included file
17806 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17807 compiler looks for a precompiled header in each directory just before it
17808 looks for the include file in that directory. The name searched for is
17809 the name specified in the @code{#include} with @samp{.gch} appended. If
17810 the precompiled header file can't be used, it is ignored.
17812 For instance, if you have @code{#include "all.h"}, and you have
17813 @file{all.h.gch} in the same directory as @file{all.h}, then the
17814 precompiled header file will be used if possible, and the original
17815 header will be used otherwise.
17817 Alternatively, you might decide to put the precompiled header file in a
17818 directory and use @option{-I} to ensure that directory is searched
17819 before (or instead of) the directory containing the original header.
17820 Then, if you want to check that the precompiled header file is always
17821 used, you can put a file of the same name as the original header in this
17822 directory containing an @code{#error} command.
17824 This also works with @option{-include}. So yet another way to use
17825 precompiled headers, good for projects not designed with precompiled
17826 header files in mind, is to simply take most of the header files used by
17827 a project, include them from another header file, precompile that header
17828 file, and @option{-include} the precompiled header. If the header files
17829 have guards against multiple inclusion, they will be skipped because
17830 they've already been included (in the precompiled header).
17832 If you need to precompile the same header file for different
17833 languages, targets, or compiler options, you can instead make a
17834 @emph{directory} named like @file{all.h.gch}, and put each precompiled
17835 header in the directory, perhaps using @option{-o}. It doesn't matter
17836 what you call the files in the directory, every precompiled header in
17837 the directory will be considered. The first precompiled header
17838 encountered in the directory that is valid for this compilation will
17839 be used; they're searched in no particular order.
17841 There are many other possibilities, limited only by your imagination,
17842 good sense, and the constraints of your build system.
17844 A precompiled header file can be used only when these conditions apply:
17848 Only one precompiled header can be used in a particular compilation.
17851 A precompiled header can't be used once the first C token is seen. You
17852 can have preprocessor directives before a precompiled header; you can
17853 even include a precompiled header from inside another header, so long as
17854 there are no C tokens before the @code{#include}.
17857 The precompiled header file must be produced for the same language as
17858 the current compilation. You can't use a C precompiled header for a C++
17862 The precompiled header file must have been produced by the same compiler
17863 binary as the current compilation is using.
17866 Any macros defined before the precompiled header is included must
17867 either be defined in the same way as when the precompiled header was
17868 generated, or must not affect the precompiled header, which usually
17869 means that they don't appear in the precompiled header at all.
17871 The @option{-D} option is one way to define a macro before a
17872 precompiled header is included; using a @code{#define} can also do it.
17873 There are also some options that define macros implicitly, like
17874 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
17877 @item If debugging information is output when using the precompiled
17878 header, using @option{-g} or similar, the same kind of debugging information
17879 must have been output when building the precompiled header. However,
17880 a precompiled header built using @option{-g} can be used in a compilation
17881 when no debugging information is being output.
17883 @item The same @option{-m} options must generally be used when building
17884 and using the precompiled header. @xref{Submodel Options},
17885 for any cases where this rule is relaxed.
17887 @item Each of the following options must be the same when building and using
17888 the precompiled header:
17890 @gccoptlist{-fexceptions}
17893 Some other command-line options starting with @option{-f},
17894 @option{-p}, or @option{-O} must be defined in the same way as when
17895 the precompiled header was generated. At present, it's not clear
17896 which options are safe to change and which are not; the safest choice
17897 is to use exactly the same options when generating and using the
17898 precompiled header. The following are known to be safe:
17900 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
17901 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
17902 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
17907 For all of these except the last, the compiler will automatically
17908 ignore the precompiled header if the conditions aren't met. If you
17909 find an option combination that doesn't work and doesn't cause the
17910 precompiled header to be ignored, please consider filing a bug report,
17913 If you do use differing options when generating and using the
17914 precompiled header, the actual behavior will be a mixture of the
17915 behavior for the options. For instance, if you use @option{-g} to
17916 generate the precompiled header but not when using it, you may or may
17917 not get debugging information for routines in the precompiled header.