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
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
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{-dr} 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.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version @@@var{file}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 -fvisibility-ms-compat @gol
196 -Wabi -Wctor-dtor-privacy @gol
197 -Wnon-virtual-dtor -Wreorder @gol
198 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
199 -Wno-non-template-friend -Wold-style-cast @gol
200 -Woverloaded-virtual -Wno-pmf-conversions @gol
203 @item Objective-C and Objective-C++ Language Options
204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 Objective-C and Objective-C++ Dialects}.
206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 -fgnu-runtime -fnext-runtime @gol
208 -fno-nil-receivers @gol
209 -fobjc-call-cxx-cdtors @gol
210 -fobjc-direct-dispatch @gol
211 -fobjc-exceptions @gol
213 -freplace-objc-classes @gol
216 -Wassign-intercept @gol
217 -Wno-protocol -Wselector @gol
218 -Wstrict-selector-match @gol
219 -Wundeclared-selector}
221 @item Language Independent Options
222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 @gccoptlist{-fmessage-length=@var{n} @gol
224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 -fdiagnostics-show-option}
227 @item Warning Options
228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
232 -Wchar-subscripts -Wclobbered -Wcomment @gol
233 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
234 -Wdisabled-optimization -Wno-div-by-zero @gol
235 -Wempty-body -Wno-endif-labels @gol
236 -Werror -Werror=* @gol
237 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
238 -Wno-format-extra-args -Wformat-nonliteral @gol
239 -Wformat-security -Wformat-y2k -Wignored-qualifiers @gol
240 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
241 -Wimport -Wno-import -Winit-self -Winline @gol
242 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
243 -Winvalid-pch -Wlarger-than-@var{len} -Wunsafe-loop-optimizations @gol
244 -Wlogical-op -Wlong-long @gol
245 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
246 -Wmissing-format-attribute -Wmissing-include-dirs @gol
247 -Wmissing-noreturn @gol
248 -Wno-multichar -Wnonnull -Wno-overflow @gol
249 -Woverlength-strings -Wpacked -Wpadded @gol
250 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
251 -Wredundant-decls @gol
252 -Wreturn-type -Wsequence-point -Wshadow @gol
253 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
254 -Wstrict-aliasing -Wstrict-aliasing=n @gol
255 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
256 -Wswitch -Wswitch-default -Wswitch-enum @gol
257 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
258 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
259 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
260 -Wunused-value -Wunused-variable @gol
261 -Wvariadic-macros -Wvla @gol
262 -Wvolatile-register-var -Wwrite-strings}
264 @item C-only Warning Options
265 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
266 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
267 -Wold-style-declaration -Wold-style-definition @gol
268 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
269 -Wdeclaration-after-statement -Wpointer-sign}
271 @item Debugging Options
272 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
273 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
274 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
275 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
276 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
277 -fdump-ipa-all -fdump-ipa-cgraph @gol
279 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
284 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-nrv -fdump-tree-vect @gol
293 -fdump-tree-sink @gol
294 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-salias @gol
296 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
298 -ftree-vectorizer-verbose=@var{n} @gol
299 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
300 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
301 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
302 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
303 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
304 -ftest-coverage -ftime-report -fvar-tracking @gol
305 -g -g@var{level} -gcoff -gdwarf-2 @gol
306 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
307 -fno-merge-debug-strings -fdebug-prefix-map=@var{old}=@var{new} @gol
308 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
309 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
310 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
311 -print-multi-directory -print-multi-lib @gol
312 -print-prog-name=@var{program} -print-search-dirs -Q @gol
313 -print-sysroot-headers-suffix @gol
316 @item Optimization Options
317 @xref{Optimize Options,,Options that Control Optimization}.
319 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
320 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
321 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
322 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
323 -fcheck-data-deps -fcprop-registers -fcrossjumping -fcse-follow-jumps @gol
324 -fcse-skip-blocks -fcx-limited-range -fdata-sections -fdce -fdce @gol
325 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
326 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
327 -ffinite-math-only -ffloat-store -fforward-propagate @gol
328 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
329 -fgcse-sm -fif-conversion -fif-conversion2 -finline-functions @gol
330 -finline-functions-called-once -finline-limit=@var{n} @gol
331 -finline-small-functions -fipa-cp -fipa-marix-reorg -fipa-pta @gol
332 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
333 -fipa-type-escape -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
334 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
335 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
336 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
337 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
338 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
339 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
340 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
341 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
342 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
343 -fprofile-generate -fprofile-use -fprofile-values -freciprocal-math @gol
344 -fregmove -frename-registers -freorder-blocks @gol
345 -freorder-blocks-and-partition -freorder-functions @gol
346 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
347 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
348 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
349 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
350 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
351 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
352 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
353 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer -ftree-ccp @gol
354 -ftree-ch -ftree-copy-prop -ftree-copyrename -ftree-dce @gol
355 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
356 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
357 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc -ftree-salias @gol
358 -ftree-sink -ftree-sra -ftree-store-ccp -ftree-ter @gol
359 -ftree-vect-loop-version -ftree-vectorize -ftree-vrp -funit-at-a-time @gol
360 -funroll-all-loops -funroll-loops -funsafe-loop-optimizations @gol
361 -funsafe-math-optimizations -funswitch-loops @gol
362 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
364 --param @var{name}=@var{value}
365 -O -O0 -O1 -O2 -O3 -Os}
367 @item Preprocessor Options
368 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
369 @gccoptlist{-A@var{question}=@var{answer} @gol
370 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
371 -C -dD -dI -dM -dN @gol
372 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
373 -idirafter @var{dir} @gol
374 -include @var{file} -imacros @var{file} @gol
375 -iprefix @var{file} -iwithprefix @var{dir} @gol
376 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
377 -imultilib @var{dir} -isysroot @var{dir} @gol
378 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
379 -P -fworking-directory -remap @gol
380 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
381 -Xpreprocessor @var{option}}
383 @item Assembler Option
384 @xref{Assembler Options,,Passing Options to the Assembler}.
385 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
388 @xref{Link Options,,Options for Linking}.
389 @gccoptlist{@var{object-file-name} -l@var{library} @gol
390 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
391 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
392 -Wl,@var{option} -Xlinker @var{option} @gol
395 @item Directory Options
396 @xref{Directory Options,,Options for Directory Search}.
397 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
398 -specs=@var{file} -I- --sysroot=@var{dir}}
401 @c I wrote this xref this way to avoid overfull hbox. -- rms
402 @xref{Target Options}.
403 @gccoptlist{-V @var{version} -b @var{machine}}
405 @item Machine Dependent Options
406 @xref{Submodel Options,,Hardware Models and Configurations}.
407 @c This list is ordered alphanumerically by subsection name.
408 @c Try and put the significant identifier (CPU or system) first,
409 @c so users have a clue at guessing where the ones they want will be.
412 @gccoptlist{-EB -EL @gol
413 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
414 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
417 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
418 -mabi=@var{name} @gol
419 -mapcs-stack-check -mno-apcs-stack-check @gol
420 -mapcs-float -mno-apcs-float @gol
421 -mapcs-reentrant -mno-apcs-reentrant @gol
422 -msched-prolog -mno-sched-prolog @gol
423 -mlittle-endian -mbig-endian -mwords-little-endian @gol
424 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
425 -mthumb-interwork -mno-thumb-interwork @gol
426 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
427 -mstructure-size-boundary=@var{n} @gol
428 -mabort-on-noreturn @gol
429 -mlong-calls -mno-long-calls @gol
430 -msingle-pic-base -mno-single-pic-base @gol
431 -mpic-register=@var{reg} @gol
432 -mnop-fun-dllimport @gol
433 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
434 -mpoke-function-name @gol
436 -mtpcs-frame -mtpcs-leaf-frame @gol
437 -mcaller-super-interworking -mcallee-super-interworking @gol
441 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
442 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
444 @emph{Blackfin Options}
445 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
446 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
447 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
448 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
449 -mno-id-shared-library -mshared-library-id=@var{n} @gol
450 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
451 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
452 -mfast-fp -minline-plt}
455 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
456 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
457 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
458 -mstack-align -mdata-align -mconst-align @gol
459 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
460 -melf -maout -melinux -mlinux -sim -sim2 @gol
461 -mmul-bug-workaround -mno-mul-bug-workaround}
464 @gccoptlist{-mmac -mpush-args}
466 @emph{Darwin Options}
467 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
468 -arch_only -bind_at_load -bundle -bundle_loader @gol
469 -client_name -compatibility_version -current_version @gol
471 -dependency-file -dylib_file -dylinker_install_name @gol
472 -dynamic -dynamiclib -exported_symbols_list @gol
473 -filelist -flat_namespace -force_cpusubtype_ALL @gol
474 -force_flat_namespace -headerpad_max_install_names @gol
476 -image_base -init -install_name -keep_private_externs @gol
477 -multi_module -multiply_defined -multiply_defined_unused @gol
478 -noall_load -no_dead_strip_inits_and_terms @gol
479 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
480 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
481 -private_bundle -read_only_relocs -sectalign @gol
482 -sectobjectsymbols -whyload -seg1addr @gol
483 -sectcreate -sectobjectsymbols -sectorder @gol
484 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
485 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
486 -segprot -segs_read_only_addr -segs_read_write_addr @gol
487 -single_module -static -sub_library -sub_umbrella @gol
488 -twolevel_namespace -umbrella -undefined @gol
489 -unexported_symbols_list -weak_reference_mismatches @gol
490 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
491 -mkernel -mone-byte-bool}
493 @emph{DEC Alpha Options}
494 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
495 -mieee -mieee-with-inexact -mieee-conformant @gol
496 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
497 -mtrap-precision=@var{mode} -mbuild-constants @gol
498 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
499 -mbwx -mmax -mfix -mcix @gol
500 -mfloat-vax -mfloat-ieee @gol
501 -mexplicit-relocs -msmall-data -mlarge-data @gol
502 -msmall-text -mlarge-text @gol
503 -mmemory-latency=@var{time}}
505 @emph{DEC Alpha/VMS Options}
506 @gccoptlist{-mvms-return-codes}
509 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
510 -mhard-float -msoft-float @gol
511 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
512 -mdouble -mno-double @gol
513 -mmedia -mno-media -mmuladd -mno-muladd @gol
514 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
515 -mlinked-fp -mlong-calls -malign-labels @gol
516 -mlibrary-pic -macc-4 -macc-8 @gol
517 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
518 -moptimize-membar -mno-optimize-membar @gol
519 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
520 -mvliw-branch -mno-vliw-branch @gol
521 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
522 -mno-nested-cond-exec -mtomcat-stats @gol
526 @emph{GNU/Linux Options}
527 @gccoptlist{-muclibc}
529 @emph{H8/300 Options}
530 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
533 @gccoptlist{-march=@var{architecture-type} @gol
534 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
535 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
536 -mfixed-range=@var{register-range} @gol
537 -mjump-in-delay -mlinker-opt -mlong-calls @gol
538 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
539 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
540 -mno-jump-in-delay -mno-long-load-store @gol
541 -mno-portable-runtime -mno-soft-float @gol
542 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
543 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
544 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
545 -munix=@var{unix-std} -nolibdld -static -threads}
547 @emph{i386 and x86-64 Options}
548 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
549 -mfpmath=@var{unit} @gol
550 -masm=@var{dialect} -mno-fancy-math-387 @gol
551 -mno-fp-ret-in-387 -msoft-float @gol
552 -mno-wide-multiply -mrtd -malign-double @gol
553 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
554 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
555 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
556 -mthreads -mno-align-stringops -minline-all-stringops @gol
557 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
558 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
559 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
560 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
561 -mcmodel=@var{code-model} @gol
562 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
563 -mfused-madd -mno-fused-madd}
566 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
567 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
568 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
569 -minline-float-divide-max-throughput @gol
570 -minline-int-divide-min-latency @gol
571 -minline-int-divide-max-throughput @gol
572 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
573 -mno-dwarf2-asm -mearly-stop-bits @gol
574 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
575 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
576 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
577 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
578 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
579 -mno-sched-prefer-non-data-spec-insns @gol
580 -mno-sched-prefer-non-control-spec-insns @gol
581 -mno-sched-count-spec-in-critical-path}
583 @emph{M32R/D Options}
584 @gccoptlist{-m32r2 -m32rx -m32r @gol
586 -malign-loops -mno-align-loops @gol
587 -missue-rate=@var{number} @gol
588 -mbranch-cost=@var{number} @gol
589 -mmodel=@var{code-size-model-type} @gol
590 -msdata=@var{sdata-type} @gol
591 -mno-flush-func -mflush-func=@var{name} @gol
592 -mno-flush-trap -mflush-trap=@var{number} @gol
596 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
598 @emph{M680x0 Options}
599 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
600 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
601 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
602 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
603 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
604 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
605 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
606 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
608 @emph{M68hc1x Options}
609 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
610 -mauto-incdec -minmax -mlong-calls -mshort @gol
611 -msoft-reg-count=@var{count}}
614 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
615 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
616 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
617 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
618 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
621 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
622 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
623 -mips16 -mno-mips16 -mflip-mips16 @gol
624 -minterlink-mips16 -mno-interlink-mips16 @gol
625 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
626 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
627 -mfp32 -mfp64 -mhard-float -msoft-float @gol
628 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
629 -msmartmips -mno-smartmips @gol
630 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
631 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
632 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
633 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
634 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
635 -membedded-data -mno-embedded-data @gol
636 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
637 -mcode-readable=@var{setting} @gol
638 -msplit-addresses -mno-split-addresses @gol
639 -mexplicit-relocs -mno-explicit-relocs @gol
640 -mcheck-zero-division -mno-check-zero-division @gol
641 -mdivide-traps -mdivide-breaks @gol
642 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
643 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
644 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
645 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
646 -mfix-sb1 -mno-fix-sb1 @gol
647 -mflush-func=@var{func} -mno-flush-func @gol
648 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
649 -mfp-exceptions -mno-fp-exceptions @gol
650 -mvr4130-align -mno-vr4130-align}
653 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
654 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
655 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
656 -mno-base-addresses -msingle-exit -mno-single-exit}
658 @emph{MN10300 Options}
659 @gccoptlist{-mmult-bug -mno-mult-bug @gol
660 -mam33 -mno-am33 @gol
661 -mam33-2 -mno-am33-2 @gol
662 -mreturn-pointer-on-d0 @gol
666 @gccoptlist{-mno-crt0 -mbacc -msim @gol
667 -march=@var{cpu-type} }
669 @emph{PDP-11 Options}
670 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
671 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
672 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
673 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
674 -mbranch-expensive -mbranch-cheap @gol
675 -msplit -mno-split -munix-asm -mdec-asm}
677 @emph{PowerPC Options}
678 See RS/6000 and PowerPC Options.
680 @emph{RS/6000 and PowerPC Options}
681 @gccoptlist{-mcpu=@var{cpu-type} @gol
682 -mtune=@var{cpu-type} @gol
683 -mpower -mno-power -mpower2 -mno-power2 @gol
684 -mpowerpc -mpowerpc64 -mno-powerpc @gol
685 -maltivec -mno-altivec @gol
686 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
687 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
688 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
689 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
690 -mnew-mnemonics -mold-mnemonics @gol
691 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
692 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
693 -malign-power -malign-natural @gol
694 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
695 -mstring -mno-string -mupdate -mno-update @gol
696 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
697 -mstrict-align -mno-strict-align -mrelocatable @gol
698 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
699 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
700 -mdynamic-no-pic -maltivec -mswdiv @gol
701 -mprioritize-restricted-insns=@var{priority} @gol
702 -msched-costly-dep=@var{dependence_type} @gol
703 -minsert-sched-nops=@var{scheme} @gol
704 -mcall-sysv -mcall-netbsd @gol
705 -maix-struct-return -msvr4-struct-return @gol
706 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
707 -misel -mno-isel @gol
708 -misel=yes -misel=no @gol
710 -mspe=yes -mspe=no @gol
712 -mvrsave -mno-vrsave @gol
713 -mmulhw -mno-mulhw @gol
714 -mdlmzb -mno-dlmzb @gol
715 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
716 -mprototype -mno-prototype @gol
717 -msim -mmvme -mads -myellowknife -memb -msdata @gol
718 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
720 @emph{S/390 and zSeries Options}
721 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
722 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
723 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
724 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
725 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
726 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
727 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
730 @gccoptlist{-meb -mel @gol
734 -mscore5 -mscore5u -mscore7 -mscore7d}
737 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
738 -m4-nofpu -m4-single-only -m4-single -m4 @gol
739 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
740 -m5-64media -m5-64media-nofpu @gol
741 -m5-32media -m5-32media-nofpu @gol
742 -m5-compact -m5-compact-nofpu @gol
743 -mb -ml -mdalign -mrelax @gol
744 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
745 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
746 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
747 -mdivsi3_libfunc=@var{name} @gol
748 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
752 @gccoptlist{-mcpu=@var{cpu-type} @gol
753 -mtune=@var{cpu-type} @gol
754 -mcmodel=@var{code-model} @gol
755 -m32 -m64 -mapp-regs -mno-app-regs @gol
756 -mfaster-structs -mno-faster-structs @gol
757 -mfpu -mno-fpu -mhard-float -msoft-float @gol
758 -mhard-quad-float -msoft-quad-float @gol
759 -mimpure-text -mno-impure-text -mlittle-endian @gol
760 -mstack-bias -mno-stack-bias @gol
761 -munaligned-doubles -mno-unaligned-doubles @gol
762 -mv8plus -mno-v8plus -mvis -mno-vis
763 -threads -pthreads -pthread}
766 @gccoptlist{-mwarn-reloc -merror-reloc @gol
767 -msafe-dma -munsafe-dma @gol
769 -msmall-mem -mlarge-mem -mstdmain @gol
770 -mfixed-range=@var{register-range}}
772 @emph{System V Options}
773 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
776 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
777 -mprolog-function -mno-prolog-function -mspace @gol
778 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
779 -mapp-regs -mno-app-regs @gol
780 -mdisable-callt -mno-disable-callt @gol
786 @gccoptlist{-mg -mgnu -munix}
788 @emph{VxWorks Options}
789 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
790 -Xbind-lazy -Xbind-now}
792 @emph{x86-64 Options}
793 See i386 and x86-64 Options.
795 @emph{Xstormy16 Options}
798 @emph{Xtensa Options}
799 @gccoptlist{-mconst16 -mno-const16 @gol
800 -mfused-madd -mno-fused-madd @gol
801 -mtext-section-literals -mno-text-section-literals @gol
802 -mtarget-align -mno-target-align @gol
803 -mlongcalls -mno-longcalls}
805 @emph{zSeries Options}
806 See S/390 and zSeries Options.
808 @item Code Generation Options
809 @xref{Code Gen Options,,Options for Code Generation Conventions}.
810 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
811 -ffixed-@var{reg} -fexceptions @gol
812 -fnon-call-exceptions -funwind-tables @gol
813 -fasynchronous-unwind-tables @gol
814 -finhibit-size-directive -finstrument-functions @gol
815 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
816 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
817 -fno-common -fno-ident @gol
818 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
819 -fno-jump-tables @gol
820 -frecord-gcc-switches @gol
821 -freg-struct-return -fshort-enums @gol
822 -fshort-double -fshort-wchar @gol
823 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
824 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
825 -fno-stack-limit -fargument-alias -fargument-noalias @gol
826 -fargument-noalias-global -fargument-noalias-anything @gol
827 -fleading-underscore -ftls-model=@var{model} @gol
828 -ftrapv -fwrapv -fbounds-check @gol
833 * Overall Options:: Controlling the kind of output:
834 an executable, object files, assembler files,
835 or preprocessed source.
836 * C Dialect Options:: Controlling the variant of C language compiled.
837 * C++ Dialect Options:: Variations on C++.
838 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
840 * Language Independent Options:: Controlling how diagnostics should be
842 * Warning Options:: How picky should the compiler be?
843 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
844 * Optimize Options:: How much optimization?
845 * Preprocessor Options:: Controlling header files and macro definitions.
846 Also, getting dependency information for Make.
847 * Assembler Options:: Passing options to the assembler.
848 * Link Options:: Specifying libraries and so on.
849 * Directory Options:: Where to find header files and libraries.
850 Where to find the compiler executable files.
851 * Spec Files:: How to pass switches to sub-processes.
852 * Target Options:: Running a cross-compiler, or an old version of GCC.
855 @node Overall Options
856 @section Options Controlling the Kind of Output
858 Compilation can involve up to four stages: preprocessing, compilation
859 proper, assembly and linking, always in that order. GCC is capable of
860 preprocessing and compiling several files either into several
861 assembler input files, or into one assembler input file; then each
862 assembler input file produces an object file, and linking combines all
863 the object files (those newly compiled, and those specified as input)
864 into an executable file.
866 @cindex file name suffix
867 For any given input file, the file name suffix determines what kind of
872 C source code which must be preprocessed.
875 C source code which should not be preprocessed.
878 C++ source code which should not be preprocessed.
881 Objective-C source code. Note that you must link with the @file{libobjc}
882 library to make an Objective-C program work.
885 Objective-C source code which should not be preprocessed.
889 Objective-C++ source code. Note that you must link with the @file{libobjc}
890 library to make an Objective-C++ program work. Note that @samp{.M} refers
891 to a literal capital M@.
894 Objective-C++ source code which should not be preprocessed.
897 C, C++, Objective-C or Objective-C++ header file to be turned into a
902 @itemx @var{file}.cxx
903 @itemx @var{file}.cpp
904 @itemx @var{file}.CPP
905 @itemx @var{file}.c++
907 C++ source code which must be preprocessed. Note that in @samp{.cxx},
908 the last two letters must both be literally @samp{x}. Likewise,
909 @samp{.C} refers to a literal capital C@.
913 Objective-C++ source code which must be preprocessed.
916 Objective-C++ source code which should not be preprocessed.
921 @itemx @var{file}.hxx
922 @itemx @var{file}.hpp
923 @itemx @var{file}.HPP
924 @itemx @var{file}.h++
925 @itemx @var{file}.tcc
926 C++ header file to be turned into a precompiled header.
929 @itemx @var{file}.for
930 @itemx @var{file}.FOR
931 Fixed form Fortran source code which should not be preprocessed.
934 @itemx @var{file}.fpp
935 @itemx @var{file}.FPP
936 Fixed form Fortran source code which must be preprocessed (with the traditional
940 @itemx @var{file}.f95
941 Free form Fortran source code which should not be preprocessed.
944 @itemx @var{file}.F95
945 Free form Fortran source code which must be preprocessed (with the
946 traditional preprocessor).
948 @c FIXME: Descriptions of Java file types.
955 Ada source code file which contains a library unit declaration (a
956 declaration of a package, subprogram, or generic, or a generic
957 instantiation), or a library unit renaming declaration (a package,
958 generic, or subprogram renaming declaration). Such files are also
961 @itemx @var{file}.adb
962 Ada source code file containing a library unit body (a subprogram or
963 package body). Such files are also called @dfn{bodies}.
965 @c GCC also knows about some suffixes for languages not yet included:
977 Assembler code which must be preprocessed.
980 An object file to be fed straight into linking.
981 Any file name with no recognized suffix is treated this way.
985 You can specify the input language explicitly with the @option{-x} option:
988 @item -x @var{language}
989 Specify explicitly the @var{language} for the following input files
990 (rather than letting the compiler choose a default based on the file
991 name suffix). This option applies to all following input files until
992 the next @option{-x} option. Possible values for @var{language} are:
994 c c-header c-cpp-output
995 c++ c++-header c++-cpp-output
996 objective-c objective-c-header objective-c-cpp-output
997 objective-c++ objective-c++-header objective-c++-cpp-output
998 assembler assembler-with-cpp
1006 Turn off any specification of a language, so that subsequent files are
1007 handled according to their file name suffixes (as they are if @option{-x}
1008 has not been used at all).
1010 @item -pass-exit-codes
1011 @opindex pass-exit-codes
1012 Normally the @command{gcc} program will exit with the code of 1 if any
1013 phase of the compiler returns a non-success return code. If you specify
1014 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1015 numerically highest error produced by any phase that returned an error
1016 indication. The C, C++, and Fortran frontends return 4, if an internal
1017 compiler error is encountered.
1020 If you only want some of the stages of compilation, you can use
1021 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1022 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1023 @command{gcc} is to stop. Note that some combinations (for example,
1024 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1029 Compile or assemble the source files, but do not link. The linking
1030 stage simply is not done. The ultimate output is in the form of an
1031 object file for each source file.
1033 By default, the object file name for a source file is made by replacing
1034 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1036 Unrecognized input files, not requiring compilation or assembly, are
1041 Stop after the stage of compilation proper; do not assemble. The output
1042 is in the form of an assembler code file for each non-assembler input
1045 By default, the assembler file name for a source file is made by
1046 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1048 Input files that don't require compilation are ignored.
1052 Stop after the preprocessing stage; do not run the compiler proper. The
1053 output is in the form of preprocessed source code, which is sent to the
1056 Input files which don't require preprocessing are ignored.
1058 @cindex output file option
1061 Place output in file @var{file}. This applies regardless to whatever
1062 sort of output is being produced, whether it be an executable file,
1063 an object file, an assembler file or preprocessed C code.
1065 If @option{-o} is not specified, the default is to put an executable
1066 file in @file{a.out}, the object file for
1067 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1068 assembler file in @file{@var{source}.s}, a precompiled header file in
1069 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1074 Print (on standard error output) the commands executed to run the stages
1075 of compilation. Also print the version number of the compiler driver
1076 program and of the preprocessor and the compiler proper.
1080 Like @option{-v} except the commands are not executed and all command
1081 arguments are quoted. This is useful for shell scripts to capture the
1082 driver-generated command lines.
1086 Use pipes rather than temporary files for communication between the
1087 various stages of compilation. This fails to work on some systems where
1088 the assembler is unable to read from a pipe; but the GNU assembler has
1093 If you are compiling multiple source files, this option tells the driver
1094 to pass all the source files to the compiler at once (for those
1095 languages for which the compiler can handle this). This will allow
1096 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1097 language for which this is supported is C@. If you pass source files for
1098 multiple languages to the driver, using this option, the driver will invoke
1099 the compiler(s) that support IMA once each, passing each compiler all the
1100 source files appropriate for it. For those languages that do not support
1101 IMA this option will be ignored, and the compiler will be invoked once for
1102 each source file in that language. If you use this option in conjunction
1103 with @option{-save-temps}, the compiler will generate multiple
1105 (one for each source file), but only one (combined) @file{.o} or
1110 Print (on the standard output) a description of the command line options
1111 understood by @command{gcc}. If the @option{-v} option is also specified
1112 then @option{--help} will also be passed on to the various processes
1113 invoked by @command{gcc}, so that they can display the command line options
1114 they accept. If the @option{-Wextra} option has also been specified
1115 (prior to the @option{--help} option), then command line options which
1116 have no documentation associated with them will also be displayed.
1119 @opindex target-help
1120 Print (on the standard output) a description of target-specific command
1121 line options for each tool. For some targets extra target-specific
1122 information may also be printed.
1124 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1125 Print (on the standard output) a description of the command line
1126 options understood by the compiler that fit into a specific class.
1127 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1128 @samp{params}, or @var{language}:
1131 @item @samp{optimizers}
1132 This will display all of the optimization options supported by the
1135 @item @samp{warnings}
1136 This will display all of the options controlling warning messages
1137 produced by the compiler.
1140 This will display target-specific options. Unlike the
1141 @option{--target-help} option however, target-specific options of the
1142 linker and assembler will not be displayed. This is because those
1143 tools do not currently support the extended @option{--help=} syntax.
1146 This will display the values recognized by the @option{--param}
1149 @item @var{language}
1150 This will display the options supported for @var{language}, where
1151 @var{language} is the name of one of the languages supported in this
1155 This will display the options that are common to all languages.
1158 It is possible to further refine the output of the @option{--help=}
1159 option by adding a comma separated list of qualifiers after the
1160 class. These can be any from the following list:
1163 @item @samp{undocumented}
1164 Display only those options which are undocumented.
1167 Display options which take an argument that appears after an equal
1168 sign in the same continuous piece of text, such as:
1169 @samp{--help=target}.
1171 @item @samp{separate}
1172 Display options which take an argument that appears as a separate word
1173 following the original option, such as: @samp{-o output-file}.
1176 Thus for example to display all the undocumented target-specific
1177 switches supported by the compiler the following can be used:
1180 --help=target,undocumented
1183 The sense of a qualifier can be inverted by prefixing it with the
1184 @var{^} character, so for example to display all binary warning
1185 options (i.e. ones that are either on or off and that do not take an
1186 argument), which have a description the following can be used:
1189 --help=warnings,^joined,^undocumented
1192 A class can also be used as a qualifier, although this usually
1193 restricts the output by so much that there is nothing to display. One
1194 case where it does work however is when one of the classes is
1195 @var{target}. So for example to display all the target-specific
1196 optimization options the following can be used:
1199 --help=target,optimizers
1202 The @option{--help=} option can be repeated on the command line. Each
1203 successive use will display its requested class of options, skipping
1204 those that have already been displayed.
1206 If the @option{-Q} option appears on the command line before the
1207 @option{--help=} option, then the descriptive text displayed by
1208 @option{--help=} is changed. Instead of describing the displayed
1209 options, an indication is given as to whether the option is enabled,
1210 disabled or set to a specific value (assuming that the compiler
1211 knows this at the point where the @option{--help=} option is used).
1213 Here is a truncated example from the ARM port of @command{gcc}:
1216 % gcc -Q -mabi=2 --help=target -c
1217 The following options are target specific:
1219 -mabort-on-noreturn [disabled]
1223 The output is sensitive to the effects of previous command line
1224 options, so for example it is possible to find out which optimizations
1225 are enabled at @option{-O2} by using:
1228 -O2 --help=optimizers
1231 Alternatively you can discover which binary optimizations are enabled
1232 by @option{-O3} by using:
1235 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1236 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1237 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1242 Display the version number and copyrights of the invoked GCC@.
1244 @include @value{srcdir}/../libiberty/at-file.texi
1248 @section Compiling C++ Programs
1250 @cindex suffixes for C++ source
1251 @cindex C++ source file suffixes
1252 C++ source files conventionally use one of the suffixes @samp{.C},
1253 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1254 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1255 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1256 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1257 files with these names and compiles them as C++ programs even if you
1258 call the compiler the same way as for compiling C programs (usually
1259 with the name @command{gcc}).
1263 However, the use of @command{gcc} does not add the C++ library.
1264 @command{g++} is a program that calls GCC and treats @samp{.c},
1265 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1266 files unless @option{-x} is used, and automatically specifies linking
1267 against the C++ library. This program is also useful when
1268 precompiling a C header file with a @samp{.h} extension for use in C++
1269 compilations. On many systems, @command{g++} is also installed with
1270 the name @command{c++}.
1272 @cindex invoking @command{g++}
1273 When you compile C++ programs, you may specify many of the same
1274 command-line options that you use for compiling programs in any
1275 language; or command-line options meaningful for C and related
1276 languages; or options that are meaningful only for C++ programs.
1277 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1278 explanations of options for languages related to C@.
1279 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1280 explanations of options that are meaningful only for C++ programs.
1282 @node C Dialect Options
1283 @section Options Controlling C Dialect
1284 @cindex dialect options
1285 @cindex language dialect options
1286 @cindex options, dialect
1288 The following options control the dialect of C (or languages derived
1289 from C, such as C++, Objective-C and Objective-C++) that the compiler
1293 @cindex ANSI support
1297 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1298 equivalent to @samp{-std=c++98}.
1300 This turns off certain features of GCC that are incompatible with ISO
1301 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1302 such as the @code{asm} and @code{typeof} keywords, and
1303 predefined macros such as @code{unix} and @code{vax} that identify the
1304 type of system you are using. It also enables the undesirable and
1305 rarely used ISO trigraph feature. For the C compiler,
1306 it disables recognition of C++ style @samp{//} comments as well as
1307 the @code{inline} keyword.
1309 The alternate keywords @code{__asm__}, @code{__extension__},
1310 @code{__inline__} and @code{__typeof__} continue to work despite
1311 @option{-ansi}. You would not want to use them in an ISO C program, of
1312 course, but it is useful to put them in header files that might be included
1313 in compilations done with @option{-ansi}. Alternate predefined macros
1314 such as @code{__unix__} and @code{__vax__} are also available, with or
1315 without @option{-ansi}.
1317 The @option{-ansi} option does not cause non-ISO programs to be
1318 rejected gratuitously. For that, @option{-pedantic} is required in
1319 addition to @option{-ansi}. @xref{Warning Options}.
1321 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1322 option is used. Some header files may notice this macro and refrain
1323 from declaring certain functions or defining certain macros that the
1324 ISO standard doesn't call for; this is to avoid interfering with any
1325 programs that might use these names for other things.
1327 Functions that would normally be built in but do not have semantics
1328 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1329 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1330 built-in functions provided by GCC}, for details of the functions
1335 Determine the language standard. @xref{Standards,,Language Standards
1336 Supported by GCC}, for details of these standard versions. This option
1337 is currently only supported when compiling C or C++.
1339 A value for this option must be provided; possible values are
1344 ISO C90 (same as @option{-ansi}).
1346 @item iso9899:199409
1347 ISO C90 as modified in amendment 1.
1353 ISO C99. Note that this standard is not yet fully supported; see
1354 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1355 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1358 Default, ISO C90 plus GNU extensions (including some C99 features).
1362 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1363 this will become the default. The name @samp{gnu9x} is deprecated.
1366 The 1998 ISO C++ standard plus amendments.
1369 The same as @option{-std=c++98} plus GNU extensions. This is the
1370 default for C++ code.
1373 The working draft of the upcoming ISO C++0x standard. This option
1374 enables experimental features that are likely to be included in
1375 C++0x. The working draft is constantly changing, and any feature that is
1376 enabled by this flag may be removed from future versions of GCC if it is
1377 not part of the C++0x standard.
1380 The same as @option{-std=c++0x} plus GNU extensions. As with
1381 @option{-std=c++0x}, this option enables experimental features that may
1382 be removed in future versions of GCC.
1385 Even when this option is not specified, you can still use some of the
1386 features of newer standards in so far as they do not conflict with
1387 previous C standards. For example, you may use @code{__restrict__} even
1388 when @option{-std=c99} is not specified.
1390 The @option{-std} options specifying some version of ISO C have the same
1391 effects as @option{-ansi}, except that features that were not in ISO C90
1392 but are in the specified version (for example, @samp{//} comments and
1393 the @code{inline} keyword in ISO C99) are not disabled.
1395 @item -fgnu89-inline
1396 @opindex fgnu89-inline
1397 The option @option{-fgnu89-inline} tells GCC to use the traditional
1398 GNU semantics for @code{inline} functions when in C99 mode.
1399 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1400 is accepted and ignored by GCC versions 4.1.3 up to but not including
1401 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1402 C99 mode. Using this option is roughly equivalent to adding the
1403 @code{gnu_inline} function attribute to all inline functions
1404 (@pxref{Function Attributes}).
1406 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1407 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1408 specifies the default behavior). This option was first supported in
1409 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1411 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1412 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1413 in effect for @code{inline} functions. @xref{Common Predefined
1414 Macros,,,cpp,The C Preprocessor}.
1416 @item -aux-info @var{filename}
1418 Output to the given filename prototyped declarations for all functions
1419 declared and/or defined in a translation unit, including those in header
1420 files. This option is silently ignored in any language other than C@.
1422 Besides declarations, the file indicates, in comments, the origin of
1423 each declaration (source file and line), whether the declaration was
1424 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1425 @samp{O} for old, respectively, in the first character after the line
1426 number and the colon), and whether it came from a declaration or a
1427 definition (@samp{C} or @samp{F}, respectively, in the following
1428 character). In the case of function definitions, a K&R-style list of
1429 arguments followed by their declarations is also provided, inside
1430 comments, after the declaration.
1434 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1435 keyword, so that code can use these words as identifiers. You can use
1436 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1437 instead. @option{-ansi} implies @option{-fno-asm}.
1439 In C++, this switch only affects the @code{typeof} keyword, since
1440 @code{asm} and @code{inline} are standard keywords. You may want to
1441 use the @option{-fno-gnu-keywords} flag instead, which has the same
1442 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1443 switch only affects the @code{asm} and @code{typeof} keywords, since
1444 @code{inline} is a standard keyword in ISO C99.
1447 @itemx -fno-builtin-@var{function}
1448 @opindex fno-builtin
1449 @cindex built-in functions
1450 Don't recognize built-in functions that do not begin with
1451 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1452 functions provided by GCC}, for details of the functions affected,
1453 including those which are not built-in functions when @option{-ansi} or
1454 @option{-std} options for strict ISO C conformance are used because they
1455 do not have an ISO standard meaning.
1457 GCC normally generates special code to handle certain built-in functions
1458 more efficiently; for instance, calls to @code{alloca} may become single
1459 instructions that adjust the stack directly, and calls to @code{memcpy}
1460 may become inline copy loops. The resulting code is often both smaller
1461 and faster, but since the function calls no longer appear as such, you
1462 cannot set a breakpoint on those calls, nor can you change the behavior
1463 of the functions by linking with a different library. In addition,
1464 when a function is recognized as a built-in function, GCC may use
1465 information about that function to warn about problems with calls to
1466 that function, or to generate more efficient code, even if the
1467 resulting code still contains calls to that function. For example,
1468 warnings are given with @option{-Wformat} for bad calls to
1469 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1470 known not to modify global memory.
1472 With the @option{-fno-builtin-@var{function}} option
1473 only the built-in function @var{function} is
1474 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1475 function is named this is not built-in in this version of GCC, this
1476 option is ignored. There is no corresponding
1477 @option{-fbuiltin-@var{function}} option; if you wish to enable
1478 built-in functions selectively when using @option{-fno-builtin} or
1479 @option{-ffreestanding}, you may define macros such as:
1482 #define abs(n) __builtin_abs ((n))
1483 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1488 @cindex hosted environment
1490 Assert that compilation takes place in a hosted environment. This implies
1491 @option{-fbuiltin}. A hosted environment is one in which the
1492 entire standard library is available, and in which @code{main} has a return
1493 type of @code{int}. Examples are nearly everything except a kernel.
1494 This is equivalent to @option{-fno-freestanding}.
1496 @item -ffreestanding
1497 @opindex ffreestanding
1498 @cindex hosted environment
1500 Assert that compilation takes place in a freestanding environment. This
1501 implies @option{-fno-builtin}. A freestanding environment
1502 is one in which the standard library may not exist, and program startup may
1503 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1504 This is equivalent to @option{-fno-hosted}.
1506 @xref{Standards,,Language Standards Supported by GCC}, for details of
1507 freestanding and hosted environments.
1511 @cindex openmp parallel
1512 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1513 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1514 compiler generates parallel code according to the OpenMP Application
1515 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1516 implies @option{-pthread}, and thus is only supported on targets that
1517 have support for @option{-pthread}.
1519 @item -fms-extensions
1520 @opindex fms-extensions
1521 Accept some non-standard constructs used in Microsoft header files.
1523 Some cases of unnamed fields in structures and unions are only
1524 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1525 fields within structs/unions}, for details.
1529 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1530 options for strict ISO C conformance) implies @option{-trigraphs}.
1532 @item -no-integrated-cpp
1533 @opindex no-integrated-cpp
1534 Performs a compilation in two passes: preprocessing and compiling. This
1535 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1536 @option{-B} option. The user supplied compilation step can then add in
1537 an additional preprocessing step after normal preprocessing but before
1538 compiling. The default is to use the integrated cpp (internal cpp)
1540 The semantics of this option will change if "cc1", "cc1plus", and
1541 "cc1obj" are merged.
1543 @cindex traditional C language
1544 @cindex C language, traditional
1546 @itemx -traditional-cpp
1547 @opindex traditional-cpp
1548 @opindex traditional
1549 Formerly, these options caused GCC to attempt to emulate a pre-standard
1550 C compiler. They are now only supported with the @option{-E} switch.
1551 The preprocessor continues to support a pre-standard mode. See the GNU
1552 CPP manual for details.
1554 @item -fcond-mismatch
1555 @opindex fcond-mismatch
1556 Allow conditional expressions with mismatched types in the second and
1557 third arguments. The value of such an expression is void. This option
1558 is not supported for C++.
1560 @item -flax-vector-conversions
1561 @opindex flax-vector-conversions
1562 Allow implicit conversions between vectors with differing numbers of
1563 elements and/or incompatible element types. This option should not be
1566 @item -funsigned-char
1567 @opindex funsigned-char
1568 Let the type @code{char} be unsigned, like @code{unsigned char}.
1570 Each kind of machine has a default for what @code{char} should
1571 be. It is either like @code{unsigned char} by default or like
1572 @code{signed char} by default.
1574 Ideally, a portable program should always use @code{signed char} or
1575 @code{unsigned char} when it depends on the signedness of an object.
1576 But many programs have been written to use plain @code{char} and
1577 expect it to be signed, or expect it to be unsigned, depending on the
1578 machines they were written for. This option, and its inverse, let you
1579 make such a program work with the opposite default.
1581 The type @code{char} is always a distinct type from each of
1582 @code{signed char} or @code{unsigned char}, even though its behavior
1583 is always just like one of those two.
1586 @opindex fsigned-char
1587 Let the type @code{char} be signed, like @code{signed char}.
1589 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1590 the negative form of @option{-funsigned-char}. Likewise, the option
1591 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1593 @item -fsigned-bitfields
1594 @itemx -funsigned-bitfields
1595 @itemx -fno-signed-bitfields
1596 @itemx -fno-unsigned-bitfields
1597 @opindex fsigned-bitfields
1598 @opindex funsigned-bitfields
1599 @opindex fno-signed-bitfields
1600 @opindex fno-unsigned-bitfields
1601 These options control whether a bit-field is signed or unsigned, when the
1602 declaration does not use either @code{signed} or @code{unsigned}. By
1603 default, such a bit-field is signed, because this is consistent: the
1604 basic integer types such as @code{int} are signed types.
1607 @node C++ Dialect Options
1608 @section Options Controlling C++ Dialect
1610 @cindex compiler options, C++
1611 @cindex C++ options, command line
1612 @cindex options, C++
1613 This section describes the command-line options that are only meaningful
1614 for C++ programs; but you can also use most of the GNU compiler options
1615 regardless of what language your program is in. For example, you
1616 might compile a file @code{firstClass.C} like this:
1619 g++ -g -frepo -O -c firstClass.C
1623 In this example, only @option{-frepo} is an option meant
1624 only for C++ programs; you can use the other options with any
1625 language supported by GCC@.
1627 Here is a list of options that are @emph{only} for compiling C++ programs:
1631 @item -fabi-version=@var{n}
1632 @opindex fabi-version
1633 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1634 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1635 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1636 the version that conforms most closely to the C++ ABI specification.
1637 Therefore, the ABI obtained using version 0 will change as ABI bugs
1640 The default is version 2.
1642 @item -fno-access-control
1643 @opindex fno-access-control
1644 Turn off all access checking. This switch is mainly useful for working
1645 around bugs in the access control code.
1649 Check that the pointer returned by @code{operator new} is non-null
1650 before attempting to modify the storage allocated. This check is
1651 normally unnecessary because the C++ standard specifies that
1652 @code{operator new} will only return @code{0} if it is declared
1653 @samp{throw()}, in which case the compiler will always check the
1654 return value even without this option. In all other cases, when
1655 @code{operator new} has a non-empty exception specification, memory
1656 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1657 @samp{new (nothrow)}.
1659 @item -fconserve-space
1660 @opindex fconserve-space
1661 Put uninitialized or runtime-initialized global variables into the
1662 common segment, as C does. This saves space in the executable at the
1663 cost of not diagnosing duplicate definitions. If you compile with this
1664 flag and your program mysteriously crashes after @code{main()} has
1665 completed, you may have an object that is being destroyed twice because
1666 two definitions were merged.
1668 This option is no longer useful on most targets, now that support has
1669 been added for putting variables into BSS without making them common.
1671 @item -ffriend-injection
1672 @opindex ffriend-injection
1673 Inject friend functions into the enclosing namespace, so that they are
1674 visible outside the scope of the class in which they are declared.
1675 Friend functions were documented to work this way in the old Annotated
1676 C++ Reference Manual, and versions of G++ before 4.1 always worked
1677 that way. However, in ISO C++ a friend function which is not declared
1678 in an enclosing scope can only be found using argument dependent
1679 lookup. This option causes friends to be injected as they were in
1682 This option is for compatibility, and may be removed in a future
1685 @item -fno-elide-constructors
1686 @opindex fno-elide-constructors
1687 The C++ standard allows an implementation to omit creating a temporary
1688 which is only used to initialize another object of the same type.
1689 Specifying this option disables that optimization, and forces G++ to
1690 call the copy constructor in all cases.
1692 @item -fno-enforce-eh-specs
1693 @opindex fno-enforce-eh-specs
1694 Don't generate code to check for violation of exception specifications
1695 at runtime. This option violates the C++ standard, but may be useful
1696 for reducing code size in production builds, much like defining
1697 @samp{NDEBUG}. This does not give user code permission to throw
1698 exceptions in violation of the exception specifications; the compiler
1699 will still optimize based on the specifications, so throwing an
1700 unexpected exception will result in undefined behavior.
1703 @itemx -fno-for-scope
1705 @opindex fno-for-scope
1706 If @option{-ffor-scope} is specified, the scope of variables declared in
1707 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1708 as specified by the C++ standard.
1709 If @option{-fno-for-scope} is specified, the scope of variables declared in
1710 a @i{for-init-statement} extends to the end of the enclosing scope,
1711 as was the case in old versions of G++, and other (traditional)
1712 implementations of C++.
1714 The default if neither flag is given to follow the standard,
1715 but to allow and give a warning for old-style code that would
1716 otherwise be invalid, or have different behavior.
1718 @item -fno-gnu-keywords
1719 @opindex fno-gnu-keywords
1720 Do not recognize @code{typeof} as a keyword, so that code can use this
1721 word as an identifier. You can use the keyword @code{__typeof__} instead.
1722 @option{-ansi} implies @option{-fno-gnu-keywords}.
1724 @item -fno-implicit-templates
1725 @opindex fno-implicit-templates
1726 Never emit code for non-inline templates which are instantiated
1727 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1728 @xref{Template Instantiation}, for more information.
1730 @item -fno-implicit-inline-templates
1731 @opindex fno-implicit-inline-templates
1732 Don't emit code for implicit instantiations of inline templates, either.
1733 The default is to handle inlines differently so that compiles with and
1734 without optimization will need the same set of explicit instantiations.
1736 @item -fno-implement-inlines
1737 @opindex fno-implement-inlines
1738 To save space, do not emit out-of-line copies of inline functions
1739 controlled by @samp{#pragma implementation}. This will cause linker
1740 errors if these functions are not inlined everywhere they are called.
1742 @item -fms-extensions
1743 @opindex fms-extensions
1744 Disable pedantic warnings about constructs used in MFC, such as implicit
1745 int and getting a pointer to member function via non-standard syntax.
1747 @item -fno-nonansi-builtins
1748 @opindex fno-nonansi-builtins
1749 Disable built-in declarations of functions that are not mandated by
1750 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1751 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1753 @item -fno-operator-names
1754 @opindex fno-operator-names
1755 Do not treat the operator name keywords @code{and}, @code{bitand},
1756 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1757 synonyms as keywords.
1759 @item -fno-optional-diags
1760 @opindex fno-optional-diags
1761 Disable diagnostics that the standard says a compiler does not need to
1762 issue. Currently, the only such diagnostic issued by G++ is the one for
1763 a name having multiple meanings within a class.
1766 @opindex fpermissive
1767 Downgrade some diagnostics about nonconformant code from errors to
1768 warnings. Thus, using @option{-fpermissive} will allow some
1769 nonconforming code to compile.
1773 Enable automatic template instantiation at link time. This option also
1774 implies @option{-fno-implicit-templates}. @xref{Template
1775 Instantiation}, for more information.
1779 Disable generation of information about every class with virtual
1780 functions for use by the C++ runtime type identification features
1781 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1782 of the language, you can save some space by using this flag. Note that
1783 exception handling uses the same information, but it will generate it as
1784 needed. The @samp{dynamic_cast} operator can still be used for casts that
1785 do not require runtime type information, i.e. casts to @code{void *} or to
1786 unambiguous base classes.
1790 Emit statistics about front-end processing at the end of the compilation.
1791 This information is generally only useful to the G++ development team.
1793 @item -ftemplate-depth-@var{n}
1794 @opindex ftemplate-depth
1795 Set the maximum instantiation depth for template classes to @var{n}.
1796 A limit on the template instantiation depth is needed to detect
1797 endless recursions during template class instantiation. ANSI/ISO C++
1798 conforming programs must not rely on a maximum depth greater than 17.
1800 @item -fno-threadsafe-statics
1801 @opindex fno-threadsafe-statics
1802 Do not emit the extra code to use the routines specified in the C++
1803 ABI for thread-safe initialization of local statics. You can use this
1804 option to reduce code size slightly in code that doesn't need to be
1807 @item -fuse-cxa-atexit
1808 @opindex fuse-cxa-atexit
1809 Register destructors for objects with static storage duration with the
1810 @code{__cxa_atexit} function rather than the @code{atexit} function.
1811 This option is required for fully standards-compliant handling of static
1812 destructors, but will only work if your C library supports
1813 @code{__cxa_atexit}.
1815 @item -fno-use-cxa-get-exception-ptr
1816 @opindex fno-use-cxa-get-exception-ptr
1817 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1818 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1819 if the runtime routine is not available.
1821 @item -fvisibility-inlines-hidden
1822 @opindex fvisibility-inlines-hidden
1823 This switch declares that the user does not attempt to compare
1824 pointers to inline methods where the addresses of the two functions
1825 were taken in different shared objects.
1827 The effect of this is that GCC may, effectively, mark inline methods with
1828 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1829 appear in the export table of a DSO and do not require a PLT indirection
1830 when used within the DSO@. Enabling this option can have a dramatic effect
1831 on load and link times of a DSO as it massively reduces the size of the
1832 dynamic export table when the library makes heavy use of templates.
1834 The behavior of this switch is not quite the same as marking the
1835 methods as hidden directly, because it does not affect static variables
1836 local to the function or cause the compiler to deduce that
1837 the function is defined in only one shared object.
1839 You may mark a method as having a visibility explicitly to negate the
1840 effect of the switch for that method. For example, if you do want to
1841 compare pointers to a particular inline method, you might mark it as
1842 having default visibility. Marking the enclosing class with explicit
1843 visibility will have no effect.
1845 Explicitly instantiated inline methods are unaffected by this option
1846 as their linkage might otherwise cross a shared library boundary.
1847 @xref{Template Instantiation}.
1849 @item -fvisibility-ms-compat
1850 @opindex fvisibility-ms-compat
1851 This flag attempts to use visibility settings to make GCC's C++
1852 linkage model compatible with that of Microsoft Visual Studio.
1854 The flag makes these changes to GCC's linkage model:
1858 It sets the default visibility to @code{hidden}, like
1859 @option{-fvisibility=hidden}.
1862 Types, but not their members, are not hidden by default.
1865 The One Definition Rule is relaxed for types without explicit
1866 visibility specifications which are defined in more than one different
1867 shared object: those declarations are permitted if they would have
1868 been permitted when this option was not used.
1871 In new code it is better to use @option{-fvisibility=hidden} and
1872 export those classes which are intended to be externally visible.
1873 Unfortunately it is possible for code to rely, perhaps accidentally,
1874 on the Visual Studio behavior.
1876 Among the consequences of these changes are that static data members
1877 of the same type with the same name but defined in different shared
1878 objects will be different, so changing one will not change the other;
1879 and that pointers to function members defined in different shared
1880 objects may not compare equal. When this flag is given, it is a
1881 violation of the ODR to define types with the same name differently.
1885 Do not use weak symbol support, even if it is provided by the linker.
1886 By default, G++ will use weak symbols if they are available. This
1887 option exists only for testing, and should not be used by end-users;
1888 it will result in inferior code and has no benefits. This option may
1889 be removed in a future release of G++.
1893 Do not search for header files in the standard directories specific to
1894 C++, but do still search the other standard directories. (This option
1895 is used when building the C++ library.)
1898 In addition, these optimization, warning, and code generation options
1899 have meanings only for C++ programs:
1902 @item -fno-default-inline
1903 @opindex fno-default-inline
1904 Do not assume @samp{inline} for functions defined inside a class scope.
1905 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1906 functions will have linkage like inline functions; they just won't be
1909 @item -Wabi @r{(C++ only)}
1911 Warn when G++ generates code that is probably not compatible with the
1912 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1913 all such cases, there are probably some cases that are not warned about,
1914 even though G++ is generating incompatible code. There may also be
1915 cases where warnings are emitted even though the code that is generated
1918 You should rewrite your code to avoid these warnings if you are
1919 concerned about the fact that code generated by G++ may not be binary
1920 compatible with code generated by other compilers.
1922 The known incompatibilities at this point include:
1927 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1928 pack data into the same byte as a base class. For example:
1931 struct A @{ virtual void f(); int f1 : 1; @};
1932 struct B : public A @{ int f2 : 1; @};
1936 In this case, G++ will place @code{B::f2} into the same byte
1937 as@code{A::f1}; other compilers will not. You can avoid this problem
1938 by explicitly padding @code{A} so that its size is a multiple of the
1939 byte size on your platform; that will cause G++ and other compilers to
1940 layout @code{B} identically.
1943 Incorrect handling of tail-padding for virtual bases. G++ does not use
1944 tail padding when laying out virtual bases. For example:
1947 struct A @{ virtual void f(); char c1; @};
1948 struct B @{ B(); char c2; @};
1949 struct C : public A, public virtual B @{@};
1953 In this case, G++ will not place @code{B} into the tail-padding for
1954 @code{A}; other compilers will. You can avoid this problem by
1955 explicitly padding @code{A} so that its size is a multiple of its
1956 alignment (ignoring virtual base classes); that will cause G++ and other
1957 compilers to layout @code{C} identically.
1960 Incorrect handling of bit-fields with declared widths greater than that
1961 of their underlying types, when the bit-fields appear in a union. For
1965 union U @{ int i : 4096; @};
1969 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1970 union too small by the number of bits in an @code{int}.
1973 Empty classes can be placed at incorrect offsets. For example:
1983 struct C : public B, public A @{@};
1987 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1988 it should be placed at offset zero. G++ mistakenly believes that the
1989 @code{A} data member of @code{B} is already at offset zero.
1992 Names of template functions whose types involve @code{typename} or
1993 template template parameters can be mangled incorrectly.
1996 template <typename Q>
1997 void f(typename Q::X) @{@}
1999 template <template <typename> class Q>
2000 void f(typename Q<int>::X) @{@}
2004 Instantiations of these templates may be mangled incorrectly.
2008 @item -Wctor-dtor-privacy @r{(C++ only)}
2009 @opindex Wctor-dtor-privacy
2010 Warn when a class seems unusable because all the constructors or
2011 destructors in that class are private, and it has neither friends nor
2012 public static member functions.
2014 @item -Wnon-virtual-dtor @r{(C++ only)}
2015 @opindex Wnon-virtual-dtor
2016 Warn when a class has virtual functions and accessible non-virtual
2017 destructor, in which case it would be possible but unsafe to delete
2018 an instance of a derived class through a pointer to the base class.
2019 This warning is also enabled if -Weffc++ is specified.
2021 @item -Wreorder @r{(C++ only)}
2023 @cindex reordering, warning
2024 @cindex warning for reordering of member initializers
2025 Warn when the order of member initializers given in the code does not
2026 match the order in which they must be executed. For instance:
2032 A(): j (0), i (1) @{ @}
2036 The compiler will rearrange the member initializers for @samp{i}
2037 and @samp{j} to match the declaration order of the members, emitting
2038 a warning to that effect. This warning is enabled by @option{-Wall}.
2041 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2044 @item -Weffc++ @r{(C++ only)}
2046 Warn about violations of the following style guidelines from Scott Meyers'
2047 @cite{Effective C++} book:
2051 Item 11: Define a copy constructor and an assignment operator for classes
2052 with dynamically allocated memory.
2055 Item 12: Prefer initialization to assignment in constructors.
2058 Item 14: Make destructors virtual in base classes.
2061 Item 15: Have @code{operator=} return a reference to @code{*this}.
2064 Item 23: Don't try to return a reference when you must return an object.
2068 Also warn about violations of the following style guidelines from
2069 Scott Meyers' @cite{More Effective C++} book:
2073 Item 6: Distinguish between prefix and postfix forms of increment and
2074 decrement operators.
2077 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2081 When selecting this option, be aware that the standard library
2082 headers do not obey all of these guidelines; use @samp{grep -v}
2083 to filter out those warnings.
2085 @item -Wno-deprecated @r{(C++ only)}
2086 @opindex Wno-deprecated
2087 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2089 @item -Wstrict-null-sentinel @r{(C++ only)}
2090 @opindex Wstrict-null-sentinel
2091 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2092 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2093 to @code{__null}. Although it is a null pointer constant not a null pointer,
2094 it is guaranteed to of the same size as a pointer. But this use is
2095 not portable across different compilers.
2097 @item -Wno-non-template-friend @r{(C++ only)}
2098 @opindex Wno-non-template-friend
2099 Disable warnings when non-templatized friend functions are declared
2100 within a template. Since the advent of explicit template specification
2101 support in G++, if the name of the friend is an unqualified-id (i.e.,
2102 @samp{friend foo(int)}), the C++ language specification demands that the
2103 friend declare or define an ordinary, nontemplate function. (Section
2104 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2105 could be interpreted as a particular specialization of a templatized
2106 function. Because this non-conforming behavior is no longer the default
2107 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2108 check existing code for potential trouble spots and is on by default.
2109 This new compiler behavior can be turned off with
2110 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2111 but disables the helpful warning.
2113 @item -Wold-style-cast @r{(C++ only)}
2114 @opindex Wold-style-cast
2115 Warn if an old-style (C-style) cast to a non-void type is used within
2116 a C++ program. The new-style casts (@samp{dynamic_cast},
2117 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2118 less vulnerable to unintended effects and much easier to search for.
2120 @item -Woverloaded-virtual @r{(C++ only)}
2121 @opindex Woverloaded-virtual
2122 @cindex overloaded virtual fn, warning
2123 @cindex warning for overloaded virtual fn
2124 Warn when a function declaration hides virtual functions from a
2125 base class. For example, in:
2132 struct B: public A @{
2137 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2145 will fail to compile.
2147 @item -Wno-pmf-conversions @r{(C++ only)}
2148 @opindex Wno-pmf-conversions
2149 Disable the diagnostic for converting a bound pointer to member function
2152 @item -Wsign-promo @r{(C++ only)}
2153 @opindex Wsign-promo
2154 Warn when overload resolution chooses a promotion from unsigned or
2155 enumerated type to a signed type, over a conversion to an unsigned type of
2156 the same size. Previous versions of G++ would try to preserve
2157 unsignedness, but the standard mandates the current behavior.
2162 A& operator = (int);
2172 In this example, G++ will synthesize a default @samp{A& operator =
2173 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2176 @node Objective-C and Objective-C++ Dialect Options
2177 @section Options Controlling Objective-C and Objective-C++ Dialects
2179 @cindex compiler options, Objective-C and Objective-C++
2180 @cindex Objective-C and Objective-C++ options, command line
2181 @cindex options, Objective-C and Objective-C++
2182 (NOTE: This manual does not describe the Objective-C and Objective-C++
2183 languages themselves. See @xref{Standards,,Language Standards
2184 Supported by GCC}, for references.)
2186 This section describes the command-line options that are only meaningful
2187 for Objective-C and Objective-C++ programs, but you can also use most of
2188 the language-independent GNU compiler options.
2189 For example, you might compile a file @code{some_class.m} like this:
2192 gcc -g -fgnu-runtime -O -c some_class.m
2196 In this example, @option{-fgnu-runtime} is an option meant only for
2197 Objective-C and Objective-C++ programs; you can use the other options with
2198 any language supported by GCC@.
2200 Note that since Objective-C is an extension of the C language, Objective-C
2201 compilations may also use options specific to the C front-end (e.g.,
2202 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2203 C++-specific options (e.g., @option{-Wabi}).
2205 Here is a list of options that are @emph{only} for compiling Objective-C
2206 and Objective-C++ programs:
2209 @item -fconstant-string-class=@var{class-name}
2210 @opindex fconstant-string-class
2211 Use @var{class-name} as the name of the class to instantiate for each
2212 literal string specified with the syntax @code{@@"@dots{}"}. The default
2213 class name is @code{NXConstantString} if the GNU runtime is being used, and
2214 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2215 @option{-fconstant-cfstrings} option, if also present, will override the
2216 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2217 to be laid out as constant CoreFoundation strings.
2220 @opindex fgnu-runtime
2221 Generate object code compatible with the standard GNU Objective-C
2222 runtime. This is the default for most types of systems.
2224 @item -fnext-runtime
2225 @opindex fnext-runtime
2226 Generate output compatible with the NeXT runtime. This is the default
2227 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2228 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2231 @item -fno-nil-receivers
2232 @opindex fno-nil-receivers
2233 Assume that all Objective-C message dispatches (e.g.,
2234 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2235 is not @code{nil}. This allows for more efficient entry points in the runtime
2236 to be used. Currently, this option is only available in conjunction with
2237 the NeXT runtime on Mac OS X 10.3 and later.
2239 @item -fobjc-call-cxx-cdtors
2240 @opindex fobjc-call-cxx-cdtors
2241 For each Objective-C class, check if any of its instance variables is a
2242 C++ object with a non-trivial default constructor. If so, synthesize a
2243 special @code{- (id) .cxx_construct} instance method that will run
2244 non-trivial default constructors on any such instance variables, in order,
2245 and then return @code{self}. Similarly, check if any instance variable
2246 is a C++ object with a non-trivial destructor, and if so, synthesize a
2247 special @code{- (void) .cxx_destruct} method that will run
2248 all such default destructors, in reverse order.
2250 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2251 thusly generated will only operate on instance variables declared in the
2252 current Objective-C class, and not those inherited from superclasses. It
2253 is the responsibility of the Objective-C runtime to invoke all such methods
2254 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2255 will be invoked by the runtime immediately after a new object
2256 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2257 be invoked immediately before the runtime deallocates an object instance.
2259 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2260 support for invoking the @code{- (id) .cxx_construct} and
2261 @code{- (void) .cxx_destruct} methods.
2263 @item -fobjc-direct-dispatch
2264 @opindex fobjc-direct-dispatch
2265 Allow fast jumps to the message dispatcher. On Darwin this is
2266 accomplished via the comm page.
2268 @item -fobjc-exceptions
2269 @opindex fobjc-exceptions
2270 Enable syntactic support for structured exception handling in Objective-C,
2271 similar to what is offered by C++ and Java. This option is
2272 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2281 @@catch (AnObjCClass *exc) @{
2288 @@catch (AnotherClass *exc) @{
2291 @@catch (id allOthers) @{
2301 The @code{@@throw} statement may appear anywhere in an Objective-C or
2302 Objective-C++ program; when used inside of a @code{@@catch} block, the
2303 @code{@@throw} may appear without an argument (as shown above), in which case
2304 the object caught by the @code{@@catch} will be rethrown.
2306 Note that only (pointers to) Objective-C objects may be thrown and
2307 caught using this scheme. When an object is thrown, it will be caught
2308 by the nearest @code{@@catch} clause capable of handling objects of that type,
2309 analogously to how @code{catch} blocks work in C++ and Java. A
2310 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2311 any and all Objective-C exceptions not caught by previous @code{@@catch}
2314 The @code{@@finally} clause, if present, will be executed upon exit from the
2315 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2316 regardless of whether any exceptions are thrown, caught or rethrown
2317 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2318 of the @code{finally} clause in Java.
2320 There are several caveats to using the new exception mechanism:
2324 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2325 idioms provided by the @code{NSException} class, the new
2326 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2327 systems, due to additional functionality needed in the (NeXT) Objective-C
2331 As mentioned above, the new exceptions do not support handling
2332 types other than Objective-C objects. Furthermore, when used from
2333 Objective-C++, the Objective-C exception model does not interoperate with C++
2334 exceptions at this time. This means you cannot @code{@@throw} an exception
2335 from Objective-C and @code{catch} it in C++, or vice versa
2336 (i.e., @code{throw @dots{} @@catch}).
2339 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2340 blocks for thread-safe execution:
2343 @@synchronized (ObjCClass *guard) @{
2348 Upon entering the @code{@@synchronized} block, a thread of execution shall
2349 first check whether a lock has been placed on the corresponding @code{guard}
2350 object by another thread. If it has, the current thread shall wait until
2351 the other thread relinquishes its lock. Once @code{guard} becomes available,
2352 the current thread will place its own lock on it, execute the code contained in
2353 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2354 making @code{guard} available to other threads).
2356 Unlike Java, Objective-C does not allow for entire methods to be marked
2357 @code{@@synchronized}. Note that throwing exceptions out of
2358 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2359 to be unlocked properly.
2363 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2365 @item -freplace-objc-classes
2366 @opindex freplace-objc-classes
2367 Emit a special marker instructing @command{ld(1)} not to statically link in
2368 the resulting object file, and allow @command{dyld(1)} to load it in at
2369 run time instead. This is used in conjunction with the Fix-and-Continue
2370 debugging mode, where the object file in question may be recompiled and
2371 dynamically reloaded in the course of program execution, without the need
2372 to restart the program itself. Currently, Fix-and-Continue functionality
2373 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2378 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2379 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2380 compile time) with static class references that get initialized at load time,
2381 which improves run-time performance. Specifying the @option{-fzero-link} flag
2382 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2383 to be retained. This is useful in Zero-Link debugging mode, since it allows
2384 for individual class implementations to be modified during program execution.
2388 Dump interface declarations for all classes seen in the source file to a
2389 file named @file{@var{sourcename}.decl}.
2391 @item -Wassign-intercept
2392 @opindex Wassign-intercept
2393 Warn whenever an Objective-C assignment is being intercepted by the
2397 @opindex Wno-protocol
2398 If a class is declared to implement a protocol, a warning is issued for
2399 every method in the protocol that is not implemented by the class. The
2400 default behavior is to issue a warning for every method not explicitly
2401 implemented in the class, even if a method implementation is inherited
2402 from the superclass. If you use the @option{-Wno-protocol} option, then
2403 methods inherited from the superclass are considered to be implemented,
2404 and no warning is issued for them.
2408 Warn if multiple methods of different types for the same selector are
2409 found during compilation. The check is performed on the list of methods
2410 in the final stage of compilation. Additionally, a check is performed
2411 for each selector appearing in a @code{@@selector(@dots{})}
2412 expression, and a corresponding method for that selector has been found
2413 during compilation. Because these checks scan the method table only at
2414 the end of compilation, these warnings are not produced if the final
2415 stage of compilation is not reached, for example because an error is
2416 found during compilation, or because the @option{-fsyntax-only} option is
2419 @item -Wstrict-selector-match
2420 @opindex Wstrict-selector-match
2421 Warn if multiple methods with differing argument and/or return types are
2422 found for a given selector when attempting to send a message using this
2423 selector to a receiver of type @code{id} or @code{Class}. When this flag
2424 is off (which is the default behavior), the compiler will omit such warnings
2425 if any differences found are confined to types which share the same size
2428 @item -Wundeclared-selector
2429 @opindex Wundeclared-selector
2430 Warn if a @code{@@selector(@dots{})} expression referring to an
2431 undeclared selector is found. A selector is considered undeclared if no
2432 method with that name has been declared before the
2433 @code{@@selector(@dots{})} expression, either explicitly in an
2434 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2435 an @code{@@implementation} section. This option always performs its
2436 checks as soon as a @code{@@selector(@dots{})} expression is found,
2437 while @option{-Wselector} only performs its checks in the final stage of
2438 compilation. This also enforces the coding style convention
2439 that methods and selectors must be declared before being used.
2441 @item -print-objc-runtime-info
2442 @opindex print-objc-runtime-info
2443 Generate C header describing the largest structure that is passed by
2448 @node Language Independent Options
2449 @section Options to Control Diagnostic Messages Formatting
2450 @cindex options to control diagnostics formatting
2451 @cindex diagnostic messages
2452 @cindex message formatting
2454 Traditionally, diagnostic messages have been formatted irrespective of
2455 the output device's aspect (e.g.@: its width, @dots{}). The options described
2456 below can be used to control the diagnostic messages formatting
2457 algorithm, e.g.@: how many characters per line, how often source location
2458 information should be reported. Right now, only the C++ front end can
2459 honor these options. However it is expected, in the near future, that
2460 the remaining front ends would be able to digest them correctly.
2463 @item -fmessage-length=@var{n}
2464 @opindex fmessage-length
2465 Try to format error messages so that they fit on lines of about @var{n}
2466 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2467 the front ends supported by GCC@. If @var{n} is zero, then no
2468 line-wrapping will be done; each error message will appear on a single
2471 @opindex fdiagnostics-show-location
2472 @item -fdiagnostics-show-location=once
2473 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2474 reporter to emit @emph{once} source location information; that is, in
2475 case the message is too long to fit on a single physical line and has to
2476 be wrapped, the source location won't be emitted (as prefix) again,
2477 over and over, in subsequent continuation lines. This is the default
2480 @item -fdiagnostics-show-location=every-line
2481 Only meaningful in line-wrapping mode. Instructs the diagnostic
2482 messages reporter to emit the same source location information (as
2483 prefix) for physical lines that result from the process of breaking
2484 a message which is too long to fit on a single line.
2486 @item -fdiagnostics-show-option
2487 @opindex fdiagnostics-show-option
2488 This option instructs the diagnostic machinery to add text to each
2489 diagnostic emitted, which indicates which command line option directly
2490 controls that diagnostic, when such an option is known to the
2491 diagnostic machinery.
2493 @item -Wcoverage-mismatch
2494 @opindex Wcoverage-mismatch
2495 Warn if feedback profiles do not match when using the
2496 @option{-fprofile-use} option.
2497 If a source file was changed between @option{-fprofile-gen} and
2498 @option{-fprofile-use}, the files with the profile feedback can fail
2499 to match the source file and GCC can not use the profile feedback
2500 information. By default, GCC emits an error message in this case.
2501 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2502 error. GCC does not use appropriate feedback profiles, so using this
2503 option can result in poorly optimized code. This option is useful
2504 only in the case of very minor changes such as bug fixes to an
2509 @node Warning Options
2510 @section Options to Request or Suppress Warnings
2511 @cindex options to control warnings
2512 @cindex warning messages
2513 @cindex messages, warning
2514 @cindex suppressing warnings
2516 Warnings are diagnostic messages that report constructions which
2517 are not inherently erroneous but which are risky or suggest there
2518 may have been an error.
2520 The following language-independent options do not enable specific
2521 warnings but control the kinds of diagnostics produced by GCC.
2524 @cindex syntax checking
2526 @opindex fsyntax-only
2527 Check the code for syntax errors, but don't do anything beyond that.
2531 Inhibit all warning messages.
2535 Make all warnings into errors.
2539 Make the specified warning into an error. The specifier for a warning
2540 is appended, for example @option{-Werror=switch} turns the warnings
2541 controlled by @option{-Wswitch} into errors. This switch takes a
2542 negative form, to be used to negate @option{-Werror} for specific
2543 warnings, for example @option{-Wno-error=switch} makes
2544 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2545 is in effect. You can use the @option{-fdiagnostics-show-option}
2546 option to have each controllable warning amended with the option which
2547 controls it, to determine what to use with this option.
2549 Note that specifying @option{-Werror=}@var{foo} automatically implies
2550 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2553 @item -Wfatal-errors
2554 @opindex Wfatal-errors
2555 This option causes the compiler to abort compilation on the first error
2556 occurred rather than trying to keep going and printing further error
2561 You can request many specific warnings with options beginning
2562 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2563 implicit declarations. Each of these specific warning options also
2564 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2565 example, @option{-Wno-implicit}. This manual lists only one of the
2566 two forms, whichever is not the default. For further,
2567 language-specific options also refer to @ref{C++ Dialect Options} and
2568 @ref{Objective-C and Objective-C++ Dialect Options}.
2573 Issue all the warnings demanded by strict ISO C and ISO C++;
2574 reject all programs that use forbidden extensions, and some other
2575 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2576 version of the ISO C standard specified by any @option{-std} option used.
2578 Valid ISO C and ISO C++ programs should compile properly with or without
2579 this option (though a rare few will require @option{-ansi} or a
2580 @option{-std} option specifying the required version of ISO C)@. However,
2581 without this option, certain GNU extensions and traditional C and C++
2582 features are supported as well. With this option, they are rejected.
2584 @option{-pedantic} does not cause warning messages for use of the
2585 alternate keywords whose names begin and end with @samp{__}. Pedantic
2586 warnings are also disabled in the expression that follows
2587 @code{__extension__}. However, only system header files should use
2588 these escape routes; application programs should avoid them.
2589 @xref{Alternate Keywords}.
2591 Some users try to use @option{-pedantic} to check programs for strict ISO
2592 C conformance. They soon find that it does not do quite what they want:
2593 it finds some non-ISO practices, but not all---only those for which
2594 ISO C @emph{requires} a diagnostic, and some others for which
2595 diagnostics have been added.
2597 A feature to report any failure to conform to ISO C might be useful in
2598 some instances, but would require considerable additional work and would
2599 be quite different from @option{-pedantic}. We don't have plans to
2600 support such a feature in the near future.
2602 Where the standard specified with @option{-std} represents a GNU
2603 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2604 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2605 extended dialect is based. Warnings from @option{-pedantic} are given
2606 where they are required by the base standard. (It would not make sense
2607 for such warnings to be given only for features not in the specified GNU
2608 C dialect, since by definition the GNU dialects of C include all
2609 features the compiler supports with the given option, and there would be
2610 nothing to warn about.)
2612 @item -pedantic-errors
2613 @opindex pedantic-errors
2614 Like @option{-pedantic}, except that errors are produced rather than
2619 This enables all the warnings about constructions that some users
2620 consider questionable, and that are easy to avoid (or modify to
2621 prevent the warning), even in conjunction with macros. This also
2622 enables some language-specific warnings described in @ref{C++ Dialect
2623 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2625 @option{-Wall} turns on the following warning flags:
2627 @gccoptlist{-Waddress @gol
2628 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2630 -Wchar-subscripts @gol
2632 -Wimplicit-function-declaration @gol
2635 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2636 -Wmissing-braces @gol
2642 -Wsequence-point @gol
2643 -Wsign-compare @r{(only in C++)} @gol
2644 -Wstrict-aliasing @gol
2645 -Wstrict-overflow=1 @gol
2648 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2649 -Wunknown-pragmas @gol
2650 -Wunused-function @gol
2653 -Wunused-variable @gol
2656 Note that some warning flags are not implied by @option{-Wall}. Some of
2657 them warn about constructions that users generally do not consider
2658 questionable, but which occasionally you might wish to check for;
2659 others warn about constructions that are necessary or hard to avoid in
2660 some cases, and there is no simple way to modify the code to suppress
2661 the warning. Some of them are enabled by @option{-Wextra} but many of
2662 them must be enabled individually.
2667 This enables some extra warning flags that are not enabled by
2668 @option{-Wall}. (This option used to be called @option{-W}. The older
2669 name is still supported, but the newer name is more descriptive.)
2671 @gccoptlist{-Wclobbered @gol
2673 -Wignored-qualifiers @gol
2674 -Wmissing-field-initializers @gol
2675 -Wmissing-parameter-type @r{(C only)} @gol
2676 -Wold-style-declaration @r{(C only)} @gol
2677 -Woverride-init @gol
2680 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2681 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2684 The option @option{-Wextra} also prints warning messages for the
2690 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2691 @samp{>}, or @samp{>=}.
2694 (C++ only) An enumerator and a non-enumerator both appear in a
2695 conditional expression.
2698 (C++ only) A non-static reference or non-static @samp{const} member
2699 appears in a class without constructors.
2702 (C++ only) Ambiguous virtual bases.
2705 (C++ only) Subscripting an array which has been declared @samp{register}.
2708 (C++ only) Taking the address of a variable which has been declared
2712 (C++ only) A base class is not initialized in a derived class' copy
2719 Inhibit warning messages about the use of @samp{#import}.
2721 @item -Wchar-subscripts
2722 @opindex Wchar-subscripts
2723 Warn if an array subscript has type @code{char}. This is a common cause
2724 of error, as programmers often forget that this type is signed on some
2726 This warning is enabled by @option{-Wall}.
2730 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2731 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2732 This warning is enabled by @option{-Wall}.
2736 @opindex ffreestanding
2737 @opindex fno-builtin
2738 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2739 the arguments supplied have types appropriate to the format string
2740 specified, and that the conversions specified in the format string make
2741 sense. This includes standard functions, and others specified by format
2742 attributes (@pxref{Function Attributes}), in the @code{printf},
2743 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2744 not in the C standard) families (or other target-specific families).
2745 Which functions are checked without format attributes having been
2746 specified depends on the standard version selected, and such checks of
2747 functions without the attribute specified are disabled by
2748 @option{-ffreestanding} or @option{-fno-builtin}.
2750 The formats are checked against the format features supported by GNU
2751 libc version 2.2. These include all ISO C90 and C99 features, as well
2752 as features from the Single Unix Specification and some BSD and GNU
2753 extensions. Other library implementations may not support all these
2754 features; GCC does not support warning about features that go beyond a
2755 particular library's limitations. However, if @option{-pedantic} is used
2756 with @option{-Wformat}, warnings will be given about format features not
2757 in the selected standard version (but not for @code{strfmon} formats,
2758 since those are not in any version of the C standard). @xref{C Dialect
2759 Options,,Options Controlling C Dialect}.
2761 Since @option{-Wformat} also checks for null format arguments for
2762 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2764 @option{-Wformat} is included in @option{-Wall}. For more control over some
2765 aspects of format checking, the options @option{-Wformat-y2k},
2766 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2767 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2768 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2771 @opindex Wformat-y2k
2772 If @option{-Wformat} is specified, also warn about @code{strftime}
2773 formats which may yield only a two-digit year.
2775 @item -Wno-format-extra-args
2776 @opindex Wno-format-extra-args
2777 If @option{-Wformat} is specified, do not warn about excess arguments to a
2778 @code{printf} or @code{scanf} format function. The C standard specifies
2779 that such arguments are ignored.
2781 Where the unused arguments lie between used arguments that are
2782 specified with @samp{$} operand number specifications, normally
2783 warnings are still given, since the implementation could not know what
2784 type to pass to @code{va_arg} to skip the unused arguments. However,
2785 in the case of @code{scanf} formats, this option will suppress the
2786 warning if the unused arguments are all pointers, since the Single
2787 Unix Specification says that such unused arguments are allowed.
2789 @item -Wno-format-zero-length
2790 @opindex Wno-format-zero-length
2791 If @option{-Wformat} is specified, do not warn about zero-length formats.
2792 The C standard specifies that zero-length formats are allowed.
2794 @item -Wformat-nonliteral
2795 @opindex Wformat-nonliteral
2796 If @option{-Wformat} is specified, also warn if the format string is not a
2797 string literal and so cannot be checked, unless the format function
2798 takes its format arguments as a @code{va_list}.
2800 @item -Wformat-security
2801 @opindex Wformat-security
2802 If @option{-Wformat} is specified, also warn about uses of format
2803 functions that represent possible security problems. At present, this
2804 warns about calls to @code{printf} and @code{scanf} functions where the
2805 format string is not a string literal and there are no format arguments,
2806 as in @code{printf (foo);}. This may be a security hole if the format
2807 string came from untrusted input and contains @samp{%n}. (This is
2808 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2809 in future warnings may be added to @option{-Wformat-security} that are not
2810 included in @option{-Wformat-nonliteral}.)
2814 Enable @option{-Wformat} plus format checks not included in
2815 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2816 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2820 Warn about passing a null pointer for arguments marked as
2821 requiring a non-null value by the @code{nonnull} function attribute.
2823 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2824 can be disabled with the @option{-Wno-nonnull} option.
2826 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2828 Warn about uninitialized variables which are initialized with themselves.
2829 Note this option can only be used with the @option{-Wuninitialized} option,
2830 which in turn only works with @option{-O1} and above.
2832 For example, GCC will warn about @code{i} being uninitialized in the
2833 following snippet only when @option{-Winit-self} has been specified:
2844 @item -Wimplicit-int
2845 @opindex Wimplicit-int
2846 Warn when a declaration does not specify a type.
2847 This warning is enabled by @option{-Wall}.
2849 @item -Wimplicit-function-declaration
2850 @opindex Wimplicit-function-declaration
2851 @opindex Wno-implicit-function-declaration
2852 Give a warning whenever a function is used before being declared. In
2853 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2854 enabled by default and it is made into an error by
2855 @option{-pedantic-errors}. This warning is also enabled by
2860 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2861 This warning is enabled by @option{-Wall}.
2863 @item -Wignored-qualifiers
2864 @opindex Wignored-qualifiers
2865 @opindex Wno-ignored-qualifiers
2866 Warn if the return type of a function has a type qualifier
2867 such as @code{const}. For ISO C such a type qualifier has no effect,
2868 since the value returned by a function is not an lvalue.
2869 For C++, the warning is only emitted for scalar types or @code{void}.
2870 ISO C prohibits qualified @code{void} return types on function
2871 definitions, so such return types always receive a warning
2872 even without this option.
2874 This warning is also enabled by @option{-Wextra}.
2878 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2879 function with external linkage, returning int, taking either zero
2880 arguments, two, or three arguments of appropriate types.
2881 This warning is enabled by @option{-Wall}.
2883 @item -Wmissing-braces
2884 @opindex Wmissing-braces
2885 Warn if an aggregate or union initializer is not fully bracketed. In
2886 the following example, the initializer for @samp{a} is not fully
2887 bracketed, but that for @samp{b} is fully bracketed.
2890 int a[2][2] = @{ 0, 1, 2, 3 @};
2891 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2894 This warning is enabled by @option{-Wall}.
2896 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2897 @opindex Wmissing-include-dirs
2898 Warn if a user-supplied include directory does not exist.
2901 @opindex Wparentheses
2902 Warn if parentheses are omitted in certain contexts, such
2903 as when there is an assignment in a context where a truth value
2904 is expected, or when operators are nested whose precedence people
2905 often get confused about.
2907 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2908 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2909 interpretation from that of ordinary mathematical notation.
2911 Also warn about constructions where there may be confusion to which
2912 @code{if} statement an @code{else} branch belongs. Here is an example of
2927 In C/C++, every @code{else} branch belongs to the innermost possible
2928 @code{if} statement, which in this example is @code{if (b)}. This is
2929 often not what the programmer expected, as illustrated in the above
2930 example by indentation the programmer chose. When there is the
2931 potential for this confusion, GCC will issue a warning when this flag
2932 is specified. To eliminate the warning, add explicit braces around
2933 the innermost @code{if} statement so there is no way the @code{else}
2934 could belong to the enclosing @code{if}. The resulting code would
2951 This warning is enabled by @option{-Wall}.
2953 @item -Wsequence-point
2954 @opindex Wsequence-point
2955 Warn about code that may have undefined semantics because of violations
2956 of sequence point rules in the C and C++ standards.
2958 The C and C++ standards defines the order in which expressions in a C/C++
2959 program are evaluated in terms of @dfn{sequence points}, which represent
2960 a partial ordering between the execution of parts of the program: those
2961 executed before the sequence point, and those executed after it. These
2962 occur after the evaluation of a full expression (one which is not part
2963 of a larger expression), after the evaluation of the first operand of a
2964 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2965 function is called (but after the evaluation of its arguments and the
2966 expression denoting the called function), and in certain other places.
2967 Other than as expressed by the sequence point rules, the order of
2968 evaluation of subexpressions of an expression is not specified. All
2969 these rules describe only a partial order rather than a total order,
2970 since, for example, if two functions are called within one expression
2971 with no sequence point between them, the order in which the functions
2972 are called is not specified. However, the standards committee have
2973 ruled that function calls do not overlap.
2975 It is not specified when between sequence points modifications to the
2976 values of objects take effect. Programs whose behavior depends on this
2977 have undefined behavior; the C and C++ standards specify that ``Between
2978 the previous and next sequence point an object shall have its stored
2979 value modified at most once by the evaluation of an expression.
2980 Furthermore, the prior value shall be read only to determine the value
2981 to be stored.''. If a program breaks these rules, the results on any
2982 particular implementation are entirely unpredictable.
2984 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2985 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2986 diagnosed by this option, and it may give an occasional false positive
2987 result, but in general it has been found fairly effective at detecting
2988 this sort of problem in programs.
2990 The standard is worded confusingly, therefore there is some debate
2991 over the precise meaning of the sequence point rules in subtle cases.
2992 Links to discussions of the problem, including proposed formal
2993 definitions, may be found on the GCC readings page, at
2994 @w{@uref{http://gcc.gnu.org/readings.html}}.
2996 This warning is enabled by @option{-Wall} for C and C++.
2999 @opindex Wreturn-type
3000 @opindex Wno-return-type
3001 Warn whenever a function is defined with a return-type that defaults
3002 to @code{int}. Also warn about any @code{return} statement with no
3003 return-value in a function whose return-type is not @code{void}
3004 (falling off the end of the function body is considered returning
3005 without a value), and about a @code{return} statement with a
3006 expression in a function whose return-type is @code{void}.
3008 For C++, a function without return type always produces a diagnostic
3009 message, even when @option{-Wno-return-type} is specified. The only
3010 exceptions are @samp{main} and functions defined in system headers.
3012 This warning is enabled by @option{-Wall}.
3016 Warn whenever a @code{switch} statement has an index of enumerated type
3017 and lacks a @code{case} for one or more of the named codes of that
3018 enumeration. (The presence of a @code{default} label prevents this
3019 warning.) @code{case} labels outside the enumeration range also
3020 provoke warnings when this option is used.
3021 This warning is enabled by @option{-Wall}.
3023 @item -Wswitch-default
3024 @opindex Wswitch-switch
3025 Warn whenever a @code{switch} statement does not have a @code{default}
3029 @opindex Wswitch-enum
3030 Warn whenever a @code{switch} statement has an index of enumerated type
3031 and lacks a @code{case} for one or more of the named codes of that
3032 enumeration. @code{case} labels outside the enumeration range also
3033 provoke warnings when this option is used.
3037 Warn if any trigraphs are encountered that might change the meaning of
3038 the program (trigraphs within comments are not warned about).
3039 This warning is enabled by @option{-Wall}.
3041 @item -Wunused-function
3042 @opindex Wunused-function
3043 Warn whenever a static function is declared but not defined or a
3044 non-inline static function is unused.
3045 This warning is enabled by @option{-Wall}.
3047 @item -Wunused-label
3048 @opindex Wunused-label
3049 Warn whenever a label is declared but not used.
3050 This warning is enabled by @option{-Wall}.
3052 To suppress this warning use the @samp{unused} attribute
3053 (@pxref{Variable Attributes}).
3055 @item -Wunused-parameter
3056 @opindex Wunused-parameter
3057 Warn whenever a function parameter is unused aside from its declaration.
3059 To suppress this warning use the @samp{unused} attribute
3060 (@pxref{Variable Attributes}).
3062 @item -Wunused-variable
3063 @opindex Wunused-variable
3064 Warn whenever a local variable or non-constant static variable is unused
3065 aside from its declaration.
3066 This warning is enabled by @option{-Wall}.
3068 To suppress this warning use the @samp{unused} attribute
3069 (@pxref{Variable Attributes}).
3071 @item -Wunused-value
3072 @opindex Wunused-value
3073 Warn whenever a statement computes a result that is explicitly not
3074 used. To suppress this warning cast the unused expression to
3075 @samp{void}. This includes an expression-statement or the left-hand
3076 side of a comma expression that contains no side effects. For example,
3077 an expression such as @samp{x[i,j]} will cause a warning, while
3078 @samp{x[(void)i,j]} will not.
3080 This warning is enabled by @option{-Wall}.
3084 All the above @option{-Wunused} options combined.
3086 In order to get a warning about an unused function parameter, you must
3087 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3088 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3090 @item -Wuninitialized
3091 @opindex Wuninitialized
3092 Warn if an automatic variable is used without first being initialized or
3093 if a variable may be clobbered by a @code{setjmp} call.
3095 These warnings are possible only in optimizing compilation,
3096 because they require data flow information that is computed only
3097 when optimizing. If you do not specify @option{-O}, you will not get
3098 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
3099 requiring @option{-O}.
3101 If you want to warn about code which uses the uninitialized value of the
3102 variable in its own initializer, use the @option{-Winit-self} option.
3104 These warnings occur for individual uninitialized or clobbered
3105 elements of structure, union or array variables as well as for
3106 variables which are uninitialized or clobbered as a whole. They do
3107 not occur for variables or elements declared @code{volatile}. Because
3108 these warnings depend on optimization, the exact variables or elements
3109 for which there are warnings will depend on the precise optimization
3110 options and version of GCC used.
3112 Note that there may be no warning about a variable that is used only
3113 to compute a value that itself is never used, because such
3114 computations may be deleted by data flow analysis before the warnings
3117 These warnings are made optional because GCC is not smart
3118 enough to see all the reasons why the code might be correct
3119 despite appearing to have an error. Here is one example of how
3140 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3141 always initialized, but GCC doesn't know this. Here is
3142 another common case:
3147 if (change_y) save_y = y, y = new_y;
3149 if (change_y) y = save_y;
3154 This has no bug because @code{save_y} is used only if it is set.
3156 @cindex @code{longjmp} warnings
3157 This option also warns when a non-volatile automatic variable might be
3158 changed by a call to @code{longjmp}. These warnings as well are possible
3159 only in optimizing compilation.
3161 The compiler sees only the calls to @code{setjmp}. It cannot know
3162 where @code{longjmp} will be called; in fact, a signal handler could
3163 call it at any point in the code. As a result, you may get a warning
3164 even when there is in fact no problem because @code{longjmp} cannot
3165 in fact be called at the place which would cause a problem.
3167 Some spurious warnings can be avoided if you declare all the functions
3168 you use that never return as @code{noreturn}. @xref{Function
3171 This warning is enabled by @option{-Wall} or @option{-Wextra} in
3172 optimizing compilations (@option{-O1} and above).
3174 @item -Wunknown-pragmas
3175 @opindex Wunknown-pragmas
3176 @cindex warning for unknown pragmas
3177 @cindex unknown pragmas, warning
3178 @cindex pragmas, warning of unknown
3179 Warn when a #pragma directive is encountered which is not understood by
3180 GCC@. If this command line option is used, warnings will even be issued
3181 for unknown pragmas in system header files. This is not the case if
3182 the warnings were only enabled by the @option{-Wall} command line option.
3185 @opindex Wno-pragmas
3187 Do not warn about misuses of pragmas, such as incorrect parameters,
3188 invalid syntax, or conflicts between pragmas. See also
3189 @samp{-Wunknown-pragmas}.
3191 @item -Wstrict-aliasing
3192 @opindex Wstrict-aliasing
3193 This option is only active when @option{-fstrict-aliasing} is active.
3194 It warns about code which might break the strict aliasing rules that the
3195 compiler is using for optimization. The warning does not catch all
3196 cases, but does attempt to catch the more common pitfalls. It is
3197 included in @option{-Wall}.
3198 It is equivalent to @option{-Wstrict-aliasing=3}
3200 @item -Wstrict-aliasing=n
3201 @opindex Wstrict-aliasing=n
3202 This option is only active when @option{-fstrict-aliasing} is active.
3203 It warns about code which might break the strict aliasing rules that the
3204 compiler is using for optimization.
3205 Higher levels correspond to higher accuracy (fewer false positives).
3206 Higher levels also correspond to more effort, similar to the way -O works.
3207 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3210 Level 1: Most aggressive, quick, least accurate.
3211 Possibly useful when higher levels
3212 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3213 false negatives. However, it has many false positives.
3214 Warns for all pointer conversions between possibly incompatible types,
3215 even if never dereferenced. Runs in the frontend only.
3217 Level 2: Aggressive, quick, not too precise.
3218 May still have many false positives (not as many as level 1 though),
3219 and few false negatives (but possibly more than level 1).
3220 Unlike level 1, it only warns when an address is taken. Warns about
3221 incomplete types. Runs in the frontend only.
3223 Level 3 (default for @option{-Wstrict-aliasing}):
3224 Should have very few false positives and few false
3225 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3226 Takes care of the common punn+dereference pattern in the frontend:
3227 @code{*(int*)&some_float}.
3228 If optimization is enabled, it also runs in the backend, where it deals
3229 with multiple statement cases using flow-sensitive points-to information.
3230 Only warns when the converted pointer is dereferenced.
3231 Does not warn about incomplete types.
3233 @item -Wstrict-overflow
3234 @itemx -Wstrict-overflow=@var{n}
3235 @opindex Wstrict-overflow
3236 This option is only active when @option{-fstrict-overflow} is active.
3237 It warns about cases where the compiler optimizes based on the
3238 assumption that signed overflow does not occur. Note that it does not
3239 warn about all cases where the code might overflow: it only warns
3240 about cases where the compiler implements some optimization. Thus
3241 this warning depends on the optimization level.
3243 An optimization which assumes that signed overflow does not occur is
3244 perfectly safe if the values of the variables involved are such that
3245 overflow never does, in fact, occur. Therefore this warning can
3246 easily give a false positive: a warning about code which is not
3247 actually a problem. To help focus on important issues, several
3248 warning levels are defined. No warnings are issued for the use of
3249 undefined signed overflow when estimating how many iterations a loop
3250 will require, in particular when determining whether a loop will be
3254 @item -Wstrict-overflow=1
3255 Warn about cases which are both questionable and easy to avoid. For
3256 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3257 compiler will simplify this to @code{1}. This level of
3258 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3259 are not, and must be explicitly requested.
3261 @item -Wstrict-overflow=2
3262 Also warn about other cases where a comparison is simplified to a
3263 constant. For example: @code{abs (x) >= 0}. This can only be
3264 simplified when @option{-fstrict-overflow} is in effect, because
3265 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3266 zero. @option{-Wstrict-overflow} (with no level) is the same as
3267 @option{-Wstrict-overflow=2}.
3269 @item -Wstrict-overflow=3
3270 Also warn about other cases where a comparison is simplified. For
3271 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3273 @item -Wstrict-overflow=4
3274 Also warn about other simplifications not covered by the above cases.
3275 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3277 @item -Wstrict-overflow=5
3278 Also warn about cases where the compiler reduces the magnitude of a
3279 constant involved in a comparison. For example: @code{x + 2 > y} will
3280 be simplified to @code{x + 1 >= y}. This is reported only at the
3281 highest warning level because this simplification applies to many
3282 comparisons, so this warning level will give a very large number of
3286 @item -Warray-bounds
3287 @opindex Wno-array-bounds
3288 @opindex Warray-bounds
3289 This option is only active when @option{-ftree-vrp} is active
3290 (default for -O2 and above). It warns about subscripts to arrays
3291 that are always out of bounds. This warning is enabled by @option{-Wall}.
3293 @item -Wno-div-by-zero
3294 @opindex Wno-div-by-zero
3295 @opindex Wdiv-by-zero
3296 Do not warn about compile-time integer division by zero. Floating point
3297 division by zero is not warned about, as it can be a legitimate way of
3298 obtaining infinities and NaNs.
3300 @item -Wsystem-headers
3301 @opindex Wsystem-headers
3302 @cindex warnings from system headers
3303 @cindex system headers, warnings from
3304 Print warning messages for constructs found in system header files.
3305 Warnings from system headers are normally suppressed, on the assumption
3306 that they usually do not indicate real problems and would only make the
3307 compiler output harder to read. Using this command line option tells
3308 GCC to emit warnings from system headers as if they occurred in user
3309 code. However, note that using @option{-Wall} in conjunction with this
3310 option will @emph{not} warn about unknown pragmas in system
3311 headers---for that, @option{-Wunknown-pragmas} must also be used.
3314 @opindex Wfloat-equal
3315 Warn if floating point values are used in equality comparisons.
3317 The idea behind this is that sometimes it is convenient (for the
3318 programmer) to consider floating-point values as approximations to
3319 infinitely precise real numbers. If you are doing this, then you need
3320 to compute (by analyzing the code, or in some other way) the maximum or
3321 likely maximum error that the computation introduces, and allow for it
3322 when performing comparisons (and when producing output, but that's a
3323 different problem). In particular, instead of testing for equality, you
3324 would check to see whether the two values have ranges that overlap; and
3325 this is done with the relational operators, so equality comparisons are
3328 @item -Wtraditional @r{(C only)}
3329 @opindex Wtraditional
3330 Warn about certain constructs that behave differently in traditional and
3331 ISO C@. Also warn about ISO C constructs that have no traditional C
3332 equivalent, and/or problematic constructs which should be avoided.
3336 Macro parameters that appear within string literals in the macro body.
3337 In traditional C macro replacement takes place within string literals,
3338 but does not in ISO C@.
3341 In traditional C, some preprocessor directives did not exist.
3342 Traditional preprocessors would only consider a line to be a directive
3343 if the @samp{#} appeared in column 1 on the line. Therefore
3344 @option{-Wtraditional} warns about directives that traditional C
3345 understands but would ignore because the @samp{#} does not appear as the
3346 first character on the line. It also suggests you hide directives like
3347 @samp{#pragma} not understood by traditional C by indenting them. Some
3348 traditional implementations would not recognize @samp{#elif}, so it
3349 suggests avoiding it altogether.
3352 A function-like macro that appears without arguments.
3355 The unary plus operator.
3358 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3359 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3360 constants.) Note, these suffixes appear in macros defined in the system
3361 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3362 Use of these macros in user code might normally lead to spurious
3363 warnings, however GCC's integrated preprocessor has enough context to
3364 avoid warning in these cases.
3367 A function declared external in one block and then used after the end of
3371 A @code{switch} statement has an operand of type @code{long}.
3374 A non-@code{static} function declaration follows a @code{static} one.
3375 This construct is not accepted by some traditional C compilers.
3378 The ISO type of an integer constant has a different width or
3379 signedness from its traditional type. This warning is only issued if
3380 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3381 typically represent bit patterns, are not warned about.
3384 Usage of ISO string concatenation is detected.
3387 Initialization of automatic aggregates.
3390 Identifier conflicts with labels. Traditional C lacks a separate
3391 namespace for labels.
3394 Initialization of unions. If the initializer is zero, the warning is
3395 omitted. This is done under the assumption that the zero initializer in
3396 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3397 initializer warnings and relies on default initialization to zero in the
3401 Conversions by prototypes between fixed/floating point values and vice
3402 versa. The absence of these prototypes when compiling with traditional
3403 C would cause serious problems. This is a subset of the possible
3404 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3407 Use of ISO C style function definitions. This warning intentionally is
3408 @emph{not} issued for prototype declarations or variadic functions
3409 because these ISO C features will appear in your code when using
3410 libiberty's traditional C compatibility macros, @code{PARAMS} and
3411 @code{VPARAMS}. This warning is also bypassed for nested functions
3412 because that feature is already a GCC extension and thus not relevant to
3413 traditional C compatibility.
3416 @item -Wtraditional-conversion @r{(C only)}
3417 @opindex Wtraditional-conversion
3418 Warn if a prototype causes a type conversion that is different from what
3419 would happen to the same argument in the absence of a prototype. This
3420 includes conversions of fixed point to floating and vice versa, and
3421 conversions changing the width or signedness of a fixed point argument
3422 except when the same as the default promotion.
3424 @item -Wdeclaration-after-statement @r{(C only)}
3425 @opindex Wdeclaration-after-statement
3426 Warn when a declaration is found after a statement in a block. This
3427 construct, known from C++, was introduced with ISO C99 and is by default
3428 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3429 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3433 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3435 @item -Wno-endif-labels
3436 @opindex Wno-endif-labels
3437 @opindex Wendif-labels
3438 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3442 Warn whenever a local variable shadows another local variable, parameter or
3443 global variable or whenever a built-in function is shadowed.
3445 @item -Wlarger-than-@var{len}
3446 @opindex Wlarger-than
3447 Warn whenever an object of larger than @var{len} bytes is defined.
3449 @item -Wunsafe-loop-optimizations
3450 @opindex Wunsafe-loop-optimizations
3451 Warn if the loop cannot be optimized because the compiler could not
3452 assume anything on the bounds of the loop indices. With
3453 @option{-funsafe-loop-optimizations} warn if the compiler made
3456 @item -Wpointer-arith
3457 @opindex Wpointer-arith
3458 Warn about anything that depends on the ``size of'' a function type or
3459 of @code{void}. GNU C assigns these types a size of 1, for
3460 convenience in calculations with @code{void *} pointers and pointers
3461 to functions. In C++, warn also when an arithmetic operation involves
3462 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3465 @opindex Wtype-limits
3466 @opindex Wno-type-limits
3467 Warn if a comparison is always true or always false due to the limited
3468 range of the data type, but do not warn for constant expressions. For
3469 example, warn if an unsigned variable is compared against zero with
3470 @samp{<} or @samp{>=}. This warning is also enabled by
3473 @item -Wbad-function-cast @r{(C only)}
3474 @opindex Wbad-function-cast
3475 Warn whenever a function call is cast to a non-matching type.
3476 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3479 Warn about ISO C constructs that are outside of the common subset of
3480 ISO C and ISO C++, e.g.@: request for implicit conversion from
3481 @code{void *} to a pointer to non-@code{void} type.
3483 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3484 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3485 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3486 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3490 Warn whenever a pointer is cast so as to remove a type qualifier from
3491 the target type. For example, warn if a @code{const char *} is cast
3492 to an ordinary @code{char *}.
3495 @opindex Wcast-align
3496 Warn whenever a pointer is cast such that the required alignment of the
3497 target is increased. For example, warn if a @code{char *} is cast to
3498 an @code{int *} on machines where integers can only be accessed at
3499 two- or four-byte boundaries.
3501 @item -Wwrite-strings
3502 @opindex Wwrite-strings
3503 When compiling C, give string constants the type @code{const
3504 char[@var{length}]} so that
3505 copying the address of one into a non-@code{const} @code{char *}
3506 pointer will get a warning; when compiling C++, warn about the
3507 deprecated conversion from string literals to @code{char *}. This
3508 warning, by default, is enabled for C++ programs.
3509 These warnings will help you find at
3510 compile time code that can try to write into a string constant, but
3511 only if you have been very careful about using @code{const} in
3512 declarations and prototypes. Otherwise, it will just be a nuisance;
3513 this is why we did not make @option{-Wall} request these warnings.
3517 Warn for variables that might be changed by @samp{longjmp} or
3518 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3521 @opindex Wconversion
3522 @opindex Wno-conversion
3523 Warn for implicit conversions that may alter a value. This includes
3524 conversions between real and integer, like @code{abs (x)} when
3525 @code{x} is @code{double}; conversions between signed and unsigned,
3526 like @code{unsigned ui = -1}; and conversions to smaller types, like
3527 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3528 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3529 changed by the conversion like in @code{abs (2.0)}. Warnings about
3530 conversions between signed and unsigned integers can be disabled by
3531 using @option{-Wno-sign-conversion}.
3533 For C++, also warn for conversions between @code{NULL} and non-pointer
3534 types; confusing overload resolution for user-defined conversions; and
3535 conversions that will never use a type conversion operator:
3536 conversions to @code{void}, the same type, a base class or a reference
3537 to them. Warnings about conversions between signed and unsigned
3538 integers are disabled by default in C++ unless
3539 @option{-Wsign-conversion} is explicitly enabled.
3542 @opindex Wempty-body
3543 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3544 while} statement. Additionally, in C++, warn when an empty body occurs
3545 in a @samp{while} or @samp{for} statement with no whitespacing before
3546 the semicolon. This warning is also enabled by @option{-Wextra}.
3548 @item -Wsign-compare
3549 @opindex Wsign-compare
3550 @cindex warning for comparison of signed and unsigned values
3551 @cindex comparison of signed and unsigned values, warning
3552 @cindex signed and unsigned values, comparison warning
3553 Warn when a comparison between signed and unsigned values could produce
3554 an incorrect result when the signed value is converted to unsigned.
3555 This warning is also enabled by @option{-Wextra}; to get the other warnings
3556 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3558 @item -Wsign-conversion
3559 @opindex Wsign-conversion
3560 @opindex Wno-sign-conversion
3561 Warn for implicit conversions that may change the sign of an integer
3562 value, like assigning a signed integer expression to an unsigned
3563 integer variable. An explicit cast silences the warning. In C, this
3564 option is enabled also by @option{-Wconversion}.
3568 @opindex Wno-address
3569 Warn about suspicious uses of memory addresses. These include using
3570 the address of a function in a conditional expression, such as
3571 @code{void func(void); if (func)}, and comparisons against the memory
3572 address of a string literal, such as @code{if (x == "abc")}. Such
3573 uses typically indicate a programmer error: the address of a function
3574 always evaluates to true, so their use in a conditional usually
3575 indicate that the programmer forgot the parentheses in a function
3576 call; and comparisons against string literals result in unspecified
3577 behavior and are not portable in C, so they usually indicate that the
3578 programmer intended to use @code{strcmp}. This warning is enabled by
3582 @opindex Wlogical-op
3583 @opindex Wno-logical-op
3584 Warn about suspicious uses of logical operators in expressions.
3585 This includes using logical operators in contexts where a
3586 bit-wise operator is likely to be expected.
3588 @item -Waggregate-return
3589 @opindex Waggregate-return
3590 Warn if any functions that return structures or unions are defined or
3591 called. (In languages where you can return an array, this also elicits
3594 @item -Wno-attributes
3595 @opindex Wno-attributes
3596 @opindex Wattributes
3597 Do not warn if an unexpected @code{__attribute__} is used, such as
3598 unrecognized attributes, function attributes applied to variables,
3599 etc. This will not stop errors for incorrect use of supported
3602 @item -Wstrict-prototypes @r{(C only)}
3603 @opindex Wstrict-prototypes
3604 Warn if a function is declared or defined without specifying the
3605 argument types. (An old-style function definition is permitted without
3606 a warning if preceded by a declaration which specifies the argument
3609 @item -Wold-style-declaration @r{(C only)}
3610 @opindex Wold-style-declaration
3611 Warn for obsolescent usages, according to the C Standard, in a
3612 declaration. For example, warn if storage-class specifiers like
3613 @code{static} are not the first things in a declaration. This warning
3614 is also enabled by @option{-Wextra}.
3616 @item -Wold-style-definition @r{(C only)}
3617 @opindex Wold-style-definition
3618 Warn if an old-style function definition is used. A warning is given
3619 even if there is a previous prototype.
3621 @item -Wmissing-parameter-type @r{(C only)}
3622 @opindex Wmissing-parameter-type
3623 A function parameter is declared without a type specifier in K&R-style
3630 This warning is also enabled by @option{-Wextra}.
3632 @item -Wmissing-prototypes @r{(C only)}
3633 @opindex Wmissing-prototypes
3634 Warn if a global function is defined without a previous prototype
3635 declaration. This warning is issued even if the definition itself
3636 provides a prototype. The aim is to detect global functions that fail
3637 to be declared in header files.
3639 @item -Wmissing-declarations @r{(C and C++ only)}
3640 @opindex Wmissing-declarations
3641 Warn if a global function is defined without a previous declaration.
3642 Do so even if the definition itself provides a prototype.
3643 Use this option to detect global functions that are not declared in
3644 header files. In C++, no warnings are issued for function templates,
3645 or for inline functions, or for functions in anonymous namespaces.
3647 @item -Wmissing-field-initializers
3648 @opindex Wmissing-field-initializers
3651 Warn if a structure's initializer has some fields missing. For
3652 example, the following code would cause such a warning, because
3653 @code{x.h} is implicitly zero:
3656 struct s @{ int f, g, h; @};
3657 struct s x = @{ 3, 4 @};
3660 This option does not warn about designated initializers, so the following
3661 modification would not trigger a warning:
3664 struct s @{ int f, g, h; @};
3665 struct s x = @{ .f = 3, .g = 4 @};
3668 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3669 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3671 @item -Wmissing-noreturn
3672 @opindex Wmissing-noreturn
3673 Warn about functions which might be candidates for attribute @code{noreturn}.
3674 Note these are only possible candidates, not absolute ones. Care should
3675 be taken to manually verify functions actually do not ever return before
3676 adding the @code{noreturn} attribute, otherwise subtle code generation
3677 bugs could be introduced. You will not get a warning for @code{main} in
3678 hosted C environments.
3680 @item -Wmissing-format-attribute
3681 @opindex Wmissing-format-attribute
3683 Warn about function pointers which might be candidates for @code{format}
3684 attributes. Note these are only possible candidates, not absolute ones.
3685 GCC will guess that function pointers with @code{format} attributes that
3686 are used in assignment, initialization, parameter passing or return
3687 statements should have a corresponding @code{format} attribute in the
3688 resulting type. I.e.@: the left-hand side of the assignment or
3689 initialization, the type of the parameter variable, or the return type
3690 of the containing function respectively should also have a @code{format}
3691 attribute to avoid the warning.
3693 GCC will also warn about function definitions which might be
3694 candidates for @code{format} attributes. Again, these are only
3695 possible candidates. GCC will guess that @code{format} attributes
3696 might be appropriate for any function that calls a function like
3697 @code{vprintf} or @code{vscanf}, but this might not always be the
3698 case, and some functions for which @code{format} attributes are
3699 appropriate may not be detected.
3701 @item -Wno-multichar
3702 @opindex Wno-multichar
3704 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3705 Usually they indicate a typo in the user's code, as they have
3706 implementation-defined values, and should not be used in portable code.
3708 @item -Wnormalized=<none|id|nfc|nfkc>
3709 @opindex Wnormalized
3712 @cindex character set, input normalization
3713 In ISO C and ISO C++, two identifiers are different if they are
3714 different sequences of characters. However, sometimes when characters
3715 outside the basic ASCII character set are used, you can have two
3716 different character sequences that look the same. To avoid confusion,
3717 the ISO 10646 standard sets out some @dfn{normalization rules} which
3718 when applied ensure that two sequences that look the same are turned into
3719 the same sequence. GCC can warn you if you are using identifiers which
3720 have not been normalized; this option controls that warning.
3722 There are four levels of warning that GCC supports. The default is
3723 @option{-Wnormalized=nfc}, which warns about any identifier which is
3724 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3725 recommended form for most uses.
3727 Unfortunately, there are some characters which ISO C and ISO C++ allow
3728 in identifiers that when turned into NFC aren't allowable as
3729 identifiers. That is, there's no way to use these symbols in portable
3730 ISO C or C++ and have all your identifiers in NFC.
3731 @option{-Wnormalized=id} suppresses the warning for these characters.
3732 It is hoped that future versions of the standards involved will correct
3733 this, which is why this option is not the default.
3735 You can switch the warning off for all characters by writing
3736 @option{-Wnormalized=none}. You would only want to do this if you
3737 were using some other normalization scheme (like ``D''), because
3738 otherwise you can easily create bugs that are literally impossible to see.
3740 Some characters in ISO 10646 have distinct meanings but look identical
3741 in some fonts or display methodologies, especially once formatting has
3742 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3743 LETTER N'', will display just like a regular @code{n} which has been
3744 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3745 normalization scheme to convert all these into a standard form as
3746 well, and GCC will warn if your code is not in NFKC if you use
3747 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3748 about every identifier that contains the letter O because it might be
3749 confused with the digit 0, and so is not the default, but may be
3750 useful as a local coding convention if the programming environment is
3751 unable to be fixed to display these characters distinctly.
3753 @item -Wno-deprecated-declarations
3754 @opindex Wno-deprecated-declarations
3755 Do not warn about uses of functions (@pxref{Function Attributes}),
3756 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3757 Attributes}) marked as deprecated by using the @code{deprecated}
3761 @opindex Wno-overflow
3762 Do not warn about compile-time overflow in constant expressions.
3764 @item -Woverride-init
3765 @opindex Woverride-init
3768 Warn if an initialized field without side effects is overridden when
3769 using designated initializers (@pxref{Designated Inits, , Designated
3772 This warning is included in @option{-Wextra}. To get other
3773 @option{-Wextra} warnings without this one, use @samp{-Wextra
3774 -Wno-override-init}.
3778 Warn if a structure is given the packed attribute, but the packed
3779 attribute has no effect on the layout or size of the structure.
3780 Such structures may be mis-aligned for little benefit. For
3781 instance, in this code, the variable @code{f.x} in @code{struct bar}
3782 will be misaligned even though @code{struct bar} does not itself
3783 have the packed attribute:
3790 @} __attribute__((packed));
3800 Warn if padding is included in a structure, either to align an element
3801 of the structure or to align the whole structure. Sometimes when this
3802 happens it is possible to rearrange the fields of the structure to
3803 reduce the padding and so make the structure smaller.
3805 @item -Wredundant-decls
3806 @opindex Wredundant-decls
3807 Warn if anything is declared more than once in the same scope, even in
3808 cases where multiple declaration is valid and changes nothing.
3810 @item -Wnested-externs @r{(C only)}
3811 @opindex Wnested-externs
3812 Warn if an @code{extern} declaration is encountered within a function.
3814 @item -Wunreachable-code
3815 @opindex Wunreachable-code
3816 Warn if the compiler detects that code will never be executed.
3818 This option is intended to warn when the compiler detects that at
3819 least a whole line of source code will never be executed, because
3820 some condition is never satisfied or because it is after a
3821 procedure that never returns.
3823 It is possible for this option to produce a warning even though there
3824 are circumstances under which part of the affected line can be executed,
3825 so care should be taken when removing apparently-unreachable code.
3827 For instance, when a function is inlined, a warning may mean that the
3828 line is unreachable in only one inlined copy of the function.
3830 This option is not made part of @option{-Wall} because in a debugging
3831 version of a program there is often substantial code which checks
3832 correct functioning of the program and is, hopefully, unreachable
3833 because the program does work. Another common use of unreachable
3834 code is to provide behavior which is selectable at compile-time.
3838 Warn if a function can not be inlined and it was declared as inline.
3839 Even with this option, the compiler will not warn about failures to
3840 inline functions declared in system headers.
3842 The compiler uses a variety of heuristics to determine whether or not
3843 to inline a function. For example, the compiler takes into account
3844 the size of the function being inlined and the amount of inlining
3845 that has already been done in the current function. Therefore,
3846 seemingly insignificant changes in the source program can cause the
3847 warnings produced by @option{-Winline} to appear or disappear.
3849 @item -Wno-invalid-offsetof @r{(C++ only)}
3850 @opindex Wno-invalid-offsetof
3851 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3852 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3853 to a non-POD type is undefined. In existing C++ implementations,
3854 however, @samp{offsetof} typically gives meaningful results even when
3855 applied to certain kinds of non-POD types. (Such as a simple
3856 @samp{struct} that fails to be a POD type only by virtue of having a
3857 constructor.) This flag is for users who are aware that they are
3858 writing nonportable code and who have deliberately chosen to ignore the
3861 The restrictions on @samp{offsetof} may be relaxed in a future version
3862 of the C++ standard.
3864 @item -Wno-int-to-pointer-cast @r{(C only)}
3865 @opindex Wno-int-to-pointer-cast
3866 Suppress warnings from casts to pointer type of an integer of a
3869 @item -Wno-pointer-to-int-cast @r{(C only)}
3870 @opindex Wno-pointer-to-int-cast
3871 Suppress warnings from casts from a pointer to an integer type of a
3875 @opindex Winvalid-pch
3876 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3877 the search path but can't be used.
3881 @opindex Wno-long-long
3882 Warn if @samp{long long} type is used. This is default. To inhibit
3883 the warning messages, use @option{-Wno-long-long}. Flags
3884 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3885 only when @option{-pedantic} flag is used.
3887 @item -Wvariadic-macros
3888 @opindex Wvariadic-macros
3889 @opindex Wno-variadic-macros
3890 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3891 alternate syntax when in pedantic ISO C99 mode. This is default.
3892 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3897 Warn if variable length array is used in the code.
3898 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3899 the variable length array.
3901 @item -Wvolatile-register-var
3902 @opindex Wvolatile-register-var
3903 @opindex Wno-volatile-register-var
3904 Warn if a register variable is declared volatile. The volatile
3905 modifier does not inhibit all optimizations that may eliminate reads
3906 and/or writes to register variables.
3908 @item -Wdisabled-optimization
3909 @opindex Wdisabled-optimization
3910 Warn if a requested optimization pass is disabled. This warning does
3911 not generally indicate that there is anything wrong with your code; it
3912 merely indicates that GCC's optimizers were unable to handle the code
3913 effectively. Often, the problem is that your code is too big or too
3914 complex; GCC will refuse to optimize programs when the optimization
3915 itself is likely to take inordinate amounts of time.
3917 @item -Wpointer-sign
3918 @opindex Wpointer-sign
3919 @opindex Wno-pointer-sign
3920 Warn for pointer argument passing or assignment with different signedness.
3921 This option is only supported for C and Objective-C@. It is implied by
3922 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3923 @option{-Wno-pointer-sign}.
3925 @item -Wstack-protector
3926 @opindex Wstack-protector
3927 This option is only active when @option{-fstack-protector} is active. It
3928 warns about functions that will not be protected against stack smashing.
3930 @item -Woverlength-strings
3931 @opindex Woverlength-strings
3932 Warn about string constants which are longer than the ``minimum
3933 maximum'' length specified in the C standard. Modern compilers
3934 generally allow string constants which are much longer than the
3935 standard's minimum limit, but very portable programs should avoid
3936 using longer strings.
3938 The limit applies @emph{after} string constant concatenation, and does
3939 not count the trailing NUL@. In C89, the limit was 509 characters; in
3940 C99, it was raised to 4095. C++98 does not specify a normative
3941 minimum maximum, so we do not diagnose overlength strings in C++@.
3943 This option is implied by @option{-pedantic}, and can be disabled with
3944 @option{-Wno-overlength-strings}.
3947 @node Debugging Options
3948 @section Options for Debugging Your Program or GCC
3949 @cindex options, debugging
3950 @cindex debugging information options
3952 GCC has various special options that are used for debugging
3953 either your program or GCC:
3958 Produce debugging information in the operating system's native format
3959 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3962 On most systems that use stabs format, @option{-g} enables use of extra
3963 debugging information that only GDB can use; this extra information
3964 makes debugging work better in GDB but will probably make other debuggers
3966 refuse to read the program. If you want to control for certain whether
3967 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3968 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3970 GCC allows you to use @option{-g} with
3971 @option{-O}. The shortcuts taken by optimized code may occasionally
3972 produce surprising results: some variables you declared may not exist
3973 at all; flow of control may briefly move where you did not expect it;
3974 some statements may not be executed because they compute constant
3975 results or their values were already at hand; some statements may
3976 execute in different places because they were moved out of loops.
3978 Nevertheless it proves possible to debug optimized output. This makes
3979 it reasonable to use the optimizer for programs that might have bugs.
3981 The following options are useful when GCC is generated with the
3982 capability for more than one debugging format.
3986 Produce debugging information for use by GDB@. This means to use the
3987 most expressive format available (DWARF 2, stabs, or the native format
3988 if neither of those are supported), including GDB extensions if at all
3993 Produce debugging information in stabs format (if that is supported),
3994 without GDB extensions. This is the format used by DBX on most BSD
3995 systems. On MIPS, Alpha and System V Release 4 systems this option
3996 produces stabs debugging output which is not understood by DBX or SDB@.
3997 On System V Release 4 systems this option requires the GNU assembler.
3999 @item -feliminate-unused-debug-symbols
4000 @opindex feliminate-unused-debug-symbols
4001 Produce debugging information in stabs format (if that is supported),
4002 for only symbols that are actually used.
4004 @item -femit-class-debug-always
4005 Instead of emitting debugging information for a C++ class in only one
4006 object file, emit it in all object files using the class. This option
4007 should be used only with debuggers that are unable to handle the way GCC
4008 normally emits debugging information for classes because using this
4009 option will increase the size of debugging information by as much as a
4014 Produce debugging information in stabs format (if that is supported),
4015 using GNU extensions understood only by the GNU debugger (GDB)@. The
4016 use of these extensions is likely to make other debuggers crash or
4017 refuse to read the program.
4021 Produce debugging information in COFF format (if that is supported).
4022 This is the format used by SDB on most System V systems prior to
4027 Produce debugging information in XCOFF format (if that is supported).
4028 This is the format used by the DBX debugger on IBM RS/6000 systems.
4032 Produce debugging information in XCOFF format (if that is supported),
4033 using GNU extensions understood only by the GNU debugger (GDB)@. The
4034 use of these extensions is likely to make other debuggers crash or
4035 refuse to read the program, and may cause assemblers other than the GNU
4036 assembler (GAS) to fail with an error.
4040 Produce debugging information in DWARF version 2 format (if that is
4041 supported). This is the format used by DBX on IRIX 6. With this
4042 option, GCC uses features of DWARF version 3 when they are useful;
4043 version 3 is upward compatible with version 2, but may still cause
4044 problems for older debuggers.
4048 Produce debugging information in VMS debug format (if that is
4049 supported). This is the format used by DEBUG on VMS systems.
4052 @itemx -ggdb@var{level}
4053 @itemx -gstabs@var{level}
4054 @itemx -gcoff@var{level}
4055 @itemx -gxcoff@var{level}
4056 @itemx -gvms@var{level}
4057 Request debugging information and also use @var{level} to specify how
4058 much information. The default level is 2.
4060 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4063 Level 1 produces minimal information, enough for making backtraces in
4064 parts of the program that you don't plan to debug. This includes
4065 descriptions of functions and external variables, but no information
4066 about local variables and no line numbers.
4068 Level 3 includes extra information, such as all the macro definitions
4069 present in the program. Some debuggers support macro expansion when
4070 you use @option{-g3}.
4072 @option{-gdwarf-2} does not accept a concatenated debug level, because
4073 GCC used to support an option @option{-gdwarf} that meant to generate
4074 debug information in version 1 of the DWARF format (which is very
4075 different from version 2), and it would have been too confusing. That
4076 debug format is long obsolete, but the option cannot be changed now.
4077 Instead use an additional @option{-g@var{level}} option to change the
4078 debug level for DWARF2.
4080 @item -feliminate-dwarf2-dups
4081 @opindex feliminate-dwarf2-dups
4082 Compress DWARF2 debugging information by eliminating duplicated
4083 information about each symbol. This option only makes sense when
4084 generating DWARF2 debugging information with @option{-gdwarf-2}.
4086 @item -femit-struct-debug-baseonly
4087 Emit debug information for struct-like types
4088 only when the base name of the compilation source file
4089 matches the base name of file in which the struct was defined.
4091 This option substantially reduces the size of debugging information,
4092 but at significant potential loss in type information to the debugger.
4093 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4094 See @option{-femit-struct-debug-detailed} for more detailed control.
4096 This option works only with DWARF 2.
4098 @item -femit-struct-debug-reduced
4099 Emit debug information for struct-like types
4100 only when the base name of the compilation source file
4101 matches the base name of file in which the type was defined,
4102 unless the struct is a template or defined in a system header.
4104 This option significantly reduces the size of debugging information,
4105 with some potential loss in type information to the debugger.
4106 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4107 See @option{-femit-struct-debug-detailed} for more detailed control.
4109 This option works only with DWARF 2.
4111 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4112 Specify the struct-like types
4113 for which the compiler will generate debug information.
4114 The intent is to reduce duplicate struct debug information
4115 between different object files within the same program.
4117 This option is a detailed version of
4118 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4119 which will serve for most needs.
4121 A specification has the syntax
4122 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4124 The optional first word limits the specification to
4125 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4126 A struct type is used directly when it is the type of a variable, member.
4127 Indirect uses arise through pointers to structs.
4128 That is, when use of an incomplete struct would be legal, the use is indirect.
4130 @samp{struct one direct; struct two * indirect;}.
4132 The optional second word limits the specification to
4133 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4134 Generic structs are a bit complicated to explain.
4135 For C++, these are non-explicit specializations of template classes,
4136 or non-template classes within the above.
4137 Other programming languages have generics,
4138 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4140 The third word specifies the source files for those
4141 structs for which the compiler will emit debug information.
4142 The values @samp{none} and @samp{any} have the normal meaning.
4143 The value @samp{base} means that
4144 the base of name of the file in which the type declaration appears
4145 must match the base of the name of the main compilation file.
4146 In practice, this means that
4147 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4148 but types declared in other header will not.
4149 The value @samp{sys} means those types satisfying @samp{base}
4150 or declared in system or compiler headers.
4152 You may need to experiment to determine the best settings for your application.
4154 The default is @samp{-femit-struct-debug-detailed=all}.
4156 This option works only with DWARF 2.
4158 @item -fno-merge-debug-strings
4159 @opindex fmerge-debug-strings
4160 @opindex fno-merge-debug-strings
4161 Direct the linker to merge together strings which are identical in
4162 different object files. This is not supported by all assemblers or
4163 linker. This decreases the size of the debug information in the
4164 output file at the cost of increasing link processing time. This is
4167 @item -fdebug-prefix-map=@var{old}=@var{new}
4168 @opindex fdebug-prefix-map
4169 When compiling files in directory @file{@var{old}}, record debugging
4170 information describing them as in @file{@var{new}} instead.
4172 @cindex @command{prof}
4175 Generate extra code to write profile information suitable for the
4176 analysis program @command{prof}. You must use this option when compiling
4177 the source files you want data about, and you must also use it when
4180 @cindex @command{gprof}
4183 Generate extra code to write profile information suitable for the
4184 analysis program @command{gprof}. You must use this option when compiling
4185 the source files you want data about, and you must also use it when
4190 Makes the compiler print out each function name as it is compiled, and
4191 print some statistics about each pass when it finishes.
4194 @opindex ftime-report
4195 Makes the compiler print some statistics about the time consumed by each
4196 pass when it finishes.
4199 @opindex fmem-report
4200 Makes the compiler print some statistics about permanent memory
4201 allocation when it finishes.
4203 @item -fpre-ipa-mem-report
4204 @opindex fpre-ipa-mem-report
4205 @item -fpost-ipa-mem-report
4206 @opindex fpost-ipa-mem-report
4207 Makes the compiler print some statistics about permanent memory
4208 allocation before or after interprocedural optimization.
4210 @item -fprofile-arcs
4211 @opindex fprofile-arcs
4212 Add code so that program flow @dfn{arcs} are instrumented. During
4213 execution the program records how many times each branch and call is
4214 executed and how many times it is taken or returns. When the compiled
4215 program exits it saves this data to a file called
4216 @file{@var{auxname}.gcda} for each source file. The data may be used for
4217 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4218 test coverage analysis (@option{-ftest-coverage}). Each object file's
4219 @var{auxname} is generated from the name of the output file, if
4220 explicitly specified and it is not the final executable, otherwise it is
4221 the basename of the source file. In both cases any suffix is removed
4222 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4223 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4224 @xref{Cross-profiling}.
4226 @cindex @command{gcov}
4230 This option is used to compile and link code instrumented for coverage
4231 analysis. The option is a synonym for @option{-fprofile-arcs}
4232 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4233 linking). See the documentation for those options for more details.
4238 Compile the source files with @option{-fprofile-arcs} plus optimization
4239 and code generation options. For test coverage analysis, use the
4240 additional @option{-ftest-coverage} option. You do not need to profile
4241 every source file in a program.
4244 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4245 (the latter implies the former).
4248 Run the program on a representative workload to generate the arc profile
4249 information. This may be repeated any number of times. You can run
4250 concurrent instances of your program, and provided that the file system
4251 supports locking, the data files will be correctly updated. Also
4252 @code{fork} calls are detected and correctly handled (double counting
4256 For profile-directed optimizations, compile the source files again with
4257 the same optimization and code generation options plus
4258 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4259 Control Optimization}).
4262 For test coverage analysis, use @command{gcov} to produce human readable
4263 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4264 @command{gcov} documentation for further information.
4268 With @option{-fprofile-arcs}, for each function of your program GCC
4269 creates a program flow graph, then finds a spanning tree for the graph.
4270 Only arcs that are not on the spanning tree have to be instrumented: the
4271 compiler adds code to count the number of times that these arcs are
4272 executed. When an arc is the only exit or only entrance to a block, the
4273 instrumentation code can be added to the block; otherwise, a new basic
4274 block must be created to hold the instrumentation code.
4277 @item -ftest-coverage
4278 @opindex ftest-coverage
4279 Produce a notes file that the @command{gcov} code-coverage utility
4280 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4281 show program coverage. Each source file's note file is called
4282 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4283 above for a description of @var{auxname} and instructions on how to
4284 generate test coverage data. Coverage data will match the source files
4285 more closely, if you do not optimize.
4287 @item -fdbg-cnt-list
4288 @opindex fdbg-cnt-list
4289 Print the name and the counter upperbound for all debug counters.
4291 @item -fdbg-cnt=@var{counter-value-list}
4293 Set the internal debug counter upperbound. @var{counter-value-list}
4294 is a comma-separated list of @var{name}:@var{value} pairs
4295 which sets the upperbound of each debug counter @var{name} to @var{value}.
4296 All debug counters have the initial upperbound of @var{UINT_MAX},
4297 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4298 e.g. With -fdbg-cnt=dce:10,tail_call:0
4299 dbg_cnt(dce) will return true only for first 10 invocations
4300 and dbg_cnt(tail_call) will return false always.
4302 @item -d@var{letters}
4303 @item -fdump-rtl-@var{pass}
4305 Says to make debugging dumps during compilation at times specified by
4306 @var{letters}. This is used for debugging the RTL-based passes of the
4307 compiler. The file names for most of the dumps are made by appending a
4308 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4309 from the name of the output file, if explicitly specified and it is not
4310 an executable, otherwise it is the basename of the source file. These
4311 switches may have different effects when @option{-E} is used for
4314 Most debug dumps can be enabled either passing a letter to the @option{-d}
4315 option, or with a long @option{-fdump-rtl} switch; here are the possible
4316 letters for use in @var{letters} and @var{pass}, and their meanings:
4321 Annotate the assembler output with miscellaneous debugging information.
4324 @itemx -fdump-rtl-bbro
4326 @opindex fdump-rtl-bbro
4327 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4330 @itemx -fdump-rtl-combine
4332 @opindex fdump-rtl-combine
4333 Dump after the RTL instruction combination pass, to the file
4334 @file{@var{file}.129r.combine}.
4337 @itemx -fdump-rtl-ce1
4338 @itemx -fdump-rtl-ce2
4340 @opindex fdump-rtl-ce1
4341 @opindex fdump-rtl-ce2
4342 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4343 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4344 and @option{-fdump-rtl-ce2} enable dumping after the second if
4345 conversion, to the file @file{@var{file}.130r.ce2}.
4348 @itemx -fdump-rtl-btl
4349 @itemx -fdump-rtl-dbr
4351 @opindex fdump-rtl-btl
4352 @opindex fdump-rtl-dbr
4353 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4354 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4355 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4356 scheduling, to @file{@var{file}.36.dbr}.
4360 Dump all macro definitions, at the end of preprocessing, in addition to
4364 @itemx -fdump-rtl-ce3
4366 @opindex fdump-rtl-ce3
4367 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4370 @itemx -fdump-rtl-cfg
4371 @itemx -fdump-rtl-life
4373 @opindex fdump-rtl-cfg
4374 @opindex fdump-rtl-life
4375 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4376 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4377 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4378 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4381 @itemx -fdump-rtl-greg
4383 @opindex fdump-rtl-greg
4384 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4387 @itemx -fdump-rtl-gcse
4388 @itemx -fdump-rtl-bypass
4390 @opindex fdump-rtl-gcse
4391 @opindex fdump-rtl-bypass
4392 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4393 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4394 enable dumping after jump bypassing and control flow optimizations, to
4395 @file{@var{file}.115r.bypass}.
4398 @itemx -fdump-rtl-eh
4400 @opindex fdump-rtl-eh
4401 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4404 @itemx -fdump-rtl-sibling
4406 @opindex fdump-rtl-sibling
4407 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4410 @itemx -fdump-rtl-jump
4412 @opindex fdump-rtl-jump
4413 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4416 @itemx -fdump-rtl-stack
4418 @opindex fdump-rtl-stack
4419 Dump after conversion from GCC's "flat register file" registers to the
4420 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4423 @itemx -fdump-rtl-lreg
4425 @opindex fdump-rtl-lreg
4426 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4429 @itemx -fdump-rtl-loop2
4431 @opindex fdump-rtl-loop2
4432 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4433 loop optimization pass, to @file{@var{file}.119r.loop2},
4434 @file{@var{file}.120r.loop2_init},
4435 @file{@var{file}.121r.loop2_invariant}, and
4436 @file{@var{file}.125r.loop2_done}.
4439 @itemx -fdump-rtl-sms
4441 @opindex fdump-rtl-sms
4442 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4445 @itemx -fdump-rtl-mach
4447 @opindex fdump-rtl-mach
4448 Dump after performing the machine dependent reorganization pass, to
4449 @file{@var{file}.155r.mach} if that pass exists.
4452 @itemx -fdump-rtl-rnreg
4454 @opindex fdump-rtl-rnreg
4455 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4458 @itemx -fdump-rtl-regmove
4460 @opindex fdump-rtl-regmove
4461 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4464 @itemx -fdump-rtl-postreload
4466 @opindex fdump-rtl-postreload
4467 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4470 @itemx -fdump-rtl-expand
4472 @opindex fdump-rtl-expand
4473 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4476 @itemx -fdump-rtl-sched2
4478 @opindex fdump-rtl-sched2
4479 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4482 @itemx -fdump-rtl-cse
4484 @opindex fdump-rtl-cse
4485 Dump after CSE (including the jump optimization that sometimes follows
4486 CSE), to @file{@var{file}.113r.cse}.
4489 @itemx -fdump-rtl-sched1
4491 @opindex fdump-rtl-sched1
4492 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4495 @itemx -fdump-rtl-cse2
4497 @opindex fdump-rtl-cse2
4498 Dump after the second CSE pass (including the jump optimization that
4499 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4502 @itemx -fdump-rtl-tracer
4504 @opindex fdump-rtl-tracer
4505 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4508 @itemx -fdump-rtl-vpt
4509 @itemx -fdump-rtl-vartrack
4511 @opindex fdump-rtl-vpt
4512 @opindex fdump-rtl-vartrack
4513 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4514 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4515 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4516 to @file{@var{file}.154r.vartrack}.
4519 @itemx -fdump-rtl-flow2
4521 @opindex fdump-rtl-flow2
4522 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4525 @itemx -fdump-rtl-peephole2
4527 @opindex fdump-rtl-peephole2
4528 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4531 @itemx -fdump-rtl-web
4533 @opindex fdump-rtl-web
4534 Dump after live range splitting, to @file{@var{file}.126r.web}.
4537 @itemx -fdump-rtl-all
4539 @opindex fdump-rtl-all
4540 Produce all the dumps listed above.
4544 Produce a core dump whenever an error occurs.
4548 Print statistics on memory usage, at the end of the run, to
4553 Annotate the assembler output with a comment indicating which
4554 pattern and alternative was used. The length of each instruction is
4559 Dump the RTL in the assembler output as a comment before each instruction.
4560 Also turns on @option{-dp} annotation.
4564 For each of the other indicated dump files (either with @option{-d} or
4565 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4566 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4570 Just generate RTL for a function instead of compiling it. Usually used
4571 with @samp{r} (@option{-fdump-rtl-expand}).
4575 Dump debugging information during parsing, to standard error.
4579 @opindex fdump-noaddr
4580 When doing debugging dumps (see @option{-d} option above), suppress
4581 address output. This makes it more feasible to use diff on debugging
4582 dumps for compiler invocations with different compiler binaries and/or
4583 different text / bss / data / heap / stack / dso start locations.
4585 @item -fdump-unnumbered
4586 @opindex fdump-unnumbered
4587 When doing debugging dumps (see @option{-d} option above), suppress instruction
4588 numbers and address output. This makes it more feasible to
4589 use diff on debugging dumps for compiler invocations with different
4590 options, in particular with and without @option{-g}.
4592 @item -fdump-translation-unit @r{(C++ only)}
4593 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4594 @opindex fdump-translation-unit
4595 Dump a representation of the tree structure for the entire translation
4596 unit to a file. The file name is made by appending @file{.tu} to the
4597 source file name. If the @samp{-@var{options}} form is used, @var{options}
4598 controls the details of the dump as described for the
4599 @option{-fdump-tree} options.
4601 @item -fdump-class-hierarchy @r{(C++ only)}
4602 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4603 @opindex fdump-class-hierarchy
4604 Dump a representation of each class's hierarchy and virtual function
4605 table layout to a file. The file name is made by appending @file{.class}
4606 to the source file name. If the @samp{-@var{options}} form is used,
4607 @var{options} controls the details of the dump as described for the
4608 @option{-fdump-tree} options.
4610 @item -fdump-ipa-@var{switch}
4612 Control the dumping at various stages of inter-procedural analysis
4613 language tree to a file. The file name is generated by appending a switch
4614 specific suffix to the source file name. The following dumps are possible:
4618 Enables all inter-procedural analysis dumps; currently the only produced
4619 dump is the @samp{cgraph} dump.
4622 Dumps information about call-graph optimization, unused function removal,
4623 and inlining decisions.
4626 @item -fdump-tree-@var{switch}
4627 @itemx -fdump-tree-@var{switch}-@var{options}
4629 Control the dumping at various stages of processing the intermediate
4630 language tree to a file. The file name is generated by appending a switch
4631 specific suffix to the source file name. If the @samp{-@var{options}}
4632 form is used, @var{options} is a list of @samp{-} separated options that
4633 control the details of the dump. Not all options are applicable to all
4634 dumps, those which are not meaningful will be ignored. The following
4635 options are available
4639 Print the address of each node. Usually this is not meaningful as it
4640 changes according to the environment and source file. Its primary use
4641 is for tying up a dump file with a debug environment.
4643 Inhibit dumping of members of a scope or body of a function merely
4644 because that scope has been reached. Only dump such items when they
4645 are directly reachable by some other path. When dumping pretty-printed
4646 trees, this option inhibits dumping the bodies of control structures.
4648 Print a raw representation of the tree. By default, trees are
4649 pretty-printed into a C-like representation.
4651 Enable more detailed dumps (not honored by every dump option).
4653 Enable dumping various statistics about the pass (not honored by every dump
4656 Enable showing basic block boundaries (disabled in raw dumps).
4658 Enable showing virtual operands for every statement.
4660 Enable showing line numbers for statements.
4662 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4664 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4667 The following tree dumps are possible:
4671 Dump before any tree based optimization, to @file{@var{file}.original}.
4674 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4677 Dump after function inlining, to @file{@var{file}.inlined}.
4680 @opindex fdump-tree-gimple
4681 Dump each function before and after the gimplification pass to a file. The
4682 file name is made by appending @file{.gimple} to the source file name.
4685 @opindex fdump-tree-cfg
4686 Dump the control flow graph of each function to a file. The file name is
4687 made by appending @file{.cfg} to the source file name.
4690 @opindex fdump-tree-vcg
4691 Dump the control flow graph of each function to a file in VCG format. The
4692 file name is made by appending @file{.vcg} to the source file name. Note
4693 that if the file contains more than one function, the generated file cannot
4694 be used directly by VCG@. You will need to cut and paste each function's
4695 graph into its own separate file first.
4698 @opindex fdump-tree-ch
4699 Dump each function after copying loop headers. The file name is made by
4700 appending @file{.ch} to the source file name.
4703 @opindex fdump-tree-ssa
4704 Dump SSA related information to a file. The file name is made by appending
4705 @file{.ssa} to the source file name.
4708 @opindex fdump-tree-salias
4709 Dump structure aliasing variable information to a file. This file name
4710 is made by appending @file{.salias} to the source file name.
4713 @opindex fdump-tree-alias
4714 Dump aliasing information for each function. The file name is made by
4715 appending @file{.alias} to the source file name.
4718 @opindex fdump-tree-ccp
4719 Dump each function after CCP@. The file name is made by appending
4720 @file{.ccp} to the source file name.
4723 @opindex fdump-tree-storeccp
4724 Dump each function after STORE-CCP. The file name is made by appending
4725 @file{.storeccp} to the source file name.
4728 @opindex fdump-tree-pre
4729 Dump trees after partial redundancy elimination. The file name is made
4730 by appending @file{.pre} to the source file name.
4733 @opindex fdump-tree-fre
4734 Dump trees after full redundancy elimination. The file name is made
4735 by appending @file{.fre} to the source file name.
4738 @opindex fdump-tree-copyprop
4739 Dump trees after copy propagation. The file name is made
4740 by appending @file{.copyprop} to the source file name.
4742 @item store_copyprop
4743 @opindex fdump-tree-store_copyprop
4744 Dump trees after store copy-propagation. The file name is made
4745 by appending @file{.store_copyprop} to the source file name.
4748 @opindex fdump-tree-dce
4749 Dump each function after dead code elimination. The file name is made by
4750 appending @file{.dce} to the source file name.
4753 @opindex fdump-tree-mudflap
4754 Dump each function after adding mudflap instrumentation. The file name is
4755 made by appending @file{.mudflap} to the source file name.
4758 @opindex fdump-tree-sra
4759 Dump each function after performing scalar replacement of aggregates. The
4760 file name is made by appending @file{.sra} to the source file name.
4763 @opindex fdump-tree-sink
4764 Dump each function after performing code sinking. The file name is made
4765 by appending @file{.sink} to the source file name.
4768 @opindex fdump-tree-dom
4769 Dump each function after applying dominator tree optimizations. The file
4770 name is made by appending @file{.dom} to the source file name.
4773 @opindex fdump-tree-dse
4774 Dump each function after applying dead store elimination. The file
4775 name is made by appending @file{.dse} to the source file name.
4778 @opindex fdump-tree-phiopt
4779 Dump each function after optimizing PHI nodes into straightline code. The file
4780 name is made by appending @file{.phiopt} to the source file name.
4783 @opindex fdump-tree-forwprop
4784 Dump each function after forward propagating single use variables. The file
4785 name is made by appending @file{.forwprop} to the source file name.
4788 @opindex fdump-tree-copyrename
4789 Dump each function after applying the copy rename optimization. The file
4790 name is made by appending @file{.copyrename} to the source file name.
4793 @opindex fdump-tree-nrv
4794 Dump each function after applying the named return value optimization on
4795 generic trees. The file name is made by appending @file{.nrv} to the source
4799 @opindex fdump-tree-vect
4800 Dump each function after applying vectorization of loops. The file name is
4801 made by appending @file{.vect} to the source file name.
4804 @opindex fdump-tree-vrp
4805 Dump each function after Value Range Propagation (VRP). The file name
4806 is made by appending @file{.vrp} to the source file name.
4809 @opindex fdump-tree-all
4810 Enable all the available tree dumps with the flags provided in this option.
4813 @item -ftree-vectorizer-verbose=@var{n}
4814 @opindex ftree-vectorizer-verbose
4815 This option controls the amount of debugging output the vectorizer prints.
4816 This information is written to standard error, unless
4817 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4818 in which case it is output to the usual dump listing file, @file{.vect}.
4819 For @var{n}=0 no diagnostic information is reported.
4820 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4821 and the total number of loops that got vectorized.
4822 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4823 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4824 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4825 level that @option{-fdump-tree-vect-stats} uses.
4826 Higher verbosity levels mean either more information dumped for each
4827 reported loop, or same amount of information reported for more loops:
4828 If @var{n}=3, alignment related information is added to the reports.
4829 If @var{n}=4, data-references related information (e.g. memory dependences,
4830 memory access-patterns) is added to the reports.
4831 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4832 that did not pass the first analysis phase (i.e. may not be countable, or
4833 may have complicated control-flow).
4834 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4835 For @var{n}=7, all the information the vectorizer generates during its
4836 analysis and transformation is reported. This is the same verbosity level
4837 that @option{-fdump-tree-vect-details} uses.
4839 @item -frandom-seed=@var{string}
4840 @opindex frandom-string
4841 This option provides a seed that GCC uses when it would otherwise use
4842 random numbers. It is used to generate certain symbol names
4843 that have to be different in every compiled file. It is also used to
4844 place unique stamps in coverage data files and the object files that
4845 produce them. You can use the @option{-frandom-seed} option to produce
4846 reproducibly identical object files.
4848 The @var{string} should be different for every file you compile.
4850 @item -fsched-verbose=@var{n}
4851 @opindex fsched-verbose
4852 On targets that use instruction scheduling, this option controls the
4853 amount of debugging output the scheduler prints. This information is
4854 written to standard error, unless @option{-dS} or @option{-dR} is
4855 specified, in which case it is output to the usual dump
4856 listing file, @file{.sched} or @file{.sched2} respectively. However
4857 for @var{n} greater than nine, the output is always printed to standard
4860 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4861 same information as @option{-dRS}. For @var{n} greater than one, it
4862 also output basic block probabilities, detailed ready list information
4863 and unit/insn info. For @var{n} greater than two, it includes RTL
4864 at abort point, control-flow and regions info. And for @var{n} over
4865 four, @option{-fsched-verbose} also includes dependence info.
4869 Store the usual ``temporary'' intermediate files permanently; place them
4870 in the current directory and name them based on the source file. Thus,
4871 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4872 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4873 preprocessed @file{foo.i} output file even though the compiler now
4874 normally uses an integrated preprocessor.
4876 When used in combination with the @option{-x} command line option,
4877 @option{-save-temps} is sensible enough to avoid over writing an
4878 input source file with the same extension as an intermediate file.
4879 The corresponding intermediate file may be obtained by renaming the
4880 source file before using @option{-save-temps}.
4884 Report the CPU time taken by each subprocess in the compilation
4885 sequence. For C source files, this is the compiler proper and assembler
4886 (plus the linker if linking is done). The output looks like this:
4893 The first number on each line is the ``user time'', that is time spent
4894 executing the program itself. The second number is ``system time'',
4895 time spent executing operating system routines on behalf of the program.
4896 Both numbers are in seconds.
4898 @item -fvar-tracking
4899 @opindex fvar-tracking
4900 Run variable tracking pass. It computes where variables are stored at each
4901 position in code. Better debugging information is then generated
4902 (if the debugging information format supports this information).
4904 It is enabled by default when compiling with optimization (@option{-Os},
4905 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4906 the debug info format supports it.
4908 @item -print-file-name=@var{library}
4909 @opindex print-file-name
4910 Print the full absolute name of the library file @var{library} that
4911 would be used when linking---and don't do anything else. With this
4912 option, GCC does not compile or link anything; it just prints the
4915 @item -print-multi-directory
4916 @opindex print-multi-directory
4917 Print the directory name corresponding to the multilib selected by any
4918 other switches present in the command line. This directory is supposed
4919 to exist in @env{GCC_EXEC_PREFIX}.
4921 @item -print-multi-lib
4922 @opindex print-multi-lib
4923 Print the mapping from multilib directory names to compiler switches
4924 that enable them. The directory name is separated from the switches by
4925 @samp{;}, and each switch starts with an @samp{@@} instead of the
4926 @samp{-}, without spaces between multiple switches. This is supposed to
4927 ease shell-processing.
4929 @item -print-prog-name=@var{program}
4930 @opindex print-prog-name
4931 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4933 @item -print-libgcc-file-name
4934 @opindex print-libgcc-file-name
4935 Same as @option{-print-file-name=libgcc.a}.
4937 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4938 but you do want to link with @file{libgcc.a}. You can do
4941 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4944 @item -print-search-dirs
4945 @opindex print-search-dirs
4946 Print the name of the configured installation directory and a list of
4947 program and library directories @command{gcc} will search---and don't do anything else.
4949 This is useful when @command{gcc} prints the error message
4950 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4951 To resolve this you either need to put @file{cpp0} and the other compiler
4952 components where @command{gcc} expects to find them, or you can set the environment
4953 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4954 Don't forget the trailing @samp{/}.
4955 @xref{Environment Variables}.
4957 @item -print-sysroot-headers-suffix
4958 @opindex print-sysroot-headers-suffix
4959 Print the suffix added to the target sysroot when searching for
4960 headers, or give an error if the compiler is not configured with such
4961 a suffix---and don't do anything else.
4964 @opindex dumpmachine
4965 Print the compiler's target machine (for example,
4966 @samp{i686-pc-linux-gnu})---and don't do anything else.
4969 @opindex dumpversion
4970 Print the compiler version (for example, @samp{3.0})---and don't do
4975 Print the compiler's built-in specs---and don't do anything else. (This
4976 is used when GCC itself is being built.) @xref{Spec Files}.
4978 @item -feliminate-unused-debug-types
4979 @opindex feliminate-unused-debug-types
4980 Normally, when producing DWARF2 output, GCC will emit debugging
4981 information for all types declared in a compilation
4982 unit, regardless of whether or not they are actually used
4983 in that compilation unit. Sometimes this is useful, such as
4984 if, in the debugger, you want to cast a value to a type that is
4985 not actually used in your program (but is declared). More often,
4986 however, this results in a significant amount of wasted space.
4987 With this option, GCC will avoid producing debug symbol output
4988 for types that are nowhere used in the source file being compiled.
4991 @node Optimize Options
4992 @section Options That Control Optimization
4993 @cindex optimize options
4994 @cindex options, optimization
4996 These options control various sorts of optimizations.
4998 Without any optimization option, the compiler's goal is to reduce the
4999 cost of compilation and to make debugging produce the expected
5000 results. Statements are independent: if you stop the program with a
5001 breakpoint between statements, you can then assign a new value to any
5002 variable or change the program counter to any other statement in the
5003 function and get exactly the results you would expect from the source
5006 Turning on optimization flags makes the compiler attempt to improve
5007 the performance and/or code size at the expense of compilation time
5008 and possibly the ability to debug the program.
5010 The compiler performs optimization based on the knowledge it has of
5011 the program. Optimization levels @option{-O} and above, in
5012 particular, enable @emph{unit-at-a-time} mode, which allows the
5013 compiler to consider information gained from later functions in
5014 the file when compiling a function. Compiling multiple files at
5015 once to a single output file in @emph{unit-at-a-time} mode allows
5016 the compiler to use information gained from all of the files when
5017 compiling each of them.
5019 Not all optimizations are controlled directly by a flag. Only
5020 optimizations that have a flag are listed.
5027 Optimize. Optimizing compilation takes somewhat more time, and a lot
5028 more memory for a large function.
5030 With @option{-O}, the compiler tries to reduce code size and execution
5031 time, without performing any optimizations that take a great deal of
5034 @option{-O} turns on the following optimization flags:
5037 -fcprop-registers @gol
5040 -fdelayed-branch @gol
5042 -fguess-branch-probability @gol
5043 -fif-conversion2 @gol
5044 -fif-conversion @gol
5045 -finline-small-functions @gol
5046 -fipa-pure-const @gol
5047 -fipa-reference @gol
5049 -fsplit-wide-types @gol
5052 -ftree-copyrename @gol
5054 -ftree-dominator-opts @gol
5061 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5062 where doing so does not interfere with debugging.
5066 Optimize even more. GCC performs nearly all supported optimizations
5067 that do not involve a space-speed tradeoff. The compiler does not
5068 perform loop unrolling or function inlining when you specify @option{-O2}.
5069 As compared to @option{-O}, this option increases both compilation time
5070 and the performance of the generated code.
5072 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5073 also turns on the following optimization flags:
5074 @gccoptlist{-fthread-jumps @gol
5075 -falign-functions -falign-jumps @gol
5076 -falign-loops -falign-labels @gol
5079 -fcse-follow-jumps -fcse-skip-blocks @gol
5080 -fdelete-null-pointer-checks @gol
5081 -fexpensive-optimizations @gol
5082 -fgcse -fgcse-lm @gol
5083 -foptimize-sibling-calls @gol
5086 -freorder-blocks -freorder-functions @gol
5087 -frerun-cse-after-loop @gol
5088 -fsched-interblock -fsched-spec @gol
5089 -fschedule-insns -fschedule-insns2 @gol
5090 -fstrict-aliasing -fstrict-overflow @gol
5094 Please note the warning under @option{-fgcse} about
5095 invoking @option{-O2} on programs that use computed gotos.
5099 Optimize yet more. @option{-O3} turns on all optimizations specified by
5100 @option{-O2} and also turns on the @option{-finline-functions},
5101 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5102 @option{-fgcse-after-reload} options.
5106 Reduce compilation time and make debugging produce the expected
5107 results. This is the default.
5111 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5112 do not typically increase code size. It also performs further
5113 optimizations designed to reduce code size.
5115 @option{-Os} disables the following optimization flags:
5116 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5117 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5118 -fprefetch-loop-arrays -ftree-vect-loop-version}
5120 If you use multiple @option{-O} options, with or without level numbers,
5121 the last such option is the one that is effective.
5124 Options of the form @option{-f@var{flag}} specify machine-independent
5125 flags. Most flags have both positive and negative forms; the negative
5126 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5127 below, only one of the forms is listed---the one you typically will
5128 use. You can figure out the other form by either removing @samp{no-}
5131 The following options control specific optimizations. They are either
5132 activated by @option{-O} options or are related to ones that are. You
5133 can use the following flags in the rare cases when ``fine-tuning'' of
5134 optimizations to be performed is desired.
5137 @item -fno-default-inline
5138 @opindex fno-default-inline
5139 Do not make member functions inline by default merely because they are
5140 defined inside the class scope (C++ only). Otherwise, when you specify
5141 @w{@option{-O}}, member functions defined inside class scope are compiled
5142 inline by default; i.e., you don't need to add @samp{inline} in front of
5143 the member function name.
5145 @item -fno-defer-pop
5146 @opindex fno-defer-pop
5147 Always pop the arguments to each function call as soon as that function
5148 returns. For machines which must pop arguments after a function call,
5149 the compiler normally lets arguments accumulate on the stack for several
5150 function calls and pops them all at once.
5152 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5154 @item -fforward-propagate
5155 @opindex fforward-propagate
5156 Perform a forward propagation pass on RTL. The pass tries to combine two
5157 instructions and checks if the result can be simplified. If loop unrolling
5158 is active, two passes are performed and the second is scheduled after
5161 This option is enabled by default at optimization levels @option{-O2},
5162 @option{-O3}, @option{-Os}.
5164 @item -fomit-frame-pointer
5165 @opindex fomit-frame-pointer
5166 Don't keep the frame pointer in a register for functions that
5167 don't need one. This avoids the instructions to save, set up and
5168 restore frame pointers; it also makes an extra register available
5169 in many functions. @strong{It also makes debugging impossible on
5172 On some machines, such as the VAX, this flag has no effect, because
5173 the standard calling sequence automatically handles the frame pointer
5174 and nothing is saved by pretending it doesn't exist. The
5175 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5176 whether a target machine supports this flag. @xref{Registers,,Register
5177 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5179 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5181 @item -foptimize-sibling-calls
5182 @opindex foptimize-sibling-calls
5183 Optimize sibling and tail recursive calls.
5185 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5189 Don't pay attention to the @code{inline} keyword. Normally this option
5190 is used to keep the compiler from expanding any functions inline.
5191 Note that if you are not optimizing, no functions can be expanded inline.
5193 @item -finline-small-functions
5194 @opindex finline-small-functions
5195 Integrate functions into their callers when their body is smaller than expected
5196 function call code (so overall size of program gets smaller). The compiler
5197 heuristically decides which functions are simple enough to be worth integrating
5200 Enabled at level @option{-O2}.
5202 @item -finline-functions
5203 @opindex finline-functions
5204 Integrate all simple functions into their callers. The compiler
5205 heuristically decides which functions are simple enough to be worth
5206 integrating in this way.
5208 If all calls to a given function are integrated, and the function is
5209 declared @code{static}, then the function is normally not output as
5210 assembler code in its own right.
5212 Enabled at level @option{-O3}.
5214 @item -finline-functions-called-once
5215 @opindex finline-functions-called-once
5216 Consider all @code{static} functions called once for inlining into their
5217 caller even if they are not marked @code{inline}. If a call to a given
5218 function is integrated, then the function is not output as assembler code
5221 Enabled if @option{-funit-at-a-time} is enabled.
5223 @item -fearly-inlining
5224 @opindex fearly-inlining
5225 Inline functions marked by @code{always_inline} and functions whose body seems
5226 smaller than the function call overhead early before doing
5227 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5228 makes profiling significantly cheaper and usually inlining faster on programs
5229 having large chains of nested wrapper functions.
5233 @item -finline-limit=@var{n}
5234 @opindex finline-limit
5235 By default, GCC limits the size of functions that can be inlined. This flag
5236 allows coarse control of this limit. @var{n} is the size of functions that
5237 can be inlined in number of pseudo instructions.
5239 Inlining is actually controlled by a number of parameters, which may be
5240 specified individually by using @option{--param @var{name}=@var{value}}.
5241 The @option{-finline-limit=@var{n}} option sets some of these parameters
5245 @item max-inline-insns-single
5246 is set to @var{n}/2.
5247 @item max-inline-insns-auto
5248 is set to @var{n}/2.
5251 See below for a documentation of the individual
5252 parameters controlling inlining and for the defaults of these parameters.
5254 @emph{Note:} there may be no value to @option{-finline-limit} that results
5255 in default behavior.
5257 @emph{Note:} pseudo instruction represents, in this particular context, an
5258 abstract measurement of function's size. In no way does it represent a count
5259 of assembly instructions and as such its exact meaning might change from one
5260 release to an another.
5262 @item -fkeep-inline-functions
5263 @opindex fkeep-inline-functions
5264 In C, emit @code{static} functions that are declared @code{inline}
5265 into the object file, even if the function has been inlined into all
5266 of its callers. This switch does not affect functions using the
5267 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5268 inline functions into the object file.
5270 @item -fkeep-static-consts
5271 @opindex fkeep-static-consts
5272 Emit variables declared @code{static const} when optimization isn't turned
5273 on, even if the variables aren't referenced.
5275 GCC enables this option by default. If you want to force the compiler to
5276 check if the variable was referenced, regardless of whether or not
5277 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5279 @item -fmerge-constants
5280 @opindex fmerge-constants
5281 Attempt to merge identical constants (string constants and floating point
5282 constants) across compilation units.
5284 This option is the default for optimized compilation if the assembler and
5285 linker support it. Use @option{-fno-merge-constants} to inhibit this
5288 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5290 @item -fmerge-all-constants
5291 @opindex fmerge-all-constants
5292 Attempt to merge identical constants and identical variables.
5294 This option implies @option{-fmerge-constants}. In addition to
5295 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5296 arrays or initialized constant variables with integral or floating point
5297 types. Languages like C or C++ require each non-automatic variable to
5298 have distinct location, so using this option will result in non-conforming
5301 @item -fmodulo-sched
5302 @opindex fmodulo-sched
5303 Perform swing modulo scheduling immediately before the first scheduling
5304 pass. This pass looks at innermost loops and reorders their
5305 instructions by overlapping different iterations.
5307 @item -fmodulo-sched-allow-regmoves
5308 @opindex fmodulo-sched-allow-regmoves
5309 Perform more aggressive SMS based modulo scheduling with register moves
5310 allowed. By setting this flag certain anti-dependences edges will be
5311 deleted which will trigger the generation of reg-moves based on the
5312 life-range analysis. This option is effective only with
5313 @option{-fmodulo-sched} enabled.
5315 @item -fno-branch-count-reg
5316 @opindex fno-branch-count-reg
5317 Do not use ``decrement and branch'' instructions on a count register,
5318 but instead generate a sequence of instructions that decrement a
5319 register, compare it against zero, then branch based upon the result.
5320 This option is only meaningful on architectures that support such
5321 instructions, which include x86, PowerPC, IA-64 and S/390.
5323 The default is @option{-fbranch-count-reg}.
5325 @item -fno-function-cse
5326 @opindex fno-function-cse
5327 Do not put function addresses in registers; make each instruction that
5328 calls a constant function contain the function's address explicitly.
5330 This option results in less efficient code, but some strange hacks
5331 that alter the assembler output may be confused by the optimizations
5332 performed when this option is not used.
5334 The default is @option{-ffunction-cse}
5336 @item -fno-zero-initialized-in-bss
5337 @opindex fno-zero-initialized-in-bss
5338 If the target supports a BSS section, GCC by default puts variables that
5339 are initialized to zero into BSS@. This can save space in the resulting
5342 This option turns off this behavior because some programs explicitly
5343 rely on variables going to the data section. E.g., so that the
5344 resulting executable can find the beginning of that section and/or make
5345 assumptions based on that.
5347 The default is @option{-fzero-initialized-in-bss}.
5349 @item -fmudflap -fmudflapth -fmudflapir
5353 @cindex bounds checking
5355 For front-ends that support it (C and C++), instrument all risky
5356 pointer/array dereferencing operations, some standard library
5357 string/heap functions, and some other associated constructs with
5358 range/validity tests. Modules so instrumented should be immune to
5359 buffer overflows, invalid heap use, and some other classes of C/C++
5360 programming errors. The instrumentation relies on a separate runtime
5361 library (@file{libmudflap}), which will be linked into a program if
5362 @option{-fmudflap} is given at link time. Run-time behavior of the
5363 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5364 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5367 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5368 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5369 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5370 instrumentation should ignore pointer reads. This produces less
5371 instrumentation (and therefore faster execution) and still provides
5372 some protection against outright memory corrupting writes, but allows
5373 erroneously read data to propagate within a program.
5375 @item -fthread-jumps
5376 @opindex fthread-jumps
5377 Perform optimizations where we check to see if a jump branches to a
5378 location where another comparison subsumed by the first is found. If
5379 so, the first branch is redirected to either the destination of the
5380 second branch or a point immediately following it, depending on whether
5381 the condition is known to be true or false.
5383 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5385 @item -fsplit-wide-types
5386 @opindex fsplit-wide-types
5387 When using a type that occupies multiple registers, such as @code{long
5388 long} on a 32-bit system, split the registers apart and allocate them
5389 independently. This normally generates better code for those types,
5390 but may make debugging more difficult.
5392 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5395 @item -fcse-follow-jumps
5396 @opindex fcse-follow-jumps
5397 In common subexpression elimination (CSE), scan through jump instructions
5398 when the target of the jump is not reached by any other path. For
5399 example, when CSE encounters an @code{if} statement with an
5400 @code{else} clause, CSE will follow the jump when the condition
5403 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5405 @item -fcse-skip-blocks
5406 @opindex fcse-skip-blocks
5407 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5408 follow jumps which conditionally skip over blocks. When CSE
5409 encounters a simple @code{if} statement with no else clause,
5410 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5411 body of the @code{if}.
5413 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5415 @item -frerun-cse-after-loop
5416 @opindex frerun-cse-after-loop
5417 Re-run common subexpression elimination after loop optimizations has been
5420 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5424 Perform a global common subexpression elimination pass.
5425 This pass also performs global constant and copy propagation.
5427 @emph{Note:} When compiling a program using computed gotos, a GCC
5428 extension, you may get better runtime performance if you disable
5429 the global common subexpression elimination pass by adding
5430 @option{-fno-gcse} to the command line.
5432 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5436 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5437 attempt to move loads which are only killed by stores into themselves. This
5438 allows a loop containing a load/store sequence to be changed to a load outside
5439 the loop, and a copy/store within the loop.
5441 Enabled by default when gcse is enabled.
5445 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5446 global common subexpression elimination. This pass will attempt to move
5447 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5448 loops containing a load/store sequence can be changed to a load before
5449 the loop and a store after the loop.
5451 Not enabled at any optimization level.
5455 When @option{-fgcse-las} is enabled, the global common subexpression
5456 elimination pass eliminates redundant loads that come after stores to the
5457 same memory location (both partial and full redundancies).
5459 Not enabled at any optimization level.
5461 @item -fgcse-after-reload
5462 @opindex fgcse-after-reload
5463 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5464 pass is performed after reload. The purpose of this pass is to cleanup
5467 @item -funsafe-loop-optimizations
5468 @opindex funsafe-loop-optimizations
5469 If given, the loop optimizer will assume that loop indices do not
5470 overflow, and that the loops with nontrivial exit condition are not
5471 infinite. This enables a wider range of loop optimizations even if
5472 the loop optimizer itself cannot prove that these assumptions are valid.
5473 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5474 if it finds this kind of loop.
5476 @item -fcrossjumping
5477 @opindex fcrossjumping
5478 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5479 resulting code may or may not perform better than without cross-jumping.
5481 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5483 @item -fauto-inc-dec
5484 @opindex fauto-inc-dec
5485 Combine increments or decrements of addresses with memory accesses.
5486 This pass is always skipped on architectures that do not have
5487 instructions to support this. Enabled by default at @option{-O} and
5488 higher on architectures that support this.
5492 Perform dead code elimination (DCE) on RTL.
5493 Enabled by default at @option{-O} and higher.
5497 Perform dead store elimination (DSE) on RTL.
5498 Enabled by default at @option{-O} and higher.
5500 @item -fif-conversion
5501 @opindex fif-conversion
5502 Attempt to transform conditional jumps into branch-less equivalents. This
5503 include use of conditional moves, min, max, set flags and abs instructions, and
5504 some tricks doable by standard arithmetics. The use of conditional execution
5505 on chips where it is available is controlled by @code{if-conversion2}.
5507 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5509 @item -fif-conversion2
5510 @opindex fif-conversion2
5511 Use conditional execution (where available) to transform conditional jumps into
5512 branch-less equivalents.
5514 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5516 @item -fdelete-null-pointer-checks
5517 @opindex fdelete-null-pointer-checks
5518 Use global dataflow analysis to identify and eliminate useless checks
5519 for null pointers. The compiler assumes that dereferencing a null
5520 pointer would have halted the program. If a pointer is checked after
5521 it has already been dereferenced, it cannot be null.
5523 In some environments, this assumption is not true, and programs can
5524 safely dereference null pointers. Use
5525 @option{-fno-delete-null-pointer-checks} to disable this optimization
5526 for programs which depend on that behavior.
5528 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5530 @item -fexpensive-optimizations
5531 @opindex fexpensive-optimizations
5532 Perform a number of minor optimizations that are relatively expensive.
5534 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5536 @item -foptimize-register-move
5538 @opindex foptimize-register-move
5540 Attempt to reassign register numbers in move instructions and as
5541 operands of other simple instructions in order to maximize the amount of
5542 register tying. This is especially helpful on machines with two-operand
5545 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5548 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5550 @item -fdelayed-branch
5551 @opindex fdelayed-branch
5552 If supported for the target machine, attempt to reorder instructions
5553 to exploit instruction slots available after delayed branch
5556 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5558 @item -fschedule-insns
5559 @opindex fschedule-insns
5560 If supported for the target machine, attempt to reorder instructions to
5561 eliminate execution stalls due to required data being unavailable. This
5562 helps machines that have slow floating point or memory load instructions
5563 by allowing other instructions to be issued until the result of the load
5564 or floating point instruction is required.
5566 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5568 @item -fschedule-insns2
5569 @opindex fschedule-insns2
5570 Similar to @option{-fschedule-insns}, but requests an additional pass of
5571 instruction scheduling after register allocation has been done. This is
5572 especially useful on machines with a relatively small number of
5573 registers and where memory load instructions take more than one cycle.
5575 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5577 @item -fno-sched-interblock
5578 @opindex fno-sched-interblock
5579 Don't schedule instructions across basic blocks. This is normally
5580 enabled by default when scheduling before register allocation, i.e.@:
5581 with @option{-fschedule-insns} or at @option{-O2} or higher.
5583 @item -fno-sched-spec
5584 @opindex fno-sched-spec
5585 Don't allow speculative motion of non-load instructions. This is normally
5586 enabled by default when scheduling before register allocation, i.e.@:
5587 with @option{-fschedule-insns} or at @option{-O2} or higher.
5589 @item -fsched-spec-load
5590 @opindex fsched-spec-load
5591 Allow speculative motion of some load instructions. This only makes
5592 sense when scheduling before register allocation, i.e.@: with
5593 @option{-fschedule-insns} or at @option{-O2} or higher.
5595 @item -fsched-spec-load-dangerous
5596 @opindex fsched-spec-load-dangerous
5597 Allow speculative motion of more load instructions. This only makes
5598 sense when scheduling before register allocation, i.e.@: with
5599 @option{-fschedule-insns} or at @option{-O2} or higher.
5601 @item -fsched-stalled-insns
5602 @itemx -fsched-stalled-insns=@var{n}
5603 @opindex fsched-stalled-insns
5604 Define how many insns (if any) can be moved prematurely from the queue
5605 of stalled insns into the ready list, during the second scheduling pass.
5606 @option{-fno-sched-stalled-insns} means that no insns will be moved
5607 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5608 on how many queued insns can be moved prematurely.
5609 @option{-fsched-stalled-insns} without a value is equivalent to
5610 @option{-fsched-stalled-insns=1}.
5612 @item -fsched-stalled-insns-dep
5613 @itemx -fsched-stalled-insns-dep=@var{n}
5614 @opindex fsched-stalled-insns-dep
5615 Define how many insn groups (cycles) will be examined for a dependency
5616 on a stalled insn that is candidate for premature removal from the queue
5617 of stalled insns. This has an effect only during the second scheduling pass,
5618 and only if @option{-fsched-stalled-insns} is used.
5619 @option{-fno-sched-stalled-insns-dep} is equivalent to
5620 @option{-fsched-stalled-insns-dep=0}.
5621 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5622 @option{-fsched-stalled-insns-dep=1}.
5624 @item -fsched2-use-superblocks
5625 @opindex fsched2-use-superblocks
5626 When scheduling after register allocation, do use superblock scheduling
5627 algorithm. Superblock scheduling allows motion across basic block boundaries
5628 resulting on faster schedules. This option is experimental, as not all machine
5629 descriptions used by GCC model the CPU closely enough to avoid unreliable
5630 results from the algorithm.
5632 This only makes sense when scheduling after register allocation, i.e.@: with
5633 @option{-fschedule-insns2} or at @option{-O2} or higher.
5635 @item -fsched2-use-traces
5636 @opindex fsched2-use-traces
5637 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5638 allocation and additionally perform code duplication in order to increase the
5639 size of superblocks using tracer pass. See @option{-ftracer} for details on
5642 This mode should produce faster but significantly longer programs. Also
5643 without @option{-fbranch-probabilities} the traces constructed may not
5644 match the reality and hurt the performance. This only makes
5645 sense when scheduling after register allocation, i.e.@: with
5646 @option{-fschedule-insns2} or at @option{-O2} or higher.
5650 Eliminate redundant sign extension instructions and move the non-redundant
5651 ones to optimal placement using lazy code motion (LCM).
5653 @item -freschedule-modulo-scheduled-loops
5654 @opindex freschedule-modulo-scheduled-loops
5655 The modulo scheduling comes before the traditional scheduling, if a loop
5656 was modulo scheduled we may want to prevent the later scheduling passes
5657 from changing its schedule, we use this option to control that.
5659 @item -fcaller-saves
5660 @opindex fcaller-saves
5661 Enable values to be allocated in registers that will be clobbered by
5662 function calls, by emitting extra instructions to save and restore the
5663 registers around such calls. Such allocation is done only when it
5664 seems to result in better code than would otherwise be produced.
5666 This option is always enabled by default on certain machines, usually
5667 those which have no call-preserved registers to use instead.
5669 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5671 @item -ftree-reassoc
5672 @opindex ftree-reassoc
5673 Perform reassociation on trees. This flag is enabled by default
5674 at @option{-O} and higher.
5678 Perform partial redundancy elimination (PRE) on trees. This flag is
5679 enabled by default at @option{-O2} and @option{-O3}.
5683 Perform full redundancy elimination (FRE) on trees. The difference
5684 between FRE and PRE is that FRE only considers expressions
5685 that are computed on all paths leading to the redundant computation.
5686 This analysis is faster than PRE, though it exposes fewer redundancies.
5687 This flag is enabled by default at @option{-O} and higher.
5689 @item -ftree-copy-prop
5690 @opindex ftree-copy-prop
5691 Perform copy propagation on trees. This pass eliminates unnecessary
5692 copy operations. This flag is enabled by default at @option{-O} and
5696 @opindex ftree-salias
5697 Perform structural alias analysis on trees. This flag
5698 is enabled by default at @option{-O} and higher.
5700 @item -fipa-pure-const
5701 @opindex fipa-pure-const
5702 Discover which functions are pure or constant.
5703 Enabled by default at @option{-O} and higher.
5705 @item -fipa-reference
5706 @opindex fipa-reference
5707 Discover which static variables do not escape cannot escape the
5709 Enabled by default at @option{-O} and higher.
5711 @item -fipa-struct-reorg
5712 @opindex fipa-struct-reorg
5713 Perform structure reorganization optimization, that change C-like structures
5714 layout in order to better utilize spatial locality. This transformation is
5715 affective for programs containing arrays of structures. Available in two
5716 compilation modes: profile-based (enabled with @option{-fprofile-generate})
5717 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
5718 to provide the safety of this transformation. It works only in whole program
5719 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
5720 enabled. Structures considered @samp{cold} by this transformation are not
5721 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
5723 With this flag, the program debug info reflects a new structure layout.
5727 Perform interprocedural pointer analysis.
5731 Perform interprocedural constant propagation.
5732 This optimization analyzes the program to determine when values passed
5733 to functions are constants and then optimizes accordingly.
5734 This optimization can substantially increase performance
5735 if the application has constants passed to functions, but
5736 because this optimization can create multiple copies of functions,
5737 it may significantly increase code size.
5739 @item -fipa-matrix-reorg
5740 @opindex fipa-matrix-reorg
5741 Perform matrix flattening and transposing.
5742 Matrix flattening tries to replace a m-dimensional matrix
5743 with its equivalent n-dimensional matrix, where n < m.
5744 This reduces the level of indirection needed for accessing the elements
5745 of the matrix. The second optimization is matrix transposing that
5746 attemps to change the order of the matrix's dimensions in order to
5747 improve cache locality.
5748 Both optimizations need fwhole-program flag.
5749 Transposing is enabled only if profiling information is avaliable.
5754 Perform forward store motion on trees. This flag is
5755 enabled by default at @option{-O} and higher.
5759 Perform sparse conditional constant propagation (CCP) on trees. This
5760 pass only operates on local scalar variables and is enabled by default
5761 at @option{-O} and higher.
5763 @item -ftree-store-ccp
5764 @opindex ftree-store-ccp
5765 Perform sparse conditional constant propagation (CCP) on trees. This
5766 pass operates on both local scalar variables and memory stores and
5767 loads (global variables, structures, arrays, etc). This flag is
5768 enabled by default at @option{-O2} and higher.
5772 Perform dead code elimination (DCE) on trees. This flag is enabled by
5773 default at @option{-O} and higher.
5775 @item -ftree-dominator-opts
5776 @opindex ftree-dominator-opts
5777 Perform a variety of simple scalar cleanups (constant/copy
5778 propagation, redundancy elimination, range propagation and expression
5779 simplification) based on a dominator tree traversal. This also
5780 performs jump threading (to reduce jumps to jumps). This flag is
5781 enabled by default at @option{-O} and higher.
5785 Perform dead store elimination (DSE) on trees. A dead store is a store into
5786 a memory location which will later be overwritten by another store without
5787 any intervening loads. In this case the earlier store can be deleted. This
5788 flag is enabled by default at @option{-O} and higher.
5792 Perform loop header copying on trees. This is beneficial since it increases
5793 effectiveness of code motion optimizations. It also saves one jump. This flag
5794 is enabled by default at @option{-O} and higher. It is not enabled
5795 for @option{-Os}, since it usually increases code size.
5797 @item -ftree-loop-optimize
5798 @opindex ftree-loop-optimize
5799 Perform loop optimizations on trees. This flag is enabled by default
5800 at @option{-O} and higher.
5802 @item -ftree-loop-linear
5803 @opindex ftree-loop-linear
5804 Perform linear loop transformations on tree. This flag can improve cache
5805 performance and allow further loop optimizations to take place.
5807 @item -fcheck-data-deps
5808 @opindex fcheck-data-deps
5809 Compare the results of several data dependence analyzers. This option
5810 is used for debugging the data dependence analyzers.
5812 @item -ftree-loop-im
5813 @opindex ftree-loop-im
5814 Perform loop invariant motion on trees. This pass moves only invariants that
5815 would be hard to handle at RTL level (function calls, operations that expand to
5816 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5817 operands of conditions that are invariant out of the loop, so that we can use
5818 just trivial invariantness analysis in loop unswitching. The pass also includes
5821 @item -ftree-loop-ivcanon
5822 @opindex ftree-loop-ivcanon
5823 Create a canonical counter for number of iterations in the loop for that
5824 determining number of iterations requires complicated analysis. Later
5825 optimizations then may determine the number easily. Useful especially
5826 in connection with unrolling.
5830 Perform induction variable optimizations (strength reduction, induction
5831 variable merging and induction variable elimination) on trees.
5833 @item -ftree-parallelize-loops=n
5834 @opindex ftree-parallelize-loops
5835 Parallelize loops, i.e., split their iteration space to run in n threads.
5836 This is only possible for loops whose iterations are independent
5837 and can be arbitrarily reordered. The optimization is only
5838 profitable on multiprocessor machines, for loops that are CPU-intensive,
5839 rather than constrained e.g. by memory bandwidth. This option
5840 implies @option{-pthread}, and thus is only supported on targets
5841 that have support for @option{-pthread}.
5845 Perform scalar replacement of aggregates. This pass replaces structure
5846 references with scalars to prevent committing structures to memory too
5847 early. This flag is enabled by default at @option{-O} and higher.
5849 @item -ftree-copyrename
5850 @opindex ftree-copyrename
5851 Perform copy renaming on trees. This pass attempts to rename compiler
5852 temporaries to other variables at copy locations, usually resulting in
5853 variable names which more closely resemble the original variables. This flag
5854 is enabled by default at @option{-O} and higher.
5858 Perform temporary expression replacement during the SSA->normal phase. Single
5859 use/single def temporaries are replaced at their use location with their
5860 defining expression. This results in non-GIMPLE code, but gives the expanders
5861 much more complex trees to work on resulting in better RTL generation. This is
5862 enabled by default at @option{-O} and higher.
5864 @item -ftree-vectorize
5865 @opindex ftree-vectorize
5866 Perform loop vectorization on trees.
5868 @item -ftree-vect-loop-version
5869 @opindex ftree-vect-loop-version
5870 Perform loop versioning when doing loop vectorization on trees. When a loop
5871 appears to be vectorizable except that data alignment or data dependence cannot
5872 be determined at compile time then vectorized and non-vectorized versions of
5873 the loop are generated along with runtime checks for alignment or dependence
5874 to control which version is executed. This option is enabled by default
5875 except at level @option{-Os} where it is disabled.
5877 @item -fvect-cost-model
5878 @opindex fvect-cost-model
5879 Enable cost model for vectorization.
5883 Perform Value Range Propagation on trees. This is similar to the
5884 constant propagation pass, but instead of values, ranges of values are
5885 propagated. This allows the optimizers to remove unnecessary range
5886 checks like array bound checks and null pointer checks. This is
5887 enabled by default at @option{-O2} and higher. Null pointer check
5888 elimination is only done if @option{-fdelete-null-pointer-checks} is
5893 Perform tail duplication to enlarge superblock size. This transformation
5894 simplifies the control flow of the function allowing other optimizations to do
5897 @item -funroll-loops
5898 @opindex funroll-loops
5899 Unroll loops whose number of iterations can be determined at compile
5900 time or upon entry to the loop. @option{-funroll-loops} implies
5901 @option{-frerun-cse-after-loop}. This option makes code larger,
5902 and may or may not make it run faster.
5904 @item -funroll-all-loops
5905 @opindex funroll-all-loops
5906 Unroll all loops, even if their number of iterations is uncertain when
5907 the loop is entered. This usually makes programs run more slowly.
5908 @option{-funroll-all-loops} implies the same options as
5909 @option{-funroll-loops},
5911 @item -fsplit-ivs-in-unroller
5912 @opindex fsplit-ivs-in-unroller
5913 Enables expressing of values of induction variables in later iterations
5914 of the unrolled loop using the value in the first iteration. This breaks
5915 long dependency chains, thus improving efficiency of the scheduling passes.
5917 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5918 same effect. However in cases the loop body is more complicated than
5919 a single basic block, this is not reliable. It also does not work at all
5920 on some of the architectures due to restrictions in the CSE pass.
5922 This optimization is enabled by default.
5924 @item -fvariable-expansion-in-unroller
5925 @opindex fvariable-expansion-in-unroller
5926 With this option, the compiler will create multiple copies of some
5927 local variables when unrolling a loop which can result in superior code.
5929 @item -fpredictive-commoning
5930 @opindex fpredictive-commoning
5931 Perform predictive commoning optimization, i.e., reusing computations
5932 (especially memory loads and stores) performed in previous
5933 iterations of loops.
5935 This option is enabled at level @option{-O3}.
5937 @item -fprefetch-loop-arrays
5938 @opindex fprefetch-loop-arrays
5939 If supported by the target machine, generate instructions to prefetch
5940 memory to improve the performance of loops that access large arrays.
5942 This option may generate better or worse code; results are highly
5943 dependent on the structure of loops within the source code.
5945 Disabled at level @option{-Os}.
5948 @itemx -fno-peephole2
5949 @opindex fno-peephole
5950 @opindex fno-peephole2
5951 Disable any machine-specific peephole optimizations. The difference
5952 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5953 are implemented in the compiler; some targets use one, some use the
5954 other, a few use both.
5956 @option{-fpeephole} is enabled by default.
5957 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5959 @item -fno-guess-branch-probability
5960 @opindex fno-guess-branch-probability
5961 Do not guess branch probabilities using heuristics.
5963 GCC will use heuristics to guess branch probabilities if they are
5964 not provided by profiling feedback (@option{-fprofile-arcs}). These
5965 heuristics are based on the control flow graph. If some branch probabilities
5966 are specified by @samp{__builtin_expect}, then the heuristics will be
5967 used to guess branch probabilities for the rest of the control flow graph,
5968 taking the @samp{__builtin_expect} info into account. The interactions
5969 between the heuristics and @samp{__builtin_expect} can be complex, and in
5970 some cases, it may be useful to disable the heuristics so that the effects
5971 of @samp{__builtin_expect} are easier to understand.
5973 The default is @option{-fguess-branch-probability} at levels
5974 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5976 @item -freorder-blocks
5977 @opindex freorder-blocks
5978 Reorder basic blocks in the compiled function in order to reduce number of
5979 taken branches and improve code locality.
5981 Enabled at levels @option{-O2}, @option{-O3}.
5983 @item -freorder-blocks-and-partition
5984 @opindex freorder-blocks-and-partition
5985 In addition to reordering basic blocks in the compiled function, in order
5986 to reduce number of taken branches, partitions hot and cold basic blocks
5987 into separate sections of the assembly and .o files, to improve
5988 paging and cache locality performance.
5990 This optimization is automatically turned off in the presence of
5991 exception handling, for linkonce sections, for functions with a user-defined
5992 section attribute and on any architecture that does not support named
5995 @item -freorder-functions
5996 @opindex freorder-functions
5997 Reorder functions in the object file in order to
5998 improve code locality. This is implemented by using special
5999 subsections @code{.text.hot} for most frequently executed functions and
6000 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6001 the linker so object file format must support named sections and linker must
6002 place them in a reasonable way.
6004 Also profile feedback must be available in to make this option effective. See
6005 @option{-fprofile-arcs} for details.
6007 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6009 @item -fstrict-aliasing
6010 @opindex fstrict-aliasing
6011 Allows the compiler to assume the strictest aliasing rules applicable to
6012 the language being compiled. For C (and C++), this activates
6013 optimizations based on the type of expressions. In particular, an
6014 object of one type is assumed never to reside at the same address as an
6015 object of a different type, unless the types are almost the same. For
6016 example, an @code{unsigned int} can alias an @code{int}, but not a
6017 @code{void*} or a @code{double}. A character type may alias any other
6020 Pay special attention to code like this:
6033 The practice of reading from a different union member than the one most
6034 recently written to (called ``type-punning'') is common. Even with
6035 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6036 is accessed through the union type. So, the code above will work as
6037 expected. However, this code might not:
6048 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6050 @item -fstrict-overflow
6051 @opindex fstrict-overflow
6052 Allow the compiler to assume strict signed overflow rules, depending
6053 on the language being compiled. For C (and C++) this means that
6054 overflow when doing arithmetic with signed numbers is undefined, which
6055 means that the compiler may assume that it will not happen. This
6056 permits various optimizations. For example, the compiler will assume
6057 that an expression like @code{i + 10 > i} will always be true for
6058 signed @code{i}. This assumption is only valid if signed overflow is
6059 undefined, as the expression is false if @code{i + 10} overflows when
6060 using twos complement arithmetic. When this option is in effect any
6061 attempt to determine whether an operation on signed numbers will
6062 overflow must be written carefully to not actually involve overflow.
6064 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6065 that signed overflow is fully defined: it wraps. When
6066 @option{-fwrapv} is used, there is no difference between
6067 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
6068 @option{-fwrapv} certain types of overflow are permitted. For
6069 example, if the compiler gets an overflow when doing arithmetic on
6070 constants, the overflowed value can still be used with
6071 @option{-fwrapv}, but not otherwise.
6073 The @option{-fstrict-overflow} option is enabled at levels
6074 @option{-O2}, @option{-O3}, @option{-Os}.
6076 @item -falign-functions
6077 @itemx -falign-functions=@var{n}
6078 @opindex falign-functions
6079 Align the start of functions to the next power-of-two greater than
6080 @var{n}, skipping up to @var{n} bytes. For instance,
6081 @option{-falign-functions=32} aligns functions to the next 32-byte
6082 boundary, but @option{-falign-functions=24} would align to the next
6083 32-byte boundary only if this can be done by skipping 23 bytes or less.
6085 @option{-fno-align-functions} and @option{-falign-functions=1} are
6086 equivalent and mean that functions will not be aligned.
6088 Some assemblers only support this flag when @var{n} is a power of two;
6089 in that case, it is rounded up.
6091 If @var{n} is not specified or is zero, use a machine-dependent default.
6093 Enabled at levels @option{-O2}, @option{-O3}.
6095 @item -falign-labels
6096 @itemx -falign-labels=@var{n}
6097 @opindex falign-labels
6098 Align all branch targets to a power-of-two boundary, skipping up to
6099 @var{n} bytes like @option{-falign-functions}. This option can easily
6100 make code slower, because it must insert dummy operations for when the
6101 branch target is reached in the usual flow of the code.
6103 @option{-fno-align-labels} and @option{-falign-labels=1} are
6104 equivalent and mean that labels will not be aligned.
6106 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6107 are greater than this value, then their values are used instead.
6109 If @var{n} is not specified or is zero, use a machine-dependent default
6110 which is very likely to be @samp{1}, meaning no alignment.
6112 Enabled at levels @option{-O2}, @option{-O3}.
6115 @itemx -falign-loops=@var{n}
6116 @opindex falign-loops
6117 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6118 like @option{-falign-functions}. The hope is that the loop will be
6119 executed many times, which will make up for any execution of the dummy
6122 @option{-fno-align-loops} and @option{-falign-loops=1} are
6123 equivalent and mean that loops will not be aligned.
6125 If @var{n} is not specified or is zero, use a machine-dependent default.
6127 Enabled at levels @option{-O2}, @option{-O3}.
6130 @itemx -falign-jumps=@var{n}
6131 @opindex falign-jumps
6132 Align branch targets to a power-of-two boundary, for branch targets
6133 where the targets can only be reached by jumping, skipping up to @var{n}
6134 bytes like @option{-falign-functions}. In this case, no dummy operations
6137 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6138 equivalent and mean that loops will not be aligned.
6140 If @var{n} is not specified or is zero, use a machine-dependent default.
6142 Enabled at levels @option{-O2}, @option{-O3}.
6144 @item -funit-at-a-time
6145 @opindex funit-at-a-time
6146 Parse the whole compilation unit before starting to produce code.
6147 This allows some extra optimizations to take place but consumes
6148 more memory (in general). There are some compatibility issues
6149 with @emph{unit-at-a-time} mode:
6152 enabling @emph{unit-at-a-time} mode may change the order
6153 in which functions, variables, and top-level @code{asm} statements
6154 are emitted, and will likely break code relying on some particular
6155 ordering. The majority of such top-level @code{asm} statements,
6156 though, can be replaced by @code{section} attributes. The
6157 @option{fno-toplevel-reorder} option may be used to keep the ordering
6158 used in the input file, at the cost of some optimizations.
6161 @emph{unit-at-a-time} mode removes unreferenced static variables
6162 and functions. This may result in undefined references
6163 when an @code{asm} statement refers directly to variables or functions
6164 that are otherwise unused. In that case either the variable/function
6165 shall be listed as an operand of the @code{asm} statement operand or,
6166 in the case of top-level @code{asm} statements the attribute @code{used}
6167 shall be used on the declaration.
6170 Static functions now can use non-standard passing conventions that
6171 may break @code{asm} statements calling functions directly. Again,
6172 attribute @code{used} will prevent this behavior.
6175 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6176 but this scheme may not be supported by future releases of GCC@.
6178 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6180 @item -fno-toplevel-reorder
6181 @opindex fno-toplevel-reorder
6182 Do not reorder top-level functions, variables, and @code{asm}
6183 statements. Output them in the same order that they appear in the
6184 input file. When this option is used, unreferenced static variables
6185 will not be removed. This option is intended to support existing code
6186 which relies on a particular ordering. For new code, it is better to
6191 Constructs webs as commonly used for register allocation purposes and assign
6192 each web individual pseudo register. This allows the register allocation pass
6193 to operate on pseudos directly, but also strengthens several other optimization
6194 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6195 however, make debugging impossible, since variables will no longer stay in a
6198 Enabled by default with @option{-funroll-loops}.
6200 @item -fwhole-program
6201 @opindex fwhole-program
6202 Assume that the current compilation unit represents whole program being
6203 compiled. All public functions and variables with the exception of @code{main}
6204 and those merged by attribute @code{externally_visible} become static functions
6205 and in a affect gets more aggressively optimized by interprocedural optimizers.
6206 While this option is equivalent to proper use of @code{static} keyword for
6207 programs consisting of single file, in combination with option
6208 @option{--combine} this flag can be used to compile most of smaller scale C
6209 programs since the functions and variables become local for the whole combined
6210 compilation unit, not for the single source file itself.
6212 This option is not supported for Fortran programs.
6214 @item -fcprop-registers
6215 @opindex fcprop-registers
6216 After register allocation and post-register allocation instruction splitting,
6217 we perform a copy-propagation pass to try to reduce scheduling dependencies
6218 and occasionally eliminate the copy.
6220 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6222 @item -fprofile-generate
6223 @opindex fprofile-generate
6225 Enable options usually used for instrumenting application to produce
6226 profile useful for later recompilation with profile feedback based
6227 optimization. You must use @option{-fprofile-generate} both when
6228 compiling and when linking your program.
6230 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6233 @opindex fprofile-use
6234 Enable profile feedback directed optimizations, and optimizations
6235 generally profitable only with profile feedback available.
6237 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6238 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6240 By default, GCC emits an error message if the feedback profiles do not
6241 match the source code. This error can be turned into a warning by using
6242 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6246 The following options control compiler behavior regarding floating
6247 point arithmetic. These options trade off between speed and
6248 correctness. All must be specifically enabled.
6252 @opindex ffloat-store
6253 Do not store floating point variables in registers, and inhibit other
6254 options that might change whether a floating point value is taken from a
6257 @cindex floating point precision
6258 This option prevents undesirable excess precision on machines such as
6259 the 68000 where the floating registers (of the 68881) keep more
6260 precision than a @code{double} is supposed to have. Similarly for the
6261 x86 architecture. For most programs, the excess precision does only
6262 good, but a few programs rely on the precise definition of IEEE floating
6263 point. Use @option{-ffloat-store} for such programs, after modifying
6264 them to store all pertinent intermediate computations into variables.
6268 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6269 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6270 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6272 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6274 This option is not turned on by any @option{-O} option since
6275 it can result in incorrect output for programs which depend on
6276 an exact implementation of IEEE or ISO rules/specifications for
6277 math functions. It may, however, yield faster code for programs
6278 that do not require the guarantees of these specifications.
6280 @item -fno-math-errno
6281 @opindex fno-math-errno
6282 Do not set ERRNO after calling math functions that are executed
6283 with a single instruction, e.g., sqrt. A program that relies on
6284 IEEE exceptions for math error handling may want to use this flag
6285 for speed while maintaining IEEE arithmetic compatibility.
6287 This option is not turned on by any @option{-O} option since
6288 it can result in incorrect output for programs which depend on
6289 an exact implementation of IEEE or ISO rules/specifications for
6290 math functions. It may, however, yield faster code for programs
6291 that do not require the guarantees of these specifications.
6293 The default is @option{-fmath-errno}.
6295 On Darwin systems, the math library never sets @code{errno}. There is
6296 therefore no reason for the compiler to consider the possibility that
6297 it might, and @option{-fno-math-errno} is the default.
6299 @item -funsafe-math-optimizations
6300 @opindex funsafe-math-optimizations
6302 Allow optimizations for floating-point arithmetic that (a) assume
6303 that arguments and results are valid and (b) may violate IEEE or
6304 ANSI standards. When used at link-time, it may include libraries
6305 or startup files that change the default FPU control word or other
6306 similar optimizations.
6308 This option is not turned on by any @option{-O} option since
6309 it can result in incorrect output for programs which depend on
6310 an exact implementation of IEEE or ISO rules/specifications for
6311 math functions. It may, however, yield faster code for programs
6312 that do not require the guarantees of these specifications.
6313 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6314 @option{-fassociative-math} and @option{-freciprocal-math}.
6316 The default is @option{-fno-unsafe-math-optimizations}.
6318 @item -fassociative-math
6319 @opindex fassociative-math
6321 Allow re-association of operands in series of floating-point operations.
6322 This violates the ISO C and C++ language standard by possibly changing
6323 computation result. NOTE: re-ordering may change the sign of zero as
6324 well as ignore NaNs and inhibit or create underflow or overflow (and
6325 thus cannot be used on a code which relies on rounding behavior like
6326 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6327 and thus may not be used when ordered comparisons are required.
6328 This option requires that both @option{-fno-signed-zeros} and
6329 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6330 much sense with @option{-frounding-math}.
6332 The default is @option{-fno-associative-math}.
6334 @item -freciprocal-math
6335 @opindex freciprocal-math
6337 Allow the reciprocal of a value to be used instead of dividing by
6338 the value if this enables optimizations. For example @code{x / y}
6339 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6340 is subject to common subexpression elimination. Note that this loses
6341 precision and increases the number of flops operating on the value.
6343 The default is @option{-fno-reciprocal-math}.
6345 @item -ffinite-math-only
6346 @opindex ffinite-math-only
6347 Allow optimizations for floating-point arithmetic that assume
6348 that arguments and results are not NaNs or +-Infs.
6350 This option is not turned on by any @option{-O} option since
6351 it can result in incorrect output for programs which depend on
6352 an exact implementation of IEEE or ISO rules/specifications for
6353 math functions. It may, however, yield faster code for programs
6354 that do not require the guarantees of these specifications.
6356 The default is @option{-fno-finite-math-only}.
6358 @item -fno-signed-zeros
6359 @opindex fno-signed-zeros
6360 Allow optimizations for floating point arithmetic that ignore the
6361 signedness of zero. IEEE arithmetic specifies the behavior of
6362 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6363 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6364 This option implies that the sign of a zero result isn't significant.
6366 The default is @option{-fsigned-zeros}.
6368 @item -fno-trapping-math
6369 @opindex fno-trapping-math
6370 Compile code assuming that floating-point operations cannot generate
6371 user-visible traps. These traps include division by zero, overflow,
6372 underflow, inexact result and invalid operation. This option requires
6373 that @option{-fno-signaling-nans} be in effect. Setting this option may
6374 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6376 This option should never be turned on by any @option{-O} option since
6377 it can result in incorrect output for programs which depend on
6378 an exact implementation of IEEE or ISO rules/specifications for
6381 The default is @option{-ftrapping-math}.
6383 @item -frounding-math
6384 @opindex frounding-math
6385 Disable transformations and optimizations that assume default floating
6386 point rounding behavior. This is round-to-zero for all floating point
6387 to integer conversions, and round-to-nearest for all other arithmetic
6388 truncations. This option should be specified for programs that change
6389 the FP rounding mode dynamically, or that may be executed with a
6390 non-default rounding mode. This option disables constant folding of
6391 floating point expressions at compile-time (which may be affected by
6392 rounding mode) and arithmetic transformations that are unsafe in the
6393 presence of sign-dependent rounding modes.
6395 The default is @option{-fno-rounding-math}.
6397 This option is experimental and does not currently guarantee to
6398 disable all GCC optimizations that are affected by rounding mode.
6399 Future versions of GCC may provide finer control of this setting
6400 using C99's @code{FENV_ACCESS} pragma. This command line option
6401 will be used to specify the default state for @code{FENV_ACCESS}.
6403 @item -frtl-abstract-sequences
6404 @opindex frtl-abstract-sequences
6405 It is a size optimization method. This option is to find identical
6406 sequences of code, which can be turned into pseudo-procedures and
6407 then replace all occurrences with calls to the newly created
6408 subroutine. It is kind of an opposite of @option{-finline-functions}.
6409 This optimization runs at RTL level.
6411 @item -fsignaling-nans
6412 @opindex fsignaling-nans
6413 Compile code assuming that IEEE signaling NaNs may generate user-visible
6414 traps during floating-point operations. Setting this option disables
6415 optimizations that may change the number of exceptions visible with
6416 signaling NaNs. This option implies @option{-ftrapping-math}.
6418 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6421 The default is @option{-fno-signaling-nans}.
6423 This option is experimental and does not currently guarantee to
6424 disable all GCC optimizations that affect signaling NaN behavior.
6426 @item -fsingle-precision-constant
6427 @opindex fsingle-precision-constant
6428 Treat floating point constant as single precision constant instead of
6429 implicitly converting it to double precision constant.
6431 @item -fcx-limited-range
6432 @opindex fcx-limited-range
6433 When enabled, this option states that a range reduction step is not
6434 needed when performing complex division. The default is
6435 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6437 This option controls the default setting of the ISO C99
6438 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6443 The following options control optimizations that may improve
6444 performance, but are not enabled by any @option{-O} options. This
6445 section includes experimental options that may produce broken code.
6448 @item -fbranch-probabilities
6449 @opindex fbranch-probabilities
6450 After running a program compiled with @option{-fprofile-arcs}
6451 (@pxref{Debugging Options,, Options for Debugging Your Program or
6452 @command{gcc}}), you can compile it a second time using
6453 @option{-fbranch-probabilities}, to improve optimizations based on
6454 the number of times each branch was taken. When the program
6455 compiled with @option{-fprofile-arcs} exits it saves arc execution
6456 counts to a file called @file{@var{sourcename}.gcda} for each source
6457 file. The information in this data file is very dependent on the
6458 structure of the generated code, so you must use the same source code
6459 and the same optimization options for both compilations.
6461 With @option{-fbranch-probabilities}, GCC puts a
6462 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6463 These can be used to improve optimization. Currently, they are only
6464 used in one place: in @file{reorg.c}, instead of guessing which path a
6465 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6466 exactly determine which path is taken more often.
6468 @item -fprofile-values
6469 @opindex fprofile-values
6470 If combined with @option{-fprofile-arcs}, it adds code so that some
6471 data about values of expressions in the program is gathered.
6473 With @option{-fbranch-probabilities}, it reads back the data gathered
6474 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6475 notes to instructions for their later usage in optimizations.
6477 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6481 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6482 a code to gather information about values of expressions.
6484 With @option{-fbranch-probabilities}, it reads back the data gathered
6485 and actually performs the optimizations based on them.
6486 Currently the optimizations include specialization of division operation
6487 using the knowledge about the value of the denominator.
6489 @item -frename-registers
6490 @opindex frename-registers
6491 Attempt to avoid false dependencies in scheduled code by making use
6492 of registers left over after register allocation. This optimization
6493 will most benefit processors with lots of registers. Depending on the
6494 debug information format adopted by the target, however, it can
6495 make debugging impossible, since variables will no longer stay in
6496 a ``home register''.
6498 Enabled by default with @option{-funroll-loops}.
6502 Perform tail duplication to enlarge superblock size. This transformation
6503 simplifies the control flow of the function allowing other optimizations to do
6506 Enabled with @option{-fprofile-use}.
6508 @item -funroll-loops
6509 @opindex funroll-loops
6510 Unroll loops whose number of iterations can be determined at compile time or
6511 upon entry to the loop. @option{-funroll-loops} implies
6512 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6513 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6514 small constant number of iterations). This option makes code larger, and may
6515 or may not make it run faster.
6517 Enabled with @option{-fprofile-use}.
6519 @item -funroll-all-loops
6520 @opindex funroll-all-loops
6521 Unroll all loops, even if their number of iterations is uncertain when
6522 the loop is entered. This usually makes programs run more slowly.
6523 @option{-funroll-all-loops} implies the same options as
6524 @option{-funroll-loops}.
6527 @opindex fpeel-loops
6528 Peels the loops for that there is enough information that they do not
6529 roll much (from profile feedback). It also turns on complete loop peeling
6530 (i.e.@: complete removal of loops with small constant number of iterations).
6532 Enabled with @option{-fprofile-use}.
6534 @item -fmove-loop-invariants
6535 @opindex fmove-loop-invariants
6536 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6537 at level @option{-O1}
6539 @item -funswitch-loops
6540 @opindex funswitch-loops
6541 Move branches with loop invariant conditions out of the loop, with duplicates
6542 of the loop on both branches (modified according to result of the condition).
6544 @item -ffunction-sections
6545 @itemx -fdata-sections
6546 @opindex ffunction-sections
6547 @opindex fdata-sections
6548 Place each function or data item into its own section in the output
6549 file if the target supports arbitrary sections. The name of the
6550 function or the name of the data item determines the section's name
6553 Use these options on systems where the linker can perform optimizations
6554 to improve locality of reference in the instruction space. Most systems
6555 using the ELF object format and SPARC processors running Solaris 2 have
6556 linkers with such optimizations. AIX may have these optimizations in
6559 Only use these options when there are significant benefits from doing
6560 so. When you specify these options, the assembler and linker will
6561 create larger object and executable files and will also be slower.
6562 You will not be able to use @code{gprof} on all systems if you
6563 specify this option and you may have problems with debugging if
6564 you specify both this option and @option{-g}.
6566 @item -fbranch-target-load-optimize
6567 @opindex fbranch-target-load-optimize
6568 Perform branch target register load optimization before prologue / epilogue
6570 The use of target registers can typically be exposed only during reload,
6571 thus hoisting loads out of loops and doing inter-block scheduling needs
6572 a separate optimization pass.
6574 @item -fbranch-target-load-optimize2
6575 @opindex fbranch-target-load-optimize2
6576 Perform branch target register load optimization after prologue / epilogue
6579 @item -fbtr-bb-exclusive
6580 @opindex fbtr-bb-exclusive
6581 When performing branch target register load optimization, don't reuse
6582 branch target registers in within any basic block.
6584 @item -fstack-protector
6585 @opindex fstack-protector
6586 Emit extra code to check for buffer overflows, such as stack smashing
6587 attacks. This is done by adding a guard variable to functions with
6588 vulnerable objects. This includes functions that call alloca, and
6589 functions with buffers larger than 8 bytes. The guards are initialized
6590 when a function is entered and then checked when the function exits.
6591 If a guard check fails, an error message is printed and the program exits.
6593 @item -fstack-protector-all
6594 @opindex fstack-protector-all
6595 Like @option{-fstack-protector} except that all functions are protected.
6597 @item -fsection-anchors
6598 @opindex fsection-anchors
6599 Try to reduce the number of symbolic address calculations by using
6600 shared ``anchor'' symbols to address nearby objects. This transformation
6601 can help to reduce the number of GOT entries and GOT accesses on some
6604 For example, the implementation of the following function @code{foo}:
6608 int foo (void) @{ return a + b + c; @}
6611 would usually calculate the addresses of all three variables, but if you
6612 compile it with @option{-fsection-anchors}, it will access the variables
6613 from a common anchor point instead. The effect is similar to the
6614 following pseudocode (which isn't valid C):
6619 register int *xr = &x;
6620 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6624 Not all targets support this option.
6626 @item --param @var{name}=@var{value}
6628 In some places, GCC uses various constants to control the amount of
6629 optimization that is done. For example, GCC will not inline functions
6630 that contain more that a certain number of instructions. You can
6631 control some of these constants on the command-line using the
6632 @option{--param} option.
6634 The names of specific parameters, and the meaning of the values, are
6635 tied to the internals of the compiler, and are subject to change
6636 without notice in future releases.
6638 In each case, the @var{value} is an integer. The allowable choices for
6639 @var{name} are given in the following table:
6642 @item salias-max-implicit-fields
6643 The maximum number of fields in a variable without direct
6644 structure accesses for which structure aliasing will consider trying
6645 to track each field. The default is 5
6647 @item salias-max-array-elements
6648 The maximum number of elements an array can have and its elements
6649 still be tracked individually by structure aliasing. The default is 4
6651 @item sra-max-structure-size
6652 The maximum structure size, in bytes, at which the scalar replacement
6653 of aggregates (SRA) optimization will perform block copies. The
6654 default value, 0, implies that GCC will select the most appropriate
6657 @item sra-field-structure-ratio
6658 The threshold ratio (as a percentage) between instantiated fields and
6659 the complete structure size. We say that if the ratio of the number
6660 of bytes in instantiated fields to the number of bytes in the complete
6661 structure exceeds this parameter, then block copies are not used. The
6664 @item struct-reorg-cold-struct-ratio
6665 The threshold ratio (as a percentage) between a structure frequency
6666 and the frequency of the hottest structure in the program. This parameter
6667 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
6668 We say that if the ratio of a structure frequency, calculated by profiling,
6669 to the hottest structure frequency in the program is less than this
6670 parameter, then structure reorganization is not applied to this structure.
6673 @item max-crossjump-edges
6674 The maximum number of incoming edges to consider for crossjumping.
6675 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6676 the number of edges incoming to each block. Increasing values mean
6677 more aggressive optimization, making the compile time increase with
6678 probably small improvement in executable size.
6680 @item min-crossjump-insns
6681 The minimum number of instructions which must be matched at the end
6682 of two blocks before crossjumping will be performed on them. This
6683 value is ignored in the case where all instructions in the block being
6684 crossjumped from are matched. The default value is 5.
6686 @item max-grow-copy-bb-insns
6687 The maximum code size expansion factor when copying basic blocks
6688 instead of jumping. The expansion is relative to a jump instruction.
6689 The default value is 8.
6691 @item max-goto-duplication-insns
6692 The maximum number of instructions to duplicate to a block that jumps
6693 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6694 passes, GCC factors computed gotos early in the compilation process,
6695 and unfactors them as late as possible. Only computed jumps at the
6696 end of a basic blocks with no more than max-goto-duplication-insns are
6697 unfactored. The default value is 8.
6699 @item max-delay-slot-insn-search
6700 The maximum number of instructions to consider when looking for an
6701 instruction to fill a delay slot. If more than this arbitrary number of
6702 instructions is searched, the time savings from filling the delay slot
6703 will be minimal so stop searching. Increasing values mean more
6704 aggressive optimization, making the compile time increase with probably
6705 small improvement in executable run time.
6707 @item max-delay-slot-live-search
6708 When trying to fill delay slots, the maximum number of instructions to
6709 consider when searching for a block with valid live register
6710 information. Increasing this arbitrarily chosen value means more
6711 aggressive optimization, increasing the compile time. This parameter
6712 should be removed when the delay slot code is rewritten to maintain the
6715 @item max-gcse-memory
6716 The approximate maximum amount of memory that will be allocated in
6717 order to perform the global common subexpression elimination
6718 optimization. If more memory than specified is required, the
6719 optimization will not be done.
6721 @item max-gcse-passes
6722 The maximum number of passes of GCSE to run. The default is 1.
6724 @item max-pending-list-length
6725 The maximum number of pending dependencies scheduling will allow
6726 before flushing the current state and starting over. Large functions
6727 with few branches or calls can create excessively large lists which
6728 needlessly consume memory and resources.
6730 @item max-inline-insns-single
6731 Several parameters control the tree inliner used in gcc.
6732 This number sets the maximum number of instructions (counted in GCC's
6733 internal representation) in a single function that the tree inliner
6734 will consider for inlining. This only affects functions declared
6735 inline and methods implemented in a class declaration (C++).
6736 The default value is 450.
6738 @item max-inline-insns-auto
6739 When you use @option{-finline-functions} (included in @option{-O3}),
6740 a lot of functions that would otherwise not be considered for inlining
6741 by the compiler will be investigated. To those functions, a different
6742 (more restrictive) limit compared to functions declared inline can
6744 The default value is 90.
6746 @item large-function-insns
6747 The limit specifying really large functions. For functions larger than this
6748 limit after inlining inlining is constrained by
6749 @option{--param large-function-growth}. This parameter is useful primarily
6750 to avoid extreme compilation time caused by non-linear algorithms used by the
6752 This parameter is ignored when @option{-funit-at-a-time} is not used.
6753 The default value is 2700.
6755 @item large-function-growth
6756 Specifies maximal growth of large function caused by inlining in percents.
6757 This parameter is ignored when @option{-funit-at-a-time} is not used.
6758 The default value is 100 which limits large function growth to 2.0 times
6761 @item large-unit-insns
6762 The limit specifying large translation unit. Growth caused by inlining of
6763 units larger than this limit is limited by @option{--param inline-unit-growth}.
6764 For small units this might be too tight (consider unit consisting of function A
6765 that is inline and B that just calls A three time. If B is small relative to
6766 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6767 large units consisting of small inlininable functions however the overall unit
6768 growth limit is needed to avoid exponential explosion of code size. Thus for
6769 smaller units, the size is increased to @option{--param large-unit-insns}
6770 before applying @option{--param inline-unit-growth}. The default is 10000
6772 @item inline-unit-growth
6773 Specifies maximal overall growth of the compilation unit caused by inlining.
6774 This parameter is ignored when @option{-funit-at-a-time} is not used.
6775 The default value is 30 which limits unit growth to 1.3 times the original
6778 @item large-stack-frame
6779 The limit specifying large stack frames. While inlining the algorithm is trying
6780 to not grow past this limit too much. Default value is 256 bytes.
6782 @item large-stack-frame-growth
6783 Specifies maximal growth of large stack frames caused by inlining in percents.
6784 The default value is 1000 which limits large stack frame growth to 11 times
6787 @item max-inline-insns-recursive
6788 @itemx max-inline-insns-recursive-auto
6789 Specifies maximum number of instructions out-of-line copy of self recursive inline
6790 function can grow into by performing recursive inlining.
6792 For functions declared inline @option{--param max-inline-insns-recursive} is
6793 taken into account. For function not declared inline, recursive inlining
6794 happens only when @option{-finline-functions} (included in @option{-O3}) is
6795 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6796 default value is 450.
6798 @item max-inline-recursive-depth
6799 @itemx max-inline-recursive-depth-auto
6800 Specifies maximum recursion depth used by the recursive inlining.
6802 For functions declared inline @option{--param max-inline-recursive-depth} is
6803 taken into account. For function not declared inline, recursive inlining
6804 happens only when @option{-finline-functions} (included in @option{-O3}) is
6805 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6808 @item min-inline-recursive-probability
6809 Recursive inlining is profitable only for function having deep recursion
6810 in average and can hurt for function having little recursion depth by
6811 increasing the prologue size or complexity of function body to other
6814 When profile feedback is available (see @option{-fprofile-generate}) the actual
6815 recursion depth can be guessed from probability that function will recurse via
6816 given call expression. This parameter limits inlining only to call expression
6817 whose probability exceeds given threshold (in percents). The default value is
6820 @item inline-call-cost
6821 Specify cost of call instruction relative to simple arithmetics operations
6822 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6823 functions and at the same time increases size of leaf function that is believed to
6824 reduce function size by being inlined. In effect it increases amount of
6825 inlining for code having large abstraction penalty (many functions that just
6826 pass the arguments to other functions) and decrease inlining for code with low
6827 abstraction penalty. The default value is 12.
6829 @item min-vect-loop-bound
6830 The minimum number of iterations under which a loop will not get vectorized
6831 when @option{-ftree-vectorize} is used. The number of iterations after
6832 vectorization needs to be greater than the value specified by this option
6833 to allow vectorization. The default value is 0.
6835 @item max-unrolled-insns
6836 The maximum number of instructions that a loop should have if that loop
6837 is unrolled, and if the loop is unrolled, it determines how many times
6838 the loop code is unrolled.
6840 @item max-average-unrolled-insns
6841 The maximum number of instructions biased by probabilities of their execution
6842 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6843 it determines how many times the loop code is unrolled.
6845 @item max-unroll-times
6846 The maximum number of unrollings of a single loop.
6848 @item max-peeled-insns
6849 The maximum number of instructions that a loop should have if that loop
6850 is peeled, and if the loop is peeled, it determines how many times
6851 the loop code is peeled.
6853 @item max-peel-times
6854 The maximum number of peelings of a single loop.
6856 @item max-completely-peeled-insns
6857 The maximum number of insns of a completely peeled loop.
6859 @item max-completely-peel-times
6860 The maximum number of iterations of a loop to be suitable for complete peeling.
6862 @item max-unswitch-insns
6863 The maximum number of insns of an unswitched loop.
6865 @item max-unswitch-level
6866 The maximum number of branches unswitched in a single loop.
6869 The minimum cost of an expensive expression in the loop invariant motion.
6871 @item iv-consider-all-candidates-bound
6872 Bound on number of candidates for induction variables below that
6873 all candidates are considered for each use in induction variable
6874 optimizations. Only the most relevant candidates are considered
6875 if there are more candidates, to avoid quadratic time complexity.
6877 @item iv-max-considered-uses
6878 The induction variable optimizations give up on loops that contain more
6879 induction variable uses.
6881 @item iv-always-prune-cand-set-bound
6882 If number of candidates in the set is smaller than this value,
6883 we always try to remove unnecessary ivs from the set during its
6884 optimization when a new iv is added to the set.
6886 @item scev-max-expr-size
6887 Bound on size of expressions used in the scalar evolutions analyzer.
6888 Large expressions slow the analyzer.
6890 @item omega-max-vars
6891 The maximum number of variables in an Omega constraint system.
6892 The default value is 128.
6894 @item omega-max-geqs
6895 The maximum number of inequalities in an Omega constraint system.
6896 The default value is 256.
6899 The maximum number of equalities in an Omega constraint system.
6900 The default value is 128.
6902 @item omega-max-wild-cards
6903 The maximum number of wildcard variables that the Omega solver will
6904 be able to insert. The default value is 18.
6906 @item omega-hash-table-size
6907 The size of the hash table in the Omega solver. The default value is
6910 @item omega-max-keys
6911 The maximal number of keys used by the Omega solver. The default
6914 @item omega-eliminate-redundant-constraints
6915 When set to 1, use expensive methods to eliminate all redundant
6916 constraints. The default value is 0.
6918 @item vect-max-version-for-alignment-checks
6919 The maximum number of runtime checks that can be performed when
6920 doing loop versioning for alignment in the vectorizer. See option
6921 ftree-vect-loop-version for more information.
6923 @item vect-max-version-for-alias-checks
6924 The maximum number of runtime checks that can be performed when
6925 doing loop versioning for alias in the vectorizer. See option
6926 ftree-vect-loop-version for more information.
6928 @item max-iterations-to-track
6930 The maximum number of iterations of a loop the brute force algorithm
6931 for analysis of # of iterations of the loop tries to evaluate.
6933 @item hot-bb-count-fraction
6934 Select fraction of the maximal count of repetitions of basic block in program
6935 given basic block needs to have to be considered hot.
6937 @item hot-bb-frequency-fraction
6938 Select fraction of the maximal frequency of executions of basic block in
6939 function given basic block needs to have to be considered hot
6941 @item max-predicted-iterations
6942 The maximum number of loop iterations we predict statically. This is useful
6943 in cases where function contain single loop with known bound and other loop
6944 with unknown. We predict the known number of iterations correctly, while
6945 the unknown number of iterations average to roughly 10. This means that the
6946 loop without bounds would appear artificially cold relative to the other one.
6948 @item align-threshold
6950 Select fraction of the maximal frequency of executions of basic block in
6951 function given basic block will get aligned.
6953 @item align-loop-iterations
6955 A loop expected to iterate at lest the selected number of iterations will get
6958 @item tracer-dynamic-coverage
6959 @itemx tracer-dynamic-coverage-feedback
6961 This value is used to limit superblock formation once the given percentage of
6962 executed instructions is covered. This limits unnecessary code size
6965 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6966 feedback is available. The real profiles (as opposed to statically estimated
6967 ones) are much less balanced allowing the threshold to be larger value.
6969 @item tracer-max-code-growth
6970 Stop tail duplication once code growth has reached given percentage. This is
6971 rather hokey argument, as most of the duplicates will be eliminated later in
6972 cross jumping, so it may be set to much higher values than is the desired code
6975 @item tracer-min-branch-ratio
6977 Stop reverse growth when the reverse probability of best edge is less than this
6978 threshold (in percent).
6980 @item tracer-min-branch-ratio
6981 @itemx tracer-min-branch-ratio-feedback
6983 Stop forward growth if the best edge do have probability lower than this
6986 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6987 compilation for profile feedback and one for compilation without. The value
6988 for compilation with profile feedback needs to be more conservative (higher) in
6989 order to make tracer effective.
6991 @item max-cse-path-length
6993 Maximum number of basic blocks on path that cse considers. The default is 10.
6996 The maximum instructions CSE process before flushing. The default is 1000.
6998 @item max-aliased-vops
7000 Maximum number of virtual operands per function allowed to represent
7001 aliases before triggering the alias partitioning heuristic. Alias
7002 partitioning reduces compile times and memory consumption needed for
7003 aliasing at the expense of precision loss in alias information. The
7004 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7007 Notice that if a function contains more memory statements than the
7008 value of this parameter, it is not really possible to achieve this
7009 reduction. In this case, the compiler will use the number of memory
7010 statements as the value for @option{max-aliased-vops}.
7012 @item avg-aliased-vops
7014 Average number of virtual operands per statement allowed to represent
7015 aliases before triggering the alias partitioning heuristic. This
7016 works in conjunction with @option{max-aliased-vops}. If a function
7017 contains more than @option{max-aliased-vops} virtual operators, then
7018 memory symbols will be grouped into memory partitions until either the
7019 total number of virtual operators is below @option{max-aliased-vops}
7020 or the average number of virtual operators per memory statement is
7021 below @option{avg-aliased-vops}. The default value for this parameter
7022 is 1 for -O1 and -O2, and 3 for -O3.
7024 @item ggc-min-expand
7026 GCC uses a garbage collector to manage its own memory allocation. This
7027 parameter specifies the minimum percentage by which the garbage
7028 collector's heap should be allowed to expand between collections.
7029 Tuning this may improve compilation speed; it has no effect on code
7032 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7033 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7034 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7035 GCC is not able to calculate RAM on a particular platform, the lower
7036 bound of 30% is used. Setting this parameter and
7037 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7038 every opportunity. This is extremely slow, but can be useful for
7041 @item ggc-min-heapsize
7043 Minimum size of the garbage collector's heap before it begins bothering
7044 to collect garbage. The first collection occurs after the heap expands
7045 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7046 tuning this may improve compilation speed, and has no effect on code
7049 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7050 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7051 with a lower bound of 4096 (four megabytes) and an upper bound of
7052 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7053 particular platform, the lower bound is used. Setting this parameter
7054 very large effectively disables garbage collection. Setting this
7055 parameter and @option{ggc-min-expand} to zero causes a full collection
7056 to occur at every opportunity.
7058 @item max-reload-search-insns
7059 The maximum number of instruction reload should look backward for equivalent
7060 register. Increasing values mean more aggressive optimization, making the
7061 compile time increase with probably slightly better performance. The default
7064 @item max-cselib-memory-locations
7065 The maximum number of memory locations cselib should take into account.
7066 Increasing values mean more aggressive optimization, making the compile time
7067 increase with probably slightly better performance. The default value is 500.
7069 @item max-flow-memory-locations
7070 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
7071 The default value is 100.
7073 @item reorder-blocks-duplicate
7074 @itemx reorder-blocks-duplicate-feedback
7076 Used by basic block reordering pass to decide whether to use unconditional
7077 branch or duplicate the code on its destination. Code is duplicated when its
7078 estimated size is smaller than this value multiplied by the estimated size of
7079 unconditional jump in the hot spots of the program.
7081 The @option{reorder-block-duplicate-feedback} is used only when profile
7082 feedback is available and may be set to higher values than
7083 @option{reorder-block-duplicate} since information about the hot spots is more
7086 @item max-sched-ready-insns
7087 The maximum number of instructions ready to be issued the scheduler should
7088 consider at any given time during the first scheduling pass. Increasing
7089 values mean more thorough searches, making the compilation time increase
7090 with probably little benefit. The default value is 100.
7092 @item max-sched-region-blocks
7093 The maximum number of blocks in a region to be considered for
7094 interblock scheduling. The default value is 10.
7096 @item max-sched-region-insns
7097 The maximum number of insns in a region to be considered for
7098 interblock scheduling. The default value is 100.
7101 The minimum probability (in percents) of reaching a source block
7102 for interblock speculative scheduling. The default value is 40.
7104 @item max-sched-extend-regions-iters
7105 The maximum number of iterations through CFG to extend regions.
7106 0 - disable region extension,
7107 N - do at most N iterations.
7108 The default value is 0.
7110 @item max-sched-insn-conflict-delay
7111 The maximum conflict delay for an insn to be considered for speculative motion.
7112 The default value is 3.
7114 @item sched-spec-prob-cutoff
7115 The minimal probability of speculation success (in percents), so that
7116 speculative insn will be scheduled.
7117 The default value is 40.
7119 @item max-last-value-rtl
7121 The maximum size measured as number of RTLs that can be recorded in an expression
7122 in combiner for a pseudo register as last known value of that register. The default
7125 @item integer-share-limit
7126 Small integer constants can use a shared data structure, reducing the
7127 compiler's memory usage and increasing its speed. This sets the maximum
7128 value of a shared integer constant's. The default value is 256.
7130 @item min-virtual-mappings
7131 Specifies the minimum number of virtual mappings in the incremental
7132 SSA updater that should be registered to trigger the virtual mappings
7133 heuristic defined by virtual-mappings-ratio. The default value is
7136 @item virtual-mappings-ratio
7137 If the number of virtual mappings is virtual-mappings-ratio bigger
7138 than the number of virtual symbols to be updated, then the incremental
7139 SSA updater switches to a full update for those symbols. The default
7142 @item ssp-buffer-size
7143 The minimum size of buffers (i.e. arrays) that will receive stack smashing
7144 protection when @option{-fstack-protection} is used.
7146 @item max-jump-thread-duplication-stmts
7147 Maximum number of statements allowed in a block that needs to be
7148 duplicated when threading jumps.
7150 @item max-fields-for-field-sensitive
7151 Maximum number of fields in a structure we will treat in
7152 a field sensitive manner during pointer analysis.
7154 @item prefetch-latency
7155 Estimate on average number of instructions that are executed before
7156 prefetch finishes. The distance we prefetch ahead is proportional
7157 to this constant. Increasing this number may also lead to less
7158 streams being prefetched (see @option{simultaneous-prefetches}).
7160 @item simultaneous-prefetches
7161 Maximum number of prefetches that can run at the same time.
7163 @item l1-cache-line-size
7164 The size of cache line in L1 cache, in bytes.
7167 The size of L1 cache, in kilobytes.
7170 The size of L2 cache, in kilobytes.
7172 @item use-canonical-types
7173 Whether the compiler should use the ``canonical'' type system. By
7174 default, this should always be 1, which uses a more efficient internal
7175 mechanism for comparing types in C++ and Objective-C++. However, if
7176 bugs in the canonical type system are causing compilation failures,
7177 set this value to 0 to disable canonical types.
7179 @item max-partial-antic-length
7180 Maximum length of the partial antic set computed during the tree
7181 partial redundancy elimination optimization (@option{-ftree-pre}) when
7182 optimizing at @option{-O3} and above. For some sorts of source code
7183 the enhanced partial redundancy elimination optimization can run away,
7184 consuming all of the memory available on the host machine. This
7185 parameter sets a limit on the length of the sets that are computed,
7186 which prevents the runaway behaviour. Setting a value of 0 for
7187 this paramter will allow an unlimited set length.
7189 @item sccvn-max-scc-size
7190 Maximum size of a strongly connected component (SCC) during SCCVN
7191 processing. If this limit is hit, SCCVN processing for the whole
7192 function will not be done and optimizations depending on it will
7193 be disabled. The default maximum SCC size is 10000.
7198 @node Preprocessor Options
7199 @section Options Controlling the Preprocessor
7200 @cindex preprocessor options
7201 @cindex options, preprocessor
7203 These options control the C preprocessor, which is run on each C source
7204 file before actual compilation.
7206 If you use the @option{-E} option, nothing is done except preprocessing.
7207 Some of these options make sense only together with @option{-E} because
7208 they cause the preprocessor output to be unsuitable for actual
7213 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7214 and pass @var{option} directly through to the preprocessor. If
7215 @var{option} contains commas, it is split into multiple options at the
7216 commas. However, many options are modified, translated or interpreted
7217 by the compiler driver before being passed to the preprocessor, and
7218 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7219 interface is undocumented and subject to change, so whenever possible
7220 you should avoid using @option{-Wp} and let the driver handle the
7223 @item -Xpreprocessor @var{option}
7224 @opindex preprocessor
7225 Pass @var{option} as an option to the preprocessor. You can use this to
7226 supply system-specific preprocessor options which GCC does not know how to
7229 If you want to pass an option that takes an argument, you must use
7230 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7233 @include cppopts.texi
7235 @node Assembler Options
7236 @section Passing Options to the Assembler
7238 @c prevent bad page break with this line
7239 You can pass options to the assembler.
7242 @item -Wa,@var{option}
7244 Pass @var{option} as an option to the assembler. If @var{option}
7245 contains commas, it is split into multiple options at the commas.
7247 @item -Xassembler @var{option}
7249 Pass @var{option} as an option to the assembler. You can use this to
7250 supply system-specific assembler options which GCC does not know how to
7253 If you want to pass an option that takes an argument, you must use
7254 @option{-Xassembler} twice, once for the option and once for the argument.
7259 @section Options for Linking
7260 @cindex link options
7261 @cindex options, linking
7263 These options come into play when the compiler links object files into
7264 an executable output file. They are meaningless if the compiler is
7265 not doing a link step.
7269 @item @var{object-file-name}
7270 A file name that does not end in a special recognized suffix is
7271 considered to name an object file or library. (Object files are
7272 distinguished from libraries by the linker according to the file
7273 contents.) If linking is done, these object files are used as input
7282 If any of these options is used, then the linker is not run, and
7283 object file names should not be used as arguments. @xref{Overall
7287 @item -l@var{library}
7288 @itemx -l @var{library}
7290 Search the library named @var{library} when linking. (The second
7291 alternative with the library as a separate argument is only for
7292 POSIX compliance and is not recommended.)
7294 It makes a difference where in the command you write this option; the
7295 linker searches and processes libraries and object files in the order they
7296 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7297 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7298 to functions in @samp{z}, those functions may not be loaded.
7300 The linker searches a standard list of directories for the library,
7301 which is actually a file named @file{lib@var{library}.a}. The linker
7302 then uses this file as if it had been specified precisely by name.
7304 The directories searched include several standard system directories
7305 plus any that you specify with @option{-L}.
7307 Normally the files found this way are library files---archive files
7308 whose members are object files. The linker handles an archive file by
7309 scanning through it for members which define symbols that have so far
7310 been referenced but not defined. But if the file that is found is an
7311 ordinary object file, it is linked in the usual fashion. The only
7312 difference between using an @option{-l} option and specifying a file name
7313 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7314 and searches several directories.
7318 You need this special case of the @option{-l} option in order to
7319 link an Objective-C or Objective-C++ program.
7322 @opindex nostartfiles
7323 Do not use the standard system startup files when linking.
7324 The standard system libraries are used normally, unless @option{-nostdlib}
7325 or @option{-nodefaultlibs} is used.
7327 @item -nodefaultlibs
7328 @opindex nodefaultlibs
7329 Do not use the standard system libraries when linking.
7330 Only the libraries you specify will be passed to the linker.
7331 The standard startup files are used normally, unless @option{-nostartfiles}
7332 is used. The compiler may generate calls to @code{memcmp},
7333 @code{memset}, @code{memcpy} and @code{memmove}.
7334 These entries are usually resolved by entries in
7335 libc. These entry points should be supplied through some other
7336 mechanism when this option is specified.
7340 Do not use the standard system startup files or libraries when linking.
7341 No startup files and only the libraries you specify will be passed to
7342 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7343 @code{memcpy} and @code{memmove}.
7344 These entries are usually resolved by entries in
7345 libc. These entry points should be supplied through some other
7346 mechanism when this option is specified.
7348 @cindex @option{-lgcc}, use with @option{-nostdlib}
7349 @cindex @option{-nostdlib} and unresolved references
7350 @cindex unresolved references and @option{-nostdlib}
7351 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7352 @cindex @option{-nodefaultlibs} and unresolved references
7353 @cindex unresolved references and @option{-nodefaultlibs}
7354 One of the standard libraries bypassed by @option{-nostdlib} and
7355 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7356 that GCC uses to overcome shortcomings of particular machines, or special
7357 needs for some languages.
7358 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7359 Collection (GCC) Internals},
7360 for more discussion of @file{libgcc.a}.)
7361 In most cases, you need @file{libgcc.a} even when you want to avoid
7362 other standard libraries. In other words, when you specify @option{-nostdlib}
7363 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7364 This ensures that you have no unresolved references to internal GCC
7365 library subroutines. (For example, @samp{__main}, used to ensure C++
7366 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7367 GNU Compiler Collection (GCC) Internals}.)
7371 Produce a position independent executable on targets which support it.
7372 For predictable results, you must also specify the same set of options
7373 that were used to generate code (@option{-fpie}, @option{-fPIE},
7374 or model suboptions) when you specify this option.
7378 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7379 that support it. This instructs the linker to add all symbols, not
7380 only used ones, to the dynamic symbol table. This option is needed
7381 for some uses of @code{dlopen} or to allow obtaining backtraces
7382 from within a program.
7386 Remove all symbol table and relocation information from the executable.
7390 On systems that support dynamic linking, this prevents linking with the shared
7391 libraries. On other systems, this option has no effect.
7395 Produce a shared object which can then be linked with other objects to
7396 form an executable. Not all systems support this option. For predictable
7397 results, you must also specify the same set of options that were used to
7398 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7399 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7400 needs to build supplementary stub code for constructors to work. On
7401 multi-libbed systems, @samp{gcc -shared} must select the correct support
7402 libraries to link against. Failing to supply the correct flags may lead
7403 to subtle defects. Supplying them in cases where they are not necessary
7406 @item -shared-libgcc
7407 @itemx -static-libgcc
7408 @opindex shared-libgcc
7409 @opindex static-libgcc
7410 On systems that provide @file{libgcc} as a shared library, these options
7411 force the use of either the shared or static version respectively.
7412 If no shared version of @file{libgcc} was built when the compiler was
7413 configured, these options have no effect.
7415 There are several situations in which an application should use the
7416 shared @file{libgcc} instead of the static version. The most common
7417 of these is when the application wishes to throw and catch exceptions
7418 across different shared libraries. In that case, each of the libraries
7419 as well as the application itself should use the shared @file{libgcc}.
7421 Therefore, the G++ and GCJ drivers automatically add
7422 @option{-shared-libgcc} whenever you build a shared library or a main
7423 executable, because C++ and Java programs typically use exceptions, so
7424 this is the right thing to do.
7426 If, instead, you use the GCC driver to create shared libraries, you may
7427 find that they will not always be linked with the shared @file{libgcc}.
7428 If GCC finds, at its configuration time, that you have a non-GNU linker
7429 or a GNU linker that does not support option @option{--eh-frame-hdr},
7430 it will link the shared version of @file{libgcc} into shared libraries
7431 by default. Otherwise, it will take advantage of the linker and optimize
7432 away the linking with the shared version of @file{libgcc}, linking with
7433 the static version of libgcc by default. This allows exceptions to
7434 propagate through such shared libraries, without incurring relocation
7435 costs at library load time.
7437 However, if a library or main executable is supposed to throw or catch
7438 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7439 for the languages used in the program, or using the option
7440 @option{-shared-libgcc}, such that it is linked with the shared
7445 Bind references to global symbols when building a shared object. Warn
7446 about any unresolved references (unless overridden by the link editor
7447 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7450 @item -Xlinker @var{option}
7452 Pass @var{option} as an option to the linker. You can use this to
7453 supply system-specific linker options which GCC does not know how to
7456 If you want to pass an option that takes an argument, you must use
7457 @option{-Xlinker} twice, once for the option and once for the argument.
7458 For example, to pass @option{-assert definitions}, you must write
7459 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7460 @option{-Xlinker "-assert definitions"}, because this passes the entire
7461 string as a single argument, which is not what the linker expects.
7463 @item -Wl,@var{option}
7465 Pass @var{option} as an option to the linker. If @var{option} contains
7466 commas, it is split into multiple options at the commas.
7468 @item -u @var{symbol}
7470 Pretend the symbol @var{symbol} is undefined, to force linking of
7471 library modules to define it. You can use @option{-u} multiple times with
7472 different symbols to force loading of additional library modules.
7475 @node Directory Options
7476 @section Options for Directory Search
7477 @cindex directory options
7478 @cindex options, directory search
7481 These options specify directories to search for header files, for
7482 libraries and for parts of the compiler:
7487 Add the directory @var{dir} to the head of the list of directories to be
7488 searched for header files. This can be used to override a system header
7489 file, substituting your own version, since these directories are
7490 searched before the system header file directories. However, you should
7491 not use this option to add directories that contain vendor-supplied
7492 system header files (use @option{-isystem} for that). If you use more than
7493 one @option{-I} option, the directories are scanned in left-to-right
7494 order; the standard system directories come after.
7496 If a standard system include directory, or a directory specified with
7497 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7498 option will be ignored. The directory will still be searched but as a
7499 system directory at its normal position in the system include chain.
7500 This is to ensure that GCC's procedure to fix buggy system headers and
7501 the ordering for the include_next directive are not inadvertently changed.
7502 If you really need to change the search order for system directories,
7503 use the @option{-nostdinc} and/or @option{-isystem} options.
7505 @item -iquote@var{dir}
7507 Add the directory @var{dir} to the head of the list of directories to
7508 be searched for header files only for the case of @samp{#include
7509 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7510 otherwise just like @option{-I}.
7514 Add directory @var{dir} to the list of directories to be searched
7517 @item -B@var{prefix}
7519 This option specifies where to find the executables, libraries,
7520 include files, and data files of the compiler itself.
7522 The compiler driver program runs one or more of the subprograms
7523 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7524 @var{prefix} as a prefix for each program it tries to run, both with and
7525 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7527 For each subprogram to be run, the compiler driver first tries the
7528 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7529 was not specified, the driver tries two standard prefixes, which are
7530 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7531 those results in a file name that is found, the unmodified program
7532 name is searched for using the directories specified in your
7533 @env{PATH} environment variable.
7535 The compiler will check to see if the path provided by the @option{-B}
7536 refers to a directory, and if necessary it will add a directory
7537 separator character at the end of the path.
7539 @option{-B} prefixes that effectively specify directory names also apply
7540 to libraries in the linker, because the compiler translates these
7541 options into @option{-L} options for the linker. They also apply to
7542 includes files in the preprocessor, because the compiler translates these
7543 options into @option{-isystem} options for the preprocessor. In this case,
7544 the compiler appends @samp{include} to the prefix.
7546 The run-time support file @file{libgcc.a} can also be searched for using
7547 the @option{-B} prefix, if needed. If it is not found there, the two
7548 standard prefixes above are tried, and that is all. The file is left
7549 out of the link if it is not found by those means.
7551 Another way to specify a prefix much like the @option{-B} prefix is to use
7552 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7555 As a special kludge, if the path provided by @option{-B} is
7556 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7557 9, then it will be replaced by @file{[dir/]include}. This is to help
7558 with boot-strapping the compiler.
7560 @item -specs=@var{file}
7562 Process @var{file} after the compiler reads in the standard @file{specs}
7563 file, in order to override the defaults that the @file{gcc} driver
7564 program uses when determining what switches to pass to @file{cc1},
7565 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7566 @option{-specs=@var{file}} can be specified on the command line, and they
7567 are processed in order, from left to right.
7569 @item --sysroot=@var{dir}
7571 Use @var{dir} as the logical root directory for headers and libraries.
7572 For example, if the compiler would normally search for headers in
7573 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7574 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7576 If you use both this option and the @option{-isysroot} option, then
7577 the @option{--sysroot} option will apply to libraries, but the
7578 @option{-isysroot} option will apply to header files.
7580 The GNU linker (beginning with version 2.16) has the necessary support
7581 for this option. If your linker does not support this option, the
7582 header file aspect of @option{--sysroot} will still work, but the
7583 library aspect will not.
7587 This option has been deprecated. Please use @option{-iquote} instead for
7588 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7589 Any directories you specify with @option{-I} options before the @option{-I-}
7590 option are searched only for the case of @samp{#include "@var{file}"};
7591 they are not searched for @samp{#include <@var{file}>}.
7593 If additional directories are specified with @option{-I} options after
7594 the @option{-I-}, these directories are searched for all @samp{#include}
7595 directives. (Ordinarily @emph{all} @option{-I} directories are used
7598 In addition, the @option{-I-} option inhibits the use of the current
7599 directory (where the current input file came from) as the first search
7600 directory for @samp{#include "@var{file}"}. There is no way to
7601 override this effect of @option{-I-}. With @option{-I.} you can specify
7602 searching the directory which was current when the compiler was
7603 invoked. That is not exactly the same as what the preprocessor does
7604 by default, but it is often satisfactory.
7606 @option{-I-} does not inhibit the use of the standard system directories
7607 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7614 @section Specifying subprocesses and the switches to pass to them
7617 @command{gcc} is a driver program. It performs its job by invoking a
7618 sequence of other programs to do the work of compiling, assembling and
7619 linking. GCC interprets its command-line parameters and uses these to
7620 deduce which programs it should invoke, and which command-line options
7621 it ought to place on their command lines. This behavior is controlled
7622 by @dfn{spec strings}. In most cases there is one spec string for each
7623 program that GCC can invoke, but a few programs have multiple spec
7624 strings to control their behavior. The spec strings built into GCC can
7625 be overridden by using the @option{-specs=} command-line switch to specify
7628 @dfn{Spec files} are plaintext files that are used to construct spec
7629 strings. They consist of a sequence of directives separated by blank
7630 lines. The type of directive is determined by the first non-whitespace
7631 character on the line and it can be one of the following:
7634 @item %@var{command}
7635 Issues a @var{command} to the spec file processor. The commands that can
7639 @item %include <@var{file}>
7641 Search for @var{file} and insert its text at the current point in the
7644 @item %include_noerr <@var{file}>
7645 @cindex %include_noerr
7646 Just like @samp{%include}, but do not generate an error message if the include
7647 file cannot be found.
7649 @item %rename @var{old_name} @var{new_name}
7651 Rename the spec string @var{old_name} to @var{new_name}.
7655 @item *[@var{spec_name}]:
7656 This tells the compiler to create, override or delete the named spec
7657 string. All lines after this directive up to the next directive or
7658 blank line are considered to be the text for the spec string. If this
7659 results in an empty string then the spec will be deleted. (Or, if the
7660 spec did not exist, then nothing will happened.) Otherwise, if the spec
7661 does not currently exist a new spec will be created. If the spec does
7662 exist then its contents will be overridden by the text of this
7663 directive, unless the first character of that text is the @samp{+}
7664 character, in which case the text will be appended to the spec.
7666 @item [@var{suffix}]:
7667 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7668 and up to the next directive or blank line are considered to make up the
7669 spec string for the indicated suffix. When the compiler encounters an
7670 input file with the named suffix, it will processes the spec string in
7671 order to work out how to compile that file. For example:
7678 This says that any input file whose name ends in @samp{.ZZ} should be
7679 passed to the program @samp{z-compile}, which should be invoked with the
7680 command-line switch @option{-input} and with the result of performing the
7681 @samp{%i} substitution. (See below.)
7683 As an alternative to providing a spec string, the text that follows a
7684 suffix directive can be one of the following:
7687 @item @@@var{language}
7688 This says that the suffix is an alias for a known @var{language}. This is
7689 similar to using the @option{-x} command-line switch to GCC to specify a
7690 language explicitly. For example:
7697 Says that .ZZ files are, in fact, C++ source files.
7700 This causes an error messages saying:
7703 @var{name} compiler not installed on this system.
7707 GCC already has an extensive list of suffixes built into it.
7708 This directive will add an entry to the end of the list of suffixes, but
7709 since the list is searched from the end backwards, it is effectively
7710 possible to override earlier entries using this technique.
7714 GCC has the following spec strings built into it. Spec files can
7715 override these strings or create their own. Note that individual
7716 targets can also add their own spec strings to this list.
7719 asm Options to pass to the assembler
7720 asm_final Options to pass to the assembler post-processor
7721 cpp Options to pass to the C preprocessor
7722 cc1 Options to pass to the C compiler
7723 cc1plus Options to pass to the C++ compiler
7724 endfile Object files to include at the end of the link
7725 link Options to pass to the linker
7726 lib Libraries to include on the command line to the linker
7727 libgcc Decides which GCC support library to pass to the linker
7728 linker Sets the name of the linker
7729 predefines Defines to be passed to the C preprocessor
7730 signed_char Defines to pass to CPP to say whether @code{char} is signed
7732 startfile Object files to include at the start of the link
7735 Here is a small example of a spec file:
7741 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7744 This example renames the spec called @samp{lib} to @samp{old_lib} and
7745 then overrides the previous definition of @samp{lib} with a new one.
7746 The new definition adds in some extra command-line options before
7747 including the text of the old definition.
7749 @dfn{Spec strings} are a list of command-line options to be passed to their
7750 corresponding program. In addition, the spec strings can contain
7751 @samp{%}-prefixed sequences to substitute variable text or to
7752 conditionally insert text into the command line. Using these constructs
7753 it is possible to generate quite complex command lines.
7755 Here is a table of all defined @samp{%}-sequences for spec
7756 strings. Note that spaces are not generated automatically around the
7757 results of expanding these sequences. Therefore you can concatenate them
7758 together or combine them with constant text in a single argument.
7762 Substitute one @samp{%} into the program name or argument.
7765 Substitute the name of the input file being processed.
7768 Substitute the basename of the input file being processed.
7769 This is the substring up to (and not including) the last period
7770 and not including the directory.
7773 This is the same as @samp{%b}, but include the file suffix (text after
7777 Marks the argument containing or following the @samp{%d} as a
7778 temporary file name, so that that file will be deleted if GCC exits
7779 successfully. Unlike @samp{%g}, this contributes no text to the
7782 @item %g@var{suffix}
7783 Substitute a file name that has suffix @var{suffix} and is chosen
7784 once per compilation, and mark the argument in the same way as
7785 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7786 name is now chosen in a way that is hard to predict even when previously
7787 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7788 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7789 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7790 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7791 was simply substituted with a file name chosen once per compilation,
7792 without regard to any appended suffix (which was therefore treated
7793 just like ordinary text), making such attacks more likely to succeed.
7795 @item %u@var{suffix}
7796 Like @samp{%g}, but generates a new temporary file name even if
7797 @samp{%u@var{suffix}} was already seen.
7799 @item %U@var{suffix}
7800 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7801 new one if there is no such last file name. In the absence of any
7802 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7803 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7804 would involve the generation of two distinct file names, one
7805 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7806 simply substituted with a file name chosen for the previous @samp{%u},
7807 without regard to any appended suffix.
7809 @item %j@var{suffix}
7810 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7811 writable, and if save-temps is off; otherwise, substitute the name
7812 of a temporary file, just like @samp{%u}. This temporary file is not
7813 meant for communication between processes, but rather as a junk
7816 @item %|@var{suffix}
7817 @itemx %m@var{suffix}
7818 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7819 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7820 all. These are the two most common ways to instruct a program that it
7821 should read from standard input or write to standard output. If you
7822 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7823 construct: see for example @file{f/lang-specs.h}.
7825 @item %.@var{SUFFIX}
7826 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7827 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7828 terminated by the next space or %.
7831 Marks the argument containing or following the @samp{%w} as the
7832 designated output file of this compilation. This puts the argument
7833 into the sequence of arguments that @samp{%o} will substitute later.
7836 Substitutes the names of all the output files, with spaces
7837 automatically placed around them. You should write spaces
7838 around the @samp{%o} as well or the results are undefined.
7839 @samp{%o} is for use in the specs for running the linker.
7840 Input files whose names have no recognized suffix are not compiled
7841 at all, but they are included among the output files, so they will
7845 Substitutes the suffix for object files. Note that this is
7846 handled specially when it immediately follows @samp{%g, %u, or %U},
7847 because of the need for those to form complete file names. The
7848 handling is such that @samp{%O} is treated exactly as if it had already
7849 been substituted, except that @samp{%g, %u, and %U} do not currently
7850 support additional @var{suffix} characters following @samp{%O} as they would
7851 following, for example, @samp{.o}.
7854 Substitutes the standard macro predefinitions for the
7855 current target machine. Use this when running @code{cpp}.
7858 Like @samp{%p}, but puts @samp{__} before and after the name of each
7859 predefined macro, except for macros that start with @samp{__} or with
7860 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7864 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7865 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7866 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7867 and @option{-imultilib} as necessary.
7870 Current argument is the name of a library or startup file of some sort.
7871 Search for that file in a standard list of directories and substitute
7872 the full name found.
7875 Print @var{str} as an error message. @var{str} is terminated by a newline.
7876 Use this when inconsistent options are detected.
7879 Substitute the contents of spec string @var{name} at this point.
7882 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7884 @item %x@{@var{option}@}
7885 Accumulate an option for @samp{%X}.
7888 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7892 Output the accumulated assembler options specified by @option{-Wa}.
7895 Output the accumulated preprocessor options specified by @option{-Wp}.
7898 Process the @code{asm} spec. This is used to compute the
7899 switches to be passed to the assembler.
7902 Process the @code{asm_final} spec. This is a spec string for
7903 passing switches to an assembler post-processor, if such a program is
7907 Process the @code{link} spec. This is the spec for computing the
7908 command line passed to the linker. Typically it will make use of the
7909 @samp{%L %G %S %D and %E} sequences.
7912 Dump out a @option{-L} option for each directory that GCC believes might
7913 contain startup files. If the target supports multilibs then the
7914 current multilib directory will be prepended to each of these paths.
7917 Process the @code{lib} spec. This is a spec string for deciding which
7918 libraries should be included on the command line to the linker.
7921 Process the @code{libgcc} spec. This is a spec string for deciding
7922 which GCC support library should be included on the command line to the linker.
7925 Process the @code{startfile} spec. This is a spec for deciding which
7926 object files should be the first ones passed to the linker. Typically
7927 this might be a file named @file{crt0.o}.
7930 Process the @code{endfile} spec. This is a spec string that specifies
7931 the last object files that will be passed to the linker.
7934 Process the @code{cpp} spec. This is used to construct the arguments
7935 to be passed to the C preprocessor.
7938 Process the @code{cc1} spec. This is used to construct the options to be
7939 passed to the actual C compiler (@samp{cc1}).
7942 Process the @code{cc1plus} spec. This is used to construct the options to be
7943 passed to the actual C++ compiler (@samp{cc1plus}).
7946 Substitute the variable part of a matched option. See below.
7947 Note that each comma in the substituted string is replaced by
7951 Remove all occurrences of @code{-S} from the command line. Note---this
7952 command is position dependent. @samp{%} commands in the spec string
7953 before this one will see @code{-S}, @samp{%} commands in the spec string
7954 after this one will not.
7956 @item %:@var{function}(@var{args})
7957 Call the named function @var{function}, passing it @var{args}.
7958 @var{args} is first processed as a nested spec string, then split
7959 into an argument vector in the usual fashion. The function returns
7960 a string which is processed as if it had appeared literally as part
7961 of the current spec.
7963 The following built-in spec functions are provided:
7967 The @code{getenv} spec function takes two arguments: an environment
7968 variable name and a string. If the environment variable is not
7969 defined, a fatal error is issued. Otherwise, the return value is the
7970 value of the environment variable concatenated with the string. For
7971 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7974 %:getenv(TOPDIR /include)
7977 expands to @file{/path/to/top/include}.
7979 @item @code{if-exists}
7980 The @code{if-exists} spec function takes one argument, an absolute
7981 pathname to a file. If the file exists, @code{if-exists} returns the
7982 pathname. Here is a small example of its usage:
7986 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7989 @item @code{if-exists-else}
7990 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7991 spec function, except that it takes two arguments. The first argument is
7992 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7993 returns the pathname. If it does not exist, it returns the second argument.
7994 This way, @code{if-exists-else} can be used to select one file or another,
7995 based on the existence of the first. Here is a small example of its usage:
7999 crt0%O%s %:if-exists(crti%O%s) \
8000 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8003 @item @code{replace-outfile}
8004 The @code{replace-outfile} spec function takes two arguments. It looks for the
8005 first argument in the outfiles array and replaces it with the second argument. Here
8006 is a small example of its usage:
8009 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8012 @item @code{print-asm-header}
8013 The @code{print-asm-header} function takes no arguments and simply
8014 prints a banner like:
8020 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8023 It is used to separate compiler options from assembler options
8024 in the @option{--target-help} output.
8028 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8029 If that switch was not specified, this substitutes nothing. Note that
8030 the leading dash is omitted when specifying this option, and it is
8031 automatically inserted if the substitution is performed. Thus the spec
8032 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8033 and would output the command line option @option{-foo}.
8035 @item %W@{@code{S}@}
8036 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8039 @item %@{@code{S}*@}
8040 Substitutes all the switches specified to GCC whose names start
8041 with @code{-S}, but which also take an argument. This is used for
8042 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8043 GCC considers @option{-o foo} as being
8044 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8045 text, including the space. Thus two arguments would be generated.
8047 @item %@{@code{S}*&@code{T}*@}
8048 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8049 (the order of @code{S} and @code{T} in the spec is not significant).
8050 There can be any number of ampersand-separated variables; for each the
8051 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8053 @item %@{@code{S}:@code{X}@}
8054 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8056 @item %@{!@code{S}:@code{X}@}
8057 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8059 @item %@{@code{S}*:@code{X}@}
8060 Substitutes @code{X} if one or more switches whose names start with
8061 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8062 once, no matter how many such switches appeared. However, if @code{%*}
8063 appears somewhere in @code{X}, then @code{X} will be substituted once
8064 for each matching switch, with the @code{%*} replaced by the part of
8065 that switch that matched the @code{*}.
8067 @item %@{.@code{S}:@code{X}@}
8068 Substitutes @code{X}, if processing a file with suffix @code{S}.
8070 @item %@{!.@code{S}:@code{X}@}
8071 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8073 @item %@{,@code{S}:@code{X}@}
8074 Substitutes @code{X}, if processing a file for language @code{S}.
8076 @item %@{!,@code{S}:@code{X}@}
8077 Substitutes @code{X}, if not processing a file for language @code{S}.
8079 @item %@{@code{S}|@code{P}:@code{X}@}
8080 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8081 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8082 @code{*} sequences as well, although they have a stronger binding than
8083 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8084 alternatives must be starred, and only the first matching alternative
8087 For example, a spec string like this:
8090 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8093 will output the following command-line options from the following input
8094 command-line options:
8099 -d fred.c -foo -baz -boggle
8100 -d jim.d -bar -baz -boggle
8103 @item %@{S:X; T:Y; :D@}
8105 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8106 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8107 be as many clauses as you need. This may be combined with @code{.},
8108 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8113 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8114 construct may contain other nested @samp{%} constructs or spaces, or
8115 even newlines. They are processed as usual, as described above.
8116 Trailing white space in @code{X} is ignored. White space may also
8117 appear anywhere on the left side of the colon in these constructs,
8118 except between @code{.} or @code{*} and the corresponding word.
8120 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8121 handled specifically in these constructs. If another value of
8122 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8123 @option{-W} switch is found later in the command line, the earlier
8124 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8125 just one letter, which passes all matching options.
8127 The character @samp{|} at the beginning of the predicate text is used to
8128 indicate that a command should be piped to the following command, but
8129 only if @option{-pipe} is specified.
8131 It is built into GCC which switches take arguments and which do not.
8132 (You might think it would be useful to generalize this to allow each
8133 compiler's spec to say which switches take arguments. But this cannot
8134 be done in a consistent fashion. GCC cannot even decide which input
8135 files have been specified without knowing which switches take arguments,
8136 and it must know which input files to compile in order to tell which
8139 GCC also knows implicitly that arguments starting in @option{-l} are to be
8140 treated as compiler output files, and passed to the linker in their
8141 proper position among the other output files.
8143 @c man begin OPTIONS
8145 @node Target Options
8146 @section Specifying Target Machine and Compiler Version
8147 @cindex target options
8148 @cindex cross compiling
8149 @cindex specifying machine version
8150 @cindex specifying compiler version and target machine
8151 @cindex compiler version, specifying
8152 @cindex target machine, specifying
8154 The usual way to run GCC is to run the executable called @file{gcc}, or
8155 @file{<machine>-gcc} when cross-compiling, or
8156 @file{<machine>-gcc-<version>} to run a version other than the one that
8157 was installed last. Sometimes this is inconvenient, so GCC provides
8158 options that will switch to another cross-compiler or version.
8161 @item -b @var{machine}
8163 The argument @var{machine} specifies the target machine for compilation.
8165 The value to use for @var{machine} is the same as was specified as the
8166 machine type when configuring GCC as a cross-compiler. For
8167 example, if a cross-compiler was configured with @samp{configure
8168 arm-elf}, meaning to compile for an arm processor with elf binaries,
8169 then you would specify @option{-b arm-elf} to run that cross compiler.
8170 Because there are other options beginning with @option{-b}, the
8171 configuration must contain a hyphen.
8173 @item -V @var{version}
8175 The argument @var{version} specifies which version of GCC to run.
8176 This is useful when multiple versions are installed. For example,
8177 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8180 The @option{-V} and @option{-b} options work by running the
8181 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8182 use them if you can just run that directly.
8184 @node Submodel Options
8185 @section Hardware Models and Configurations
8186 @cindex submodel options
8187 @cindex specifying hardware config
8188 @cindex hardware models and configurations, specifying
8189 @cindex machine dependent options
8191 Earlier we discussed the standard option @option{-b} which chooses among
8192 different installed compilers for completely different target
8193 machines, such as VAX vs.@: 68000 vs.@: 80386.
8195 In addition, each of these target machine types can have its own
8196 special options, starting with @samp{-m}, to choose among various
8197 hardware models or configurations---for example, 68010 vs 68020,
8198 floating coprocessor or none. A single installed version of the
8199 compiler can compile for any model or configuration, according to the
8202 Some configurations of the compiler also support additional special
8203 options, usually for compatibility with other compilers on the same
8206 @c This list is ordered alphanumerically by subsection name.
8207 @c It should be the same order and spelling as these options are listed
8208 @c in Machine Dependent Options
8214 * Blackfin Options::
8218 * DEC Alpha Options::
8219 * DEC Alpha/VMS Options::
8221 * GNU/Linux Options::
8224 * i386 and x86-64 Options::
8237 * RS/6000 and PowerPC Options::
8238 * S/390 and zSeries Options::
8243 * System V Options::
8248 * Xstormy16 Options::
8254 @subsection ARC Options
8257 These options are defined for ARC implementations:
8262 Compile code for little endian mode. This is the default.
8266 Compile code for big endian mode.
8269 @opindex mmangle-cpu
8270 Prepend the name of the cpu to all public symbol names.
8271 In multiple-processor systems, there are many ARC variants with different
8272 instruction and register set characteristics. This flag prevents code
8273 compiled for one cpu to be linked with code compiled for another.
8274 No facility exists for handling variants that are ``almost identical''.
8275 This is an all or nothing option.
8277 @item -mcpu=@var{cpu}
8279 Compile code for ARC variant @var{cpu}.
8280 Which variants are supported depend on the configuration.
8281 All variants support @option{-mcpu=base}, this is the default.
8283 @item -mtext=@var{text-section}
8284 @itemx -mdata=@var{data-section}
8285 @itemx -mrodata=@var{readonly-data-section}
8289 Put functions, data, and readonly data in @var{text-section},
8290 @var{data-section}, and @var{readonly-data-section} respectively
8291 by default. This can be overridden with the @code{section} attribute.
8292 @xref{Variable Attributes}.
8297 @subsection ARM Options
8300 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8304 @item -mabi=@var{name}
8306 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8307 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8310 @opindex mapcs-frame
8311 Generate a stack frame that is compliant with the ARM Procedure Call
8312 Standard for all functions, even if this is not strictly necessary for
8313 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8314 with this option will cause the stack frames not to be generated for
8315 leaf functions. The default is @option{-mno-apcs-frame}.
8319 This is a synonym for @option{-mapcs-frame}.
8322 @c not currently implemented
8323 @item -mapcs-stack-check
8324 @opindex mapcs-stack-check
8325 Generate code to check the amount of stack space available upon entry to
8326 every function (that actually uses some stack space). If there is
8327 insufficient space available then either the function
8328 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8329 called, depending upon the amount of stack space required. The run time
8330 system is required to provide these functions. The default is
8331 @option{-mno-apcs-stack-check}, since this produces smaller code.
8333 @c not currently implemented
8335 @opindex mapcs-float
8336 Pass floating point arguments using the float point registers. This is
8337 one of the variants of the APCS@. This option is recommended if the
8338 target hardware has a floating point unit or if a lot of floating point
8339 arithmetic is going to be performed by the code. The default is
8340 @option{-mno-apcs-float}, since integer only code is slightly increased in
8341 size if @option{-mapcs-float} is used.
8343 @c not currently implemented
8344 @item -mapcs-reentrant
8345 @opindex mapcs-reentrant
8346 Generate reentrant, position independent code. The default is
8347 @option{-mno-apcs-reentrant}.
8350 @item -mthumb-interwork
8351 @opindex mthumb-interwork
8352 Generate code which supports calling between the ARM and Thumb
8353 instruction sets. Without this option the two instruction sets cannot
8354 be reliably used inside one program. The default is
8355 @option{-mno-thumb-interwork}, since slightly larger code is generated
8356 when @option{-mthumb-interwork} is specified.
8358 @item -mno-sched-prolog
8359 @opindex mno-sched-prolog
8360 Prevent the reordering of instructions in the function prolog, or the
8361 merging of those instruction with the instructions in the function's
8362 body. This means that all functions will start with a recognizable set
8363 of instructions (or in fact one of a choice from a small set of
8364 different function prologues), and this information can be used to
8365 locate the start if functions inside an executable piece of code. The
8366 default is @option{-msched-prolog}.
8369 @opindex mhard-float
8370 Generate output containing floating point instructions. This is the
8374 @opindex msoft-float
8375 Generate output containing library calls for floating point.
8376 @strong{Warning:} the requisite libraries are not available for all ARM
8377 targets. Normally the facilities of the machine's usual C compiler are
8378 used, but this cannot be done directly in cross-compilation. You must make
8379 your own arrangements to provide suitable library functions for
8382 @option{-msoft-float} changes the calling convention in the output file;
8383 therefore, it is only useful if you compile @emph{all} of a program with
8384 this option. In particular, you need to compile @file{libgcc.a}, the
8385 library that comes with GCC, with @option{-msoft-float} in order for
8388 @item -mfloat-abi=@var{name}
8390 Specifies which ABI to use for floating point values. Permissible values
8391 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8393 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8394 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8395 of floating point instructions, but still uses the soft-float calling
8398 @item -mlittle-endian
8399 @opindex mlittle-endian
8400 Generate code for a processor running in little-endian mode. This is
8401 the default for all standard configurations.
8404 @opindex mbig-endian
8405 Generate code for a processor running in big-endian mode; the default is
8406 to compile code for a little-endian processor.
8408 @item -mwords-little-endian
8409 @opindex mwords-little-endian
8410 This option only applies when generating code for big-endian processors.
8411 Generate code for a little-endian word order but a big-endian byte
8412 order. That is, a byte order of the form @samp{32107654}. Note: this
8413 option should only be used if you require compatibility with code for
8414 big-endian ARM processors generated by versions of the compiler prior to
8417 @item -mcpu=@var{name}
8419 This specifies the name of the target ARM processor. GCC uses this name
8420 to determine what kind of instructions it can emit when generating
8421 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8422 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8423 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8424 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8425 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8426 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8427 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8428 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8429 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8430 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8431 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8432 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8433 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8434 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8435 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8436 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8438 @itemx -mtune=@var{name}
8440 This option is very similar to the @option{-mcpu=} option, except that
8441 instead of specifying the actual target processor type, and hence
8442 restricting which instructions can be used, it specifies that GCC should
8443 tune the performance of the code as if the target were of the type
8444 specified in this option, but still choosing the instructions that it
8445 will generate based on the cpu specified by a @option{-mcpu=} option.
8446 For some ARM implementations better performance can be obtained by using
8449 @item -march=@var{name}
8451 This specifies the name of the target ARM architecture. GCC uses this
8452 name to determine what kind of instructions it can emit when generating
8453 assembly code. This option can be used in conjunction with or instead
8454 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8455 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8456 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8457 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8458 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8460 @item -mfpu=@var{name}
8461 @itemx -mfpe=@var{number}
8462 @itemx -mfp=@var{number}
8466 This specifies what floating point hardware (or hardware emulation) is
8467 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8468 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8469 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8470 with older versions of GCC@.
8472 If @option{-msoft-float} is specified this specifies the format of
8473 floating point values.
8475 @item -mstructure-size-boundary=@var{n}
8476 @opindex mstructure-size-boundary
8477 The size of all structures and unions will be rounded up to a multiple
8478 of the number of bits set by this option. Permissible values are 8, 32
8479 and 64. The default value varies for different toolchains. For the COFF
8480 targeted toolchain the default value is 8. A value of 64 is only allowed
8481 if the underlying ABI supports it.
8483 Specifying the larger number can produce faster, more efficient code, but
8484 can also increase the size of the program. Different values are potentially
8485 incompatible. Code compiled with one value cannot necessarily expect to
8486 work with code or libraries compiled with another value, if they exchange
8487 information using structures or unions.
8489 @item -mabort-on-noreturn
8490 @opindex mabort-on-noreturn
8491 Generate a call to the function @code{abort} at the end of a
8492 @code{noreturn} function. It will be executed if the function tries to
8496 @itemx -mno-long-calls
8497 @opindex mlong-calls
8498 @opindex mno-long-calls
8499 Tells the compiler to perform function calls by first loading the
8500 address of the function into a register and then performing a subroutine
8501 call on this register. This switch is needed if the target function
8502 will lie outside of the 64 megabyte addressing range of the offset based
8503 version of subroutine call instruction.
8505 Even if this switch is enabled, not all function calls will be turned
8506 into long calls. The heuristic is that static functions, functions
8507 which have the @samp{short-call} attribute, functions that are inside
8508 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8509 definitions have already been compiled within the current compilation
8510 unit, will not be turned into long calls. The exception to this rule is
8511 that weak function definitions, functions with the @samp{long-call}
8512 attribute or the @samp{section} attribute, and functions that are within
8513 the scope of a @samp{#pragma long_calls} directive, will always be
8514 turned into long calls.
8516 This feature is not enabled by default. Specifying
8517 @option{-mno-long-calls} will restore the default behavior, as will
8518 placing the function calls within the scope of a @samp{#pragma
8519 long_calls_off} directive. Note these switches have no effect on how
8520 the compiler generates code to handle function calls via function
8523 @item -mnop-fun-dllimport
8524 @opindex mnop-fun-dllimport
8525 Disable support for the @code{dllimport} attribute.
8527 @item -msingle-pic-base
8528 @opindex msingle-pic-base
8529 Treat the register used for PIC addressing as read-only, rather than
8530 loading it in the prologue for each function. The run-time system is
8531 responsible for initializing this register with an appropriate value
8532 before execution begins.
8534 @item -mpic-register=@var{reg}
8535 @opindex mpic-register
8536 Specify the register to be used for PIC addressing. The default is R10
8537 unless stack-checking is enabled, when R9 is used.
8539 @item -mcirrus-fix-invalid-insns
8540 @opindex mcirrus-fix-invalid-insns
8541 @opindex mno-cirrus-fix-invalid-insns
8542 Insert NOPs into the instruction stream to in order to work around
8543 problems with invalid Maverick instruction combinations. This option
8544 is only valid if the @option{-mcpu=ep9312} option has been used to
8545 enable generation of instructions for the Cirrus Maverick floating
8546 point co-processor. This option is not enabled by default, since the
8547 problem is only present in older Maverick implementations. The default
8548 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8551 @item -mpoke-function-name
8552 @opindex mpoke-function-name
8553 Write the name of each function into the text section, directly
8554 preceding the function prologue. The generated code is similar to this:
8558 .ascii "arm_poke_function_name", 0
8561 .word 0xff000000 + (t1 - t0)
8562 arm_poke_function_name
8564 stmfd sp!, @{fp, ip, lr, pc@}
8568 When performing a stack backtrace, code can inspect the value of
8569 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8570 location @code{pc - 12} and the top 8 bits are set, then we know that
8571 there is a function name embedded immediately preceding this location
8572 and has length @code{((pc[-3]) & 0xff000000)}.
8576 Generate code for the Thumb instruction set. The default is to
8577 use the 32-bit ARM instruction set.
8578 This option automatically enables either 16-bit Thumb-1 or
8579 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8580 and @option{-march=@var{name}} options.
8583 @opindex mtpcs-frame
8584 Generate a stack frame that is compliant with the Thumb Procedure Call
8585 Standard for all non-leaf functions. (A leaf function is one that does
8586 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8588 @item -mtpcs-leaf-frame
8589 @opindex mtpcs-leaf-frame
8590 Generate a stack frame that is compliant with the Thumb Procedure Call
8591 Standard for all leaf functions. (A leaf function is one that does
8592 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8594 @item -mcallee-super-interworking
8595 @opindex mcallee-super-interworking
8596 Gives all externally visible functions in the file being compiled an ARM
8597 instruction set header which switches to Thumb mode before executing the
8598 rest of the function. This allows these functions to be called from
8599 non-interworking code.
8601 @item -mcaller-super-interworking
8602 @opindex mcaller-super-interworking
8603 Allows calls via function pointers (including virtual functions) to
8604 execute correctly regardless of whether the target code has been
8605 compiled for interworking or not. There is a small overhead in the cost
8606 of executing a function pointer if this option is enabled.
8608 @item -mtp=@var{name}
8610 Specify the access model for the thread local storage pointer. The valid
8611 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8612 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8613 (supported in the arm6k architecture), and @option{auto}, which uses the
8614 best available method for the selected processor. The default setting is
8620 @subsection AVR Options
8623 These options are defined for AVR implementations:
8626 @item -mmcu=@var{mcu}
8628 Specify ATMEL AVR instruction set or MCU type.
8630 Instruction set avr1 is for the minimal AVR core, not supported by the C
8631 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8632 attiny11, attiny12, attiny15, attiny28).
8634 Instruction set avr2 (default) is for the classic AVR core with up to
8635 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8636 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8637 at90c8534, at90s8535).
8639 Instruction set avr3 is for the classic AVR core with up to 128K program
8640 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8642 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8643 memory space (MCU types: atmega8, atmega83, atmega85).
8645 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8646 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8647 atmega64, atmega128, at43usb355, at94k).
8651 Output instruction sizes to the asm file.
8653 @item -minit-stack=@var{N}
8654 @opindex minit-stack
8655 Specify the initial stack address, which may be a symbol or numeric value,
8656 @samp{__stack} is the default.
8658 @item -mno-interrupts
8659 @opindex mno-interrupts
8660 Generated code is not compatible with hardware interrupts.
8661 Code size will be smaller.
8663 @item -mcall-prologues
8664 @opindex mcall-prologues
8665 Functions prologues/epilogues expanded as call to appropriate
8666 subroutines. Code size will be smaller.
8668 @item -mno-tablejump
8669 @opindex mno-tablejump
8670 Do not generate tablejump insns which sometimes increase code size.
8673 @opindex mtiny-stack
8674 Change only the low 8 bits of the stack pointer.
8678 Assume int to be 8 bit integer. This affects the sizes of all types: A
8679 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8680 and long long will be 4 bytes. Please note that this option does not
8681 comply to the C standards, but it will provide you with smaller code
8685 @node Blackfin Options
8686 @subsection Blackfin Options
8687 @cindex Blackfin Options
8690 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8692 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8693 can be one of @samp{bf522}, @samp{bf523}, @samp{bf524},
8694 @samp{bf525}, @samp{bf526}, @samp{bf527},
8695 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8696 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8697 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
8699 The optional @var{sirevision} specifies the silicon revision of the target
8700 Blackfin processor. Any workarounds available for the targeted silicon revision
8701 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8702 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8703 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8704 hexadecimal digits representing the major and minor numbers in the silicon
8705 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8706 is not defined. If @var{sirevision} is @samp{any}, the
8707 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8708 If this optional @var{sirevision} is not used, GCC assumes the latest known
8709 silicon revision of the targeted Blackfin processor.
8711 Support for @samp{bf561} is incomplete. For @samp{bf561},
8712 Only the processor macro is defined.
8713 Without this option, @samp{bf532} is used as the processor by default.
8714 The corresponding predefined processor macros for @var{cpu} is to
8715 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8716 provided by libgloss to be linked in if @option{-msim} is not given.
8720 Specifies that the program will be run on the simulator. This causes
8721 the simulator BSP provided by libgloss to be linked in. This option
8722 has effect only for @samp{bfin-elf} toolchain.
8723 Certain other options, such as @option{-mid-shared-library} and
8724 @option{-mfdpic}, imply @option{-msim}.
8726 @item -momit-leaf-frame-pointer
8727 @opindex momit-leaf-frame-pointer
8728 Don't keep the frame pointer in a register for leaf functions. This
8729 avoids the instructions to save, set up and restore frame pointers and
8730 makes an extra register available in leaf functions. The option
8731 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8732 which might make debugging harder.
8734 @item -mspecld-anomaly
8735 @opindex mspecld-anomaly
8736 When enabled, the compiler will ensure that the generated code does not
8737 contain speculative loads after jump instructions. If this option is used,
8738 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8740 @item -mno-specld-anomaly
8741 @opindex mno-specld-anomaly
8742 Don't generate extra code to prevent speculative loads from occurring.
8744 @item -mcsync-anomaly
8745 @opindex mcsync-anomaly
8746 When enabled, the compiler will ensure that the generated code does not
8747 contain CSYNC or SSYNC instructions too soon after conditional branches.
8748 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8750 @item -mno-csync-anomaly
8751 @opindex mno-csync-anomaly
8752 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8753 occurring too soon after a conditional branch.
8757 When enabled, the compiler is free to take advantage of the knowledge that
8758 the entire program fits into the low 64k of memory.
8761 @opindex mno-low-64k
8762 Assume that the program is arbitrarily large. This is the default.
8764 @item -mstack-check-l1
8765 @opindex mstack-check-l1
8766 Do stack checking using information placed into L1 scratchpad memory by the
8769 @item -mid-shared-library
8770 @opindex mid-shared-library
8771 Generate code that supports shared libraries via the library ID method.
8772 This allows for execute in place and shared libraries in an environment
8773 without virtual memory management. This option implies @option{-fPIC}.
8774 With a @samp{bfin-elf} target, this option implies @option{-msim}.
8776 @item -mno-id-shared-library
8777 @opindex mno-id-shared-library
8778 Generate code that doesn't assume ID based shared libraries are being used.
8779 This is the default.
8781 @item -mleaf-id-shared-library
8782 @opindex mleaf-id-shared-library
8783 Generate code that supports shared libraries via the library ID method,
8784 but assumes that this library or executable won't link against any other
8785 ID shared libraries. That allows the compiler to use faster code for jumps
8788 @item -mno-leaf-id-shared-library
8789 @opindex mno-leaf-id-shared-library
8790 Do not assume that the code being compiled won't link against any ID shared
8791 libraries. Slower code will be generated for jump and call insns.
8793 @item -mshared-library-id=n
8794 @opindex mshared-library-id
8795 Specified the identification number of the ID based shared library being
8796 compiled. Specifying a value of 0 will generate more compact code, specifying
8797 other values will force the allocation of that number to the current
8798 library but is no more space or time efficient than omitting this option.
8802 Generate code that allows the data segment to be located in a different
8803 area of memory from the text segment. This allows for execute in place in
8804 an environment without virtual memory management by eliminating relocations
8805 against the text section.
8808 @opindex mno-sep-data
8809 Generate code that assumes that the data segment follows the text segment.
8810 This is the default.
8813 @itemx -mno-long-calls
8814 @opindex mlong-calls
8815 @opindex mno-long-calls
8816 Tells the compiler to perform function calls by first loading the
8817 address of the function into a register and then performing a subroutine
8818 call on this register. This switch is needed if the target function
8819 will lie outside of the 24 bit addressing range of the offset based
8820 version of subroutine call instruction.
8822 This feature is not enabled by default. Specifying
8823 @option{-mno-long-calls} will restore the default behavior. Note these
8824 switches have no effect on how the compiler generates code to handle
8825 function calls via function pointers.
8829 Link with the fast floating-point library. This library relaxes some of
8830 the IEEE floating-point standard's rules for checking inputs against
8831 Not-a-Number (NAN), in the interest of performance.
8834 @opindex minline-plt
8835 Enable inlining of PLT entries in function calls to functions that are
8836 not known to bind locally. It has no effect without @option{-mfdpic}.
8840 @subsection CRIS Options
8841 @cindex CRIS Options
8843 These options are defined specifically for the CRIS ports.
8846 @item -march=@var{architecture-type}
8847 @itemx -mcpu=@var{architecture-type}
8850 Generate code for the specified architecture. The choices for
8851 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8852 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8853 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8856 @item -mtune=@var{architecture-type}
8858 Tune to @var{architecture-type} everything applicable about the generated
8859 code, except for the ABI and the set of available instructions. The
8860 choices for @var{architecture-type} are the same as for
8861 @option{-march=@var{architecture-type}}.
8863 @item -mmax-stack-frame=@var{n}
8864 @opindex mmax-stack-frame
8865 Warn when the stack frame of a function exceeds @var{n} bytes.
8867 @item -melinux-stacksize=@var{n}
8868 @opindex melinux-stacksize
8869 Only available with the @samp{cris-axis-aout} target. Arranges for
8870 indications in the program to the kernel loader that the stack of the
8871 program should be set to @var{n} bytes.
8877 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8878 @option{-march=v3} and @option{-march=v8} respectively.
8880 @item -mmul-bug-workaround
8881 @itemx -mno-mul-bug-workaround
8882 @opindex mmul-bug-workaround
8883 @opindex mno-mul-bug-workaround
8884 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8885 models where it applies. This option is active by default.
8889 Enable CRIS-specific verbose debug-related information in the assembly
8890 code. This option also has the effect to turn off the @samp{#NO_APP}
8891 formatted-code indicator to the assembler at the beginning of the
8896 Do not use condition-code results from previous instruction; always emit
8897 compare and test instructions before use of condition codes.
8899 @item -mno-side-effects
8900 @opindex mno-side-effects
8901 Do not emit instructions with side-effects in addressing modes other than
8905 @itemx -mno-stack-align
8907 @itemx -mno-data-align
8908 @itemx -mconst-align
8909 @itemx -mno-const-align
8910 @opindex mstack-align
8911 @opindex mno-stack-align
8912 @opindex mdata-align
8913 @opindex mno-data-align
8914 @opindex mconst-align
8915 @opindex mno-const-align
8916 These options (no-options) arranges (eliminate arrangements) for the
8917 stack-frame, individual data and constants to be aligned for the maximum
8918 single data access size for the chosen CPU model. The default is to
8919 arrange for 32-bit alignment. ABI details such as structure layout are
8920 not affected by these options.
8928 Similar to the stack- data- and const-align options above, these options
8929 arrange for stack-frame, writable data and constants to all be 32-bit,
8930 16-bit or 8-bit aligned. The default is 32-bit alignment.
8932 @item -mno-prologue-epilogue
8933 @itemx -mprologue-epilogue
8934 @opindex mno-prologue-epilogue
8935 @opindex mprologue-epilogue
8936 With @option{-mno-prologue-epilogue}, the normal function prologue and
8937 epilogue that sets up the stack-frame are omitted and no return
8938 instructions or return sequences are generated in the code. Use this
8939 option only together with visual inspection of the compiled code: no
8940 warnings or errors are generated when call-saved registers must be saved,
8941 or storage for local variable needs to be allocated.
8947 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8948 instruction sequences that load addresses for functions from the PLT part
8949 of the GOT rather than (traditional on other architectures) calls to the
8950 PLT@. The default is @option{-mgotplt}.
8954 Legacy no-op option only recognized with the cris-axis-aout target.
8958 Legacy no-op option only recognized with the cris-axis-elf and
8959 cris-axis-linux-gnu targets.
8963 Only recognized with the cris-axis-aout target, where it selects a
8964 GNU/linux-like multilib, include files and instruction set for
8969 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8973 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8974 to link with input-output functions from a simulator library. Code,
8975 initialized data and zero-initialized data are allocated consecutively.
8979 Like @option{-sim}, but pass linker options to locate initialized data at
8980 0x40000000 and zero-initialized data at 0x80000000.
8984 @subsection CRX Options
8987 These options are defined specifically for the CRX ports.
8993 Enable the use of multiply-accumulate instructions. Disabled by default.
8997 Push instructions will be used to pass outgoing arguments when functions
8998 are called. Enabled by default.
9001 @node Darwin Options
9002 @subsection Darwin Options
9003 @cindex Darwin options
9005 These options are defined for all architectures running the Darwin operating
9008 FSF GCC on Darwin does not create ``fat'' object files; it will create
9009 an object file for the single architecture that it was built to
9010 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9011 @option{-arch} options are used; it does so by running the compiler or
9012 linker multiple times and joining the results together with
9015 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9016 @samp{i686}) is determined by the flags that specify the ISA
9017 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9018 @option{-force_cpusubtype_ALL} option can be used to override this.
9020 The Darwin tools vary in their behavior when presented with an ISA
9021 mismatch. The assembler, @file{as}, will only permit instructions to
9022 be used that are valid for the subtype of the file it is generating,
9023 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9024 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9025 and print an error if asked to create a shared library with a less
9026 restrictive subtype than its input files (for instance, trying to put
9027 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9028 for executables, @file{ld}, will quietly give the executable the most
9029 restrictive subtype of any of its input files.
9034 Add the framework directory @var{dir} to the head of the list of
9035 directories to be searched for header files. These directories are
9036 interleaved with those specified by @option{-I} options and are
9037 scanned in a left-to-right order.
9039 A framework directory is a directory with frameworks in it. A
9040 framework is a directory with a @samp{"Headers"} and/or
9041 @samp{"PrivateHeaders"} directory contained directly in it that ends
9042 in @samp{".framework"}. The name of a framework is the name of this
9043 directory excluding the @samp{".framework"}. Headers associated with
9044 the framework are found in one of those two directories, with
9045 @samp{"Headers"} being searched first. A subframework is a framework
9046 directory that is in a framework's @samp{"Frameworks"} directory.
9047 Includes of subframework headers can only appear in a header of a
9048 framework that contains the subframework, or in a sibling subframework
9049 header. Two subframeworks are siblings if they occur in the same
9050 framework. A subframework should not have the same name as a
9051 framework, a warning will be issued if this is violated. Currently a
9052 subframework cannot have subframeworks, in the future, the mechanism
9053 may be extended to support this. The standard frameworks can be found
9054 in @samp{"/System/Library/Frameworks"} and
9055 @samp{"/Library/Frameworks"}. An example include looks like
9056 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9057 the name of the framework and header.h is found in the
9058 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9060 @item -iframework@var{dir}
9062 Like @option{-F} except the directory is a treated as a system
9063 directory. The main difference between this @option{-iframework} and
9064 @option{-F} is that with @option{-iframework} the compiler does not
9065 warn about constructs contained within header files found via
9066 @var{dir}. This option is valid only for the C family of languages.
9070 Emit debugging information for symbols that are used. For STABS
9071 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9072 This is by default ON@.
9076 Emit debugging information for all symbols and types.
9078 @item -mmacosx-version-min=@var{version}
9079 The earliest version of MacOS X that this executable will run on
9080 is @var{version}. Typical values of @var{version} include @code{10.1},
9081 @code{10.2}, and @code{10.3.9}.
9083 If the compiler was built to use the system's headers by default,
9084 then the default for this option is the system version on which the
9085 compiler is running, otherwise the default is to make choices which
9086 are compatible with as many systems and code bases as possible.
9090 Enable kernel development mode. The @option{-mkernel} option sets
9091 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9092 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9093 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9094 applicable. This mode also sets @option{-mno-altivec},
9095 @option{-msoft-float}, @option{-fno-builtin} and
9096 @option{-mlong-branch} for PowerPC targets.
9098 @item -mone-byte-bool
9099 @opindex mone-byte-bool
9100 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9101 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9102 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9103 option has no effect on x86.
9105 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9106 to generate code that is not binary compatible with code generated
9107 without that switch. Using this switch may require recompiling all
9108 other modules in a program, including system libraries. Use this
9109 switch to conform to a non-default data model.
9111 @item -mfix-and-continue
9112 @itemx -ffix-and-continue
9113 @itemx -findirect-data
9114 @opindex mfix-and-continue
9115 @opindex ffix-and-continue
9116 @opindex findirect-data
9117 Generate code suitable for fast turn around development. Needed to
9118 enable gdb to dynamically load @code{.o} files into already running
9119 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9120 are provided for backwards compatibility.
9124 Loads all members of static archive libraries.
9125 See man ld(1) for more information.
9127 @item -arch_errors_fatal
9128 @opindex arch_errors_fatal
9129 Cause the errors having to do with files that have the wrong architecture
9133 @opindex bind_at_load
9134 Causes the output file to be marked such that the dynamic linker will
9135 bind all undefined references when the file is loaded or launched.
9139 Produce a Mach-o bundle format file.
9140 See man ld(1) for more information.
9142 @item -bundle_loader @var{executable}
9143 @opindex bundle_loader
9144 This option specifies the @var{executable} that will be loading the build
9145 output file being linked. See man ld(1) for more information.
9149 When passed this option, GCC will produce a dynamic library instead of
9150 an executable when linking, using the Darwin @file{libtool} command.
9152 @item -force_cpusubtype_ALL
9153 @opindex force_cpusubtype_ALL
9154 This causes GCC's output file to have the @var{ALL} subtype, instead of
9155 one controlled by the @option{-mcpu} or @option{-march} option.
9157 @item -allowable_client @var{client_name}
9159 @itemx -compatibility_version
9160 @itemx -current_version
9162 @itemx -dependency-file
9164 @itemx -dylinker_install_name
9166 @itemx -exported_symbols_list
9168 @itemx -flat_namespace
9169 @itemx -force_flat_namespace
9170 @itemx -headerpad_max_install_names
9173 @itemx -install_name
9174 @itemx -keep_private_externs
9175 @itemx -multi_module
9176 @itemx -multiply_defined
9177 @itemx -multiply_defined_unused
9179 @itemx -no_dead_strip_inits_and_terms
9180 @itemx -nofixprebinding
9183 @itemx -noseglinkedit
9184 @itemx -pagezero_size
9186 @itemx -prebind_all_twolevel_modules
9187 @itemx -private_bundle
9188 @itemx -read_only_relocs
9190 @itemx -sectobjectsymbols
9194 @itemx -sectobjectsymbols
9197 @itemx -segs_read_only_addr
9198 @itemx -segs_read_write_addr
9199 @itemx -seg_addr_table
9200 @itemx -seg_addr_table_filename
9203 @itemx -segs_read_only_addr
9204 @itemx -segs_read_write_addr
9205 @itemx -single_module
9208 @itemx -sub_umbrella
9209 @itemx -twolevel_namespace
9212 @itemx -unexported_symbols_list
9213 @itemx -weak_reference_mismatches
9216 @opindex allowable_client
9217 @opindex client_name
9218 @opindex compatibility_version
9219 @opindex current_version
9221 @opindex dependency-file
9223 @opindex dylinker_install_name
9225 @opindex exported_symbols_list
9227 @opindex flat_namespace
9228 @opindex force_flat_namespace
9229 @opindex headerpad_max_install_names
9232 @opindex install_name
9233 @opindex keep_private_externs
9234 @opindex multi_module
9235 @opindex multiply_defined
9236 @opindex multiply_defined_unused
9238 @opindex no_dead_strip_inits_and_terms
9239 @opindex nofixprebinding
9240 @opindex nomultidefs
9242 @opindex noseglinkedit
9243 @opindex pagezero_size
9245 @opindex prebind_all_twolevel_modules
9246 @opindex private_bundle
9247 @opindex read_only_relocs
9249 @opindex sectobjectsymbols
9253 @opindex sectobjectsymbols
9256 @opindex segs_read_only_addr
9257 @opindex segs_read_write_addr
9258 @opindex seg_addr_table
9259 @opindex seg_addr_table_filename
9260 @opindex seglinkedit
9262 @opindex segs_read_only_addr
9263 @opindex segs_read_write_addr
9264 @opindex single_module
9266 @opindex sub_library
9267 @opindex sub_umbrella
9268 @opindex twolevel_namespace
9271 @opindex unexported_symbols_list
9272 @opindex weak_reference_mismatches
9273 @opindex whatsloaded
9275 These options are passed to the Darwin linker. The Darwin linker man page
9276 describes them in detail.
9279 @node DEC Alpha Options
9280 @subsection DEC Alpha Options
9282 These @samp{-m} options are defined for the DEC Alpha implementations:
9285 @item -mno-soft-float
9287 @opindex mno-soft-float
9288 @opindex msoft-float
9289 Use (do not use) the hardware floating-point instructions for
9290 floating-point operations. When @option{-msoft-float} is specified,
9291 functions in @file{libgcc.a} will be used to perform floating-point
9292 operations. Unless they are replaced by routines that emulate the
9293 floating-point operations, or compiled in such a way as to call such
9294 emulations routines, these routines will issue floating-point
9295 operations. If you are compiling for an Alpha without floating-point
9296 operations, you must ensure that the library is built so as not to call
9299 Note that Alpha implementations without floating-point operations are
9300 required to have floating-point registers.
9305 @opindex mno-fp-regs
9306 Generate code that uses (does not use) the floating-point register set.
9307 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9308 register set is not used, floating point operands are passed in integer
9309 registers as if they were integers and floating-point results are passed
9310 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9311 so any function with a floating-point argument or return value called by code
9312 compiled with @option{-mno-fp-regs} must also be compiled with that
9315 A typical use of this option is building a kernel that does not use,
9316 and hence need not save and restore, any floating-point registers.
9320 The Alpha architecture implements floating-point hardware optimized for
9321 maximum performance. It is mostly compliant with the IEEE floating
9322 point standard. However, for full compliance, software assistance is
9323 required. This option generates code fully IEEE compliant code
9324 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9325 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9326 defined during compilation. The resulting code is less efficient but is
9327 able to correctly support denormalized numbers and exceptional IEEE
9328 values such as not-a-number and plus/minus infinity. Other Alpha
9329 compilers call this option @option{-ieee_with_no_inexact}.
9331 @item -mieee-with-inexact
9332 @opindex mieee-with-inexact
9333 This is like @option{-mieee} except the generated code also maintains
9334 the IEEE @var{inexact-flag}. Turning on this option causes the
9335 generated code to implement fully-compliant IEEE math. In addition to
9336 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9337 macro. On some Alpha implementations the resulting code may execute
9338 significantly slower than the code generated by default. Since there is
9339 very little code that depends on the @var{inexact-flag}, you should
9340 normally not specify this option. Other Alpha compilers call this
9341 option @option{-ieee_with_inexact}.
9343 @item -mfp-trap-mode=@var{trap-mode}
9344 @opindex mfp-trap-mode
9345 This option controls what floating-point related traps are enabled.
9346 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9347 The trap mode can be set to one of four values:
9351 This is the default (normal) setting. The only traps that are enabled
9352 are the ones that cannot be disabled in software (e.g., division by zero
9356 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9360 Like @samp{u}, but the instructions are marked to be safe for software
9361 completion (see Alpha architecture manual for details).
9364 Like @samp{su}, but inexact traps are enabled as well.
9367 @item -mfp-rounding-mode=@var{rounding-mode}
9368 @opindex mfp-rounding-mode
9369 Selects the IEEE rounding mode. Other Alpha compilers call this option
9370 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9375 Normal IEEE rounding mode. Floating point numbers are rounded towards
9376 the nearest machine number or towards the even machine number in case
9380 Round towards minus infinity.
9383 Chopped rounding mode. Floating point numbers are rounded towards zero.
9386 Dynamic rounding mode. A field in the floating point control register
9387 (@var{fpcr}, see Alpha architecture reference manual) controls the
9388 rounding mode in effect. The C library initializes this register for
9389 rounding towards plus infinity. Thus, unless your program modifies the
9390 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9393 @item -mtrap-precision=@var{trap-precision}
9394 @opindex mtrap-precision
9395 In the Alpha architecture, floating point traps are imprecise. This
9396 means without software assistance it is impossible to recover from a
9397 floating trap and program execution normally needs to be terminated.
9398 GCC can generate code that can assist operating system trap handlers
9399 in determining the exact location that caused a floating point trap.
9400 Depending on the requirements of an application, different levels of
9401 precisions can be selected:
9405 Program precision. This option is the default and means a trap handler
9406 can only identify which program caused a floating point exception.
9409 Function precision. The trap handler can determine the function that
9410 caused a floating point exception.
9413 Instruction precision. The trap handler can determine the exact
9414 instruction that caused a floating point exception.
9417 Other Alpha compilers provide the equivalent options called
9418 @option{-scope_safe} and @option{-resumption_safe}.
9420 @item -mieee-conformant
9421 @opindex mieee-conformant
9422 This option marks the generated code as IEEE conformant. You must not
9423 use this option unless you also specify @option{-mtrap-precision=i} and either
9424 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9425 is to emit the line @samp{.eflag 48} in the function prologue of the
9426 generated assembly file. Under DEC Unix, this has the effect that
9427 IEEE-conformant math library routines will be linked in.
9429 @item -mbuild-constants
9430 @opindex mbuild-constants
9431 Normally GCC examines a 32- or 64-bit integer constant to
9432 see if it can construct it from smaller constants in two or three
9433 instructions. If it cannot, it will output the constant as a literal and
9434 generate code to load it from the data segment at runtime.
9436 Use this option to require GCC to construct @emph{all} integer constants
9437 using code, even if it takes more instructions (the maximum is six).
9439 You would typically use this option to build a shared library dynamic
9440 loader. Itself a shared library, it must relocate itself in memory
9441 before it can find the variables and constants in its own data segment.
9447 Select whether to generate code to be assembled by the vendor-supplied
9448 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9466 Indicate whether GCC should generate code to use the optional BWX,
9467 CIX, FIX and MAX instruction sets. The default is to use the instruction
9468 sets supported by the CPU type specified via @option{-mcpu=} option or that
9469 of the CPU on which GCC was built if none was specified.
9474 @opindex mfloat-ieee
9475 Generate code that uses (does not use) VAX F and G floating point
9476 arithmetic instead of IEEE single and double precision.
9478 @item -mexplicit-relocs
9479 @itemx -mno-explicit-relocs
9480 @opindex mexplicit-relocs
9481 @opindex mno-explicit-relocs
9482 Older Alpha assemblers provided no way to generate symbol relocations
9483 except via assembler macros. Use of these macros does not allow
9484 optimal instruction scheduling. GNU binutils as of version 2.12
9485 supports a new syntax that allows the compiler to explicitly mark
9486 which relocations should apply to which instructions. This option
9487 is mostly useful for debugging, as GCC detects the capabilities of
9488 the assembler when it is built and sets the default accordingly.
9492 @opindex msmall-data
9493 @opindex mlarge-data
9494 When @option{-mexplicit-relocs} is in effect, static data is
9495 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9496 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9497 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9498 16-bit relocations off of the @code{$gp} register. This limits the
9499 size of the small data area to 64KB, but allows the variables to be
9500 directly accessed via a single instruction.
9502 The default is @option{-mlarge-data}. With this option the data area
9503 is limited to just below 2GB@. Programs that require more than 2GB of
9504 data must use @code{malloc} or @code{mmap} to allocate the data in the
9505 heap instead of in the program's data segment.
9507 When generating code for shared libraries, @option{-fpic} implies
9508 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9512 @opindex msmall-text
9513 @opindex mlarge-text
9514 When @option{-msmall-text} is used, the compiler assumes that the
9515 code of the entire program (or shared library) fits in 4MB, and is
9516 thus reachable with a branch instruction. When @option{-msmall-data}
9517 is used, the compiler can assume that all local symbols share the
9518 same @code{$gp} value, and thus reduce the number of instructions
9519 required for a function call from 4 to 1.
9521 The default is @option{-mlarge-text}.
9523 @item -mcpu=@var{cpu_type}
9525 Set the instruction set and instruction scheduling parameters for
9526 machine type @var{cpu_type}. You can specify either the @samp{EV}
9527 style name or the corresponding chip number. GCC supports scheduling
9528 parameters for the EV4, EV5 and EV6 family of processors and will
9529 choose the default values for the instruction set from the processor
9530 you specify. If you do not specify a processor type, GCC will default
9531 to the processor on which the compiler was built.
9533 Supported values for @var{cpu_type} are
9539 Schedules as an EV4 and has no instruction set extensions.
9543 Schedules as an EV5 and has no instruction set extensions.
9547 Schedules as an EV5 and supports the BWX extension.
9552 Schedules as an EV5 and supports the BWX and MAX extensions.
9556 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9560 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9563 @item -mtune=@var{cpu_type}
9565 Set only the instruction scheduling parameters for machine type
9566 @var{cpu_type}. The instruction set is not changed.
9568 @item -mmemory-latency=@var{time}
9569 @opindex mmemory-latency
9570 Sets the latency the scheduler should assume for typical memory
9571 references as seen by the application. This number is highly
9572 dependent on the memory access patterns used by the application
9573 and the size of the external cache on the machine.
9575 Valid options for @var{time} are
9579 A decimal number representing clock cycles.
9585 The compiler contains estimates of the number of clock cycles for
9586 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9587 (also called Dcache, Scache, and Bcache), as well as to main memory.
9588 Note that L3 is only valid for EV5.
9593 @node DEC Alpha/VMS Options
9594 @subsection DEC Alpha/VMS Options
9596 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9599 @item -mvms-return-codes
9600 @opindex mvms-return-codes
9601 Return VMS condition codes from main. The default is to return POSIX
9602 style condition (e.g.@: error) codes.
9606 @subsection FRV Options
9613 Only use the first 32 general purpose registers.
9618 Use all 64 general purpose registers.
9623 Use only the first 32 floating point registers.
9628 Use all 64 floating point registers
9631 @opindex mhard-float
9633 Use hardware instructions for floating point operations.
9636 @opindex msoft-float
9638 Use library routines for floating point operations.
9643 Dynamically allocate condition code registers.
9648 Do not try to dynamically allocate condition code registers, only
9649 use @code{icc0} and @code{fcc0}.
9654 Change ABI to use double word insns.
9659 Do not use double word instructions.
9664 Use floating point double instructions.
9669 Do not use floating point double instructions.
9674 Use media instructions.
9679 Do not use media instructions.
9684 Use multiply and add/subtract instructions.
9689 Do not use multiply and add/subtract instructions.
9694 Select the FDPIC ABI, that uses function descriptors to represent
9695 pointers to functions. Without any PIC/PIE-related options, it
9696 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9697 assumes GOT entries and small data are within a 12-bit range from the
9698 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9699 are computed with 32 bits.
9700 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9703 @opindex minline-plt
9705 Enable inlining of PLT entries in function calls to functions that are
9706 not known to bind locally. It has no effect without @option{-mfdpic}.
9707 It's enabled by default if optimizing for speed and compiling for
9708 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9709 optimization option such as @option{-O3} or above is present in the
9715 Assume a large TLS segment when generating thread-local code.
9720 Do not assume a large TLS segment when generating thread-local code.
9725 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9726 that is known to be in read-only sections. It's enabled by default,
9727 except for @option{-fpic} or @option{-fpie}: even though it may help
9728 make the global offset table smaller, it trades 1 instruction for 4.
9729 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9730 one of which may be shared by multiple symbols, and it avoids the need
9731 for a GOT entry for the referenced symbol, so it's more likely to be a
9732 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9734 @item -multilib-library-pic
9735 @opindex multilib-library-pic
9737 Link with the (library, not FD) pic libraries. It's implied by
9738 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9739 @option{-fpic} without @option{-mfdpic}. You should never have to use
9745 Follow the EABI requirement of always creating a frame pointer whenever
9746 a stack frame is allocated. This option is enabled by default and can
9747 be disabled with @option{-mno-linked-fp}.
9750 @opindex mlong-calls
9752 Use indirect addressing to call functions outside the current
9753 compilation unit. This allows the functions to be placed anywhere
9754 within the 32-bit address space.
9756 @item -malign-labels
9757 @opindex malign-labels
9759 Try to align labels to an 8-byte boundary by inserting nops into the
9760 previous packet. This option only has an effect when VLIW packing
9761 is enabled. It doesn't create new packets; it merely adds nops to
9765 @opindex mlibrary-pic
9767 Generate position-independent EABI code.
9772 Use only the first four media accumulator registers.
9777 Use all eight media accumulator registers.
9782 Pack VLIW instructions.
9787 Do not pack VLIW instructions.
9792 Do not mark ABI switches in e_flags.
9797 Enable the use of conditional-move instructions (default).
9799 This switch is mainly for debugging the compiler and will likely be removed
9800 in a future version.
9802 @item -mno-cond-move
9803 @opindex mno-cond-move
9805 Disable the use of conditional-move instructions.
9807 This switch is mainly for debugging the compiler and will likely be removed
9808 in a future version.
9813 Enable the use of conditional set instructions (default).
9815 This switch is mainly for debugging the compiler and will likely be removed
9816 in a future version.
9821 Disable the use of conditional set instructions.
9823 This switch is mainly for debugging the compiler and will likely be removed
9824 in a future version.
9829 Enable the use of conditional execution (default).
9831 This switch is mainly for debugging the compiler and will likely be removed
9832 in a future version.
9834 @item -mno-cond-exec
9835 @opindex mno-cond-exec
9837 Disable the use of conditional execution.
9839 This switch is mainly for debugging the compiler and will likely be removed
9840 in a future version.
9843 @opindex mvliw-branch
9845 Run a pass to pack branches into VLIW instructions (default).
9847 This switch is mainly for debugging the compiler and will likely be removed
9848 in a future version.
9850 @item -mno-vliw-branch
9851 @opindex mno-vliw-branch
9853 Do not run a pass to pack branches into VLIW instructions.
9855 This switch is mainly for debugging the compiler and will likely be removed
9856 in a future version.
9858 @item -mmulti-cond-exec
9859 @opindex mmulti-cond-exec
9861 Enable optimization of @code{&&} and @code{||} in conditional execution
9864 This switch is mainly for debugging the compiler and will likely be removed
9865 in a future version.
9867 @item -mno-multi-cond-exec
9868 @opindex mno-multi-cond-exec
9870 Disable optimization of @code{&&} and @code{||} in conditional execution.
9872 This switch is mainly for debugging the compiler and will likely be removed
9873 in a future version.
9875 @item -mnested-cond-exec
9876 @opindex mnested-cond-exec
9878 Enable nested conditional execution optimizations (default).
9880 This switch is mainly for debugging the compiler and will likely be removed
9881 in a future version.
9883 @item -mno-nested-cond-exec
9884 @opindex mno-nested-cond-exec
9886 Disable nested conditional execution optimizations.
9888 This switch is mainly for debugging the compiler and will likely be removed
9889 in a future version.
9891 @item -moptimize-membar
9892 @opindex moptimize-membar
9894 This switch removes redundant @code{membar} instructions from the
9895 compiler generated code. It is enabled by default.
9897 @item -mno-optimize-membar
9898 @opindex mno-optimize-membar
9900 This switch disables the automatic removal of redundant @code{membar}
9901 instructions from the generated code.
9903 @item -mtomcat-stats
9904 @opindex mtomcat-stats
9906 Cause gas to print out tomcat statistics.
9908 @item -mcpu=@var{cpu}
9911 Select the processor type for which to generate code. Possible values are
9912 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9913 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9917 @node GNU/Linux Options
9918 @subsection GNU/Linux Options
9920 These @samp{-m} options are defined for GNU/Linux targets:
9925 Use the GNU C library instead of uClibc. This is the default except
9926 on @samp{*-*-linux-*uclibc*} targets.
9930 Use uClibc instead of the GNU C library. This is the default on
9931 @samp{*-*-linux-*uclibc*} targets.
9934 @node H8/300 Options
9935 @subsection H8/300 Options
9937 These @samp{-m} options are defined for the H8/300 implementations:
9942 Shorten some address references at link time, when possible; uses the
9943 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9944 ld, Using ld}, for a fuller description.
9948 Generate code for the H8/300H@.
9952 Generate code for the H8S@.
9956 Generate code for the H8S and H8/300H in the normal mode. This switch
9957 must be used either with @option{-mh} or @option{-ms}.
9961 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9965 Make @code{int} data 32 bits by default.
9969 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9970 The default for the H8/300H and H8S is to align longs and floats on 4
9972 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9973 This option has no effect on the H8/300.
9977 @subsection HPPA Options
9978 @cindex HPPA Options
9980 These @samp{-m} options are defined for the HPPA family of computers:
9983 @item -march=@var{architecture-type}
9985 Generate code for the specified architecture. The choices for
9986 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9987 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9988 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9989 architecture option for your machine. Code compiled for lower numbered
9990 architectures will run on higher numbered architectures, but not the
9994 @itemx -mpa-risc-1-1
9995 @itemx -mpa-risc-2-0
9996 @opindex mpa-risc-1-0
9997 @opindex mpa-risc-1-1
9998 @opindex mpa-risc-2-0
9999 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10002 @opindex mbig-switch
10003 Generate code suitable for big switch tables. Use this option only if
10004 the assembler/linker complain about out of range branches within a switch
10007 @item -mjump-in-delay
10008 @opindex mjump-in-delay
10009 Fill delay slots of function calls with unconditional jump instructions
10010 by modifying the return pointer for the function call to be the target
10011 of the conditional jump.
10013 @item -mdisable-fpregs
10014 @opindex mdisable-fpregs
10015 Prevent floating point registers from being used in any manner. This is
10016 necessary for compiling kernels which perform lazy context switching of
10017 floating point registers. If you use this option and attempt to perform
10018 floating point operations, the compiler will abort.
10020 @item -mdisable-indexing
10021 @opindex mdisable-indexing
10022 Prevent the compiler from using indexing address modes. This avoids some
10023 rather obscure problems when compiling MIG generated code under MACH@.
10025 @item -mno-space-regs
10026 @opindex mno-space-regs
10027 Generate code that assumes the target has no space registers. This allows
10028 GCC to generate faster indirect calls and use unscaled index address modes.
10030 Such code is suitable for level 0 PA systems and kernels.
10032 @item -mfast-indirect-calls
10033 @opindex mfast-indirect-calls
10034 Generate code that assumes calls never cross space boundaries. This
10035 allows GCC to emit code which performs faster indirect calls.
10037 This option will not work in the presence of shared libraries or nested
10040 @item -mfixed-range=@var{register-range}
10041 @opindex mfixed-range
10042 Generate code treating the given register range as fixed registers.
10043 A fixed register is one that the register allocator can not use. This is
10044 useful when compiling kernel code. A register range is specified as
10045 two registers separated by a dash. Multiple register ranges can be
10046 specified separated by a comma.
10048 @item -mlong-load-store
10049 @opindex mlong-load-store
10050 Generate 3-instruction load and store sequences as sometimes required by
10051 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10054 @item -mportable-runtime
10055 @opindex mportable-runtime
10056 Use the portable calling conventions proposed by HP for ELF systems.
10060 Enable the use of assembler directives only GAS understands.
10062 @item -mschedule=@var{cpu-type}
10064 Schedule code according to the constraints for the machine type
10065 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10066 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10067 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10068 proper scheduling option for your machine. The default scheduling is
10072 @opindex mlinker-opt
10073 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10074 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10075 linkers in which they give bogus error messages when linking some programs.
10078 @opindex msoft-float
10079 Generate output containing library calls for floating point.
10080 @strong{Warning:} the requisite libraries are not available for all HPPA
10081 targets. Normally the facilities of the machine's usual C compiler are
10082 used, but this cannot be done directly in cross-compilation. You must make
10083 your own arrangements to provide suitable library functions for
10084 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
10085 does provide software floating point support.
10087 @option{-msoft-float} changes the calling convention in the output file;
10088 therefore, it is only useful if you compile @emph{all} of a program with
10089 this option. In particular, you need to compile @file{libgcc.a}, the
10090 library that comes with GCC, with @option{-msoft-float} in order for
10095 Generate the predefine, @code{_SIO}, for server IO@. The default is
10096 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10097 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10098 options are available under HP-UX and HI-UX@.
10102 Use GNU ld specific options. This passes @option{-shared} to ld when
10103 building a shared library. It is the default when GCC is configured,
10104 explicitly or implicitly, with the GNU linker. This option does not
10105 have any affect on which ld is called, it only changes what parameters
10106 are passed to that ld. The ld that is called is determined by the
10107 @option{--with-ld} configure option, GCC's program search path, and
10108 finally by the user's @env{PATH}. The linker used by GCC can be printed
10109 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10110 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
10114 Use HP ld specific options. This passes @option{-b} to ld when building
10115 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10116 links. It is the default when GCC is configured, explicitly or
10117 implicitly, with the HP linker. This option does not have any affect on
10118 which ld is called, it only changes what parameters are passed to that
10119 ld. The ld that is called is determined by the @option{--with-ld}
10120 configure option, GCC's program search path, and finally by the user's
10121 @env{PATH}. The linker used by GCC can be printed using @samp{which
10122 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10123 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
10126 @opindex mno-long-calls
10127 Generate code that uses long call sequences. This ensures that a call
10128 is always able to reach linker generated stubs. The default is to generate
10129 long calls only when the distance from the call site to the beginning
10130 of the function or translation unit, as the case may be, exceeds a
10131 predefined limit set by the branch type being used. The limits for
10132 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10133 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10136 Distances are measured from the beginning of functions when using the
10137 @option{-ffunction-sections} option, or when using the @option{-mgas}
10138 and @option{-mno-portable-runtime} options together under HP-UX with
10141 It is normally not desirable to use this option as it will degrade
10142 performance. However, it may be useful in large applications,
10143 particularly when partial linking is used to build the application.
10145 The types of long calls used depends on the capabilities of the
10146 assembler and linker, and the type of code being generated. The
10147 impact on systems that support long absolute calls, and long pic
10148 symbol-difference or pc-relative calls should be relatively small.
10149 However, an indirect call is used on 32-bit ELF systems in pic code
10150 and it is quite long.
10152 @item -munix=@var{unix-std}
10154 Generate compiler predefines and select a startfile for the specified
10155 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10156 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10157 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10158 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10159 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10162 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10163 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10164 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10165 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10166 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10167 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10169 It is @emph{important} to note that this option changes the interfaces
10170 for various library routines. It also affects the operational behavior
10171 of the C library. Thus, @emph{extreme} care is needed in using this
10174 Library code that is intended to operate with more than one UNIX
10175 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10176 as appropriate. Most GNU software doesn't provide this capability.
10180 Suppress the generation of link options to search libdld.sl when the
10181 @option{-static} option is specified on HP-UX 10 and later.
10185 The HP-UX implementation of setlocale in libc has a dependency on
10186 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10187 when the @option{-static} option is specified, special link options
10188 are needed to resolve this dependency.
10190 On HP-UX 10 and later, the GCC driver adds the necessary options to
10191 link with libdld.sl when the @option{-static} option is specified.
10192 This causes the resulting binary to be dynamic. On the 64-bit port,
10193 the linkers generate dynamic binaries by default in any case. The
10194 @option{-nolibdld} option can be used to prevent the GCC driver from
10195 adding these link options.
10199 Add support for multithreading with the @dfn{dce thread} library
10200 under HP-UX@. This option sets flags for both the preprocessor and
10204 @node i386 and x86-64 Options
10205 @subsection Intel 386 and AMD x86-64 Options
10206 @cindex i386 Options
10207 @cindex x86-64 Options
10208 @cindex Intel 386 Options
10209 @cindex AMD x86-64 Options
10211 These @samp{-m} options are defined for the i386 and x86-64 family of
10215 @item -mtune=@var{cpu-type}
10217 Tune to @var{cpu-type} everything applicable about the generated code, except
10218 for the ABI and the set of available instructions. The choices for
10219 @var{cpu-type} are:
10222 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10223 If you know the CPU on which your code will run, then you should use
10224 the corresponding @option{-mtune} option instead of
10225 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10226 of your application will have, then you should use this option.
10228 As new processors are deployed in the marketplace, the behavior of this
10229 option will change. Therefore, if you upgrade to a newer version of
10230 GCC, the code generated option will change to reflect the processors
10231 that were most common when that version of GCC was released.
10233 There is no @option{-march=generic} option because @option{-march}
10234 indicates the instruction set the compiler can use, and there is no
10235 generic instruction set applicable to all processors. In contrast,
10236 @option{-mtune} indicates the processor (or, in this case, collection of
10237 processors) for which the code is optimized.
10239 This selects the CPU to tune for at compilation time by determining
10240 the processor type of the compiling machine. Using @option{-mtune=native}
10241 will produce code optimized for the local machine under the constraints
10242 of the selected instruction set. Using @option{-march=native} will
10243 enable all instruction subsets supported by the local machine (hence
10244 the result might not run on different machines).
10246 Original Intel's i386 CPU@.
10248 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10249 @item i586, pentium
10250 Intel Pentium CPU with no MMX support.
10252 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10254 Intel PentiumPro CPU@.
10256 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10257 instruction set will be used, so the code will run on all i686 family chips.
10259 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10260 @item pentium3, pentium3m
10261 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10264 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10265 support. Used by Centrino notebooks.
10266 @item pentium4, pentium4m
10267 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10269 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10272 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10273 SSE2 and SSE3 instruction set support.
10275 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10276 instruction set support.
10278 AMD K6 CPU with MMX instruction set support.
10280 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
10281 @item athlon, athlon-tbird
10282 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
10284 @item athlon-4, athlon-xp, athlon-mp
10285 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
10286 instruction set support.
10287 @item k8, opteron, athlon64, athlon-fx
10288 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10289 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
10290 @item k8-sse3, opteron-sse3, athlon64-sse3
10291 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10292 @item amdfam10, barcelona
10293 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10294 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10295 instruction set extensions.)
10297 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10300 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
10301 instruction set support.
10303 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
10304 implemented for this chip.)
10306 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10307 implemented for this chip.)
10309 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10312 While picking a specific @var{cpu-type} will schedule things appropriately
10313 for that particular chip, the compiler will not generate any code that
10314 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10317 @item -march=@var{cpu-type}
10319 Generate instructions for the machine type @var{cpu-type}. The choices
10320 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10321 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10323 @item -mcpu=@var{cpu-type}
10325 A deprecated synonym for @option{-mtune}.
10327 @item -mfpmath=@var{unit}
10329 Generate floating point arithmetics for selected unit @var{unit}. The choices
10330 for @var{unit} are:
10334 Use the standard 387 floating point coprocessor present majority of chips and
10335 emulated otherwise. Code compiled with this option will run almost everywhere.
10336 The temporary results are computed in 80bit precision instead of precision
10337 specified by the type resulting in slightly different results compared to most
10338 of other chips. See @option{-ffloat-store} for more detailed description.
10340 This is the default choice for i386 compiler.
10343 Use scalar floating point instructions present in the SSE instruction set.
10344 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10345 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10346 instruction set supports only single precision arithmetics, thus the double and
10347 extended precision arithmetics is still done using 387. Later version, present
10348 only in Pentium4 and the future AMD x86-64 chips supports double precision
10351 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10352 or @option{-msse2} switches to enable SSE extensions and make this option
10353 effective. For the x86-64 compiler, these extensions are enabled by default.
10355 The resulting code should be considerably faster in the majority of cases and avoid
10356 the numerical instability problems of 387 code, but may break some existing
10357 code that expects temporaries to be 80bit.
10359 This is the default choice for the x86-64 compiler.
10362 Attempt to utilize both instruction sets at once. This effectively double the
10363 amount of available registers and on chips with separate execution units for
10364 387 and SSE the execution resources too. Use this option with care, as it is
10365 still experimental, because the GCC register allocator does not model separate
10366 functional units well resulting in instable performance.
10369 @item -masm=@var{dialect}
10370 @opindex masm=@var{dialect}
10371 Output asm instructions using selected @var{dialect}. Supported
10372 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10373 not support @samp{intel}.
10376 @itemx -mno-ieee-fp
10378 @opindex mno-ieee-fp
10379 Control whether or not the compiler uses IEEE floating point
10380 comparisons. These handle correctly the case where the result of a
10381 comparison is unordered.
10384 @opindex msoft-float
10385 Generate output containing library calls for floating point.
10386 @strong{Warning:} the requisite libraries are not part of GCC@.
10387 Normally the facilities of the machine's usual C compiler are used, but
10388 this can't be done directly in cross-compilation. You must make your
10389 own arrangements to provide suitable library functions for
10392 On machines where a function returns floating point results in the 80387
10393 register stack, some floating point opcodes may be emitted even if
10394 @option{-msoft-float} is used.
10396 @item -mno-fp-ret-in-387
10397 @opindex mno-fp-ret-in-387
10398 Do not use the FPU registers for return values of functions.
10400 The usual calling convention has functions return values of types
10401 @code{float} and @code{double} in an FPU register, even if there
10402 is no FPU@. The idea is that the operating system should emulate
10405 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10406 in ordinary CPU registers instead.
10408 @item -mno-fancy-math-387
10409 @opindex mno-fancy-math-387
10410 Some 387 emulators do not support the @code{sin}, @code{cos} and
10411 @code{sqrt} instructions for the 387. Specify this option to avoid
10412 generating those instructions. This option is the default on FreeBSD,
10413 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10414 indicates that the target cpu will always have an FPU and so the
10415 instruction will not need emulation. As of revision 2.6.1, these
10416 instructions are not generated unless you also use the
10417 @option{-funsafe-math-optimizations} switch.
10419 @item -malign-double
10420 @itemx -mno-align-double
10421 @opindex malign-double
10422 @opindex mno-align-double
10423 Control whether GCC aligns @code{double}, @code{long double}, and
10424 @code{long long} variables on a two word boundary or a one word
10425 boundary. Aligning @code{double} variables on a two word boundary will
10426 produce code that runs somewhat faster on a @samp{Pentium} at the
10427 expense of more memory.
10429 On x86-64, @option{-malign-double} is enabled by default.
10431 @strong{Warning:} if you use the @option{-malign-double} switch,
10432 structures containing the above types will be aligned differently than
10433 the published application binary interface specifications for the 386
10434 and will not be binary compatible with structures in code compiled
10435 without that switch.
10437 @item -m96bit-long-double
10438 @itemx -m128bit-long-double
10439 @opindex m96bit-long-double
10440 @opindex m128bit-long-double
10441 These switches control the size of @code{long double} type. The i386
10442 application binary interface specifies the size to be 96 bits,
10443 so @option{-m96bit-long-double} is the default in 32 bit mode.
10445 Modern architectures (Pentium and newer) would prefer @code{long double}
10446 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10447 conforming to the ABI, this would not be possible. So specifying a
10448 @option{-m128bit-long-double} will align @code{long double}
10449 to a 16 byte boundary by padding the @code{long double} with an additional
10452 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10453 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10455 Notice that neither of these options enable any extra precision over the x87
10456 standard of 80 bits for a @code{long double}.
10458 @strong{Warning:} if you override the default value for your target ABI, the
10459 structures and arrays containing @code{long double} variables will change
10460 their size as well as function calling convention for function taking
10461 @code{long double} will be modified. Hence they will not be binary
10462 compatible with arrays or structures in code compiled without that switch.
10464 @item -mmlarge-data-threshold=@var{number}
10465 @opindex mlarge-data-threshold=@var{number}
10466 When @option{-mcmodel=medium} is specified, the data greater than
10467 @var{threshold} are placed in large data section. This value must be the
10468 same across all object linked into the binary and defaults to 65535.
10472 Use a different function-calling convention, in which functions that
10473 take a fixed number of arguments return with the @code{ret} @var{num}
10474 instruction, which pops their arguments while returning. This saves one
10475 instruction in the caller since there is no need to pop the arguments
10478 You can specify that an individual function is called with this calling
10479 sequence with the function attribute @samp{stdcall}. You can also
10480 override the @option{-mrtd} option by using the function attribute
10481 @samp{cdecl}. @xref{Function Attributes}.
10483 @strong{Warning:} this calling convention is incompatible with the one
10484 normally used on Unix, so you cannot use it if you need to call
10485 libraries compiled with the Unix compiler.
10487 Also, you must provide function prototypes for all functions that
10488 take variable numbers of arguments (including @code{printf});
10489 otherwise incorrect code will be generated for calls to those
10492 In addition, seriously incorrect code will result if you call a
10493 function with too many arguments. (Normally, extra arguments are
10494 harmlessly ignored.)
10496 @item -mregparm=@var{num}
10498 Control how many registers are used to pass integer arguments. By
10499 default, no registers are used to pass arguments, and at most 3
10500 registers can be used. You can control this behavior for a specific
10501 function by using the function attribute @samp{regparm}.
10502 @xref{Function Attributes}.
10504 @strong{Warning:} if you use this switch, and
10505 @var{num} is nonzero, then you must build all modules with the same
10506 value, including any libraries. This includes the system libraries and
10510 @opindex msseregparm
10511 Use SSE register passing conventions for float and double arguments
10512 and return values. You can control this behavior for a specific
10513 function by using the function attribute @samp{sseregparm}.
10514 @xref{Function Attributes}.
10516 @strong{Warning:} if you use this switch then you must build all
10517 modules with the same value, including any libraries. This includes
10518 the system libraries and startup modules.
10527 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10528 is specified, the significands of results of floating-point operations are
10529 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10530 significands of results of floating-point operations to 53 bits (double
10531 precision) and @option{-mpc80} rounds the significands of results of
10532 floating-point operations to 64 bits (extended double precision), which is
10533 the default. When this option is used, floating-point operations in higher
10534 precisions are not available to the programmer without setting the FPU
10535 control word explicitly.
10537 Setting the rounding of floating-point operations to less than the default
10538 80 bits can speed some programs by 2% or more. Note that some mathematical
10539 libraries assume that extended precision (80 bit) floating-point operations
10540 are enabled by default; routines in such libraries could suffer significant
10541 loss of accuracy, typically through so-called "catastrophic cancellation",
10542 when this option is used to set the precision to less than extended precision.
10544 @item -mstackrealign
10545 @opindex mstackrealign
10546 Realign the stack at entry. On the Intel x86, the
10547 @option{-mstackrealign} option will generate an alternate prologue and
10548 epilogue that realigns the runtime stack. This supports mixing legacy
10549 codes that keep a 4-byte aligned stack with modern codes that keep a
10550 16-byte stack for SSE compatibility. The alternate prologue and
10551 epilogue are slower and bigger than the regular ones, and the
10552 alternate prologue requires an extra scratch register; this lowers the
10553 number of registers available if used in conjunction with the
10554 @code{regparm} attribute. The @option{-mstackrealign} option is
10555 incompatible with the nested function prologue; this is considered a
10556 hard error. See also the attribute @code{force_align_arg_pointer},
10557 applicable to individual functions.
10559 @item -mpreferred-stack-boundary=@var{num}
10560 @opindex mpreferred-stack-boundary
10561 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10562 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10563 the default is 4 (16 bytes or 128 bits).
10565 On Pentium and PentiumPro, @code{double} and @code{long double} values
10566 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10567 suffer significant run time performance penalties. On Pentium III, the
10568 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10569 properly if it is not 16 byte aligned.
10571 To ensure proper alignment of this values on the stack, the stack boundary
10572 must be as aligned as that required by any value stored on the stack.
10573 Further, every function must be generated such that it keeps the stack
10574 aligned. Thus calling a function compiled with a higher preferred
10575 stack boundary from a function compiled with a lower preferred stack
10576 boundary will most likely misalign the stack. It is recommended that
10577 libraries that use callbacks always use the default setting.
10579 This extra alignment does consume extra stack space, and generally
10580 increases code size. Code that is sensitive to stack space usage, such
10581 as embedded systems and operating system kernels, may want to reduce the
10582 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10616 These switches enable or disable the use of instructions in the MMX,
10617 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, SSE5, ABM or 3DNow! extended
10619 These extensions are also available as built-in functions: see
10620 @ref{X86 Built-in Functions}, for details of the functions enabled and
10621 disabled by these switches.
10623 To have SSE/SSE2 instructions generated automatically from floating-point
10624 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10626 These options will enable GCC to use these extended instructions in
10627 generated code, even without @option{-mfpmath=sse}. Applications which
10628 perform runtime CPU detection must compile separate files for each
10629 supported architecture, using the appropriate flags. In particular,
10630 the file containing the CPU detection code should be compiled without
10635 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10636 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10637 data types. This is useful for high resolution counters that could be updated
10638 by multiple processors (or cores). This instruction is generated as part of
10639 atomic built-in functions: see @ref{Atomic Builtins} for details.
10643 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10644 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10645 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10646 SAHF are load and store instructions, respectively, for certain status flags.
10647 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10648 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10652 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10653 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
10654 to increase precision instead of DIVSS and SQRTSS (and their vectorized
10655 variants) for single precision floating point arguments. These instructions
10656 are generated only when @option{-funsafe-math-optimizations} is enabled
10657 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
10658 Note that while the throughput of the sequence is higher than the throughput
10659 of the non-reciprocal instruction, the precision of the sequence can be
10660 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
10662 @item -mveclibabi=@var{type}
10663 @opindex mveclibabi
10664 Specifies the ABI type to use for vectorizing intrinsics using an
10665 external library. Supported types are @code{acml} for the AMD
10666 math core library style of interfacing. GCC will currently emit
10667 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10668 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10669 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10670 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10671 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10672 is enabled. A ACML ABI compatible library will have to be specified
10676 @itemx -mno-push-args
10677 @opindex mpush-args
10678 @opindex mno-push-args
10679 Use PUSH operations to store outgoing parameters. This method is shorter
10680 and usually equally fast as method using SUB/MOV operations and is enabled
10681 by default. In some cases disabling it may improve performance because of
10682 improved scheduling and reduced dependencies.
10684 @item -maccumulate-outgoing-args
10685 @opindex maccumulate-outgoing-args
10686 If enabled, the maximum amount of space required for outgoing arguments will be
10687 computed in the function prologue. This is faster on most modern CPUs
10688 because of reduced dependencies, improved scheduling and reduced stack usage
10689 when preferred stack boundary is not equal to 2. The drawback is a notable
10690 increase in code size. This switch implies @option{-mno-push-args}.
10694 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10695 on thread-safe exception handling must compile and link all code with the
10696 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10697 @option{-D_MT}; when linking, it links in a special thread helper library
10698 @option{-lmingwthrd} which cleans up per thread exception handling data.
10700 @item -mno-align-stringops
10701 @opindex mno-align-stringops
10702 Do not align destination of inlined string operations. This switch reduces
10703 code size and improves performance in case the destination is already aligned,
10704 but GCC doesn't know about it.
10706 @item -minline-all-stringops
10707 @opindex minline-all-stringops
10708 By default GCC inlines string operations only when destination is known to be
10709 aligned at least to 4 byte boundary. This enables more inlining, increase code
10710 size, but may improve performance of code that depends on fast memcpy, strlen
10711 and memset for short lengths.
10713 @item -minline-stringops-dynamically
10714 @opindex minline-stringops-dynamically
10715 For string operation of unknown size, inline runtime checks so for small
10716 blocks inline code is used, while for large blocks library call is used.
10718 @item -mstringop-strategy=@var{alg}
10719 @opindex mstringop-strategy=@var{alg}
10720 Overwrite internal decision heuristic about particular algorithm to inline
10721 string operation with. The allowed values are @code{rep_byte},
10722 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10723 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10724 expanding inline loop, @code{libcall} for always expanding library call.
10726 @item -momit-leaf-frame-pointer
10727 @opindex momit-leaf-frame-pointer
10728 Don't keep the frame pointer in a register for leaf functions. This
10729 avoids the instructions to save, set up and restore frame pointers and
10730 makes an extra register available in leaf functions. The option
10731 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10732 which might make debugging harder.
10734 @item -mtls-direct-seg-refs
10735 @itemx -mno-tls-direct-seg-refs
10736 @opindex mtls-direct-seg-refs
10737 Controls whether TLS variables may be accessed with offsets from the
10738 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10739 or whether the thread base pointer must be added. Whether or not this
10740 is legal depends on the operating system, and whether it maps the
10741 segment to cover the entire TLS area.
10743 For systems that use GNU libc, the default is on.
10746 @itemx -mno-fused-madd
10747 @opindex mfused-madd
10748 Enable automatic generation of fused floating point multiply-add instructions
10749 if the ISA supports such instructions. The -mfused-madd option is on by
10750 default. The fused multiply-add instructions have a different
10751 rounding behavior compared to executing a multiply followed by an add.
10754 These @samp{-m} switches are supported in addition to the above
10755 on AMD x86-64 processors in 64-bit environments.
10762 Generate code for a 32-bit or 64-bit environment.
10763 The 32-bit environment sets int, long and pointer to 32 bits and
10764 generates code that runs on any i386 system.
10765 The 64-bit environment sets int to 32 bits and long and pointer
10766 to 64 bits and generates code for AMD's x86-64 architecture. For
10767 darwin only the -m64 option turns off the @option{-fno-pic} and
10768 @option{-mdynamic-no-pic} options.
10770 @item -mno-red-zone
10771 @opindex no-red-zone
10772 Do not use a so called red zone for x86-64 code. The red zone is mandated
10773 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10774 stack pointer that will not be modified by signal or interrupt handlers
10775 and therefore can be used for temporary data without adjusting the stack
10776 pointer. The flag @option{-mno-red-zone} disables this red zone.
10778 @item -mcmodel=small
10779 @opindex mcmodel=small
10780 Generate code for the small code model: the program and its symbols must
10781 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10782 Programs can be statically or dynamically linked. This is the default
10785 @item -mcmodel=kernel
10786 @opindex mcmodel=kernel
10787 Generate code for the kernel code model. The kernel runs in the
10788 negative 2 GB of the address space.
10789 This model has to be used for Linux kernel code.
10791 @item -mcmodel=medium
10792 @opindex mcmodel=medium
10793 Generate code for the medium model: The program is linked in the lower 2
10794 GB of the address space but symbols can be located anywhere in the
10795 address space. Programs can be statically or dynamically linked, but
10796 building of shared libraries are not supported with the medium model.
10798 @item -mcmodel=large
10799 @opindex mcmodel=large
10800 Generate code for the large model: This model makes no assumptions
10801 about addresses and sizes of sections.
10804 @node IA-64 Options
10805 @subsection IA-64 Options
10806 @cindex IA-64 Options
10808 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10812 @opindex mbig-endian
10813 Generate code for a big endian target. This is the default for HP-UX@.
10815 @item -mlittle-endian
10816 @opindex mlittle-endian
10817 Generate code for a little endian target. This is the default for AIX5
10823 @opindex mno-gnu-as
10824 Generate (or don't) code for the GNU assembler. This is the default.
10825 @c Also, this is the default if the configure option @option{--with-gnu-as}
10831 @opindex mno-gnu-ld
10832 Generate (or don't) code for the GNU linker. This is the default.
10833 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10838 Generate code that does not use a global pointer register. The result
10839 is not position independent code, and violates the IA-64 ABI@.
10841 @item -mvolatile-asm-stop
10842 @itemx -mno-volatile-asm-stop
10843 @opindex mvolatile-asm-stop
10844 @opindex mno-volatile-asm-stop
10845 Generate (or don't) a stop bit immediately before and after volatile asm
10848 @item -mregister-names
10849 @itemx -mno-register-names
10850 @opindex mregister-names
10851 @opindex mno-register-names
10852 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10853 the stacked registers. This may make assembler output more readable.
10859 Disable (or enable) optimizations that use the small data section. This may
10860 be useful for working around optimizer bugs.
10862 @item -mconstant-gp
10863 @opindex mconstant-gp
10864 Generate code that uses a single constant global pointer value. This is
10865 useful when compiling kernel code.
10869 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10870 This is useful when compiling firmware code.
10872 @item -minline-float-divide-min-latency
10873 @opindex minline-float-divide-min-latency
10874 Generate code for inline divides of floating point values
10875 using the minimum latency algorithm.
10877 @item -minline-float-divide-max-throughput
10878 @opindex minline-float-divide-max-throughput
10879 Generate code for inline divides of floating point values
10880 using the maximum throughput algorithm.
10882 @item -minline-int-divide-min-latency
10883 @opindex minline-int-divide-min-latency
10884 Generate code for inline divides of integer values
10885 using the minimum latency algorithm.
10887 @item -minline-int-divide-max-throughput
10888 @opindex minline-int-divide-max-throughput
10889 Generate code for inline divides of integer values
10890 using the maximum throughput algorithm.
10892 @item -minline-sqrt-min-latency
10893 @opindex minline-sqrt-min-latency
10894 Generate code for inline square roots
10895 using the minimum latency algorithm.
10897 @item -minline-sqrt-max-throughput
10898 @opindex minline-sqrt-max-throughput
10899 Generate code for inline square roots
10900 using the maximum throughput algorithm.
10902 @item -mno-dwarf2-asm
10903 @itemx -mdwarf2-asm
10904 @opindex mno-dwarf2-asm
10905 @opindex mdwarf2-asm
10906 Don't (or do) generate assembler code for the DWARF2 line number debugging
10907 info. This may be useful when not using the GNU assembler.
10909 @item -mearly-stop-bits
10910 @itemx -mno-early-stop-bits
10911 @opindex mearly-stop-bits
10912 @opindex mno-early-stop-bits
10913 Allow stop bits to be placed earlier than immediately preceding the
10914 instruction that triggered the stop bit. This can improve instruction
10915 scheduling, but does not always do so.
10917 @item -mfixed-range=@var{register-range}
10918 @opindex mfixed-range
10919 Generate code treating the given register range as fixed registers.
10920 A fixed register is one that the register allocator can not use. This is
10921 useful when compiling kernel code. A register range is specified as
10922 two registers separated by a dash. Multiple register ranges can be
10923 specified separated by a comma.
10925 @item -mtls-size=@var{tls-size}
10927 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10930 @item -mtune=@var{cpu-type}
10932 Tune the instruction scheduling for a particular CPU, Valid values are
10933 itanium, itanium1, merced, itanium2, and mckinley.
10939 Add support for multithreading using the POSIX threads library. This
10940 option sets flags for both the preprocessor and linker. It does
10941 not affect the thread safety of object code produced by the compiler or
10942 that of libraries supplied with it. These are HP-UX specific flags.
10948 Generate code for a 32-bit or 64-bit environment.
10949 The 32-bit environment sets int, long and pointer to 32 bits.
10950 The 64-bit environment sets int to 32 bits and long and pointer
10951 to 64 bits. These are HP-UX specific flags.
10953 @item -mno-sched-br-data-spec
10954 @itemx -msched-br-data-spec
10955 @opindex mno-sched-br-data-spec
10956 @opindex msched-br-data-spec
10957 (Dis/En)able data speculative scheduling before reload.
10958 This will result in generation of the ld.a instructions and
10959 the corresponding check instructions (ld.c / chk.a).
10960 The default is 'disable'.
10962 @item -msched-ar-data-spec
10963 @itemx -mno-sched-ar-data-spec
10964 @opindex msched-ar-data-spec
10965 @opindex mno-sched-ar-data-spec
10966 (En/Dis)able data speculative scheduling after reload.
10967 This will result in generation of the ld.a instructions and
10968 the corresponding check instructions (ld.c / chk.a).
10969 The default is 'enable'.
10971 @item -mno-sched-control-spec
10972 @itemx -msched-control-spec
10973 @opindex mno-sched-control-spec
10974 @opindex msched-control-spec
10975 (Dis/En)able control speculative scheduling. This feature is
10976 available only during region scheduling (i.e. before reload).
10977 This will result in generation of the ld.s instructions and
10978 the corresponding check instructions chk.s .
10979 The default is 'disable'.
10981 @item -msched-br-in-data-spec
10982 @itemx -mno-sched-br-in-data-spec
10983 @opindex msched-br-in-data-spec
10984 @opindex mno-sched-br-in-data-spec
10985 (En/Dis)able speculative scheduling of the instructions that
10986 are dependent on the data speculative loads before reload.
10987 This is effective only with @option{-msched-br-data-spec} enabled.
10988 The default is 'enable'.
10990 @item -msched-ar-in-data-spec
10991 @itemx -mno-sched-ar-in-data-spec
10992 @opindex msched-ar-in-data-spec
10993 @opindex mno-sched-ar-in-data-spec
10994 (En/Dis)able speculative scheduling of the instructions that
10995 are dependent on the data speculative loads after reload.
10996 This is effective only with @option{-msched-ar-data-spec} enabled.
10997 The default is 'enable'.
10999 @item -msched-in-control-spec
11000 @itemx -mno-sched-in-control-spec
11001 @opindex msched-in-control-spec
11002 @opindex mno-sched-in-control-spec
11003 (En/Dis)able speculative scheduling of the instructions that
11004 are dependent on the control speculative loads.
11005 This is effective only with @option{-msched-control-spec} enabled.
11006 The default is 'enable'.
11009 @itemx -mno-sched-ldc
11010 @opindex msched-ldc
11011 @opindex mno-sched-ldc
11012 (En/Dis)able use of simple data speculation checks ld.c .
11013 If disabled, only chk.a instructions will be emitted to check
11014 data speculative loads.
11015 The default is 'enable'.
11017 @item -mno-sched-control-ldc
11018 @itemx -msched-control-ldc
11019 @opindex mno-sched-control-ldc
11020 @opindex msched-control-ldc
11021 (Dis/En)able use of ld.c instructions to check control speculative loads.
11022 If enabled, in case of control speculative load with no speculatively
11023 scheduled dependent instructions this load will be emitted as ld.sa and
11024 ld.c will be used to check it.
11025 The default is 'disable'.
11027 @item -mno-sched-spec-verbose
11028 @itemx -msched-spec-verbose
11029 @opindex mno-sched-spec-verbose
11030 @opindex msched-spec-verbose
11031 (Dis/En)able printing of the information about speculative motions.
11033 @item -mno-sched-prefer-non-data-spec-insns
11034 @itemx -msched-prefer-non-data-spec-insns
11035 @opindex mno-sched-prefer-non-data-spec-insns
11036 @opindex msched-prefer-non-data-spec-insns
11037 If enabled, data speculative instructions will be chosen for schedule
11038 only if there are no other choices at the moment. This will make
11039 the use of the data speculation much more conservative.
11040 The default is 'disable'.
11042 @item -mno-sched-prefer-non-control-spec-insns
11043 @itemx -msched-prefer-non-control-spec-insns
11044 @opindex mno-sched-prefer-non-control-spec-insns
11045 @opindex msched-prefer-non-control-spec-insns
11046 If enabled, control speculative instructions will be chosen for schedule
11047 only if there are no other choices at the moment. This will make
11048 the use of the control speculation much more conservative.
11049 The default is 'disable'.
11051 @item -mno-sched-count-spec-in-critical-path
11052 @itemx -msched-count-spec-in-critical-path
11053 @opindex mno-sched-count-spec-in-critical-path
11054 @opindex msched-count-spec-in-critical-path
11055 If enabled, speculative dependencies will be considered during
11056 computation of the instructions priorities. This will make the use of the
11057 speculation a bit more conservative.
11058 The default is 'disable'.
11063 @subsection M32C Options
11064 @cindex M32C options
11067 @item -mcpu=@var{name}
11069 Select the CPU for which code is generated. @var{name} may be one of
11070 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11071 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11072 the M32C/80 series.
11076 Specifies that the program will be run on the simulator. This causes
11077 an alternate runtime library to be linked in which supports, for
11078 example, file I/O. You must not use this option when generating
11079 programs that will run on real hardware; you must provide your own
11080 runtime library for whatever I/O functions are needed.
11082 @item -memregs=@var{number}
11084 Specifies the number of memory-based pseudo-registers GCC will use
11085 during code generation. These pseudo-registers will be used like real
11086 registers, so there is a tradeoff between GCC's ability to fit the
11087 code into available registers, and the performance penalty of using
11088 memory instead of registers. Note that all modules in a program must
11089 be compiled with the same value for this option. Because of that, you
11090 must not use this option with the default runtime libraries gcc
11095 @node M32R/D Options
11096 @subsection M32R/D Options
11097 @cindex M32R/D options
11099 These @option{-m} options are defined for Renesas M32R/D architectures:
11104 Generate code for the M32R/2@.
11108 Generate code for the M32R/X@.
11112 Generate code for the M32R@. This is the default.
11114 @item -mmodel=small
11115 @opindex mmodel=small
11116 Assume all objects live in the lower 16MB of memory (so that their addresses
11117 can be loaded with the @code{ld24} instruction), and assume all subroutines
11118 are reachable with the @code{bl} instruction.
11119 This is the default.
11121 The addressability of a particular object can be set with the
11122 @code{model} attribute.
11124 @item -mmodel=medium
11125 @opindex mmodel=medium
11126 Assume objects may be anywhere in the 32-bit address space (the compiler
11127 will generate @code{seth/add3} instructions to load their addresses), and
11128 assume all subroutines are reachable with the @code{bl} instruction.
11130 @item -mmodel=large
11131 @opindex mmodel=large
11132 Assume objects may be anywhere in the 32-bit address space (the compiler
11133 will generate @code{seth/add3} instructions to load their addresses), and
11134 assume subroutines may not be reachable with the @code{bl} instruction
11135 (the compiler will generate the much slower @code{seth/add3/jl}
11136 instruction sequence).
11139 @opindex msdata=none
11140 Disable use of the small data area. Variables will be put into
11141 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11142 @code{section} attribute has been specified).
11143 This is the default.
11145 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11146 Objects may be explicitly put in the small data area with the
11147 @code{section} attribute using one of these sections.
11149 @item -msdata=sdata
11150 @opindex msdata=sdata
11151 Put small global and static data in the small data area, but do not
11152 generate special code to reference them.
11155 @opindex msdata=use
11156 Put small global and static data in the small data area, and generate
11157 special instructions to reference them.
11161 @cindex smaller data references
11162 Put global and static objects less than or equal to @var{num} bytes
11163 into the small data or bss sections instead of the normal data or bss
11164 sections. The default value of @var{num} is 8.
11165 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11166 for this option to have any effect.
11168 All modules should be compiled with the same @option{-G @var{num}} value.
11169 Compiling with different values of @var{num} may or may not work; if it
11170 doesn't the linker will give an error message---incorrect code will not be
11175 Makes the M32R specific code in the compiler display some statistics
11176 that might help in debugging programs.
11178 @item -malign-loops
11179 @opindex malign-loops
11180 Align all loops to a 32-byte boundary.
11182 @item -mno-align-loops
11183 @opindex mno-align-loops
11184 Do not enforce a 32-byte alignment for loops. This is the default.
11186 @item -missue-rate=@var{number}
11187 @opindex missue-rate=@var{number}
11188 Issue @var{number} instructions per cycle. @var{number} can only be 1
11191 @item -mbranch-cost=@var{number}
11192 @opindex mbranch-cost=@var{number}
11193 @var{number} can only be 1 or 2. If it is 1 then branches will be
11194 preferred over conditional code, if it is 2, then the opposite will
11197 @item -mflush-trap=@var{number}
11198 @opindex mflush-trap=@var{number}
11199 Specifies the trap number to use to flush the cache. The default is
11200 12. Valid numbers are between 0 and 15 inclusive.
11202 @item -mno-flush-trap
11203 @opindex mno-flush-trap
11204 Specifies that the cache cannot be flushed by using a trap.
11206 @item -mflush-func=@var{name}
11207 @opindex mflush-func=@var{name}
11208 Specifies the name of the operating system function to call to flush
11209 the cache. The default is @emph{_flush_cache}, but a function call
11210 will only be used if a trap is not available.
11212 @item -mno-flush-func
11213 @opindex mno-flush-func
11214 Indicates that there is no OS function for flushing the cache.
11218 @node M680x0 Options
11219 @subsection M680x0 Options
11220 @cindex M680x0 options
11222 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11223 The default settings depend on which architecture was selected when
11224 the compiler was configured; the defaults for the most common choices
11228 @item -march=@var{arch}
11230 Generate code for a specific M680x0 or ColdFire instruction set
11231 architecture. Permissible values of @var{arch} for M680x0
11232 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11233 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11234 architectures are selected according to Freescale's ISA classification
11235 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11236 @samp{isab} and @samp{isac}.
11238 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11239 code for a ColdFire target. The @var{arch} in this macro is one of the
11240 @option{-march} arguments given above.
11242 When used together, @option{-march} and @option{-mtune} select code
11243 that runs on a family of similar processors but that is optimized
11244 for a particular microarchitecture.
11246 @item -mcpu=@var{cpu}
11248 Generate code for a specific M680x0 or ColdFire processor.
11249 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11250 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11251 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11252 below, which also classifies the CPUs into families:
11254 @multitable @columnfractions 0.20 0.80
11255 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11256 @item @samp{51qe} @tab @samp{51qe}
11257 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11258 @item @samp{5206e} @tab @samp{5206e}
11259 @item @samp{5208} @tab @samp{5207} @samp{5208}
11260 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11261 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11262 @item @samp{5216} @tab @samp{5214} @samp{5216}
11263 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11264 @item @samp{5225} @tab @samp{5224} @samp{5225}
11265 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11266 @item @samp{5249} @tab @samp{5249}
11267 @item @samp{5250} @tab @samp{5250}
11268 @item @samp{5271} @tab @samp{5270} @samp{5271}
11269 @item @samp{5272} @tab @samp{5272}
11270 @item @samp{5275} @tab @samp{5274} @samp{5275}
11271 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11272 @item @samp{5307} @tab @samp{5307}
11273 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11274 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11275 @item @samp{5407} @tab @samp{5407}
11276 @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}
11279 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11280 @var{arch} is compatible with @var{cpu}. Other combinations of
11281 @option{-mcpu} and @option{-march} are rejected.
11283 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11284 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11285 where the value of @var{family} is given by the table above.
11287 @item -mtune=@var{tune}
11289 Tune the code for a particular microarchitecture, within the
11290 constraints set by @option{-march} and @option{-mcpu}.
11291 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11292 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11293 and @samp{cpu32}. The ColdFire microarchitectures
11294 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11296 You can also use @option{-mtune=68020-40} for code that needs
11297 to run relatively well on 68020, 68030 and 68040 targets.
11298 @option{-mtune=68020-60} is similar but includes 68060 targets
11299 as well. These two options select the same tuning decisions as
11300 @option{-m68020-40} and @option{-m68020-60} respectively.
11302 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11303 when tuning for 680x0 architecture @var{arch}. It also defines
11304 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11305 option is used. If gcc is tuning for a range of architectures,
11306 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11307 it defines the macros for every architecture in the range.
11309 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11310 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11311 of the arguments given above.
11317 Generate output for a 68000. This is the default
11318 when the compiler is configured for 68000-based systems.
11319 It is equivalent to @option{-march=68000}.
11321 Use this option for microcontrollers with a 68000 or EC000 core,
11322 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11326 Generate output for a 68010. This is the default
11327 when the compiler is configured for 68010-based systems.
11328 It is equivalent to @option{-march=68010}.
11334 Generate output for a 68020. This is the default
11335 when the compiler is configured for 68020-based systems.
11336 It is equivalent to @option{-march=68020}.
11340 Generate output for a 68030. This is the default when the compiler is
11341 configured for 68030-based systems. It is equivalent to
11342 @option{-march=68030}.
11346 Generate output for a 68040. This is the default when the compiler is
11347 configured for 68040-based systems. It is equivalent to
11348 @option{-march=68040}.
11350 This option inhibits the use of 68881/68882 instructions that have to be
11351 emulated by software on the 68040. Use this option if your 68040 does not
11352 have code to emulate those instructions.
11356 Generate output for a 68060. This is the default when the compiler is
11357 configured for 68060-based systems. It is equivalent to
11358 @option{-march=68060}.
11360 This option inhibits the use of 68020 and 68881/68882 instructions that
11361 have to be emulated by software on the 68060. Use this option if your 68060
11362 does not have code to emulate those instructions.
11366 Generate output for a CPU32. This is the default
11367 when the compiler is configured for CPU32-based systems.
11368 It is equivalent to @option{-march=cpu32}.
11370 Use this option for microcontrollers with a
11371 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11372 68336, 68340, 68341, 68349 and 68360.
11376 Generate output for a 520X ColdFire CPU. This is the default
11377 when the compiler is configured for 520X-based systems.
11378 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11379 in favor of that option.
11381 Use this option for microcontroller with a 5200 core, including
11382 the MCF5202, MCF5203, MCF5204 and MCF5206.
11386 Generate output for a 5206e ColdFire CPU. The option is now
11387 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11391 Generate output for a member of the ColdFire 528X family.
11392 The option is now deprecated in favor of the equivalent
11393 @option{-mcpu=528x}.
11397 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11398 in favor of the equivalent @option{-mcpu=5307}.
11402 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11403 in favor of the equivalent @option{-mcpu=5407}.
11407 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11408 This includes use of hardware floating point instructions.
11409 The option is equivalent to @option{-mcpu=547x}, and is now
11410 deprecated in favor of that option.
11414 Generate output for a 68040, without using any of the new instructions.
11415 This results in code which can run relatively efficiently on either a
11416 68020/68881 or a 68030 or a 68040. The generated code does use the
11417 68881 instructions that are emulated on the 68040.
11419 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11423 Generate output for a 68060, without using any of the new instructions.
11424 This results in code which can run relatively efficiently on either a
11425 68020/68881 or a 68030 or a 68040. The generated code does use the
11426 68881 instructions that are emulated on the 68060.
11428 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11432 @opindex mhard-float
11434 Generate floating-point instructions. This is the default for 68020
11435 and above, and for ColdFire devices that have an FPU. It defines the
11436 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11437 on ColdFire targets.
11440 @opindex msoft-float
11441 Do not generate floating-point instructions; use library calls instead.
11442 This is the default for 68000, 68010, and 68832 targets. It is also
11443 the default for ColdFire devices that have no FPU.
11449 Generate (do not generate) ColdFire hardware divide and remainder
11450 instructions. If @option{-march} is used without @option{-mcpu},
11451 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11452 architectures. Otherwise, the default is taken from the target CPU
11453 (either the default CPU, or the one specified by @option{-mcpu}). For
11454 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11455 @option{-mcpu=5206e}.
11457 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11461 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11462 Additionally, parameters passed on the stack are also aligned to a
11463 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11467 Do not consider type @code{int} to be 16 bits wide. This is the default.
11470 @itemx -mno-bitfield
11471 @opindex mnobitfield
11472 @opindex mno-bitfield
11473 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11474 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11478 Do use the bit-field instructions. The @option{-m68020} option implies
11479 @option{-mbitfield}. This is the default if you use a configuration
11480 designed for a 68020.
11484 Use a different function-calling convention, in which functions
11485 that take a fixed number of arguments return with the @code{rtd}
11486 instruction, which pops their arguments while returning. This
11487 saves one instruction in the caller since there is no need to pop
11488 the arguments there.
11490 This calling convention is incompatible with the one normally
11491 used on Unix, so you cannot use it if you need to call libraries
11492 compiled with the Unix compiler.
11494 Also, you must provide function prototypes for all functions that
11495 take variable numbers of arguments (including @code{printf});
11496 otherwise incorrect code will be generated for calls to those
11499 In addition, seriously incorrect code will result if you call a
11500 function with too many arguments. (Normally, extra arguments are
11501 harmlessly ignored.)
11503 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11504 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11508 Do not use the calling conventions selected by @option{-mrtd}.
11509 This is the default.
11512 @itemx -mno-align-int
11513 @opindex malign-int
11514 @opindex mno-align-int
11515 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11516 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11517 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11518 Aligning variables on 32-bit boundaries produces code that runs somewhat
11519 faster on processors with 32-bit busses at the expense of more memory.
11521 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11522 align structures containing the above types differently than
11523 most published application binary interface specifications for the m68k.
11527 Use the pc-relative addressing mode of the 68000 directly, instead of
11528 using a global offset table. At present, this option implies @option{-fpic},
11529 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11530 not presently supported with @option{-mpcrel}, though this could be supported for
11531 68020 and higher processors.
11533 @item -mno-strict-align
11534 @itemx -mstrict-align
11535 @opindex mno-strict-align
11536 @opindex mstrict-align
11537 Do not (do) assume that unaligned memory references will be handled by
11541 Generate code that allows the data segment to be located in a different
11542 area of memory from the text segment. This allows for execute in place in
11543 an environment without virtual memory management. This option implies
11546 @item -mno-sep-data
11547 Generate code that assumes that the data segment follows the text segment.
11548 This is the default.
11550 @item -mid-shared-library
11551 Generate code that supports shared libraries via the library ID method.
11552 This allows for execute in place and shared libraries in an environment
11553 without virtual memory management. This option implies @option{-fPIC}.
11555 @item -mno-id-shared-library
11556 Generate code that doesn't assume ID based shared libraries are being used.
11557 This is the default.
11559 @item -mshared-library-id=n
11560 Specified the identification number of the ID based shared library being
11561 compiled. Specifying a value of 0 will generate more compact code, specifying
11562 other values will force the allocation of that number to the current
11563 library but is no more space or time efficient than omitting this option.
11567 @node M68hc1x Options
11568 @subsection M68hc1x Options
11569 @cindex M68hc1x options
11571 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11572 microcontrollers. The default values for these options depends on
11573 which style of microcontroller was selected when the compiler was configured;
11574 the defaults for the most common choices are given below.
11581 Generate output for a 68HC11. This is the default
11582 when the compiler is configured for 68HC11-based systems.
11588 Generate output for a 68HC12. This is the default
11589 when the compiler is configured for 68HC12-based systems.
11595 Generate output for a 68HCS12.
11597 @item -mauto-incdec
11598 @opindex mauto-incdec
11599 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11606 Enable the use of 68HC12 min and max instructions.
11609 @itemx -mno-long-calls
11610 @opindex mlong-calls
11611 @opindex mno-long-calls
11612 Treat all calls as being far away (near). If calls are assumed to be
11613 far away, the compiler will use the @code{call} instruction to
11614 call a function and the @code{rtc} instruction for returning.
11618 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11620 @item -msoft-reg-count=@var{count}
11621 @opindex msoft-reg-count
11622 Specify the number of pseudo-soft registers which are used for the
11623 code generation. The maximum number is 32. Using more pseudo-soft
11624 register may or may not result in better code depending on the program.
11625 The default is 4 for 68HC11 and 2 for 68HC12.
11629 @node MCore Options
11630 @subsection MCore Options
11631 @cindex MCore options
11633 These are the @samp{-m} options defined for the Motorola M*Core
11639 @itemx -mno-hardlit
11641 @opindex mno-hardlit
11642 Inline constants into the code stream if it can be done in two
11643 instructions or less.
11649 Use the divide instruction. (Enabled by default).
11651 @item -mrelax-immediate
11652 @itemx -mno-relax-immediate
11653 @opindex mrelax-immediate
11654 @opindex mno-relax-immediate
11655 Allow arbitrary sized immediates in bit operations.
11657 @item -mwide-bitfields
11658 @itemx -mno-wide-bitfields
11659 @opindex mwide-bitfields
11660 @opindex mno-wide-bitfields
11661 Always treat bit-fields as int-sized.
11663 @item -m4byte-functions
11664 @itemx -mno-4byte-functions
11665 @opindex m4byte-functions
11666 @opindex mno-4byte-functions
11667 Force all functions to be aligned to a four byte boundary.
11669 @item -mcallgraph-data
11670 @itemx -mno-callgraph-data
11671 @opindex mcallgraph-data
11672 @opindex mno-callgraph-data
11673 Emit callgraph information.
11676 @itemx -mno-slow-bytes
11677 @opindex mslow-bytes
11678 @opindex mno-slow-bytes
11679 Prefer word access when reading byte quantities.
11681 @item -mlittle-endian
11682 @itemx -mbig-endian
11683 @opindex mlittle-endian
11684 @opindex mbig-endian
11685 Generate code for a little endian target.
11691 Generate code for the 210 processor.
11695 @subsection MIPS Options
11696 @cindex MIPS options
11702 Generate big-endian code.
11706 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11709 @item -march=@var{arch}
11711 Generate code that will run on @var{arch}, which can be the name of a
11712 generic MIPS ISA, or the name of a particular processor.
11714 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11715 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11716 The processor names are:
11717 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11718 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11719 @samp{5kc}, @samp{5kf},
11721 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11722 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11723 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11724 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11727 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11728 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11729 @samp{rm7000}, @samp{rm9000},
11732 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11733 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11734 The special value @samp{from-abi} selects the
11735 most compatible architecture for the selected ABI (that is,
11736 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11738 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11739 (for example, @samp{-march=r2k}). Prefixes are optional, and
11740 @samp{vr} may be written @samp{r}.
11742 Names of the form @samp{@var{n}f2_1} refer to processors with
11743 FPUs clocked at half the rate of the core, names of the form
11744 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11745 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11746 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11747 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11748 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11749 accepted as synonyms for @samp{@var{n}f1_1}.
11751 GCC defines two macros based on the value of this option. The first
11752 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11753 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11754 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11755 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11756 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11758 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11759 above. In other words, it will have the full prefix and will not
11760 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11761 the macro names the resolved architecture (either @samp{"mips1"} or
11762 @samp{"mips3"}). It names the default architecture when no
11763 @option{-march} option is given.
11765 @item -mtune=@var{arch}
11767 Optimize for @var{arch}. Among other things, this option controls
11768 the way instructions are scheduled, and the perceived cost of arithmetic
11769 operations. The list of @var{arch} values is the same as for
11772 When this option is not used, GCC will optimize for the processor
11773 specified by @option{-march}. By using @option{-march} and
11774 @option{-mtune} together, it is possible to generate code that will
11775 run on a family of processors, but optimize the code for one
11776 particular member of that family.
11778 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11779 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11780 @samp{-march} ones described above.
11784 Equivalent to @samp{-march=mips1}.
11788 Equivalent to @samp{-march=mips2}.
11792 Equivalent to @samp{-march=mips3}.
11796 Equivalent to @samp{-march=mips4}.
11800 Equivalent to @samp{-march=mips32}.
11804 Equivalent to @samp{-march=mips32r2}.
11808 Equivalent to @samp{-march=mips64}.
11813 @opindex mno-mips16
11814 Generate (do not generate) MIPS16 code. If GCC is targetting a
11815 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11817 MIPS16 code generation can also be controlled on a per-function basis
11818 by means of @code{mips16} and @code{nomips16} attributes.
11819 @xref{Function Attributes}, for more information.
11821 @item -mflip-mips16
11822 @opindex mflip-mips16
11823 Generate MIPS16 code on alternating functions. This option is provided
11824 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11825 not intended for ordinary use in compiling user code.
11827 @item -minterlink-mips16
11828 @itemx -mno-interlink-mips16
11829 @opindex minterlink-mips16
11830 @opindex mno-interlink-mips16
11831 Require (do not require) that non-MIPS16 code be link-compatible with
11834 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
11835 it must either use a call or an indirect jump. @option{-minterlink-mips16}
11836 therefore disables direct jumps unless GCC knows that the target of the
11837 jump is not MIPS16.
11849 Generate code for the given ABI@.
11851 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11852 generates 64-bit code when you select a 64-bit architecture, but you
11853 can use @option{-mgp32} to get 32-bit code instead.
11855 For information about the O64 ABI, see
11856 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11858 GCC supports a variant of the o32 ABI in which floating-point registers
11859 are 64 rather than 32 bits wide. You can select this combination with
11860 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11861 and @samp{mfhc1} instructions and is therefore only supported for
11862 MIPS32R2 processors.
11864 The register assignments for arguments and return values remain the
11865 same, but each scalar value is passed in a single 64-bit register
11866 rather than a pair of 32-bit registers. For example, scalar
11867 floating-point values are returned in @samp{$f0} only, not a
11868 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11869 remains the same, but all 64 bits are saved.
11872 @itemx -mno-abicalls
11874 @opindex mno-abicalls
11875 Generate (do not generate) code that is suitable for SVR4-style
11876 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11881 Generate (do not generate) code that is fully position-independent,
11882 and that can therefore be linked into shared libraries. This option
11883 only affects @option{-mabicalls}.
11885 All @option{-mabicalls} code has traditionally been position-independent,
11886 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11887 as an extension, the GNU toolchain allows executables to use absolute
11888 accesses for locally-binding symbols. It can also use shorter GP
11889 initialization sequences and generate direct calls to locally-defined
11890 functions. This mode is selected by @option{-mno-shared}.
11892 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11893 objects that can only be linked by the GNU linker. However, the option
11894 does not affect the ABI of the final executable; it only affects the ABI
11895 of relocatable objects. Using @option{-mno-shared} will generally make
11896 executables both smaller and quicker.
11898 @option{-mshared} is the default.
11904 Lift (do not lift) the usual restrictions on the size of the global
11907 GCC normally uses a single instruction to load values from the GOT@.
11908 While this is relatively efficient, it will only work if the GOT
11909 is smaller than about 64k. Anything larger will cause the linker
11910 to report an error such as:
11912 @cindex relocation truncated to fit (MIPS)
11914 relocation truncated to fit: R_MIPS_GOT16 foobar
11917 If this happens, you should recompile your code with @option{-mxgot}.
11918 It should then work with very large GOTs, although it will also be
11919 less efficient, since it will take three instructions to fetch the
11920 value of a global symbol.
11922 Note that some linkers can create multiple GOTs. If you have such a
11923 linker, you should only need to use @option{-mxgot} when a single object
11924 file accesses more than 64k's worth of GOT entries. Very few do.
11926 These options have no effect unless GCC is generating position
11931 Assume that general-purpose registers are 32 bits wide.
11935 Assume that general-purpose registers are 64 bits wide.
11939 Assume that floating-point registers are 32 bits wide.
11943 Assume that floating-point registers are 64 bits wide.
11946 @opindex mhard-float
11947 Use floating-point coprocessor instructions.
11950 @opindex msoft-float
11951 Do not use floating-point coprocessor instructions. Implement
11952 floating-point calculations using library calls instead.
11954 @item -msingle-float
11955 @opindex msingle-float
11956 Assume that the floating-point coprocessor only supports single-precision
11959 @item -mdouble-float
11960 @opindex mdouble-float
11961 Assume that the floating-point coprocessor supports double-precision
11962 operations. This is the default.
11968 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
11969 implement atomic memory built-in functions. When neither option is
11970 specified, GCC will use the instructions if the target architecture
11973 @option{-mllsc} is useful if the runtime environment can emulate the
11974 instructions and @option{-mno-llsc} can be useful when compiling for
11975 nonstandard ISAs. You can make either option the default by
11976 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
11977 respectively. @option{--with-llsc} is the default for some
11978 configurations; see the installation documentation for details.
11984 Use (do not use) revision 1 of the MIPS DSP ASE.
11985 @xref{MIPS DSP Built-in Functions}. This option defines the
11986 preprocessor macro @samp{__mips_dsp}. It also defines
11987 @samp{__mips_dsp_rev} to 1.
11993 Use (do not use) revision 2 of the MIPS DSP ASE.
11994 @xref{MIPS DSP Built-in Functions}. This option defines the
11995 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
11996 It also defines @samp{__mips_dsp_rev} to 2.
11999 @itemx -mno-smartmips
12000 @opindex msmartmips
12001 @opindex mno-smartmips
12002 Use (do not use) the MIPS SmartMIPS ASE.
12004 @item -mpaired-single
12005 @itemx -mno-paired-single
12006 @opindex mpaired-single
12007 @opindex mno-paired-single
12008 Use (do not use) paired-single floating-point instructions.
12009 @xref{MIPS Paired-Single Support}. This option requires
12010 hardware floating-point support to be enabled.
12016 Use (do not use) MIPS Digital Media Extension instructions.
12017 This option can only be used when generating 64-bit code and requires
12018 hardware floating-point support to be enabled.
12023 @opindex mno-mips3d
12024 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12025 The option @option{-mips3d} implies @option{-mpaired-single}.
12031 Use (do not use) MT Multithreading instructions.
12035 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12036 an explanation of the default and the way that the pointer size is
12041 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12043 The default size of @code{int}s, @code{long}s and pointers depends on
12044 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12045 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12046 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12047 or the same size as integer registers, whichever is smaller.
12053 Assume (do not assume) that all symbols have 32-bit values, regardless
12054 of the selected ABI@. This option is useful in combination with
12055 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12056 to generate shorter and faster references to symbolic addresses.
12060 Put definitions of externally-visible data in a small data section
12061 if that data is no bigger than @var{num} bytes. GCC can then access
12062 the data more efficiently; see @option{-mgpopt} for details.
12064 The default @option{-G} option depends on the configuration.
12066 @item -mlocal-sdata
12067 @itemx -mno-local-sdata
12068 @opindex mlocal-sdata
12069 @opindex mno-local-sdata
12070 Extend (do not extend) the @option{-G} behavior to local data too,
12071 such as to static variables in C. @option{-mlocal-sdata} is the
12072 default for all configurations.
12074 If the linker complains that an application is using too much small data,
12075 you might want to try rebuilding the less performance-critical parts with
12076 @option{-mno-local-sdata}. You might also want to build large
12077 libraries with @option{-mno-local-sdata}, so that the libraries leave
12078 more room for the main program.
12080 @item -mextern-sdata
12081 @itemx -mno-extern-sdata
12082 @opindex mextern-sdata
12083 @opindex mno-extern-sdata
12084 Assume (do not assume) that externally-defined data will be in
12085 a small data section if that data is within the @option{-G} limit.
12086 @option{-mextern-sdata} is the default for all configurations.
12088 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12089 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12090 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12091 is placed in a small data section. If @var{Var} is defined by another
12092 module, you must either compile that module with a high-enough
12093 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12094 definition. If @var{Var} is common, you must link the application
12095 with a high-enough @option{-G} setting.
12097 The easiest way of satisfying these restrictions is to compile
12098 and link every module with the same @option{-G} option. However,
12099 you may wish to build a library that supports several different
12100 small data limits. You can do this by compiling the library with
12101 the highest supported @option{-G} setting and additionally using
12102 @option{-mno-extern-sdata} to stop the library from making assumptions
12103 about externally-defined data.
12109 Use (do not use) GP-relative accesses for symbols that are known to be
12110 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12111 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12114 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12115 might not hold the value of @code{_gp}. For example, if the code is
12116 part of a library that might be used in a boot monitor, programs that
12117 call boot monitor routines will pass an unknown value in @code{$gp}.
12118 (In such situations, the boot monitor itself would usually be compiled
12119 with @option{-G0}.)
12121 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12122 @option{-mno-extern-sdata}.
12124 @item -membedded-data
12125 @itemx -mno-embedded-data
12126 @opindex membedded-data
12127 @opindex mno-embedded-data
12128 Allocate variables to the read-only data section first if possible, then
12129 next in the small data section if possible, otherwise in data. This gives
12130 slightly slower code than the default, but reduces the amount of RAM required
12131 when executing, and thus may be preferred for some embedded systems.
12133 @item -muninit-const-in-rodata
12134 @itemx -mno-uninit-const-in-rodata
12135 @opindex muninit-const-in-rodata
12136 @opindex mno-uninit-const-in-rodata
12137 Put uninitialized @code{const} variables in the read-only data section.
12138 This option is only meaningful in conjunction with @option{-membedded-data}.
12140 @item -mcode-readable=@var{setting}
12141 @opindex mcode-readable
12142 Specify whether GCC may generate code that reads from executable sections.
12143 There are three possible settings:
12146 @item -mcode-readable=yes
12147 Instructions may freely access executable sections. This is the
12150 @item -mcode-readable=pcrel
12151 MIPS16 PC-relative load instructions can access executable sections,
12152 but other instructions must not do so. This option is useful on 4KSc
12153 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12154 It is also useful on processors that can be configured to have a dual
12155 instruction/data SRAM interface and that, like the M4K, automatically
12156 redirect PC-relative loads to the instruction RAM.
12158 @item -mcode-readable=no
12159 Instructions must not access executable sections. This option can be
12160 useful on targets that are configured to have a dual instruction/data
12161 SRAM interface but that (unlike the M4K) do not automatically redirect
12162 PC-relative loads to the instruction RAM.
12165 @item -msplit-addresses
12166 @itemx -mno-split-addresses
12167 @opindex msplit-addresses
12168 @opindex mno-split-addresses
12169 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12170 relocation operators. This option has been superseded by
12171 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12173 @item -mexplicit-relocs
12174 @itemx -mno-explicit-relocs
12175 @opindex mexplicit-relocs
12176 @opindex mno-explicit-relocs
12177 Use (do not use) assembler relocation operators when dealing with symbolic
12178 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12179 is to use assembler macros instead.
12181 @option{-mexplicit-relocs} is the default if GCC was configured
12182 to use an assembler that supports relocation operators.
12184 @item -mcheck-zero-division
12185 @itemx -mno-check-zero-division
12186 @opindex mcheck-zero-division
12187 @opindex mno-check-zero-division
12188 Trap (do not trap) on integer division by zero.
12190 The default is @option{-mcheck-zero-division}.
12192 @item -mdivide-traps
12193 @itemx -mdivide-breaks
12194 @opindex mdivide-traps
12195 @opindex mdivide-breaks
12196 MIPS systems check for division by zero by generating either a
12197 conditional trap or a break instruction. Using traps results in
12198 smaller code, but is only supported on MIPS II and later. Also, some
12199 versions of the Linux kernel have a bug that prevents trap from
12200 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12201 allow conditional traps on architectures that support them and
12202 @option{-mdivide-breaks} to force the use of breaks.
12204 The default is usually @option{-mdivide-traps}, but this can be
12205 overridden at configure time using @option{--with-divide=breaks}.
12206 Divide-by-zero checks can be completely disabled using
12207 @option{-mno-check-zero-division}.
12212 @opindex mno-memcpy
12213 Force (do not force) the use of @code{memcpy()} for non-trivial block
12214 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12215 most constant-sized copies.
12218 @itemx -mno-long-calls
12219 @opindex mlong-calls
12220 @opindex mno-long-calls
12221 Disable (do not disable) use of the @code{jal} instruction. Calling
12222 functions using @code{jal} is more efficient but requires the caller
12223 and callee to be in the same 256 megabyte segment.
12225 This option has no effect on abicalls code. The default is
12226 @option{-mno-long-calls}.
12232 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12233 instructions, as provided by the R4650 ISA@.
12236 @itemx -mno-fused-madd
12237 @opindex mfused-madd
12238 @opindex mno-fused-madd
12239 Enable (disable) use of the floating point multiply-accumulate
12240 instructions, when they are available. The default is
12241 @option{-mfused-madd}.
12243 When multiply-accumulate instructions are used, the intermediate
12244 product is calculated to infinite precision and is not subject to
12245 the FCSR Flush to Zero bit. This may be undesirable in some
12250 Tell the MIPS assembler to not run its preprocessor over user
12251 assembler files (with a @samp{.s} suffix) when assembling them.
12254 @itemx -mno-fix-r4000
12255 @opindex mfix-r4000
12256 @opindex mno-fix-r4000
12257 Work around certain R4000 CPU errata:
12260 A double-word or a variable shift may give an incorrect result if executed
12261 immediately after starting an integer division.
12263 A double-word or a variable shift may give an incorrect result if executed
12264 while an integer multiplication is in progress.
12266 An integer division may give an incorrect result if started in a delay slot
12267 of a taken branch or a jump.
12271 @itemx -mno-fix-r4400
12272 @opindex mfix-r4400
12273 @opindex mno-fix-r4400
12274 Work around certain R4400 CPU errata:
12277 A double-word or a variable shift may give an incorrect result if executed
12278 immediately after starting an integer division.
12282 @itemx -mno-fix-vr4120
12283 @opindex mfix-vr4120
12284 Work around certain VR4120 errata:
12287 @code{dmultu} does not always produce the correct result.
12289 @code{div} and @code{ddiv} do not always produce the correct result if one
12290 of the operands is negative.
12292 The workarounds for the division errata rely on special functions in
12293 @file{libgcc.a}. At present, these functions are only provided by
12294 the @code{mips64vr*-elf} configurations.
12296 Other VR4120 errata require a nop to be inserted between certain pairs of
12297 instructions. These errata are handled by the assembler, not by GCC itself.
12300 @opindex mfix-vr4130
12301 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12302 workarounds are implemented by the assembler rather than by GCC,
12303 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12304 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12305 instructions are available instead.
12308 @itemx -mno-fix-sb1
12310 Work around certain SB-1 CPU core errata.
12311 (This flag currently works around the SB-1 revision 2
12312 ``F1'' and ``F2'' floating point errata.)
12314 @item -mflush-func=@var{func}
12315 @itemx -mno-flush-func
12316 @opindex mflush-func
12317 Specifies the function to call to flush the I and D caches, or to not
12318 call any such function. If called, the function must take the same
12319 arguments as the common @code{_flush_func()}, that is, the address of the
12320 memory range for which the cache is being flushed, the size of the
12321 memory range, and the number 3 (to flush both caches). The default
12322 depends on the target GCC was configured for, but commonly is either
12323 @samp{_flush_func} or @samp{__cpu_flush}.
12325 @item mbranch-cost=@var{num}
12326 @opindex mbranch-cost
12327 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12328 This cost is only a heuristic and is not guaranteed to produce
12329 consistent results across releases. A zero cost redundantly selects
12330 the default, which is based on the @option{-mtune} setting.
12332 @item -mbranch-likely
12333 @itemx -mno-branch-likely
12334 @opindex mbranch-likely
12335 @opindex mno-branch-likely
12336 Enable or disable use of Branch Likely instructions, regardless of the
12337 default for the selected architecture. By default, Branch Likely
12338 instructions may be generated if they are supported by the selected
12339 architecture. An exception is for the MIPS32 and MIPS64 architectures
12340 and processors which implement those architectures; for those, Branch
12341 Likely instructions will not be generated by default because the MIPS32
12342 and MIPS64 architectures specifically deprecate their use.
12344 @item -mfp-exceptions
12345 @itemx -mno-fp-exceptions
12346 @opindex mfp-exceptions
12347 Specifies whether FP exceptions are enabled. This affects how we schedule
12348 FP instructions for some processors. The default is that FP exceptions are
12351 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12352 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12355 @item -mvr4130-align
12356 @itemx -mno-vr4130-align
12357 @opindex mvr4130-align
12358 The VR4130 pipeline is two-way superscalar, but can only issue two
12359 instructions together if the first one is 8-byte aligned. When this
12360 option is enabled, GCC will align pairs of instructions that it
12361 thinks should execute in parallel.
12363 This option only has an effect when optimizing for the VR4130.
12364 It normally makes code faster, but at the expense of making it bigger.
12365 It is enabled by default at optimization level @option{-O3}.
12369 @subsection MMIX Options
12370 @cindex MMIX Options
12372 These options are defined for the MMIX:
12376 @itemx -mno-libfuncs
12378 @opindex mno-libfuncs
12379 Specify that intrinsic library functions are being compiled, passing all
12380 values in registers, no matter the size.
12383 @itemx -mno-epsilon
12385 @opindex mno-epsilon
12386 Generate floating-point comparison instructions that compare with respect
12387 to the @code{rE} epsilon register.
12389 @item -mabi=mmixware
12391 @opindex mabi-mmixware
12393 Generate code that passes function parameters and return values that (in
12394 the called function) are seen as registers @code{$0} and up, as opposed to
12395 the GNU ABI which uses global registers @code{$231} and up.
12397 @item -mzero-extend
12398 @itemx -mno-zero-extend
12399 @opindex mzero-extend
12400 @opindex mno-zero-extend
12401 When reading data from memory in sizes shorter than 64 bits, use (do not
12402 use) zero-extending load instructions by default, rather than
12403 sign-extending ones.
12406 @itemx -mno-knuthdiv
12408 @opindex mno-knuthdiv
12409 Make the result of a division yielding a remainder have the same sign as
12410 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12411 remainder follows the sign of the dividend. Both methods are
12412 arithmetically valid, the latter being almost exclusively used.
12414 @item -mtoplevel-symbols
12415 @itemx -mno-toplevel-symbols
12416 @opindex mtoplevel-symbols
12417 @opindex mno-toplevel-symbols
12418 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12419 code can be used with the @code{PREFIX} assembly directive.
12423 Generate an executable in the ELF format, rather than the default
12424 @samp{mmo} format used by the @command{mmix} simulator.
12426 @item -mbranch-predict
12427 @itemx -mno-branch-predict
12428 @opindex mbranch-predict
12429 @opindex mno-branch-predict
12430 Use (do not use) the probable-branch instructions, when static branch
12431 prediction indicates a probable branch.
12433 @item -mbase-addresses
12434 @itemx -mno-base-addresses
12435 @opindex mbase-addresses
12436 @opindex mno-base-addresses
12437 Generate (do not generate) code that uses @emph{base addresses}. Using a
12438 base address automatically generates a request (handled by the assembler
12439 and the linker) for a constant to be set up in a global register. The
12440 register is used for one or more base address requests within the range 0
12441 to 255 from the value held in the register. The generally leads to short
12442 and fast code, but the number of different data items that can be
12443 addressed is limited. This means that a program that uses lots of static
12444 data may require @option{-mno-base-addresses}.
12446 @item -msingle-exit
12447 @itemx -mno-single-exit
12448 @opindex msingle-exit
12449 @opindex mno-single-exit
12450 Force (do not force) generated code to have a single exit point in each
12454 @node MN10300 Options
12455 @subsection MN10300 Options
12456 @cindex MN10300 options
12458 These @option{-m} options are defined for Matsushita MN10300 architectures:
12463 Generate code to avoid bugs in the multiply instructions for the MN10300
12464 processors. This is the default.
12466 @item -mno-mult-bug
12467 @opindex mno-mult-bug
12468 Do not generate code to avoid bugs in the multiply instructions for the
12469 MN10300 processors.
12473 Generate code which uses features specific to the AM33 processor.
12477 Do not generate code which uses features specific to the AM33 processor. This
12480 @item -mreturn-pointer-on-d0
12481 @opindex mreturn-pointer-on-d0
12482 When generating a function which returns a pointer, return the pointer
12483 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12484 only in a0, and attempts to call such functions without a prototype
12485 would result in errors. Note that this option is on by default; use
12486 @option{-mno-return-pointer-on-d0} to disable it.
12490 Do not link in the C run-time initialization object file.
12494 Indicate to the linker that it should perform a relaxation optimization pass
12495 to shorten branches, calls and absolute memory addresses. This option only
12496 has an effect when used on the command line for the final link step.
12498 This option makes symbolic debugging impossible.
12502 @subsection MT Options
12505 These @option{-m} options are defined for Morpho MT architectures:
12509 @item -march=@var{cpu-type}
12511 Generate code that will run on @var{cpu-type}, which is the name of a system
12512 representing a certain processor type. Possible values for
12513 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12514 @samp{ms1-16-003} and @samp{ms2}.
12516 When this option is not used, the default is @option{-march=ms1-16-002}.
12520 Use byte loads and stores when generating code.
12524 Do not use byte loads and stores when generating code.
12528 Use simulator runtime
12532 Do not link in the C run-time initialization object file
12533 @file{crti.o}. Other run-time initialization and termination files
12534 such as @file{startup.o} and @file{exit.o} are still included on the
12535 linker command line.
12539 @node PDP-11 Options
12540 @subsection PDP-11 Options
12541 @cindex PDP-11 Options
12543 These options are defined for the PDP-11:
12548 Use hardware FPP floating point. This is the default. (FIS floating
12549 point on the PDP-11/40 is not supported.)
12552 @opindex msoft-float
12553 Do not use hardware floating point.
12557 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12561 Return floating-point results in memory. This is the default.
12565 Generate code for a PDP-11/40.
12569 Generate code for a PDP-11/45. This is the default.
12573 Generate code for a PDP-11/10.
12575 @item -mbcopy-builtin
12576 @opindex bcopy-builtin
12577 Use inline @code{movmemhi} patterns for copying memory. This is the
12582 Do not use inline @code{movmemhi} patterns for copying memory.
12588 Use 16-bit @code{int}. This is the default.
12594 Use 32-bit @code{int}.
12597 @itemx -mno-float32
12599 @opindex mno-float32
12600 Use 64-bit @code{float}. This is the default.
12603 @itemx -mno-float64
12605 @opindex mno-float64
12606 Use 32-bit @code{float}.
12610 Use @code{abshi2} pattern. This is the default.
12614 Do not use @code{abshi2} pattern.
12616 @item -mbranch-expensive
12617 @opindex mbranch-expensive
12618 Pretend that branches are expensive. This is for experimenting with
12619 code generation only.
12621 @item -mbranch-cheap
12622 @opindex mbranch-cheap
12623 Do not pretend that branches are expensive. This is the default.
12627 Generate code for a system with split I&D@.
12631 Generate code for a system without split I&D@. This is the default.
12635 Use Unix assembler syntax. This is the default when configured for
12636 @samp{pdp11-*-bsd}.
12640 Use DEC assembler syntax. This is the default when configured for any
12641 PDP-11 target other than @samp{pdp11-*-bsd}.
12644 @node PowerPC Options
12645 @subsection PowerPC Options
12646 @cindex PowerPC options
12648 These are listed under @xref{RS/6000 and PowerPC Options}.
12650 @node RS/6000 and PowerPC Options
12651 @subsection IBM RS/6000 and PowerPC Options
12652 @cindex RS/6000 and PowerPC Options
12653 @cindex IBM RS/6000 and PowerPC Options
12655 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12662 @itemx -mno-powerpc
12663 @itemx -mpowerpc-gpopt
12664 @itemx -mno-powerpc-gpopt
12665 @itemx -mpowerpc-gfxopt
12666 @itemx -mno-powerpc-gfxopt
12668 @itemx -mno-powerpc64
12672 @itemx -mno-popcntb
12680 @itemx -mno-hard-dfp
12684 @opindex mno-power2
12686 @opindex mno-powerpc
12687 @opindex mpowerpc-gpopt
12688 @opindex mno-powerpc-gpopt
12689 @opindex mpowerpc-gfxopt
12690 @opindex mno-powerpc-gfxopt
12691 @opindex mpowerpc64
12692 @opindex mno-powerpc64
12696 @opindex mno-popcntb
12702 @opindex mno-mfpgpr
12704 @opindex mno-hard-dfp
12705 GCC supports two related instruction set architectures for the
12706 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12707 instructions supported by the @samp{rios} chip set used in the original
12708 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12709 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12710 the IBM 4xx, 6xx, and follow-on microprocessors.
12712 Neither architecture is a subset of the other. However there is a
12713 large common subset of instructions supported by both. An MQ
12714 register is included in processors supporting the POWER architecture.
12716 You use these options to specify which instructions are available on the
12717 processor you are using. The default value of these options is
12718 determined when configuring GCC@. Specifying the
12719 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12720 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12721 rather than the options listed above.
12723 The @option{-mpower} option allows GCC to generate instructions that
12724 are found only in the POWER architecture and to use the MQ register.
12725 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12726 to generate instructions that are present in the POWER2 architecture but
12727 not the original POWER architecture.
12729 The @option{-mpowerpc} option allows GCC to generate instructions that
12730 are found only in the 32-bit subset of the PowerPC architecture.
12731 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12732 GCC to use the optional PowerPC architecture instructions in the
12733 General Purpose group, including floating-point square root. Specifying
12734 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12735 use the optional PowerPC architecture instructions in the Graphics
12736 group, including floating-point select.
12738 The @option{-mmfcrf} option allows GCC to generate the move from
12739 condition register field instruction implemented on the POWER4
12740 processor and other processors that support the PowerPC V2.01
12742 The @option{-mpopcntb} option allows GCC to generate the popcount and
12743 double precision FP reciprocal estimate instruction implemented on the
12744 POWER5 processor and other processors that support the PowerPC V2.02
12746 The @option{-mfprnd} option allows GCC to generate the FP round to
12747 integer instructions implemented on the POWER5+ processor and other
12748 processors that support the PowerPC V2.03 architecture.
12749 The @option{-mcmpb} option allows GCC to generate the compare bytes
12750 instruction implemented on the POWER6 processor and other processors
12751 that support the PowerPC V2.05 architecture.
12752 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12753 general purpose register instructions implemented on the POWER6X
12754 processor and other processors that support the extended PowerPC V2.05
12756 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12757 point instructions implemented on some POWER processors.
12759 The @option{-mpowerpc64} option allows GCC to generate the additional
12760 64-bit instructions that are found in the full PowerPC64 architecture
12761 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12762 @option{-mno-powerpc64}.
12764 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12765 will use only the instructions in the common subset of both
12766 architectures plus some special AIX common-mode calls, and will not use
12767 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12768 permits GCC to use any instruction from either architecture and to
12769 allow use of the MQ register; specify this for the Motorola MPC601.
12771 @item -mnew-mnemonics
12772 @itemx -mold-mnemonics
12773 @opindex mnew-mnemonics
12774 @opindex mold-mnemonics
12775 Select which mnemonics to use in the generated assembler code. With
12776 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12777 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12778 assembler mnemonics defined for the POWER architecture. Instructions
12779 defined in only one architecture have only one mnemonic; GCC uses that
12780 mnemonic irrespective of which of these options is specified.
12782 GCC defaults to the mnemonics appropriate for the architecture in
12783 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12784 value of these option. Unless you are building a cross-compiler, you
12785 should normally not specify either @option{-mnew-mnemonics} or
12786 @option{-mold-mnemonics}, but should instead accept the default.
12788 @item -mcpu=@var{cpu_type}
12790 Set architecture type, register usage, choice of mnemonics, and
12791 instruction scheduling parameters for machine type @var{cpu_type}.
12792 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12793 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12794 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12795 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12796 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12797 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12798 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12799 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12800 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12801 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12803 @option{-mcpu=common} selects a completely generic processor. Code
12804 generated under this option will run on any POWER or PowerPC processor.
12805 GCC will use only the instructions in the common subset of both
12806 architectures, and will not use the MQ register. GCC assumes a generic
12807 processor model for scheduling purposes.
12809 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12810 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12811 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12812 types, with an appropriate, generic processor model assumed for
12813 scheduling purposes.
12815 The other options specify a specific processor. Code generated under
12816 those options will run best on that processor, and may not run at all on
12819 The @option{-mcpu} options automatically enable or disable the
12822 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12823 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12824 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12826 The particular options set for any particular CPU will vary between
12827 compiler versions, depending on what setting seems to produce optimal
12828 code for that CPU; it doesn't necessarily reflect the actual hardware's
12829 capabilities. If you wish to set an individual option to a particular
12830 value, you may specify it after the @option{-mcpu} option, like
12831 @samp{-mcpu=970 -mno-altivec}.
12833 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12834 not enabled or disabled by the @option{-mcpu} option at present because
12835 AIX does not have full support for these options. You may still
12836 enable or disable them individually if you're sure it'll work in your
12839 @item -mtune=@var{cpu_type}
12841 Set the instruction scheduling parameters for machine type
12842 @var{cpu_type}, but do not set the architecture type, register usage, or
12843 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12844 values for @var{cpu_type} are used for @option{-mtune} as for
12845 @option{-mcpu}. If both are specified, the code generated will use the
12846 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12847 scheduling parameters set by @option{-mtune}.
12853 Generate code to compute division as reciprocal estimate and iterative
12854 refinement, creating opportunities for increased throughput. This
12855 feature requires: optional PowerPC Graphics instruction set for single
12856 precision and FRE instruction for double precision, assuming divides
12857 cannot generate user-visible traps, and the domain values not include
12858 Infinities, denormals or zero denominator.
12861 @itemx -mno-altivec
12863 @opindex mno-altivec
12864 Generate code that uses (does not use) AltiVec instructions, and also
12865 enable the use of built-in functions that allow more direct access to
12866 the AltiVec instruction set. You may also need to set
12867 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12873 @opindex mno-vrsave
12874 Generate VRSAVE instructions when generating AltiVec code.
12877 @opindex msecure-plt
12878 Generate code that allows ld and ld.so to build executables and shared
12879 libraries with non-exec .plt and .got sections. This is a PowerPC
12880 32-bit SYSV ABI option.
12884 Generate code that uses a BSS .plt section that ld.so fills in, and
12885 requires .plt and .got sections that are both writable and executable.
12886 This is a PowerPC 32-bit SYSV ABI option.
12892 This switch enables or disables the generation of ISEL instructions.
12894 @item -misel=@var{yes/no}
12895 This switch has been deprecated. Use @option{-misel} and
12896 @option{-mno-isel} instead.
12902 This switch enables or disables the generation of SPE simd
12908 @opindex mno-paired
12909 This switch enables or disables the generation of PAIRED simd
12912 @item -mspe=@var{yes/no}
12913 This option has been deprecated. Use @option{-mspe} and
12914 @option{-mno-spe} instead.
12916 @item -mfloat-gprs=@var{yes/single/double/no}
12917 @itemx -mfloat-gprs
12918 @opindex mfloat-gprs
12919 This switch enables or disables the generation of floating point
12920 operations on the general purpose registers for architectures that
12923 The argument @var{yes} or @var{single} enables the use of
12924 single-precision floating point operations.
12926 The argument @var{double} enables the use of single and
12927 double-precision floating point operations.
12929 The argument @var{no} disables floating point operations on the
12930 general purpose registers.
12932 This option is currently only available on the MPC854x.
12938 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12939 targets (including GNU/Linux). The 32-bit environment sets int, long
12940 and pointer to 32 bits and generates code that runs on any PowerPC
12941 variant. The 64-bit environment sets int to 32 bits and long and
12942 pointer to 64 bits, and generates code for PowerPC64, as for
12943 @option{-mpowerpc64}.
12946 @itemx -mno-fp-in-toc
12947 @itemx -mno-sum-in-toc
12948 @itemx -mminimal-toc
12950 @opindex mno-fp-in-toc
12951 @opindex mno-sum-in-toc
12952 @opindex mminimal-toc
12953 Modify generation of the TOC (Table Of Contents), which is created for
12954 every executable file. The @option{-mfull-toc} option is selected by
12955 default. In that case, GCC will allocate at least one TOC entry for
12956 each unique non-automatic variable reference in your program. GCC
12957 will also place floating-point constants in the TOC@. However, only
12958 16,384 entries are available in the TOC@.
12960 If you receive a linker error message that saying you have overflowed
12961 the available TOC space, you can reduce the amount of TOC space used
12962 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12963 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12964 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12965 generate code to calculate the sum of an address and a constant at
12966 run-time instead of putting that sum into the TOC@. You may specify one
12967 or both of these options. Each causes GCC to produce very slightly
12968 slower and larger code at the expense of conserving TOC space.
12970 If you still run out of space in the TOC even when you specify both of
12971 these options, specify @option{-mminimal-toc} instead. This option causes
12972 GCC to make only one TOC entry for every file. When you specify this
12973 option, GCC will produce code that is slower and larger but which
12974 uses extremely little TOC space. You may wish to use this option
12975 only on files that contain less frequently executed code.
12981 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12982 @code{long} type, and the infrastructure needed to support them.
12983 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12984 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12985 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12988 @itemx -mno-xl-compat
12989 @opindex mxl-compat
12990 @opindex mno-xl-compat
12991 Produce code that conforms more closely to IBM XL compiler semantics
12992 when using AIX-compatible ABI. Pass floating-point arguments to
12993 prototyped functions beyond the register save area (RSA) on the stack
12994 in addition to argument FPRs. Do not assume that most significant
12995 double in 128-bit long double value is properly rounded when comparing
12996 values and converting to double. Use XL symbol names for long double
12999 The AIX calling convention was extended but not initially documented to
13000 handle an obscure K&R C case of calling a function that takes the
13001 address of its arguments with fewer arguments than declared. IBM XL
13002 compilers access floating point arguments which do not fit in the
13003 RSA from the stack when a subroutine is compiled without
13004 optimization. Because always storing floating-point arguments on the
13005 stack is inefficient and rarely needed, this option is not enabled by
13006 default and only is necessary when calling subroutines compiled by IBM
13007 XL compilers without optimization.
13011 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13012 application written to use message passing with special startup code to
13013 enable the application to run. The system must have PE installed in the
13014 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13015 must be overridden with the @option{-specs=} option to specify the
13016 appropriate directory location. The Parallel Environment does not
13017 support threads, so the @option{-mpe} option and the @option{-pthread}
13018 option are incompatible.
13020 @item -malign-natural
13021 @itemx -malign-power
13022 @opindex malign-natural
13023 @opindex malign-power
13024 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13025 @option{-malign-natural} overrides the ABI-defined alignment of larger
13026 types, such as floating-point doubles, on their natural size-based boundary.
13027 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13028 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13030 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13034 @itemx -mhard-float
13035 @opindex msoft-float
13036 @opindex mhard-float
13037 Generate code that does not use (uses) the floating-point register set.
13038 Software floating point emulation is provided if you use the
13039 @option{-msoft-float} option, and pass the option to GCC when linking.
13042 @itemx -mno-multiple
13044 @opindex mno-multiple
13045 Generate code that uses (does not use) the load multiple word
13046 instructions and the store multiple word instructions. These
13047 instructions are generated by default on POWER systems, and not
13048 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13049 endian PowerPC systems, since those instructions do not work when the
13050 processor is in little endian mode. The exceptions are PPC740 and
13051 PPC750 which permit the instructions usage in little endian mode.
13056 @opindex mno-string
13057 Generate code that uses (does not use) the load string instructions
13058 and the store string word instructions to save multiple registers and
13059 do small block moves. These instructions are generated by default on
13060 POWER systems, and not generated on PowerPC systems. Do not use
13061 @option{-mstring} on little endian PowerPC systems, since those
13062 instructions do not work when the processor is in little endian mode.
13063 The exceptions are PPC740 and PPC750 which permit the instructions
13064 usage in little endian mode.
13069 @opindex mno-update
13070 Generate code that uses (does not use) the load or store instructions
13071 that update the base register to the address of the calculated memory
13072 location. These instructions are generated by default. If you use
13073 @option{-mno-update}, there is a small window between the time that the
13074 stack pointer is updated and the address of the previous frame is
13075 stored, which means code that walks the stack frame across interrupts or
13076 signals may get corrupted data.
13079 @itemx -mno-fused-madd
13080 @opindex mfused-madd
13081 @opindex mno-fused-madd
13082 Generate code that uses (does not use) the floating point multiply and
13083 accumulate instructions. These instructions are generated by default if
13084 hardware floating is used.
13090 Generate code that uses (does not use) the half-word multiply and
13091 multiply-accumulate instructions on the IBM 405 and 440 processors.
13092 These instructions are generated by default when targetting those
13099 Generate code that uses (does not use) the string-search @samp{dlmzb}
13100 instruction on the IBM 405 and 440 processors. This instruction is
13101 generated by default when targetting those processors.
13103 @item -mno-bit-align
13105 @opindex mno-bit-align
13106 @opindex mbit-align
13107 On System V.4 and embedded PowerPC systems do not (do) force structures
13108 and unions that contain bit-fields to be aligned to the base type of the
13111 For example, by default a structure containing nothing but 8
13112 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13113 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13114 the structure would be aligned to a 1 byte boundary and be one byte in
13117 @item -mno-strict-align
13118 @itemx -mstrict-align
13119 @opindex mno-strict-align
13120 @opindex mstrict-align
13121 On System V.4 and embedded PowerPC systems do not (do) assume that
13122 unaligned memory references will be handled by the system.
13124 @item -mrelocatable
13125 @itemx -mno-relocatable
13126 @opindex mrelocatable
13127 @opindex mno-relocatable
13128 On embedded PowerPC systems generate code that allows (does not allow)
13129 the program to be relocated to a different address at runtime. If you
13130 use @option{-mrelocatable} on any module, all objects linked together must
13131 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13133 @item -mrelocatable-lib
13134 @itemx -mno-relocatable-lib
13135 @opindex mrelocatable-lib
13136 @opindex mno-relocatable-lib
13137 On embedded PowerPC systems generate code that allows (does not allow)
13138 the program to be relocated to a different address at runtime. Modules
13139 compiled with @option{-mrelocatable-lib} can be linked with either modules
13140 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13141 with modules compiled with the @option{-mrelocatable} options.
13147 On System V.4 and embedded PowerPC systems do not (do) assume that
13148 register 2 contains a pointer to a global area pointing to the addresses
13149 used in the program.
13152 @itemx -mlittle-endian
13154 @opindex mlittle-endian
13155 On System V.4 and embedded PowerPC systems compile code for the
13156 processor in little endian mode. The @option{-mlittle-endian} option is
13157 the same as @option{-mlittle}.
13160 @itemx -mbig-endian
13162 @opindex mbig-endian
13163 On System V.4 and embedded PowerPC systems compile code for the
13164 processor in big endian mode. The @option{-mbig-endian} option is
13165 the same as @option{-mbig}.
13167 @item -mdynamic-no-pic
13168 @opindex mdynamic-no-pic
13169 On Darwin and Mac OS X systems, compile code so that it is not
13170 relocatable, but that its external references are relocatable. The
13171 resulting code is suitable for applications, but not shared
13174 @item -mprioritize-restricted-insns=@var{priority}
13175 @opindex mprioritize-restricted-insns
13176 This option controls the priority that is assigned to
13177 dispatch-slot restricted instructions during the second scheduling
13178 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13179 @var{no/highest/second-highest} priority to dispatch slot restricted
13182 @item -msched-costly-dep=@var{dependence_type}
13183 @opindex msched-costly-dep
13184 This option controls which dependences are considered costly
13185 by the target during instruction scheduling. The argument
13186 @var{dependence_type} takes one of the following values:
13187 @var{no}: no dependence is costly,
13188 @var{all}: all dependences are costly,
13189 @var{true_store_to_load}: a true dependence from store to load is costly,
13190 @var{store_to_load}: any dependence from store to load is costly,
13191 @var{number}: any dependence which latency >= @var{number} is costly.
13193 @item -minsert-sched-nops=@var{scheme}
13194 @opindex minsert-sched-nops
13195 This option controls which nop insertion scheme will be used during
13196 the second scheduling pass. The argument @var{scheme} takes one of the
13198 @var{no}: Don't insert nops.
13199 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13200 according to the scheduler's grouping.
13201 @var{regroup_exact}: Insert nops to force costly dependent insns into
13202 separate groups. Insert exactly as many nops as needed to force an insn
13203 to a new group, according to the estimated processor grouping.
13204 @var{number}: Insert nops to force costly dependent insns into
13205 separate groups. Insert @var{number} nops to force an insn to a new group.
13208 @opindex mcall-sysv
13209 On System V.4 and embedded PowerPC systems compile code using calling
13210 conventions that adheres to the March 1995 draft of the System V
13211 Application Binary Interface, PowerPC processor supplement. This is the
13212 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13214 @item -mcall-sysv-eabi
13215 @opindex mcall-sysv-eabi
13216 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13218 @item -mcall-sysv-noeabi
13219 @opindex mcall-sysv-noeabi
13220 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13222 @item -mcall-solaris
13223 @opindex mcall-solaris
13224 On System V.4 and embedded PowerPC systems compile code for the Solaris
13228 @opindex mcall-linux
13229 On System V.4 and embedded PowerPC systems compile code for the
13230 Linux-based GNU system.
13234 On System V.4 and embedded PowerPC systems compile code for the
13235 Hurd-based GNU system.
13237 @item -mcall-netbsd
13238 @opindex mcall-netbsd
13239 On System V.4 and embedded PowerPC systems compile code for the
13240 NetBSD operating system.
13242 @item -maix-struct-return
13243 @opindex maix-struct-return
13244 Return all structures in memory (as specified by the AIX ABI)@.
13246 @item -msvr4-struct-return
13247 @opindex msvr4-struct-return
13248 Return structures smaller than 8 bytes in registers (as specified by the
13251 @item -mabi=@var{abi-type}
13253 Extend the current ABI with a particular extension, or remove such extension.
13254 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13255 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13259 Extend the current ABI with SPE ABI extensions. This does not change
13260 the default ABI, instead it adds the SPE ABI extensions to the current
13264 @opindex mabi=no-spe
13265 Disable Booke SPE ABI extensions for the current ABI@.
13267 @item -mabi=ibmlongdouble
13268 @opindex mabi=ibmlongdouble
13269 Change the current ABI to use IBM extended precision long double.
13270 This is a PowerPC 32-bit SYSV ABI option.
13272 @item -mabi=ieeelongdouble
13273 @opindex mabi=ieeelongdouble
13274 Change the current ABI to use IEEE extended precision long double.
13275 This is a PowerPC 32-bit Linux ABI option.
13278 @itemx -mno-prototype
13279 @opindex mprototype
13280 @opindex mno-prototype
13281 On System V.4 and embedded PowerPC systems assume that all calls to
13282 variable argument functions are properly prototyped. Otherwise, the
13283 compiler must insert an instruction before every non prototyped call to
13284 set or clear bit 6 of the condition code register (@var{CR}) to
13285 indicate whether floating point values were passed in the floating point
13286 registers in case the function takes a variable arguments. With
13287 @option{-mprototype}, only calls to prototyped variable argument functions
13288 will set or clear the bit.
13292 On embedded PowerPC systems, assume that the startup module is called
13293 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13294 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
13299 On embedded PowerPC systems, assume that the startup module is called
13300 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13305 On embedded PowerPC systems, assume that the startup module is called
13306 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13309 @item -myellowknife
13310 @opindex myellowknife
13311 On embedded PowerPC systems, assume that the startup module is called
13312 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13317 On System V.4 and embedded PowerPC systems, specify that you are
13318 compiling for a VxWorks system.
13322 Specify that you are compiling for the WindISS simulation environment.
13326 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13327 header to indicate that @samp{eabi} extended relocations are used.
13333 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13334 Embedded Applications Binary Interface (eabi) which is a set of
13335 modifications to the System V.4 specifications. Selecting @option{-meabi}
13336 means that the stack is aligned to an 8 byte boundary, a function
13337 @code{__eabi} is called to from @code{main} to set up the eabi
13338 environment, and the @option{-msdata} option can use both @code{r2} and
13339 @code{r13} to point to two separate small data areas. Selecting
13340 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13341 do not call an initialization function from @code{main}, and the
13342 @option{-msdata} option will only use @code{r13} to point to a single
13343 small data area. The @option{-meabi} option is on by default if you
13344 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13347 @opindex msdata=eabi
13348 On System V.4 and embedded PowerPC systems, put small initialized
13349 @code{const} global and static data in the @samp{.sdata2} section, which
13350 is pointed to by register @code{r2}. Put small initialized
13351 non-@code{const} global and static data in the @samp{.sdata} section,
13352 which is pointed to by register @code{r13}. Put small uninitialized
13353 global and static data in the @samp{.sbss} section, which is adjacent to
13354 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13355 incompatible with the @option{-mrelocatable} option. The
13356 @option{-msdata=eabi} option also sets the @option{-memb} option.
13359 @opindex msdata=sysv
13360 On System V.4 and embedded PowerPC systems, put small global and static
13361 data in the @samp{.sdata} section, which is pointed to by register
13362 @code{r13}. Put small uninitialized global and static data in the
13363 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13364 The @option{-msdata=sysv} option is incompatible with the
13365 @option{-mrelocatable} option.
13367 @item -msdata=default
13369 @opindex msdata=default
13371 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13372 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13373 same as @option{-msdata=sysv}.
13376 @opindex msdata-data
13377 On System V.4 and embedded PowerPC systems, put small global
13378 data in the @samp{.sdata} section. Put small uninitialized global
13379 data in the @samp{.sbss} section. Do not use register @code{r13}
13380 to address small data however. This is the default behavior unless
13381 other @option{-msdata} options are used.
13385 @opindex msdata=none
13387 On embedded PowerPC systems, put all initialized global and static data
13388 in the @samp{.data} section, and all uninitialized data in the
13389 @samp{.bss} section.
13393 @cindex smaller data references (PowerPC)
13394 @cindex .sdata/.sdata2 references (PowerPC)
13395 On embedded PowerPC systems, put global and static items less than or
13396 equal to @var{num} bytes into the small data or bss sections instead of
13397 the normal data or bss section. By default, @var{num} is 8. The
13398 @option{-G @var{num}} switch is also passed to the linker.
13399 All modules should be compiled with the same @option{-G @var{num}} value.
13402 @itemx -mno-regnames
13404 @opindex mno-regnames
13405 On System V.4 and embedded PowerPC systems do (do not) emit register
13406 names in the assembly language output using symbolic forms.
13409 @itemx -mno-longcall
13411 @opindex mno-longcall
13412 By default assume that all calls are far away so that a longer more
13413 expensive calling sequence is required. This is required for calls
13414 further than 32 megabytes (33,554,432 bytes) from the current location.
13415 A short call will be generated if the compiler knows
13416 the call cannot be that far away. This setting can be overridden by
13417 the @code{shortcall} function attribute, or by @code{#pragma
13420 Some linkers are capable of detecting out-of-range calls and generating
13421 glue code on the fly. On these systems, long calls are unnecessary and
13422 generate slower code. As of this writing, the AIX linker can do this,
13423 as can the GNU linker for PowerPC/64. It is planned to add this feature
13424 to the GNU linker for 32-bit PowerPC systems as well.
13426 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13427 callee, L42'', plus a ``branch island'' (glue code). The two target
13428 addresses represent the callee and the ``branch island''. The
13429 Darwin/PPC linker will prefer the first address and generate a ``bl
13430 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13431 otherwise, the linker will generate ``bl L42'' to call the ``branch
13432 island''. The ``branch island'' is appended to the body of the
13433 calling function; it computes the full 32-bit address of the callee
13436 On Mach-O (Darwin) systems, this option directs the compiler emit to
13437 the glue for every direct call, and the Darwin linker decides whether
13438 to use or discard it.
13440 In the future, we may cause GCC to ignore all longcall specifications
13441 when the linker is known to generate glue.
13445 Adds support for multithreading with the @dfn{pthreads} library.
13446 This option sets flags for both the preprocessor and linker.
13450 @node S/390 and zSeries Options
13451 @subsection S/390 and zSeries Options
13452 @cindex S/390 and zSeries Options
13454 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13458 @itemx -msoft-float
13459 @opindex mhard-float
13460 @opindex msoft-float
13461 Use (do not use) the hardware floating-point instructions and registers
13462 for floating-point operations. When @option{-msoft-float} is specified,
13463 functions in @file{libgcc.a} will be used to perform floating-point
13464 operations. When @option{-mhard-float} is specified, the compiler
13465 generates IEEE floating-point instructions. This is the default.
13467 @item -mlong-double-64
13468 @itemx -mlong-double-128
13469 @opindex mlong-double-64
13470 @opindex mlong-double-128
13471 These switches control the size of @code{long double} type. A size
13472 of 64bit makes the @code{long double} type equivalent to the @code{double}
13473 type. This is the default.
13476 @itemx -mno-backchain
13477 @opindex mbackchain
13478 @opindex mno-backchain
13479 Store (do not store) the address of the caller's frame as backchain pointer
13480 into the callee's stack frame.
13481 A backchain may be needed to allow debugging using tools that do not understand
13482 DWARF-2 call frame information.
13483 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13484 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13485 the backchain is placed into the topmost word of the 96/160 byte register
13488 In general, code compiled with @option{-mbackchain} is call-compatible with
13489 code compiled with @option{-mmo-backchain}; however, use of the backchain
13490 for debugging purposes usually requires that the whole binary is built with
13491 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13492 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13493 to build a linux kernel use @option{-msoft-float}.
13495 The default is to not maintain the backchain.
13497 @item -mpacked-stack
13498 @item -mno-packed-stack
13499 @opindex mpacked-stack
13500 @opindex mno-packed-stack
13501 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13502 specified, the compiler uses the all fields of the 96/160 byte register save
13503 area only for their default purpose; unused fields still take up stack space.
13504 When @option{-mpacked-stack} is specified, register save slots are densely
13505 packed at the top of the register save area; unused space is reused for other
13506 purposes, allowing for more efficient use of the available stack space.
13507 However, when @option{-mbackchain} is also in effect, the topmost word of
13508 the save area is always used to store the backchain, and the return address
13509 register is always saved two words below the backchain.
13511 As long as the stack frame backchain is not used, code generated with
13512 @option{-mpacked-stack} is call-compatible with code generated with
13513 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13514 S/390 or zSeries generated code that uses the stack frame backchain at run
13515 time, not just for debugging purposes. Such code is not call-compatible
13516 with code compiled with @option{-mpacked-stack}. Also, note that the
13517 combination of @option{-mbackchain},
13518 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13519 to build a linux kernel use @option{-msoft-float}.
13521 The default is to not use the packed stack layout.
13524 @itemx -mno-small-exec
13525 @opindex msmall-exec
13526 @opindex mno-small-exec
13527 Generate (or do not generate) code using the @code{bras} instruction
13528 to do subroutine calls.
13529 This only works reliably if the total executable size does not
13530 exceed 64k. The default is to use the @code{basr} instruction instead,
13531 which does not have this limitation.
13537 When @option{-m31} is specified, generate code compliant to the
13538 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13539 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13540 particular to generate 64-bit instructions. For the @samp{s390}
13541 targets, the default is @option{-m31}, while the @samp{s390x}
13542 targets default to @option{-m64}.
13548 When @option{-mzarch} is specified, generate code using the
13549 instructions available on z/Architecture.
13550 When @option{-mesa} is specified, generate code using the
13551 instructions available on ESA/390. Note that @option{-mesa} is
13552 not possible with @option{-m64}.
13553 When generating code compliant to the GNU/Linux for S/390 ABI,
13554 the default is @option{-mesa}. When generating code compliant
13555 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13561 Generate (or do not generate) code using the @code{mvcle} instruction
13562 to perform block moves. When @option{-mno-mvcle} is specified,
13563 use a @code{mvc} loop instead. This is the default unless optimizing for
13570 Print (or do not print) additional debug information when compiling.
13571 The default is to not print debug information.
13573 @item -march=@var{cpu-type}
13575 Generate code that will run on @var{cpu-type}, which is the name of a system
13576 representing a certain processor type. Possible values for
13577 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13578 When generating code using the instructions available on z/Architecture,
13579 the default is @option{-march=z900}. Otherwise, the default is
13580 @option{-march=g5}.
13582 @item -mtune=@var{cpu-type}
13584 Tune to @var{cpu-type} everything applicable about the generated code,
13585 except for the ABI and the set of available instructions.
13586 The list of @var{cpu-type} values is the same as for @option{-march}.
13587 The default is the value used for @option{-march}.
13590 @itemx -mno-tpf-trace
13591 @opindex mtpf-trace
13592 @opindex mno-tpf-trace
13593 Generate code that adds (does not add) in TPF OS specific branches to trace
13594 routines in the operating system. This option is off by default, even
13595 when compiling for the TPF OS@.
13598 @itemx -mno-fused-madd
13599 @opindex mfused-madd
13600 @opindex mno-fused-madd
13601 Generate code that uses (does not use) the floating point multiply and
13602 accumulate instructions. These instructions are generated by default if
13603 hardware floating point is used.
13605 @item -mwarn-framesize=@var{framesize}
13606 @opindex mwarn-framesize
13607 Emit a warning if the current function exceeds the given frame size. Because
13608 this is a compile time check it doesn't need to be a real problem when the program
13609 runs. It is intended to identify functions which most probably cause
13610 a stack overflow. It is useful to be used in an environment with limited stack
13611 size e.g.@: the linux kernel.
13613 @item -mwarn-dynamicstack
13614 @opindex mwarn-dynamicstack
13615 Emit a warning if the function calls alloca or uses dynamically
13616 sized arrays. This is generally a bad idea with a limited stack size.
13618 @item -mstack-guard=@var{stack-guard}
13619 @item -mstack-size=@var{stack-size}
13620 @opindex mstack-guard
13621 @opindex mstack-size
13622 If these options are provided the s390 back end emits additional instructions in
13623 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13624 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13625 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13626 the frame size of the compiled function is chosen.
13627 These options are intended to be used to help debugging stack overflow problems.
13628 The additionally emitted code causes only little overhead and hence can also be
13629 used in production like systems without greater performance degradation. The given
13630 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13631 @var{stack-guard} without exceeding 64k.
13632 In order to be efficient the extra code makes the assumption that the stack starts
13633 at an address aligned to the value given by @var{stack-size}.
13634 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13637 @node Score Options
13638 @subsection Score Options
13639 @cindex Score Options
13641 These options are defined for Score implementations:
13646 Compile code for big endian mode. This is the default.
13650 Compile code for little endian mode.
13654 Disable generate bcnz instruction.
13658 Enable generate unaligned load and store instruction.
13662 Enable the use of multiply-accumulate instructions. Disabled by default.
13666 Specify the SCORE5 as the target architecture.
13670 Specify the SCORE5U of the target architecture.
13674 Specify the SCORE7 as the target architecture. This is the default.
13678 Specify the SCORE7D as the target architecture.
13682 @subsection SH Options
13684 These @samp{-m} options are defined for the SH implementations:
13689 Generate code for the SH1.
13693 Generate code for the SH2.
13696 Generate code for the SH2e.
13700 Generate code for the SH3.
13704 Generate code for the SH3e.
13708 Generate code for the SH4 without a floating-point unit.
13710 @item -m4-single-only
13711 @opindex m4-single-only
13712 Generate code for the SH4 with a floating-point unit that only
13713 supports single-precision arithmetic.
13717 Generate code for the SH4 assuming the floating-point unit is in
13718 single-precision mode by default.
13722 Generate code for the SH4.
13726 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13727 floating-point unit is not used.
13729 @item -m4a-single-only
13730 @opindex m4a-single-only
13731 Generate code for the SH4a, in such a way that no double-precision
13732 floating point operations are used.
13735 @opindex m4a-single
13736 Generate code for the SH4a assuming the floating-point unit is in
13737 single-precision mode by default.
13741 Generate code for the SH4a.
13745 Same as @option{-m4a-nofpu}, except that it implicitly passes
13746 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13747 instructions at the moment.
13751 Compile code for the processor in big endian mode.
13755 Compile code for the processor in little endian mode.
13759 Align doubles at 64-bit boundaries. Note that this changes the calling
13760 conventions, and thus some functions from the standard C library will
13761 not work unless you recompile it first with @option{-mdalign}.
13765 Shorten some address references at link time, when possible; uses the
13766 linker option @option{-relax}.
13770 Use 32-bit offsets in @code{switch} tables. The default is to use
13775 Enable the use of the instruction @code{fmovd}.
13779 Comply with the calling conventions defined by Renesas.
13783 Comply with the calling conventions defined by Renesas.
13787 Comply with the calling conventions defined for GCC before the Renesas
13788 conventions were available. This option is the default for all
13789 targets of the SH toolchain except for @samp{sh-symbianelf}.
13792 @opindex mnomacsave
13793 Mark the @code{MAC} register as call-clobbered, even if
13794 @option{-mhitachi} is given.
13798 Increase IEEE-compliance of floating-point code.
13799 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13800 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13801 comparisons of NANs / infinities incurs extra overhead in every
13802 floating point comparison, therefore the default is set to
13803 @option{-ffinite-math-only}.
13805 @item -minline-ic_invalidate
13806 @opindex minline-ic_invalidate
13807 Inline code to invalidate instruction cache entries after setting up
13808 nested function trampolines.
13809 This option has no effect if -musermode is in effect and the selected
13810 code generation option (e.g. -m4) does not allow the use of the icbi
13812 If the selected code generation option does not allow the use of the icbi
13813 instruction, and -musermode is not in effect, the inlined code will
13814 manipulate the instruction cache address array directly with an associative
13815 write. This not only requires privileged mode, but it will also
13816 fail if the cache line had been mapped via the TLB and has become unmapped.
13820 Dump instruction size and location in the assembly code.
13823 @opindex mpadstruct
13824 This option is deprecated. It pads structures to multiple of 4 bytes,
13825 which is incompatible with the SH ABI@.
13829 Optimize for space instead of speed. Implied by @option{-Os}.
13832 @opindex mprefergot
13833 When generating position-independent code, emit function calls using
13834 the Global Offset Table instead of the Procedure Linkage Table.
13838 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13839 if the inlined code would not work in user mode.
13840 This is the default when the target is @code{sh-*-linux*}.
13842 @item -multcost=@var{number}
13843 @opindex multcost=@var{number}
13844 Set the cost to assume for a multiply insn.
13846 @item -mdiv=@var{strategy}
13847 @opindex mdiv=@var{strategy}
13848 Set the division strategy to use for SHmedia code. @var{strategy} must be
13849 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13850 inv:call2, inv:fp .
13851 "fp" performs the operation in floating point. This has a very high latency,
13852 but needs only a few instructions, so it might be a good choice if
13853 your code has enough easily exploitable ILP to allow the compiler to
13854 schedule the floating point instructions together with other instructions.
13855 Division by zero causes a floating point exception.
13856 "inv" uses integer operations to calculate the inverse of the divisor,
13857 and then multiplies the dividend with the inverse. This strategy allows
13858 cse and hoisting of the inverse calculation. Division by zero calculates
13859 an unspecified result, but does not trap.
13860 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13861 have been found, or if the entire operation has been hoisted to the same
13862 place, the last stages of the inverse calculation are intertwined with the
13863 final multiply to reduce the overall latency, at the expense of using a few
13864 more instructions, and thus offering fewer scheduling opportunities with
13866 "call" calls a library function that usually implements the inv:minlat
13868 This gives high code density for m5-*media-nofpu compilations.
13869 "call2" uses a different entry point of the same library function, where it
13870 assumes that a pointer to a lookup table has already been set up, which
13871 exposes the pointer load to cse / code hoisting optimizations.
13872 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13873 code generation, but if the code stays unoptimized, revert to the "call",
13874 "call2", or "fp" strategies, respectively. Note that the
13875 potentially-trapping side effect of division by zero is carried by a
13876 separate instruction, so it is possible that all the integer instructions
13877 are hoisted out, but the marker for the side effect stays where it is.
13878 A recombination to fp operations or a call is not possible in that case.
13879 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13880 that the inverse calculation was nor separated from the multiply, they speed
13881 up division where the dividend fits into 20 bits (plus sign where applicable),
13882 by inserting a test to skip a number of operations in this case; this test
13883 slows down the case of larger dividends. inv20u assumes the case of a such
13884 a small dividend to be unlikely, and inv20l assumes it to be likely.
13886 @item -mdivsi3_libfunc=@var{name}
13887 @opindex mdivsi3_libfunc=@var{name}
13888 Set the name of the library function used for 32 bit signed division to
13889 @var{name}. This only affect the name used in the call and inv:call
13890 division strategies, and the compiler will still expect the same
13891 sets of input/output/clobbered registers as if this option was not present.
13893 @item -madjust-unroll
13894 @opindex madjust-unroll
13895 Throttle unrolling to avoid thrashing target registers.
13896 This option only has an effect if the gcc code base supports the
13897 TARGET_ADJUST_UNROLL_MAX target hook.
13899 @item -mindexed-addressing
13900 @opindex mindexed-addressing
13901 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13902 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13903 semantics for the indexed addressing mode. The architecture allows the
13904 implementation of processors with 64 bit MMU, which the OS could use to
13905 get 32 bit addressing, but since no current hardware implementation supports
13906 this or any other way to make the indexed addressing mode safe to use in
13907 the 32 bit ABI, the default is -mno-indexed-addressing.
13909 @item -mgettrcost=@var{number}
13910 @opindex mgettrcost=@var{number}
13911 Set the cost assumed for the gettr instruction to @var{number}.
13912 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13916 Assume pt* instructions won't trap. This will generally generate better
13917 scheduled code, but is unsafe on current hardware. The current architecture
13918 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13919 This has the unintentional effect of making it unsafe to schedule ptabs /
13920 ptrel before a branch, or hoist it out of a loop. For example,
13921 __do_global_ctors, a part of libgcc that runs constructors at program
13922 startup, calls functions in a list which is delimited by @minus{}1. With the
13923 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13924 That means that all the constructors will be run a bit quicker, but when
13925 the loop comes to the end of the list, the program crashes because ptabs
13926 loads @minus{}1 into a target register. Since this option is unsafe for any
13927 hardware implementing the current architecture specification, the default
13928 is -mno-pt-fixed. Unless the user specifies a specific cost with
13929 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13930 this deters register allocation using target registers for storing
13933 @item -minvalid-symbols
13934 @opindex minvalid-symbols
13935 Assume symbols might be invalid. Ordinary function symbols generated by
13936 the compiler will always be valid to load with movi/shori/ptabs or
13937 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13938 to generate symbols that will cause ptabs / ptrel to trap.
13939 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13940 It will then prevent cross-basic-block cse, hoisting and most scheduling
13941 of symbol loads. The default is @option{-mno-invalid-symbols}.
13944 @node SPARC Options
13945 @subsection SPARC Options
13946 @cindex SPARC options
13948 These @samp{-m} options are supported on the SPARC:
13951 @item -mno-app-regs
13953 @opindex mno-app-regs
13955 Specify @option{-mapp-regs} to generate output using the global registers
13956 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13959 To be fully SVR4 ABI compliant at the cost of some performance loss,
13960 specify @option{-mno-app-regs}. You should compile libraries and system
13961 software with this option.
13964 @itemx -mhard-float
13966 @opindex mhard-float
13967 Generate output containing floating point instructions. This is the
13971 @itemx -msoft-float
13973 @opindex msoft-float
13974 Generate output containing library calls for floating point.
13975 @strong{Warning:} the requisite libraries are not available for all SPARC
13976 targets. Normally the facilities of the machine's usual C compiler are
13977 used, but this cannot be done directly in cross-compilation. You must make
13978 your own arrangements to provide suitable library functions for
13979 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13980 @samp{sparclite-*-*} do provide software floating point support.
13982 @option{-msoft-float} changes the calling convention in the output file;
13983 therefore, it is only useful if you compile @emph{all} of a program with
13984 this option. In particular, you need to compile @file{libgcc.a}, the
13985 library that comes with GCC, with @option{-msoft-float} in order for
13988 @item -mhard-quad-float
13989 @opindex mhard-quad-float
13990 Generate output containing quad-word (long double) floating point
13993 @item -msoft-quad-float
13994 @opindex msoft-quad-float
13995 Generate output containing library calls for quad-word (long double)
13996 floating point instructions. The functions called are those specified
13997 in the SPARC ABI@. This is the default.
13999 As of this writing, there are no SPARC implementations that have hardware
14000 support for the quad-word floating point instructions. They all invoke
14001 a trap handler for one of these instructions, and then the trap handler
14002 emulates the effect of the instruction. Because of the trap handler overhead,
14003 this is much slower than calling the ABI library routines. Thus the
14004 @option{-msoft-quad-float} option is the default.
14006 @item -mno-unaligned-doubles
14007 @itemx -munaligned-doubles
14008 @opindex mno-unaligned-doubles
14009 @opindex munaligned-doubles
14010 Assume that doubles have 8 byte alignment. This is the default.
14012 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14013 alignment only if they are contained in another type, or if they have an
14014 absolute address. Otherwise, it assumes they have 4 byte alignment.
14015 Specifying this option avoids some rare compatibility problems with code
14016 generated by other compilers. It is not the default because it results
14017 in a performance loss, especially for floating point code.
14019 @item -mno-faster-structs
14020 @itemx -mfaster-structs
14021 @opindex mno-faster-structs
14022 @opindex mfaster-structs
14023 With @option{-mfaster-structs}, the compiler assumes that structures
14024 should have 8 byte alignment. This enables the use of pairs of
14025 @code{ldd} and @code{std} instructions for copies in structure
14026 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14027 However, the use of this changed alignment directly violates the SPARC
14028 ABI@. Thus, it's intended only for use on targets where the developer
14029 acknowledges that their resulting code will not be directly in line with
14030 the rules of the ABI@.
14032 @item -mimpure-text
14033 @opindex mimpure-text
14034 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14035 the compiler to not pass @option{-z text} to the linker when linking a
14036 shared object. Using this option, you can link position-dependent
14037 code into a shared object.
14039 @option{-mimpure-text} suppresses the ``relocations remain against
14040 allocatable but non-writable sections'' linker error message.
14041 However, the necessary relocations will trigger copy-on-write, and the
14042 shared object is not actually shared across processes. Instead of
14043 using @option{-mimpure-text}, you should compile all source code with
14044 @option{-fpic} or @option{-fPIC}.
14046 This option is only available on SunOS and Solaris.
14048 @item -mcpu=@var{cpu_type}
14050 Set the instruction set, register set, and instruction scheduling parameters
14051 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14052 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14053 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14054 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14055 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14057 Default instruction scheduling parameters are used for values that select
14058 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14059 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14061 Here is a list of each supported architecture and their supported
14066 v8: supersparc, hypersparc
14067 sparclite: f930, f934, sparclite86x
14069 v9: ultrasparc, ultrasparc3, niagara, niagara2
14072 By default (unless configured otherwise), GCC generates code for the V7
14073 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14074 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14075 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14076 SPARCStation 1, 2, IPX etc.
14078 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14079 architecture. The only difference from V7 code is that the compiler emits
14080 the integer multiply and integer divide instructions which exist in SPARC-V8
14081 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14082 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14085 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14086 the SPARC architecture. This adds the integer multiply, integer divide step
14087 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14088 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14089 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14090 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14091 MB86934 chip, which is the more recent SPARClite with FPU@.
14093 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14094 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14095 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14096 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14097 optimizes it for the TEMIC SPARClet chip.
14099 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14100 architecture. This adds 64-bit integer and floating-point move instructions,
14101 3 additional floating-point condition code registers and conditional move
14102 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14103 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14104 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14105 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14106 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14107 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14108 additionally optimizes it for Sun UltraSPARC T2 chips.
14110 @item -mtune=@var{cpu_type}
14112 Set the instruction scheduling parameters for machine type
14113 @var{cpu_type}, but do not set the instruction set or register set that the
14114 option @option{-mcpu=@var{cpu_type}} would.
14116 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14117 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14118 that select a particular cpu implementation. Those are @samp{cypress},
14119 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14120 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14121 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14126 @opindex mno-v8plus
14127 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14128 difference from the V8 ABI is that the global and out registers are
14129 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14130 mode for all SPARC-V9 processors.
14136 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14137 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14140 These @samp{-m} options are supported in addition to the above
14141 on SPARC-V9 processors in 64-bit environments:
14144 @item -mlittle-endian
14145 @opindex mlittle-endian
14146 Generate code for a processor running in little-endian mode. It is only
14147 available for a few configurations and most notably not on Solaris and Linux.
14153 Generate code for a 32-bit or 64-bit environment.
14154 The 32-bit environment sets int, long and pointer to 32 bits.
14155 The 64-bit environment sets int to 32 bits and long and pointer
14158 @item -mcmodel=medlow
14159 @opindex mcmodel=medlow
14160 Generate code for the Medium/Low code model: 64-bit addresses, programs
14161 must be linked in the low 32 bits of memory. Programs can be statically
14162 or dynamically linked.
14164 @item -mcmodel=medmid
14165 @opindex mcmodel=medmid
14166 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14167 must be linked in the low 44 bits of memory, the text and data segments must
14168 be less than 2GB in size and the data segment must be located within 2GB of
14171 @item -mcmodel=medany
14172 @opindex mcmodel=medany
14173 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14174 may be linked anywhere in memory, the text and data segments must be less
14175 than 2GB in size and the data segment must be located within 2GB of the
14178 @item -mcmodel=embmedany
14179 @opindex mcmodel=embmedany
14180 Generate code for the Medium/Anywhere code model for embedded systems:
14181 64-bit addresses, the text and data segments must be less than 2GB in
14182 size, both starting anywhere in memory (determined at link time). The
14183 global register %g4 points to the base of the data segment. Programs
14184 are statically linked and PIC is not supported.
14187 @itemx -mno-stack-bias
14188 @opindex mstack-bias
14189 @opindex mno-stack-bias
14190 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14191 frame pointer if present, are offset by @minus{}2047 which must be added back
14192 when making stack frame references. This is the default in 64-bit mode.
14193 Otherwise, assume no such offset is present.
14196 These switches are supported in addition to the above on Solaris:
14201 Add support for multithreading using the Solaris threads library. This
14202 option sets flags for both the preprocessor and linker. This option does
14203 not affect the thread safety of object code produced by the compiler or
14204 that of libraries supplied with it.
14208 Add support for multithreading using the POSIX threads library. This
14209 option sets flags for both the preprocessor and linker. This option does
14210 not affect the thread safety of object code produced by the compiler or
14211 that of libraries supplied with it.
14215 This is a synonym for @option{-pthreads}.
14219 @subsection SPU Options
14220 @cindex SPU options
14222 These @samp{-m} options are supported on the SPU:
14226 @itemx -merror-reloc
14227 @opindex mwarn-reloc
14228 @opindex merror-reloc
14230 The loader for SPU does not handle dynamic relocations. By default, GCC
14231 will give an error when it generates code that requires a dynamic
14232 relocation. @option{-mno-error-reloc} disables the error,
14233 @option{-mwarn-reloc} will generate a warning instead.
14236 @itemx -munsafe-dma
14238 @opindex munsafe-dma
14240 Instructions which initiate or test completion of DMA must not be
14241 reordered with respect to loads and stores of the memory which is being
14242 accessed. Users typically address this problem using the volatile
14243 keyword, but that can lead to inefficient code in places where the
14244 memory is known to not change. Rather than mark the memory as volatile
14245 we treat the DMA instructions as potentially effecting all memory. With
14246 @option{-munsafe-dma} users must use the volatile keyword to protect
14249 @item -mbranch-hints
14250 @opindex mbranch-hints
14252 By default, GCC will generate a branch hint instruction to avoid
14253 pipeline stalls for always taken or probably taken branches. A hint
14254 will not be generated closer than 8 instructions away from its branch.
14255 There is little reason to disable them, except for debugging purposes,
14256 or to make an object a little bit smaller.
14260 @opindex msmall-mem
14261 @opindex mlarge-mem
14263 By default, GCC generates code assuming that addresses are never larger
14264 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14265 a full 32 bit address.
14270 By default, GCC links against startup code that assumes the SPU-style
14271 main function interface (which has an unconventional parameter list).
14272 With @option{-mstdmain}, GCC will link your program against startup
14273 code that assumes a C99-style interface to @code{main}, including a
14274 local copy of @code{argv} strings.
14276 @item -mfixed-range=@var{register-range}
14277 @opindex mfixed-range
14278 Generate code treating the given register range as fixed registers.
14279 A fixed register is one that the register allocator can not use. This is
14280 useful when compiling kernel code. A register range is specified as
14281 two registers separated by a dash. Multiple register ranges can be
14282 specified separated by a comma.
14286 @node System V Options
14287 @subsection Options for System V
14289 These additional options are available on System V Release 4 for
14290 compatibility with other compilers on those systems:
14295 Create a shared object.
14296 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14300 Identify the versions of each tool used by the compiler, in a
14301 @code{.ident} assembler directive in the output.
14305 Refrain from adding @code{.ident} directives to the output file (this is
14308 @item -YP,@var{dirs}
14310 Search the directories @var{dirs}, and no others, for libraries
14311 specified with @option{-l}.
14313 @item -Ym,@var{dir}
14315 Look in the directory @var{dir} to find the M4 preprocessor.
14316 The assembler uses this option.
14317 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14318 @c the generic assembler that comes with Solaris takes just -Ym.
14322 @subsection V850 Options
14323 @cindex V850 Options
14325 These @samp{-m} options are defined for V850 implementations:
14329 @itemx -mno-long-calls
14330 @opindex mlong-calls
14331 @opindex mno-long-calls
14332 Treat all calls as being far away (near). If calls are assumed to be
14333 far away, the compiler will always load the functions address up into a
14334 register, and call indirect through the pointer.
14340 Do not optimize (do optimize) basic blocks that use the same index
14341 pointer 4 or more times to copy pointer into the @code{ep} register, and
14342 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14343 option is on by default if you optimize.
14345 @item -mno-prolog-function
14346 @itemx -mprolog-function
14347 @opindex mno-prolog-function
14348 @opindex mprolog-function
14349 Do not use (do use) external functions to save and restore registers
14350 at the prologue and epilogue of a function. The external functions
14351 are slower, but use less code space if more than one function saves
14352 the same number of registers. The @option{-mprolog-function} option
14353 is on by default if you optimize.
14357 Try to make the code as small as possible. At present, this just turns
14358 on the @option{-mep} and @option{-mprolog-function} options.
14360 @item -mtda=@var{n}
14362 Put static or global variables whose size is @var{n} bytes or less into
14363 the tiny data area that register @code{ep} points to. The tiny data
14364 area can hold up to 256 bytes in total (128 bytes for byte references).
14366 @item -msda=@var{n}
14368 Put static or global variables whose size is @var{n} bytes or less into
14369 the small data area that register @code{gp} points to. The small data
14370 area can hold up to 64 kilobytes.
14372 @item -mzda=@var{n}
14374 Put static or global variables whose size is @var{n} bytes or less into
14375 the first 32 kilobytes of memory.
14379 Specify that the target processor is the V850.
14382 @opindex mbig-switch
14383 Generate code suitable for big switch tables. Use this option only if
14384 the assembler/linker complain about out of range branches within a switch
14389 This option will cause r2 and r5 to be used in the code generated by
14390 the compiler. This setting is the default.
14392 @item -mno-app-regs
14393 @opindex mno-app-regs
14394 This option will cause r2 and r5 to be treated as fixed registers.
14398 Specify that the target processor is the V850E1. The preprocessor
14399 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14400 this option is used.
14404 Specify that the target processor is the V850E@. The preprocessor
14405 constant @samp{__v850e__} will be defined if this option is used.
14407 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14408 are defined then a default target processor will be chosen and the
14409 relevant @samp{__v850*__} preprocessor constant will be defined.
14411 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14412 defined, regardless of which processor variant is the target.
14414 @item -mdisable-callt
14415 @opindex mdisable-callt
14416 This option will suppress generation of the CALLT instruction for the
14417 v850e and v850e1 flavors of the v850 architecture. The default is
14418 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14423 @subsection VAX Options
14424 @cindex VAX options
14426 These @samp{-m} options are defined for the VAX:
14431 Do not output certain jump instructions (@code{aobleq} and so on)
14432 that the Unix assembler for the VAX cannot handle across long
14437 Do output those jump instructions, on the assumption that you
14438 will assemble with the GNU assembler.
14442 Output code for g-format floating point numbers instead of d-format.
14445 @node VxWorks Options
14446 @subsection VxWorks Options
14447 @cindex VxWorks Options
14449 The options in this section are defined for all VxWorks targets.
14450 Options specific to the target hardware are listed with the other
14451 options for that target.
14456 GCC can generate code for both VxWorks kernels and real time processes
14457 (RTPs). This option switches from the former to the latter. It also
14458 defines the preprocessor macro @code{__RTP__}.
14461 @opindex non-static
14462 Link an RTP executable against shared libraries rather than static
14463 libraries. The options @option{-static} and @option{-shared} can
14464 also be used for RTPs (@pxref{Link Options}); @option{-static}
14471 These options are passed down to the linker. They are defined for
14472 compatibility with Diab.
14475 @opindex Xbind-lazy
14476 Enable lazy binding of function calls. This option is equivalent to
14477 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14481 Disable lazy binding of function calls. This option is the default and
14482 is defined for compatibility with Diab.
14485 @node x86-64 Options
14486 @subsection x86-64 Options
14487 @cindex x86-64 options
14489 These are listed under @xref{i386 and x86-64 Options}.
14491 @node Xstormy16 Options
14492 @subsection Xstormy16 Options
14493 @cindex Xstormy16 Options
14495 These options are defined for Xstormy16:
14500 Choose startup files and linker script suitable for the simulator.
14503 @node Xtensa Options
14504 @subsection Xtensa Options
14505 @cindex Xtensa Options
14507 These options are supported for Xtensa targets:
14511 @itemx -mno-const16
14513 @opindex mno-const16
14514 Enable or disable use of @code{CONST16} instructions for loading
14515 constant values. The @code{CONST16} instruction is currently not a
14516 standard option from Tensilica. When enabled, @code{CONST16}
14517 instructions are always used in place of the standard @code{L32R}
14518 instructions. The use of @code{CONST16} is enabled by default only if
14519 the @code{L32R} instruction is not available.
14522 @itemx -mno-fused-madd
14523 @opindex mfused-madd
14524 @opindex mno-fused-madd
14525 Enable or disable use of fused multiply/add and multiply/subtract
14526 instructions in the floating-point option. This has no effect if the
14527 floating-point option is not also enabled. Disabling fused multiply/add
14528 and multiply/subtract instructions forces the compiler to use separate
14529 instructions for the multiply and add/subtract operations. This may be
14530 desirable in some cases where strict IEEE 754-compliant results are
14531 required: the fused multiply add/subtract instructions do not round the
14532 intermediate result, thereby producing results with @emph{more} bits of
14533 precision than specified by the IEEE standard. Disabling fused multiply
14534 add/subtract instructions also ensures that the program output is not
14535 sensitive to the compiler's ability to combine multiply and add/subtract
14538 @item -mtext-section-literals
14539 @itemx -mno-text-section-literals
14540 @opindex mtext-section-literals
14541 @opindex mno-text-section-literals
14542 Control the treatment of literal pools. The default is
14543 @option{-mno-text-section-literals}, which places literals in a separate
14544 section in the output file. This allows the literal pool to be placed
14545 in a data RAM/ROM, and it also allows the linker to combine literal
14546 pools from separate object files to remove redundant literals and
14547 improve code size. With @option{-mtext-section-literals}, the literals
14548 are interspersed in the text section in order to keep them as close as
14549 possible to their references. This may be necessary for large assembly
14552 @item -mtarget-align
14553 @itemx -mno-target-align
14554 @opindex mtarget-align
14555 @opindex mno-target-align
14556 When this option is enabled, GCC instructs the assembler to
14557 automatically align instructions to reduce branch penalties at the
14558 expense of some code density. The assembler attempts to widen density
14559 instructions to align branch targets and the instructions following call
14560 instructions. If there are not enough preceding safe density
14561 instructions to align a target, no widening will be performed. The
14562 default is @option{-mtarget-align}. These options do not affect the
14563 treatment of auto-aligned instructions like @code{LOOP}, which the
14564 assembler will always align, either by widening density instructions or
14565 by inserting no-op instructions.
14568 @itemx -mno-longcalls
14569 @opindex mlongcalls
14570 @opindex mno-longcalls
14571 When this option is enabled, GCC instructs the assembler to translate
14572 direct calls to indirect calls unless it can determine that the target
14573 of a direct call is in the range allowed by the call instruction. This
14574 translation typically occurs for calls to functions in other source
14575 files. Specifically, the assembler translates a direct @code{CALL}
14576 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14577 The default is @option{-mno-longcalls}. This option should be used in
14578 programs where the call target can potentially be out of range. This
14579 option is implemented in the assembler, not the compiler, so the
14580 assembly code generated by GCC will still show direct call
14581 instructions---look at the disassembled object code to see the actual
14582 instructions. Note that the assembler will use an indirect call for
14583 every cross-file call, not just those that really will be out of range.
14586 @node zSeries Options
14587 @subsection zSeries Options
14588 @cindex zSeries options
14590 These are listed under @xref{S/390 and zSeries Options}.
14592 @node Code Gen Options
14593 @section Options for Code Generation Conventions
14594 @cindex code generation conventions
14595 @cindex options, code generation
14596 @cindex run-time options
14598 These machine-independent options control the interface conventions
14599 used in code generation.
14601 Most of them have both positive and negative forms; the negative form
14602 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14603 one of the forms is listed---the one which is not the default. You
14604 can figure out the other form by either removing @samp{no-} or adding
14608 @item -fbounds-check
14609 @opindex fbounds-check
14610 For front-ends that support it, generate additional code to check that
14611 indices used to access arrays are within the declared range. This is
14612 currently only supported by the Java and Fortran front-ends, where
14613 this option defaults to true and false respectively.
14617 This option generates traps for signed overflow on addition, subtraction,
14618 multiplication operations.
14622 This option instructs the compiler to assume that signed arithmetic
14623 overflow of addition, subtraction and multiplication wraps around
14624 using twos-complement representation. This flag enables some optimizations
14625 and disables others. This option is enabled by default for the Java
14626 front-end, as required by the Java language specification.
14629 @opindex fexceptions
14630 Enable exception handling. Generates extra code needed to propagate
14631 exceptions. For some targets, this implies GCC will generate frame
14632 unwind information for all functions, which can produce significant data
14633 size overhead, although it does not affect execution. If you do not
14634 specify this option, GCC will enable it by default for languages like
14635 C++ which normally require exception handling, and disable it for
14636 languages like C that do not normally require it. However, you may need
14637 to enable this option when compiling C code that needs to interoperate
14638 properly with exception handlers written in C++. You may also wish to
14639 disable this option if you are compiling older C++ programs that don't
14640 use exception handling.
14642 @item -fnon-call-exceptions
14643 @opindex fnon-call-exceptions
14644 Generate code that allows trapping instructions to throw exceptions.
14645 Note that this requires platform-specific runtime support that does
14646 not exist everywhere. Moreover, it only allows @emph{trapping}
14647 instructions to throw exceptions, i.e.@: memory references or floating
14648 point instructions. It does not allow exceptions to be thrown from
14649 arbitrary signal handlers such as @code{SIGALRM}.
14651 @item -funwind-tables
14652 @opindex funwind-tables
14653 Similar to @option{-fexceptions}, except that it will just generate any needed
14654 static data, but will not affect the generated code in any other way.
14655 You will normally not enable this option; instead, a language processor
14656 that needs this handling would enable it on your behalf.
14658 @item -fasynchronous-unwind-tables
14659 @opindex fasynchronous-unwind-tables
14660 Generate unwind table in dwarf2 format, if supported by target machine. The
14661 table is exact at each instruction boundary, so it can be used for stack
14662 unwinding from asynchronous events (such as debugger or garbage collector).
14664 @item -fpcc-struct-return
14665 @opindex fpcc-struct-return
14666 Return ``short'' @code{struct} and @code{union} values in memory like
14667 longer ones, rather than in registers. This convention is less
14668 efficient, but it has the advantage of allowing intercallability between
14669 GCC-compiled files and files compiled with other compilers, particularly
14670 the Portable C Compiler (pcc).
14672 The precise convention for returning structures in memory depends
14673 on the target configuration macros.
14675 Short structures and unions are those whose size and alignment match
14676 that of some integer type.
14678 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14679 switch is not binary compatible with code compiled with the
14680 @option{-freg-struct-return} switch.
14681 Use it to conform to a non-default application binary interface.
14683 @item -freg-struct-return
14684 @opindex freg-struct-return
14685 Return @code{struct} and @code{union} values in registers when possible.
14686 This is more efficient for small structures than
14687 @option{-fpcc-struct-return}.
14689 If you specify neither @option{-fpcc-struct-return} nor
14690 @option{-freg-struct-return}, GCC defaults to whichever convention is
14691 standard for the target. If there is no standard convention, GCC
14692 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14693 the principal compiler. In those cases, we can choose the standard, and
14694 we chose the more efficient register return alternative.
14696 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14697 switch is not binary compatible with code compiled with the
14698 @option{-fpcc-struct-return} switch.
14699 Use it to conform to a non-default application binary interface.
14701 @item -fshort-enums
14702 @opindex fshort-enums
14703 Allocate to an @code{enum} type only as many bytes as it needs for the
14704 declared range of possible values. Specifically, the @code{enum} type
14705 will be equivalent to the smallest integer type which has enough room.
14707 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14708 code that is not binary compatible with code generated without that switch.
14709 Use it to conform to a non-default application binary interface.
14711 @item -fshort-double
14712 @opindex fshort-double
14713 Use the same size for @code{double} as for @code{float}.
14715 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14716 code that is not binary compatible with code generated without that switch.
14717 Use it to conform to a non-default application binary interface.
14719 @item -fshort-wchar
14720 @opindex fshort-wchar
14721 Override the underlying type for @samp{wchar_t} to be @samp{short
14722 unsigned int} instead of the default for the target. This option is
14723 useful for building programs to run under WINE@.
14725 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14726 code that is not binary compatible with code generated without that switch.
14727 Use it to conform to a non-default application binary interface.
14730 @opindex fno-common
14731 In C, allocate even uninitialized global variables in the data section of the
14732 object file, rather than generating them as common blocks. This has the
14733 effect that if the same variable is declared (without @code{extern}) in
14734 two different compilations, you will get an error when you link them.
14735 The only reason this might be useful is if you wish to verify that the
14736 program will work on other systems which always work this way.
14740 Ignore the @samp{#ident} directive.
14742 @item -finhibit-size-directive
14743 @opindex finhibit-size-directive
14744 Don't output a @code{.size} assembler directive, or anything else that
14745 would cause trouble if the function is split in the middle, and the
14746 two halves are placed at locations far apart in memory. This option is
14747 used when compiling @file{crtstuff.c}; you should not need to use it
14750 @item -fverbose-asm
14751 @opindex fverbose-asm
14752 Put extra commentary information in the generated assembly code to
14753 make it more readable. This option is generally only of use to those
14754 who actually need to read the generated assembly code (perhaps while
14755 debugging the compiler itself).
14757 @option{-fno-verbose-asm}, the default, causes the
14758 extra information to be omitted and is useful when comparing two assembler
14761 @item -frecord-gcc-switches
14762 @opindex frecord-gcc-switches
14763 This switch causes the command line that was used to invoke the
14764 compiler to be recorded into the object file that is being created.
14765 This switch is only implemented on some targets and the exact format
14766 of the recording is target and binary file format dependent, but it
14767 usually takes the form of a section containing ASCII text. This
14768 switch is related to the @option{-fverbose-asm} switch, but that
14769 switch only records information in the assembler output file as
14770 comments, so it never reaches the object file.
14774 @cindex global offset table
14776 Generate position-independent code (PIC) suitable for use in a shared
14777 library, if supported for the target machine. Such code accesses all
14778 constant addresses through a global offset table (GOT)@. The dynamic
14779 loader resolves the GOT entries when the program starts (the dynamic
14780 loader is not part of GCC; it is part of the operating system). If
14781 the GOT size for the linked executable exceeds a machine-specific
14782 maximum size, you get an error message from the linker indicating that
14783 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14784 instead. (These maximums are 8k on the SPARC and 32k
14785 on the m68k and RS/6000. The 386 has no such limit.)
14787 Position-independent code requires special support, and therefore works
14788 only on certain machines. For the 386, GCC supports PIC for System V
14789 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14790 position-independent.
14792 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14797 If supported for the target machine, emit position-independent code,
14798 suitable for dynamic linking and avoiding any limit on the size of the
14799 global offset table. This option makes a difference on the m68k,
14800 PowerPC and SPARC@.
14802 Position-independent code requires special support, and therefore works
14803 only on certain machines.
14805 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14812 These options are similar to @option{-fpic} and @option{-fPIC}, but
14813 generated position independent code can be only linked into executables.
14814 Usually these options are used when @option{-pie} GCC option will be
14815 used during linking.
14817 @option{-fpie} and @option{-fPIE} both define the macros
14818 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14819 for @option{-fpie} and 2 for @option{-fPIE}.
14821 @item -fno-jump-tables
14822 @opindex fno-jump-tables
14823 Do not use jump tables for switch statements even where it would be
14824 more efficient than other code generation strategies. This option is
14825 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14826 building code which forms part of a dynamic linker and cannot
14827 reference the address of a jump table. On some targets, jump tables
14828 do not require a GOT and this option is not needed.
14830 @item -ffixed-@var{reg}
14832 Treat the register named @var{reg} as a fixed register; generated code
14833 should never refer to it (except perhaps as a stack pointer, frame
14834 pointer or in some other fixed role).
14836 @var{reg} must be the name of a register. The register names accepted
14837 are machine-specific and are defined in the @code{REGISTER_NAMES}
14838 macro in the machine description macro file.
14840 This flag does not have a negative form, because it specifies a
14843 @item -fcall-used-@var{reg}
14844 @opindex fcall-used
14845 Treat the register named @var{reg} as an allocable register that is
14846 clobbered by function calls. It may be allocated for temporaries or
14847 variables that do not live across a call. Functions compiled this way
14848 will not save and restore the register @var{reg}.
14850 It is an error to used this flag with the frame pointer or stack pointer.
14851 Use of this flag for other registers that have fixed pervasive roles in
14852 the machine's execution model will produce disastrous results.
14854 This flag does not have a negative form, because it specifies a
14857 @item -fcall-saved-@var{reg}
14858 @opindex fcall-saved
14859 Treat the register named @var{reg} as an allocable register saved by
14860 functions. It may be allocated even for temporaries or variables that
14861 live across a call. Functions compiled this way will save and restore
14862 the register @var{reg} if they use it.
14864 It is an error to used this flag with the frame pointer or stack pointer.
14865 Use of this flag for other registers that have fixed pervasive roles in
14866 the machine's execution model will produce disastrous results.
14868 A different sort of disaster will result from the use of this flag for
14869 a register in which function values may be returned.
14871 This flag does not have a negative form, because it specifies a
14874 @item -fpack-struct[=@var{n}]
14875 @opindex fpack-struct
14876 Without a value specified, pack all structure members together without
14877 holes. When a value is specified (which must be a small power of two), pack
14878 structure members according to this value, representing the maximum
14879 alignment (that is, objects with default alignment requirements larger than
14880 this will be output potentially unaligned at the next fitting location.
14882 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14883 code that is not binary compatible with code generated without that switch.
14884 Additionally, it makes the code suboptimal.
14885 Use it to conform to a non-default application binary interface.
14887 @item -finstrument-functions
14888 @opindex finstrument-functions
14889 Generate instrumentation calls for entry and exit to functions. Just
14890 after function entry and just before function exit, the following
14891 profiling functions will be called with the address of the current
14892 function and its call site. (On some platforms,
14893 @code{__builtin_return_address} does not work beyond the current
14894 function, so the call site information may not be available to the
14895 profiling functions otherwise.)
14898 void __cyg_profile_func_enter (void *this_fn,
14900 void __cyg_profile_func_exit (void *this_fn,
14904 The first argument is the address of the start of the current function,
14905 which may be looked up exactly in the symbol table.
14907 This instrumentation is also done for functions expanded inline in other
14908 functions. The profiling calls will indicate where, conceptually, the
14909 inline function is entered and exited. This means that addressable
14910 versions of such functions must be available. If all your uses of a
14911 function are expanded inline, this may mean an additional expansion of
14912 code size. If you use @samp{extern inline} in your C code, an
14913 addressable version of such functions must be provided. (This is
14914 normally the case anyways, but if you get lucky and the optimizer always
14915 expands the functions inline, you might have gotten away without
14916 providing static copies.)
14918 A function may be given the attribute @code{no_instrument_function}, in
14919 which case this instrumentation will not be done. This can be used, for
14920 example, for the profiling functions listed above, high-priority
14921 interrupt routines, and any functions from which the profiling functions
14922 cannot safely be called (perhaps signal handlers, if the profiling
14923 routines generate output or allocate memory).
14925 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
14926 @opindex finstrument-functions-exclude-file-list
14928 Set the list of functions that are excluded from instrumentation (see
14929 the description of @code{-finstrument-functions}). If the file that
14930 contains a function definition matches with one of @var{file}, then
14931 that function is not instrumented. The match is done on substrings:
14932 if the @var{file} parameter is a substring of the file name, it is
14933 considered to be a match.
14936 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
14937 will exclude any inline function defined in files whose pathnames
14938 contain @code{/bits/stl} or @code{include/sys}.
14940 If, for some reason, you want to include letter @code{','} in one of
14941 @var{sym}, write @code{'\,'}. For example,
14942 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
14943 (note the single quote surrounding the option).
14945 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
14946 @opindex finstrument-functions-exclude-function-list
14948 This is similar to @code{-finstrument-functions-exclude-file-list},
14949 but this option sets the list of function names to be excluded from
14950 instrumentation. The function name to be matched is its user-visible
14951 name, such as @code{vector<int> blah(const vector<int> &)}, not the
14952 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
14953 match is done on substrings: if the @var{sym} parameter is a substring
14954 of the function name, it is considered to be a match.
14956 @item -fstack-check
14957 @opindex fstack-check
14958 Generate code to verify that you do not go beyond the boundary of the
14959 stack. You should specify this flag if you are running in an
14960 environment with multiple threads, but only rarely need to specify it in
14961 a single-threaded environment since stack overflow is automatically
14962 detected on nearly all systems if there is only one stack.
14964 Note that this switch does not actually cause checking to be done; the
14965 operating system must do that. The switch causes generation of code
14966 to ensure that the operating system sees the stack being extended.
14968 @item -fstack-limit-register=@var{reg}
14969 @itemx -fstack-limit-symbol=@var{sym}
14970 @itemx -fno-stack-limit
14971 @opindex fstack-limit-register
14972 @opindex fstack-limit-symbol
14973 @opindex fno-stack-limit
14974 Generate code to ensure that the stack does not grow beyond a certain value,
14975 either the value of a register or the address of a symbol. If the stack
14976 would grow beyond the value, a signal is raised. For most targets,
14977 the signal is raised before the stack overruns the boundary, so
14978 it is possible to catch the signal without taking special precautions.
14980 For instance, if the stack starts at absolute address @samp{0x80000000}
14981 and grows downwards, you can use the flags
14982 @option{-fstack-limit-symbol=__stack_limit} and
14983 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14984 of 128KB@. Note that this may only work with the GNU linker.
14986 @cindex aliasing of parameters
14987 @cindex parameters, aliased
14988 @item -fargument-alias
14989 @itemx -fargument-noalias
14990 @itemx -fargument-noalias-global
14991 @itemx -fargument-noalias-anything
14992 @opindex fargument-alias
14993 @opindex fargument-noalias
14994 @opindex fargument-noalias-global
14995 @opindex fargument-noalias-anything
14996 Specify the possible relationships among parameters and between
14997 parameters and global data.
14999 @option{-fargument-alias} specifies that arguments (parameters) may
15000 alias each other and may alias global storage.@*
15001 @option{-fargument-noalias} specifies that arguments do not alias
15002 each other, but may alias global storage.@*
15003 @option{-fargument-noalias-global} specifies that arguments do not
15004 alias each other and do not alias global storage.
15005 @option{-fargument-noalias-anything} specifies that arguments do not
15006 alias any other storage.
15008 Each language will automatically use whatever option is required by
15009 the language standard. You should not need to use these options yourself.
15011 @item -fleading-underscore
15012 @opindex fleading-underscore
15013 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15014 change the way C symbols are represented in the object file. One use
15015 is to help link with legacy assembly code.
15017 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15018 generate code that is not binary compatible with code generated without that
15019 switch. Use it to conform to a non-default application binary interface.
15020 Not all targets provide complete support for this switch.
15022 @item -ftls-model=@var{model}
15023 @opindex ftls-model
15024 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15025 The @var{model} argument should be one of @code{global-dynamic},
15026 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15028 The default without @option{-fpic} is @code{initial-exec}; with
15029 @option{-fpic} the default is @code{global-dynamic}.
15031 @item -fvisibility=@var{default|internal|hidden|protected}
15032 @opindex fvisibility
15033 Set the default ELF image symbol visibility to the specified option---all
15034 symbols will be marked with this unless overridden within the code.
15035 Using this feature can very substantially improve linking and
15036 load times of shared object libraries, produce more optimized
15037 code, provide near-perfect API export and prevent symbol clashes.
15038 It is @strong{strongly} recommended that you use this in any shared objects
15041 Despite the nomenclature, @code{default} always means public ie;
15042 available to be linked against from outside the shared object.
15043 @code{protected} and @code{internal} are pretty useless in real-world
15044 usage so the only other commonly used option will be @code{hidden}.
15045 The default if @option{-fvisibility} isn't specified is
15046 @code{default}, i.e., make every
15047 symbol public---this causes the same behavior as previous versions of
15050 A good explanation of the benefits offered by ensuring ELF
15051 symbols have the correct visibility is given by ``How To Write
15052 Shared Libraries'' by Ulrich Drepper (which can be found at
15053 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15054 solution made possible by this option to marking things hidden when
15055 the default is public is to make the default hidden and mark things
15056 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15057 and @code{__attribute__ ((visibility("default")))} instead of
15058 @code{__declspec(dllexport)} you get almost identical semantics with
15059 identical syntax. This is a great boon to those working with
15060 cross-platform projects.
15062 For those adding visibility support to existing code, you may find
15063 @samp{#pragma GCC visibility} of use. This works by you enclosing
15064 the declarations you wish to set visibility for with (for example)
15065 @samp{#pragma GCC visibility push(hidden)} and
15066 @samp{#pragma GCC visibility pop}.
15067 Bear in mind that symbol visibility should be viewed @strong{as
15068 part of the API interface contract} and thus all new code should
15069 always specify visibility when it is not the default ie; declarations
15070 only for use within the local DSO should @strong{always} be marked explicitly
15071 as hidden as so to avoid PLT indirection overheads---making this
15072 abundantly clear also aids readability and self-documentation of the code.
15073 Note that due to ISO C++ specification requirements, operator new and
15074 operator delete must always be of default visibility.
15076 Be aware that headers from outside your project, in particular system
15077 headers and headers from any other library you use, may not be
15078 expecting to be compiled with visibility other than the default. You
15079 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15080 before including any such headers.
15082 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15083 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15084 no modifications. However, this means that calls to @samp{extern}
15085 functions with no explicit visibility will use the PLT, so it is more
15086 effective to use @samp{__attribute ((visibility))} and/or
15087 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15088 declarations should be treated as hidden.
15090 Note that @samp{-fvisibility} does affect C++ vague linkage
15091 entities. This means that, for instance, an exception class that will
15092 be thrown between DSOs must be explicitly marked with default
15093 visibility so that the @samp{type_info} nodes will be unified between
15096 An overview of these techniques, their benefits and how to use them
15097 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15103 @node Environment Variables
15104 @section Environment Variables Affecting GCC
15105 @cindex environment variables
15107 @c man begin ENVIRONMENT
15108 This section describes several environment variables that affect how GCC
15109 operates. Some of them work by specifying directories or prefixes to use
15110 when searching for various kinds of files. Some are used to specify other
15111 aspects of the compilation environment.
15113 Note that you can also specify places to search using options such as
15114 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15115 take precedence over places specified using environment variables, which
15116 in turn take precedence over those specified by the configuration of GCC@.
15117 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15118 GNU Compiler Collection (GCC) Internals}.
15123 @c @itemx LC_COLLATE
15125 @c @itemx LC_MONETARY
15126 @c @itemx LC_NUMERIC
15131 @c @findex LC_COLLATE
15132 @findex LC_MESSAGES
15133 @c @findex LC_MONETARY
15134 @c @findex LC_NUMERIC
15138 These environment variables control the way that GCC uses
15139 localization information that allow GCC to work with different
15140 national conventions. GCC inspects the locale categories
15141 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15142 so. These locale categories can be set to any value supported by your
15143 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15144 Kingdom encoded in UTF-8.
15146 The @env{LC_CTYPE} environment variable specifies character
15147 classification. GCC uses it to determine the character boundaries in
15148 a string; this is needed for some multibyte encodings that contain quote
15149 and escape characters that would otherwise be interpreted as a string
15152 The @env{LC_MESSAGES} environment variable specifies the language to
15153 use in diagnostic messages.
15155 If the @env{LC_ALL} environment variable is set, it overrides the value
15156 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15157 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15158 environment variable. If none of these variables are set, GCC
15159 defaults to traditional C English behavior.
15163 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15164 files. GCC uses temporary files to hold the output of one stage of
15165 compilation which is to be used as input to the next stage: for example,
15166 the output of the preprocessor, which is the input to the compiler
15169 @item GCC_EXEC_PREFIX
15170 @findex GCC_EXEC_PREFIX
15171 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15172 names of the subprograms executed by the compiler. No slash is added
15173 when this prefix is combined with the name of a subprogram, but you can
15174 specify a prefix that ends with a slash if you wish.
15176 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15177 an appropriate prefix to use based on the pathname it was invoked with.
15179 If GCC cannot find the subprogram using the specified prefix, it
15180 tries looking in the usual places for the subprogram.
15182 The default value of @env{GCC_EXEC_PREFIX} is
15183 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15184 the installed compiler. In many cases @var{prefix} is the value
15185 of @code{prefix} when you ran the @file{configure} script.
15187 Other prefixes specified with @option{-B} take precedence over this prefix.
15189 This prefix is also used for finding files such as @file{crt0.o} that are
15192 In addition, the prefix is used in an unusual way in finding the
15193 directories to search for header files. For each of the standard
15194 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15195 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15196 replacing that beginning with the specified prefix to produce an
15197 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15198 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15199 These alternate directories are searched first; the standard directories
15200 come next. If a standard directory begins with the configured
15201 @var{prefix} then the value of @var{prefix} is replaced by
15202 @env{GCC_EXEC_PREFIX} when looking for header files.
15204 @item COMPILER_PATH
15205 @findex COMPILER_PATH
15206 The value of @env{COMPILER_PATH} is a colon-separated list of
15207 directories, much like @env{PATH}. GCC tries the directories thus
15208 specified when searching for subprograms, if it can't find the
15209 subprograms using @env{GCC_EXEC_PREFIX}.
15212 @findex LIBRARY_PATH
15213 The value of @env{LIBRARY_PATH} is a colon-separated list of
15214 directories, much like @env{PATH}. When configured as a native compiler,
15215 GCC tries the directories thus specified when searching for special
15216 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15217 using GCC also uses these directories when searching for ordinary
15218 libraries for the @option{-l} option (but directories specified with
15219 @option{-L} come first).
15223 @cindex locale definition
15224 This variable is used to pass locale information to the compiler. One way in
15225 which this information is used is to determine the character set to be used
15226 when character literals, string literals and comments are parsed in C and C++.
15227 When the compiler is configured to allow multibyte characters,
15228 the following values for @env{LANG} are recognized:
15232 Recognize JIS characters.
15234 Recognize SJIS characters.
15236 Recognize EUCJP characters.
15239 If @env{LANG} is not defined, or if it has some other value, then the
15240 compiler will use mblen and mbtowc as defined by the default locale to
15241 recognize and translate multibyte characters.
15245 Some additional environments variables affect the behavior of the
15248 @include cppenv.texi
15252 @node Precompiled Headers
15253 @section Using Precompiled Headers
15254 @cindex precompiled headers
15255 @cindex speed of compilation
15257 Often large projects have many header files that are included in every
15258 source file. The time the compiler takes to process these header files
15259 over and over again can account for nearly all of the time required to
15260 build the project. To make builds faster, GCC allows users to
15261 `precompile' a header file; then, if builds can use the precompiled
15262 header file they will be much faster.
15264 To create a precompiled header file, simply compile it as you would any
15265 other file, if necessary using the @option{-x} option to make the driver
15266 treat it as a C or C++ header file. You will probably want to use a
15267 tool like @command{make} to keep the precompiled header up-to-date when
15268 the headers it contains change.
15270 A precompiled header file will be searched for when @code{#include} is
15271 seen in the compilation. As it searches for the included file
15272 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15273 compiler looks for a precompiled header in each directory just before it
15274 looks for the include file in that directory. The name searched for is
15275 the name specified in the @code{#include} with @samp{.gch} appended. If
15276 the precompiled header file can't be used, it is ignored.
15278 For instance, if you have @code{#include "all.h"}, and you have
15279 @file{all.h.gch} in the same directory as @file{all.h}, then the
15280 precompiled header file will be used if possible, and the original
15281 header will be used otherwise.
15283 Alternatively, you might decide to put the precompiled header file in a
15284 directory and use @option{-I} to ensure that directory is searched
15285 before (or instead of) the directory containing the original header.
15286 Then, if you want to check that the precompiled header file is always
15287 used, you can put a file of the same name as the original header in this
15288 directory containing an @code{#error} command.
15290 This also works with @option{-include}. So yet another way to use
15291 precompiled headers, good for projects not designed with precompiled
15292 header files in mind, is to simply take most of the header files used by
15293 a project, include them from another header file, precompile that header
15294 file, and @option{-include} the precompiled header. If the header files
15295 have guards against multiple inclusion, they will be skipped because
15296 they've already been included (in the precompiled header).
15298 If you need to precompile the same header file for different
15299 languages, targets, or compiler options, you can instead make a
15300 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15301 header in the directory, perhaps using @option{-o}. It doesn't matter
15302 what you call the files in the directory, every precompiled header in
15303 the directory will be considered. The first precompiled header
15304 encountered in the directory that is valid for this compilation will
15305 be used; they're searched in no particular order.
15307 There are many other possibilities, limited only by your imagination,
15308 good sense, and the constraints of your build system.
15310 A precompiled header file can be used only when these conditions apply:
15314 Only one precompiled header can be used in a particular compilation.
15317 A precompiled header can't be used once the first C token is seen. You
15318 can have preprocessor directives before a precompiled header; you can
15319 even include a precompiled header from inside another header, so long as
15320 there are no C tokens before the @code{#include}.
15323 The precompiled header file must be produced for the same language as
15324 the current compilation. You can't use a C precompiled header for a C++
15328 The precompiled header file must have been produced by the same compiler
15329 binary as the current compilation is using.
15332 Any macros defined before the precompiled header is included must
15333 either be defined in the same way as when the precompiled header was
15334 generated, or must not affect the precompiled header, which usually
15335 means that they don't appear in the precompiled header at all.
15337 The @option{-D} option is one way to define a macro before a
15338 precompiled header is included; using a @code{#define} can also do it.
15339 There are also some options that define macros implicitly, like
15340 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15343 @item If debugging information is output when using the precompiled
15344 header, using @option{-g} or similar, the same kind of debugging information
15345 must have been output when building the precompiled header. However,
15346 a precompiled header built using @option{-g} can be used in a compilation
15347 when no debugging information is being output.
15349 @item The same @option{-m} options must generally be used when building
15350 and using the precompiled header. @xref{Submodel Options},
15351 for any cases where this rule is relaxed.
15353 @item Each of the following options must be the same when building and using
15354 the precompiled header:
15356 @gccoptlist{-fexceptions -funit-at-a-time}
15359 Some other command-line options starting with @option{-f},
15360 @option{-p}, or @option{-O} must be defined in the same way as when
15361 the precompiled header was generated. At present, it's not clear
15362 which options are safe to change and which are not; the safest choice
15363 is to use exactly the same options when generating and using the
15364 precompiled header. The following are known to be safe:
15366 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15367 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15368 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15373 For all of these except the last, the compiler will automatically
15374 ignore the precompiled header if the conditions aren't met. If you
15375 find an option combination that doesn't work and doesn't cause the
15376 precompiled header to be ignored, please consider filing a bug report,
15379 If you do use differing options when generating and using the
15380 precompiled header, the actual behavior will be a mixture of the
15381 behavior for the options. For instance, if you use @option{-g} to
15382 generate the precompiled header but not when using it, you may or may
15383 not get debugging information for routines in the precompiled header.
15385 @node Running Protoize
15386 @section Running Protoize
15388 The program @code{protoize} is an optional part of GCC@. You can use
15389 it to add prototypes to a program, thus converting the program to ISO
15390 C in one respect. The companion program @code{unprotoize} does the
15391 reverse: it removes argument types from any prototypes that are found.
15393 When you run these programs, you must specify a set of source files as
15394 command line arguments. The conversion programs start out by compiling
15395 these files to see what functions they define. The information gathered
15396 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15398 After scanning comes actual conversion. The specified files are all
15399 eligible to be converted; any files they include (whether sources or
15400 just headers) are eligible as well.
15402 But not all the eligible files are converted. By default,
15403 @code{protoize} and @code{unprotoize} convert only source and header
15404 files in the current directory. You can specify additional directories
15405 whose files should be converted with the @option{-d @var{directory}}
15406 option. You can also specify particular files to exclude with the
15407 @option{-x @var{file}} option. A file is converted if it is eligible, its
15408 directory name matches one of the specified directory names, and its
15409 name within the directory has not been excluded.
15411 Basic conversion with @code{protoize} consists of rewriting most
15412 function definitions and function declarations to specify the types of
15413 the arguments. The only ones not rewritten are those for varargs
15416 @code{protoize} optionally inserts prototype declarations at the
15417 beginning of the source file, to make them available for any calls that
15418 precede the function's definition. Or it can insert prototype
15419 declarations with block scope in the blocks where undeclared functions
15422 Basic conversion with @code{unprotoize} consists of rewriting most
15423 function declarations to remove any argument types, and rewriting
15424 function definitions to the old-style pre-ISO form.
15426 Both conversion programs print a warning for any function declaration or
15427 definition that they can't convert. You can suppress these warnings
15430 The output from @code{protoize} or @code{unprotoize} replaces the
15431 original source file. The original file is renamed to a name ending
15432 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15433 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15434 for DOS) file already exists, then the source file is simply discarded.
15436 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15437 scan the program and collect information about the functions it uses.
15438 So neither of these programs will work until GCC is installed.
15440 Here is a table of the options you can use with @code{protoize} and
15441 @code{unprotoize}. Each option works with both programs unless
15445 @item -B @var{directory}
15446 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15447 usual directory (normally @file{/usr/local/lib}). This file contains
15448 prototype information about standard system functions. This option
15449 applies only to @code{protoize}.
15451 @item -c @var{compilation-options}
15452 Use @var{compilation-options} as the options when running @command{gcc} to
15453 produce the @samp{.X} files. The special option @option{-aux-info} is
15454 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15456 Note that the compilation options must be given as a single argument to
15457 @code{protoize} or @code{unprotoize}. If you want to specify several
15458 @command{gcc} options, you must quote the entire set of compilation options
15459 to make them a single word in the shell.
15461 There are certain @command{gcc} arguments that you cannot use, because they
15462 would produce the wrong kind of output. These include @option{-g},
15463 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15464 the @var{compilation-options}, they are ignored.
15467 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15468 systems) instead of @samp{.c}. This is convenient if you are converting
15469 a C program to C++. This option applies only to @code{protoize}.
15472 Add explicit global declarations. This means inserting explicit
15473 declarations at the beginning of each source file for each function
15474 that is called in the file and was not declared. These declarations
15475 precede the first function definition that contains a call to an
15476 undeclared function. This option applies only to @code{protoize}.
15478 @item -i @var{string}
15479 Indent old-style parameter declarations with the string @var{string}.
15480 This option applies only to @code{protoize}.
15482 @code{unprotoize} converts prototyped function definitions to old-style
15483 function definitions, where the arguments are declared between the
15484 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15485 uses five spaces as the indentation. If you want to indent with just
15486 one space instead, use @option{-i " "}.
15489 Keep the @samp{.X} files. Normally, they are deleted after conversion
15493 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15494 a prototype declaration for each function in each block which calls the
15495 function without any declaration. This option applies only to
15499 Make no real changes. This mode just prints information about the conversions
15500 that would have been done without @option{-n}.
15503 Make no @samp{.save} files. The original files are simply deleted.
15504 Use this option with caution.
15506 @item -p @var{program}
15507 Use the program @var{program} as the compiler. Normally, the name
15508 @file{gcc} is used.
15511 Work quietly. Most warnings are suppressed.
15514 Print the version number, just like @option{-v} for @command{gcc}.
15517 If you need special compiler options to compile one of your program's
15518 source files, then you should generate that file's @samp{.X} file
15519 specially, by running @command{gcc} on that source file with the
15520 appropriate options and the option @option{-aux-info}. Then run
15521 @code{protoize} on the entire set of files. @code{protoize} will use
15522 the existing @samp{.X} file because it is newer than the source file.
15526 gcc -Dfoo=bar file1.c -aux-info file1.X
15531 You need to include the special files along with the rest in the
15532 @code{protoize} command, even though their @samp{.X} files already
15533 exist, because otherwise they won't get converted.
15535 @xref{Protoize Caveats}, for more information on how to use
15536 @code{protoize} successfully.