1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
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,
15 2007 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 @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 -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 -fforce-addr -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-pta -fipa-pure-const -fipa-reference @gol
332 -fipa-type-escape -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
333 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
334 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
335 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
336 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
337 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
338 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
339 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
340 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
341 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
342 -fprofile-generate -fprofile-use -fprofile-values -freciprocal-math @gol
343 -fregmove -frename-registers -freorder-blocks @gol
344 -freorder-blocks-and-partition -freorder-functions @gol
345 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
346 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
347 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
348 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
349 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
350 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
351 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
352 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer -ftree-ccp @gol
353 -ftree-ch -ftree-copy-prop -ftree-copyrename -ftree-dce @gol
354 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
355 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
356 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc -ftree-salias @gol
357 -ftree-sink -ftree-sra -ftree-store-ccp -ftree-ter @gol
358 -ftree-vect-loop-version -ftree-vectorize -ftree-vrp -funit-at-a-time @gol
359 -funroll-all-loops -funroll-loops -funsafe-loop-optimizations @gol
360 -funsafe-math-optimizations -funswitch-loops @gol
361 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
363 --param @var{name}=@var{value}
364 -O -O0 -O1 -O2 -O3 -Os}
366 @item Preprocessor Options
367 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
368 @gccoptlist{-A@var{question}=@var{answer} @gol
369 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
370 -C -dD -dI -dM -dN @gol
371 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
372 -idirafter @var{dir} @gol
373 -include @var{file} -imacros @var{file} @gol
374 -iprefix @var{file} -iwithprefix @var{dir} @gol
375 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
376 -imultilib @var{dir} -isysroot @var{dir} @gol
377 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
378 -P -fworking-directory -remap @gol
379 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
380 -Xpreprocessor @var{option}}
382 @item Assembler Option
383 @xref{Assembler Options,,Passing Options to the Assembler}.
384 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
387 @xref{Link Options,,Options for Linking}.
388 @gccoptlist{@var{object-file-name} -l@var{library} @gol
389 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
390 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
391 -Wl,@var{option} -Xlinker @var{option} @gol
394 @item Directory Options
395 @xref{Directory Options,,Options for Directory Search}.
396 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
397 -specs=@var{file} -I- --sysroot=@var{dir}}
400 @c I wrote this xref this way to avoid overfull hbox. -- rms
401 @xref{Target Options}.
402 @gccoptlist{-V @var{version} -b @var{machine}}
404 @item Machine Dependent Options
405 @xref{Submodel Options,,Hardware Models and Configurations}.
406 @c This list is ordered alphanumerically by subsection name.
407 @c Try and put the significant identifier (CPU or system) first,
408 @c so users have a clue at guessing where the ones they want will be.
411 @gccoptlist{-EB -EL @gol
412 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
413 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
416 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
417 -mabi=@var{name} @gol
418 -mapcs-stack-check -mno-apcs-stack-check @gol
419 -mapcs-float -mno-apcs-float @gol
420 -mapcs-reentrant -mno-apcs-reentrant @gol
421 -msched-prolog -mno-sched-prolog @gol
422 -mlittle-endian -mbig-endian -mwords-little-endian @gol
423 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
424 -mthumb-interwork -mno-thumb-interwork @gol
425 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
426 -mstructure-size-boundary=@var{n} @gol
427 -mabort-on-noreturn @gol
428 -mlong-calls -mno-long-calls @gol
429 -msingle-pic-base -mno-single-pic-base @gol
430 -mpic-register=@var{reg} @gol
431 -mnop-fun-dllimport @gol
432 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
433 -mpoke-function-name @gol
435 -mtpcs-frame -mtpcs-leaf-frame @gol
436 -mcaller-super-interworking -mcallee-super-interworking @gol
440 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
441 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
443 @emph{Blackfin Options}
444 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
445 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
446 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
447 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
448 -mno-id-shared-library -mshared-library-id=@var{n} @gol
449 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
450 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
451 -mfast-fp -minline-plt}
454 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
455 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
456 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
457 -mstack-align -mdata-align -mconst-align @gol
458 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
459 -melf -maout -melinux -mlinux -sim -sim2 @gol
460 -mmul-bug-workaround -mno-mul-bug-workaround}
463 @gccoptlist{-mmac -mpush-args}
465 @emph{Darwin Options}
466 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
467 -arch_only -bind_at_load -bundle -bundle_loader @gol
468 -client_name -compatibility_version -current_version @gol
470 -dependency-file -dylib_file -dylinker_install_name @gol
471 -dynamic -dynamiclib -exported_symbols_list @gol
472 -filelist -flat_namespace -force_cpusubtype_ALL @gol
473 -force_flat_namespace -headerpad_max_install_names @gol
475 -image_base -init -install_name -keep_private_externs @gol
476 -multi_module -multiply_defined -multiply_defined_unused @gol
477 -noall_load -no_dead_strip_inits_and_terms @gol
478 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
479 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
480 -private_bundle -read_only_relocs -sectalign @gol
481 -sectobjectsymbols -whyload -seg1addr @gol
482 -sectcreate -sectobjectsymbols -sectorder @gol
483 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
484 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
485 -segprot -segs_read_only_addr -segs_read_write_addr @gol
486 -single_module -static -sub_library -sub_umbrella @gol
487 -twolevel_namespace -umbrella -undefined @gol
488 -unexported_symbols_list -weak_reference_mismatches @gol
489 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
490 -mkernel -mone-byte-bool}
492 @emph{DEC Alpha Options}
493 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
494 -mieee -mieee-with-inexact -mieee-conformant @gol
495 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
496 -mtrap-precision=@var{mode} -mbuild-constants @gol
497 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
498 -mbwx -mmax -mfix -mcix @gol
499 -mfloat-vax -mfloat-ieee @gol
500 -mexplicit-relocs -msmall-data -mlarge-data @gol
501 -msmall-text -mlarge-text @gol
502 -mmemory-latency=@var{time}}
504 @emph{DEC Alpha/VMS Options}
505 @gccoptlist{-mvms-return-codes}
508 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
509 -mhard-float -msoft-float @gol
510 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
511 -mdouble -mno-double @gol
512 -mmedia -mno-media -mmuladd -mno-muladd @gol
513 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
514 -mlinked-fp -mlong-calls -malign-labels @gol
515 -mlibrary-pic -macc-4 -macc-8 @gol
516 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
517 -moptimize-membar -mno-optimize-membar @gol
518 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
519 -mvliw-branch -mno-vliw-branch @gol
520 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
521 -mno-nested-cond-exec -mtomcat-stats @gol
525 @emph{GNU/Linux Options}
526 @gccoptlist{-muclibc}
528 @emph{H8/300 Options}
529 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
532 @gccoptlist{-march=@var{architecture-type} @gol
533 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
534 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
535 -mfixed-range=@var{register-range} @gol
536 -mjump-in-delay -mlinker-opt -mlong-calls @gol
537 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
538 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
539 -mno-jump-in-delay -mno-long-load-store @gol
540 -mno-portable-runtime -mno-soft-float @gol
541 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
542 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
543 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
544 -munix=@var{unix-std} -nolibdld -static -threads}
546 @emph{i386 and x86-64 Options}
547 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
548 -mfpmath=@var{unit} @gol
549 -masm=@var{dialect} -mno-fancy-math-387 @gol
550 -mno-fp-ret-in-387 -msoft-float @gol
551 -mno-wide-multiply -mrtd -malign-double @gol
552 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
553 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
554 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
555 -mthreads -mno-align-stringops -minline-all-stringops @gol
556 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
557 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
558 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
559 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
560 -mcmodel=@var{code-model} @gol
561 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
562 -mfused-madd -mno-fused-madd}
565 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
566 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
567 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
568 -minline-float-divide-max-throughput @gol
569 -minline-int-divide-min-latency @gol
570 -minline-int-divide-max-throughput @gol
571 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
572 -mno-dwarf2-asm -mearly-stop-bits @gol
573 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
574 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
575 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
576 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
577 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
578 -mno-sched-prefer-non-data-spec-insns @gol
579 -mno-sched-prefer-non-control-spec-insns @gol
580 -mno-sched-count-spec-in-critical-path}
582 @emph{M32R/D Options}
583 @gccoptlist{-m32r2 -m32rx -m32r @gol
585 -malign-loops -mno-align-loops @gol
586 -missue-rate=@var{number} @gol
587 -mbranch-cost=@var{number} @gol
588 -mmodel=@var{code-size-model-type} @gol
589 -msdata=@var{sdata-type} @gol
590 -mno-flush-func -mflush-func=@var{name} @gol
591 -mno-flush-trap -mflush-trap=@var{number} @gol
595 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
597 @emph{M680x0 Options}
598 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
599 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
600 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
601 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
602 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
603 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
604 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
605 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
607 @emph{M68hc1x Options}
608 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
609 -mauto-incdec -minmax -mlong-calls -mshort @gol
610 -msoft-reg-count=@var{count}}
613 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
614 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
615 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
616 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
617 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
620 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
621 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
622 -mips16 -mno-mips16 -mflip-mips16 @gol
623 -minterlink-mips16 -mno-interlink-mips16 @gol
624 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
625 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
626 -mfp32 -mfp64 -mhard-float -msoft-float @gol
627 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
628 -msmartmips -mno-smartmips @gol
629 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
630 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
631 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
632 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
633 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
634 -membedded-data -mno-embedded-data @gol
635 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
636 -mcode-readable=@var{setting} @gol
637 -msplit-addresses -mno-split-addresses @gol
638 -mexplicit-relocs -mno-explicit-relocs @gol
639 -mcheck-zero-division -mno-check-zero-division @gol
640 -mdivide-traps -mdivide-breaks @gol
641 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
642 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
643 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
644 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
645 -mfix-sb1 -mno-fix-sb1 @gol
646 -mflush-func=@var{func} -mno-flush-func @gol
647 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
648 -mfp-exceptions -mno-fp-exceptions @gol
649 -mvr4130-align -mno-vr4130-align}
652 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
653 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
654 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
655 -mno-base-addresses -msingle-exit -mno-single-exit}
657 @emph{MN10300 Options}
658 @gccoptlist{-mmult-bug -mno-mult-bug @gol
659 -mam33 -mno-am33 @gol
660 -mam33-2 -mno-am33-2 @gol
661 -mreturn-pointer-on-d0 @gol
665 @gccoptlist{-mno-crt0 -mbacc -msim @gol
666 -march=@var{cpu-type} }
668 @emph{PDP-11 Options}
669 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
670 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
671 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
672 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
673 -mbranch-expensive -mbranch-cheap @gol
674 -msplit -mno-split -munix-asm -mdec-asm}
676 @emph{PowerPC Options}
677 See RS/6000 and PowerPC Options.
679 @emph{RS/6000 and PowerPC Options}
680 @gccoptlist{-mcpu=@var{cpu-type} @gol
681 -mtune=@var{cpu-type} @gol
682 -mpower -mno-power -mpower2 -mno-power2 @gol
683 -mpowerpc -mpowerpc64 -mno-powerpc @gol
684 -maltivec -mno-altivec @gol
685 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
686 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
687 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
688 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
689 -mnew-mnemonics -mold-mnemonics @gol
690 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
691 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
692 -malign-power -malign-natural @gol
693 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
694 -mstring -mno-string -mupdate -mno-update @gol
695 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
696 -mstrict-align -mno-strict-align -mrelocatable @gol
697 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
698 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
699 -mdynamic-no-pic -maltivec -mswdiv @gol
700 -mprioritize-restricted-insns=@var{priority} @gol
701 -msched-costly-dep=@var{dependence_type} @gol
702 -minsert-sched-nops=@var{scheme} @gol
703 -mcall-sysv -mcall-netbsd @gol
704 -maix-struct-return -msvr4-struct-return @gol
705 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
706 -misel -mno-isel @gol
707 -misel=yes -misel=no @gol
709 -mspe=yes -mspe=no @gol
711 -mvrsave -mno-vrsave @gol
712 -mmulhw -mno-mulhw @gol
713 -mdlmzb -mno-dlmzb @gol
714 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
715 -mprototype -mno-prototype @gol
716 -msim -mmvme -mads -myellowknife -memb -msdata @gol
717 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
719 @emph{S/390 and zSeries Options}
720 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
721 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
722 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
723 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
724 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
725 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
726 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
729 @gccoptlist{-meb -mel @gol
733 -mscore5 -mscore5u -mscore7 -mscore7d}
736 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
737 -m4-nofpu -m4-single-only -m4-single -m4 @gol
738 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
739 -m5-64media -m5-64media-nofpu @gol
740 -m5-32media -m5-32media-nofpu @gol
741 -m5-compact -m5-compact-nofpu @gol
742 -mb -ml -mdalign -mrelax @gol
743 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
744 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
745 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
746 -mdivsi3_libfunc=@var{name} @gol
747 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
751 @gccoptlist{-mcpu=@var{cpu-type} @gol
752 -mtune=@var{cpu-type} @gol
753 -mcmodel=@var{code-model} @gol
754 -m32 -m64 -mapp-regs -mno-app-regs @gol
755 -mfaster-structs -mno-faster-structs @gol
756 -mfpu -mno-fpu -mhard-float -msoft-float @gol
757 -mhard-quad-float -msoft-quad-float @gol
758 -mimpure-text -mno-impure-text -mlittle-endian @gol
759 -mstack-bias -mno-stack-bias @gol
760 -munaligned-doubles -mno-unaligned-doubles @gol
761 -mv8plus -mno-v8plus -mvis -mno-vis
762 -threads -pthreads -pthread}
765 @gccoptlist{-mwarn-reloc -merror-reloc @gol
766 -msafe-dma -munsafe-dma @gol
768 -msmall-mem -mlarge-mem -mstdmain @gol
769 -mfixed-range=@var{register-range}}
771 @emph{System V Options}
772 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
774 @emph{TMS320C3x/C4x Options}
775 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
776 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
777 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
778 -mparallel-insns -mparallel-mpy -mpreserve-float}
781 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
782 -mprolog-function -mno-prolog-function -mspace @gol
783 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
784 -mapp-regs -mno-app-regs @gol
785 -mdisable-callt -mno-disable-callt @gol
791 @gccoptlist{-mg -mgnu -munix}
793 @emph{VxWorks Options}
794 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
795 -Xbind-lazy -Xbind-now}
797 @emph{x86-64 Options}
798 See i386 and x86-64 Options.
800 @emph{Xstormy16 Options}
803 @emph{Xtensa Options}
804 @gccoptlist{-mconst16 -mno-const16 @gol
805 -mfused-madd -mno-fused-madd @gol
806 -mtext-section-literals -mno-text-section-literals @gol
807 -mtarget-align -mno-target-align @gol
808 -mlongcalls -mno-longcalls}
810 @emph{zSeries Options}
811 See S/390 and zSeries Options.
813 @item Code Generation Options
814 @xref{Code Gen Options,,Options for Code Generation Conventions}.
815 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
816 -ffixed-@var{reg} -fexceptions @gol
817 -fnon-call-exceptions -funwind-tables @gol
818 -fasynchronous-unwind-tables @gol
819 -finhibit-size-directive -finstrument-functions @gol
820 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
821 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
822 -fno-common -fno-ident @gol
823 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
824 -fno-jump-tables @gol
825 -frecord-gcc-switches @gol
826 -freg-struct-return -fshort-enums @gol
827 -fshort-double -fshort-wchar @gol
828 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
829 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
830 -fno-stack-limit -fargument-alias -fargument-noalias @gol
831 -fargument-noalias-global -fargument-noalias-anything @gol
832 -fleading-underscore -ftls-model=@var{model} @gol
833 -ftrapv -fwrapv -fbounds-check @gol
838 * Overall Options:: Controlling the kind of output:
839 an executable, object files, assembler files,
840 or preprocessed source.
841 * C Dialect Options:: Controlling the variant of C language compiled.
842 * C++ Dialect Options:: Variations on C++.
843 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
845 * Language Independent Options:: Controlling how diagnostics should be
847 * Warning Options:: How picky should the compiler be?
848 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
849 * Optimize Options:: How much optimization?
850 * Preprocessor Options:: Controlling header files and macro definitions.
851 Also, getting dependency information for Make.
852 * Assembler Options:: Passing options to the assembler.
853 * Link Options:: Specifying libraries and so on.
854 * Directory Options:: Where to find header files and libraries.
855 Where to find the compiler executable files.
856 * Spec Files:: How to pass switches to sub-processes.
857 * Target Options:: Running a cross-compiler, or an old version of GCC.
860 @node Overall Options
861 @section Options Controlling the Kind of Output
863 Compilation can involve up to four stages: preprocessing, compilation
864 proper, assembly and linking, always in that order. GCC is capable of
865 preprocessing and compiling several files either into several
866 assembler input files, or into one assembler input file; then each
867 assembler input file produces an object file, and linking combines all
868 the object files (those newly compiled, and those specified as input)
869 into an executable file.
871 @cindex file name suffix
872 For any given input file, the file name suffix determines what kind of
877 C source code which must be preprocessed.
880 C source code which should not be preprocessed.
883 C++ source code which should not be preprocessed.
886 Objective-C source code. Note that you must link with the @file{libobjc}
887 library to make an Objective-C program work.
890 Objective-C source code which should not be preprocessed.
894 Objective-C++ source code. Note that you must link with the @file{libobjc}
895 library to make an Objective-C++ program work. Note that @samp{.M} refers
896 to a literal capital M@.
899 Objective-C++ source code which should not be preprocessed.
902 C, C++, Objective-C or Objective-C++ header file to be turned into a
907 @itemx @var{file}.cxx
908 @itemx @var{file}.cpp
909 @itemx @var{file}.CPP
910 @itemx @var{file}.c++
912 C++ source code which must be preprocessed. Note that in @samp{.cxx},
913 the last two letters must both be literally @samp{x}. Likewise,
914 @samp{.C} refers to a literal capital C@.
918 Objective-C++ source code which must be preprocessed.
921 Objective-C++ source code which should not be preprocessed.
926 @itemx @var{file}.hxx
927 @itemx @var{file}.hpp
928 @itemx @var{file}.HPP
929 @itemx @var{file}.h++
930 @itemx @var{file}.tcc
931 C++ header file to be turned into a precompiled header.
934 @itemx @var{file}.for
935 @itemx @var{file}.FOR
936 Fixed form Fortran source code which should not be preprocessed.
939 @itemx @var{file}.fpp
940 @itemx @var{file}.FPP
941 Fixed form Fortran source code which must be preprocessed (with the traditional
945 @itemx @var{file}.f95
946 Free form Fortran source code which should not be preprocessed.
949 @itemx @var{file}.F95
950 Free form Fortran source code which must be preprocessed (with the
951 traditional preprocessor).
953 @c FIXME: Descriptions of Java file types.
960 Ada source code file which contains a library unit declaration (a
961 declaration of a package, subprogram, or generic, or a generic
962 instantiation), or a library unit renaming declaration (a package,
963 generic, or subprogram renaming declaration). Such files are also
966 @itemx @var{file}.adb
967 Ada source code file containing a library unit body (a subprogram or
968 package body). Such files are also called @dfn{bodies}.
970 @c GCC also knows about some suffixes for languages not yet included:
982 Assembler code which must be preprocessed.
985 An object file to be fed straight into linking.
986 Any file name with no recognized suffix is treated this way.
990 You can specify the input language explicitly with the @option{-x} option:
993 @item -x @var{language}
994 Specify explicitly the @var{language} for the following input files
995 (rather than letting the compiler choose a default based on the file
996 name suffix). This option applies to all following input files until
997 the next @option{-x} option. Possible values for @var{language} are:
999 c c-header c-cpp-output
1000 c++ c++-header c++-cpp-output
1001 objective-c objective-c-header objective-c-cpp-output
1002 objective-c++ objective-c++-header objective-c++-cpp-output
1003 assembler assembler-with-cpp
1011 Turn off any specification of a language, so that subsequent files are
1012 handled according to their file name suffixes (as they are if @option{-x}
1013 has not been used at all).
1015 @item -pass-exit-codes
1016 @opindex pass-exit-codes
1017 Normally the @command{gcc} program will exit with the code of 1 if any
1018 phase of the compiler returns a non-success return code. If you specify
1019 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1020 numerically highest error produced by any phase that returned an error
1021 indication. The C, C++, and Fortran frontends return 4, if an internal
1022 compiler error is encountered.
1025 If you only want some of the stages of compilation, you can use
1026 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1027 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1028 @command{gcc} is to stop. Note that some combinations (for example,
1029 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1034 Compile or assemble the source files, but do not link. The linking
1035 stage simply is not done. The ultimate output is in the form of an
1036 object file for each source file.
1038 By default, the object file name for a source file is made by replacing
1039 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1041 Unrecognized input files, not requiring compilation or assembly, are
1046 Stop after the stage of compilation proper; do not assemble. The output
1047 is in the form of an assembler code file for each non-assembler input
1050 By default, the assembler file name for a source file is made by
1051 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1053 Input files that don't require compilation are ignored.
1057 Stop after the preprocessing stage; do not run the compiler proper. The
1058 output is in the form of preprocessed source code, which is sent to the
1061 Input files which don't require preprocessing are ignored.
1063 @cindex output file option
1066 Place output in file @var{file}. This applies regardless to whatever
1067 sort of output is being produced, whether it be an executable file,
1068 an object file, an assembler file or preprocessed C code.
1070 If @option{-o} is not specified, the default is to put an executable
1071 file in @file{a.out}, the object file for
1072 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1073 assembler file in @file{@var{source}.s}, a precompiled header file in
1074 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1079 Print (on standard error output) the commands executed to run the stages
1080 of compilation. Also print the version number of the compiler driver
1081 program and of the preprocessor and the compiler proper.
1085 Like @option{-v} except the commands are not executed and all command
1086 arguments are quoted. This is useful for shell scripts to capture the
1087 driver-generated command lines.
1091 Use pipes rather than temporary files for communication between the
1092 various stages of compilation. This fails to work on some systems where
1093 the assembler is unable to read from a pipe; but the GNU assembler has
1098 If you are compiling multiple source files, this option tells the driver
1099 to pass all the source files to the compiler at once (for those
1100 languages for which the compiler can handle this). This will allow
1101 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1102 language for which this is supported is C@. If you pass source files for
1103 multiple languages to the driver, using this option, the driver will invoke
1104 the compiler(s) that support IMA once each, passing each compiler all the
1105 source files appropriate for it. For those languages that do not support
1106 IMA this option will be ignored, and the compiler will be invoked once for
1107 each source file in that language. If you use this option in conjunction
1108 with @option{-save-temps}, the compiler will generate multiple
1110 (one for each source file), but only one (combined) @file{.o} or
1115 Print (on the standard output) a description of the command line options
1116 understood by @command{gcc}. If the @option{-v} option is also specified
1117 then @option{--help} will also be passed on to the various processes
1118 invoked by @command{gcc}, so that they can display the command line options
1119 they accept. If the @option{-Wextra} option has also been specified
1120 (prior to the @option{--help} option), then command line options which
1121 have no documentation associated with them will also be displayed.
1124 @opindex target-help
1125 Print (on the standard output) a description of target-specific command
1126 line options for each tool. For some targets extra target-specific
1127 information may also be printed.
1129 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1130 Print (on the standard output) a description of the command line
1131 options understood by the compiler that fit into a specific class.
1132 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1133 @samp{params}, or @var{language}:
1136 @item @samp{optimizers}
1137 This will display all of the optimization options supported by the
1140 @item @samp{warnings}
1141 This will display all of the options controlling warning messages
1142 produced by the compiler.
1145 This will display target-specific options. Unlike the
1146 @option{--target-help} option however, target-specific options of the
1147 linker and assembler will not be displayed. This is because those
1148 tools do not currently support the extended @option{--help=} syntax.
1151 This will display the values recognized by the @option{--param}
1154 @item @var{language}
1155 This will display the options supported for @var{language}, where
1156 @var{language} is the name of one of the languages supported in this
1160 This will display the options that are common to all languages.
1163 It is possible to further refine the output of the @option{--help=}
1164 option by adding a comma separated list of qualifiers after the
1165 class. These can be any from the following list:
1168 @item @samp{undocumented}
1169 Display only those options which are undocumented.
1172 Display options which take an argument that appears after an equal
1173 sign in the same continuous piece of text, such as:
1174 @samp{--help=target}.
1176 @item @samp{separate}
1177 Display options which take an argument that appears as a separate word
1178 following the original option, such as: @samp{-o output-file}.
1181 Thus for example to display all the undocumented target-specific
1182 switches supported by the compiler the following can be used:
1185 --help=target,undocumented
1188 The sense of a qualifier can be inverted by prefixing it with the
1189 @var{^} character, so for example to display all binary warning
1190 options (i.e. ones that are either on or off and that do not take an
1191 argument), which have a description the following can be used:
1194 --help=warnings,^joined,^undocumented
1197 A class can also be used as a qualifier, although this usually
1198 restricts the output by so much that there is nothing to display. One
1199 case where it does work however is when one of the classes is
1200 @var{target}. So for example to display all the target-specific
1201 optimization options the following can be used:
1204 --help=target,optimizers
1207 The @option{--help=} option can be repeated on the command line. Each
1208 successive use will display its requested class of options, skipping
1209 those that have already been displayed.
1211 If the @option{-Q} option appears on the command line before the
1212 @option{--help=} option, then the descriptive text displayed by
1213 @option{--help=} is changed. Instead of describing the displayed
1214 options, an indication is given as to whether the option is enabled,
1215 disabled or set to a specific value (assuming that the compiler
1216 knows this at the point where the @option{--help=} option is used).
1218 Here is a truncated example from the ARM port of @command{gcc}:
1221 % gcc -Q -mabi=2 --help=target -c
1222 The following options are target specific:
1224 -mabort-on-noreturn [disabled]
1228 The output is sensitive to the effects of previous command line
1229 options, so for example it is possible to find out which optimizations
1230 are enabled at @option{-O2} by using:
1233 -O2 --help=optimizers
1236 Alternatively you can discover which binary optimizations are enabled
1237 by @option{-O3} by using:
1240 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1241 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1242 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1247 Display the version number and copyrights of the invoked GCC@.
1249 @include @value{srcdir}/../libiberty/at-file.texi
1253 @section Compiling C++ Programs
1255 @cindex suffixes for C++ source
1256 @cindex C++ source file suffixes
1257 C++ source files conventionally use one of the suffixes @samp{.C},
1258 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1259 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1260 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1261 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1262 files with these names and compiles them as C++ programs even if you
1263 call the compiler the same way as for compiling C programs (usually
1264 with the name @command{gcc}).
1268 However, the use of @command{gcc} does not add the C++ library.
1269 @command{g++} is a program that calls GCC and treats @samp{.c},
1270 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1271 files unless @option{-x} is used, and automatically specifies linking
1272 against the C++ library. This program is also useful when
1273 precompiling a C header file with a @samp{.h} extension for use in C++
1274 compilations. On many systems, @command{g++} is also installed with
1275 the name @command{c++}.
1277 @cindex invoking @command{g++}
1278 When you compile C++ programs, you may specify many of the same
1279 command-line options that you use for compiling programs in any
1280 language; or command-line options meaningful for C and related
1281 languages; or options that are meaningful only for C++ programs.
1282 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1283 explanations of options for languages related to C@.
1284 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1285 explanations of options that are meaningful only for C++ programs.
1287 @node C Dialect Options
1288 @section Options Controlling C Dialect
1289 @cindex dialect options
1290 @cindex language dialect options
1291 @cindex options, dialect
1293 The following options control the dialect of C (or languages derived
1294 from C, such as C++, Objective-C and Objective-C++) that the compiler
1298 @cindex ANSI support
1302 In C mode, support all ISO C90 programs. In C++ mode,
1303 remove GNU extensions that conflict with ISO C++.
1305 This turns off certain features of GCC that are incompatible with ISO
1306 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1307 such as the @code{asm} and @code{typeof} keywords, and
1308 predefined macros such as @code{unix} and @code{vax} that identify the
1309 type of system you are using. It also enables the undesirable and
1310 rarely used ISO trigraph feature. For the C compiler,
1311 it disables recognition of C++ style @samp{//} comments as well as
1312 the @code{inline} keyword.
1314 The alternate keywords @code{__asm__}, @code{__extension__},
1315 @code{__inline__} and @code{__typeof__} continue to work despite
1316 @option{-ansi}. You would not want to use them in an ISO C program, of
1317 course, but it is useful to put them in header files that might be included
1318 in compilations done with @option{-ansi}. Alternate predefined macros
1319 such as @code{__unix__} and @code{__vax__} are also available, with or
1320 without @option{-ansi}.
1322 The @option{-ansi} option does not cause non-ISO programs to be
1323 rejected gratuitously. For that, @option{-pedantic} is required in
1324 addition to @option{-ansi}. @xref{Warning Options}.
1326 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1327 option is used. Some header files may notice this macro and refrain
1328 from declaring certain functions or defining certain macros that the
1329 ISO standard doesn't call for; this is to avoid interfering with any
1330 programs that might use these names for other things.
1332 Functions which would normally be built in but do not have semantics
1333 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1334 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1335 built-in functions provided by GCC}, for details of the functions
1340 Determine the language standard. This option is currently only
1341 supported when compiling C or C++. A value for this option must be
1342 provided; possible values are
1347 ISO C90 (same as @option{-ansi}).
1349 @item iso9899:199409
1350 ISO C90 as modified in amendment 1.
1356 ISO C99. Note that this standard is not yet fully supported; see
1357 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1358 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1361 Default, ISO C90 plus GNU extensions (including some C99 features).
1365 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1366 this will become the default. The name @samp{gnu9x} is deprecated.
1369 The 1998 ISO C++ standard plus amendments.
1372 The same as @option{-std=c++98} plus GNU extensions. This is the
1373 default for C++ code.
1376 The working draft of the upcoming ISO C++0x standard. This option
1377 enables experimental features that are likely to be included in
1378 C++0x. The working draft is constantly changing, and any feature that is
1379 enabled by this flag may be removed from future versions of GCC if it is
1380 not part of the C++0x standard.
1383 The same as @option{-std=c++0x} plus GNU extensions. As with
1384 @option{-std=c++0x}, this option enables experimental features that may
1385 be removed in future versions of GCC.
1388 Even when this option is not specified, you can still use some of the
1389 features of newer standards in so far as they do not conflict with
1390 previous C standards. For example, you may use @code{__restrict__} even
1391 when @option{-std=c99} is not specified.
1393 The @option{-std} options specifying some version of ISO C have the same
1394 effects as @option{-ansi}, except that features that were not in ISO C90
1395 but are in the specified version (for example, @samp{//} comments and
1396 the @code{inline} keyword in ISO C99) are not disabled.
1398 @xref{Standards,,Language Standards Supported by GCC}, for details of
1399 these standard versions.
1401 @item -fgnu89-inline
1402 @opindex fgnu89-inline
1403 The option @option{-fgnu89-inline} tells GCC to use the traditional
1404 GNU semantics for @code{inline} functions when in C99 mode.
1405 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1406 is accepted and ignored by GCC versions 4.1.3 up to but not including
1407 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1408 C99 mode. Using this option is roughly equivalent to adding the
1409 @code{gnu_inline} function attribute to all inline functions
1410 (@pxref{Function Attributes}).
1412 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1413 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1414 specifies the default behavior). This option was first supported in
1415 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1417 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1418 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1419 in effect for @code{inline} functions. @xref{Common Predefined
1420 Macros,,,cpp,The C Preprocessor}.
1422 @item -aux-info @var{filename}
1424 Output to the given filename prototyped declarations for all functions
1425 declared and/or defined in a translation unit, including those in header
1426 files. This option is silently ignored in any language other than C@.
1428 Besides declarations, the file indicates, in comments, the origin of
1429 each declaration (source file and line), whether the declaration was
1430 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1431 @samp{O} for old, respectively, in the first character after the line
1432 number and the colon), and whether it came from a declaration or a
1433 definition (@samp{C} or @samp{F}, respectively, in the following
1434 character). In the case of function definitions, a K&R-style list of
1435 arguments followed by their declarations is also provided, inside
1436 comments, after the declaration.
1440 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1441 keyword, so that code can use these words as identifiers. You can use
1442 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1443 instead. @option{-ansi} implies @option{-fno-asm}.
1445 In C++, this switch only affects the @code{typeof} keyword, since
1446 @code{asm} and @code{inline} are standard keywords. You may want to
1447 use the @option{-fno-gnu-keywords} flag instead, which has the same
1448 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1449 switch only affects the @code{asm} and @code{typeof} keywords, since
1450 @code{inline} is a standard keyword in ISO C99.
1453 @itemx -fno-builtin-@var{function}
1454 @opindex fno-builtin
1455 @cindex built-in functions
1456 Don't recognize built-in functions that do not begin with
1457 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1458 functions provided by GCC}, for details of the functions affected,
1459 including those which are not built-in functions when @option{-ansi} or
1460 @option{-std} options for strict ISO C conformance are used because they
1461 do not have an ISO standard meaning.
1463 GCC normally generates special code to handle certain built-in functions
1464 more efficiently; for instance, calls to @code{alloca} may become single
1465 instructions that adjust the stack directly, and calls to @code{memcpy}
1466 may become inline copy loops. The resulting code is often both smaller
1467 and faster, but since the function calls no longer appear as such, you
1468 cannot set a breakpoint on those calls, nor can you change the behavior
1469 of the functions by linking with a different library. In addition,
1470 when a function is recognized as a built-in function, GCC may use
1471 information about that function to warn about problems with calls to
1472 that function, or to generate more efficient code, even if the
1473 resulting code still contains calls to that function. For example,
1474 warnings are given with @option{-Wformat} for bad calls to
1475 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1476 known not to modify global memory.
1478 With the @option{-fno-builtin-@var{function}} option
1479 only the built-in function @var{function} is
1480 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1481 function is named this is not built-in in this version of GCC, this
1482 option is ignored. There is no corresponding
1483 @option{-fbuiltin-@var{function}} option; if you wish to enable
1484 built-in functions selectively when using @option{-fno-builtin} or
1485 @option{-ffreestanding}, you may define macros such as:
1488 #define abs(n) __builtin_abs ((n))
1489 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1494 @cindex hosted environment
1496 Assert that compilation takes place in a hosted environment. This implies
1497 @option{-fbuiltin}. A hosted environment is one in which the
1498 entire standard library is available, and in which @code{main} has a return
1499 type of @code{int}. Examples are nearly everything except a kernel.
1500 This is equivalent to @option{-fno-freestanding}.
1502 @item -ffreestanding
1503 @opindex ffreestanding
1504 @cindex hosted environment
1506 Assert that compilation takes place in a freestanding environment. This
1507 implies @option{-fno-builtin}. A freestanding environment
1508 is one in which the standard library may not exist, and program startup may
1509 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1510 This is equivalent to @option{-fno-hosted}.
1512 @xref{Standards,,Language Standards Supported by GCC}, for details of
1513 freestanding and hosted environments.
1517 @cindex openmp parallel
1518 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1519 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1520 compiler generates parallel code according to the OpenMP Application
1521 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1523 @item -fms-extensions
1524 @opindex fms-extensions
1525 Accept some non-standard constructs used in Microsoft header files.
1527 Some cases of unnamed fields in structures and unions are only
1528 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1529 fields within structs/unions}, for details.
1533 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1534 options for strict ISO C conformance) implies @option{-trigraphs}.
1536 @item -no-integrated-cpp
1537 @opindex no-integrated-cpp
1538 Performs a compilation in two passes: preprocessing and compiling. This
1539 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1540 @option{-B} option. The user supplied compilation step can then add in
1541 an additional preprocessing step after normal preprocessing but before
1542 compiling. The default is to use the integrated cpp (internal cpp)
1544 The semantics of this option will change if "cc1", "cc1plus", and
1545 "cc1obj" are merged.
1547 @cindex traditional C language
1548 @cindex C language, traditional
1550 @itemx -traditional-cpp
1551 @opindex traditional-cpp
1552 @opindex traditional
1553 Formerly, these options caused GCC to attempt to emulate a pre-standard
1554 C compiler. They are now only supported with the @option{-E} switch.
1555 The preprocessor continues to support a pre-standard mode. See the GNU
1556 CPP manual for details.
1558 @item -fcond-mismatch
1559 @opindex fcond-mismatch
1560 Allow conditional expressions with mismatched types in the second and
1561 third arguments. The value of such an expression is void. This option
1562 is not supported for C++.
1564 @item -flax-vector-conversions
1565 @opindex flax-vector-conversions
1566 Allow implicit conversions between vectors with differing numbers of
1567 elements and/or incompatible element types. This option should not be
1570 @item -funsigned-char
1571 @opindex funsigned-char
1572 Let the type @code{char} be unsigned, like @code{unsigned char}.
1574 Each kind of machine has a default for what @code{char} should
1575 be. It is either like @code{unsigned char} by default or like
1576 @code{signed char} by default.
1578 Ideally, a portable program should always use @code{signed char} or
1579 @code{unsigned char} when it depends on the signedness of an object.
1580 But many programs have been written to use plain @code{char} and
1581 expect it to be signed, or expect it to be unsigned, depending on the
1582 machines they were written for. This option, and its inverse, let you
1583 make such a program work with the opposite default.
1585 The type @code{char} is always a distinct type from each of
1586 @code{signed char} or @code{unsigned char}, even though its behavior
1587 is always just like one of those two.
1590 @opindex fsigned-char
1591 Let the type @code{char} be signed, like @code{signed char}.
1593 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1594 the negative form of @option{-funsigned-char}. Likewise, the option
1595 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1597 @item -fsigned-bitfields
1598 @itemx -funsigned-bitfields
1599 @itemx -fno-signed-bitfields
1600 @itemx -fno-unsigned-bitfields
1601 @opindex fsigned-bitfields
1602 @opindex funsigned-bitfields
1603 @opindex fno-signed-bitfields
1604 @opindex fno-unsigned-bitfields
1605 These options control whether a bit-field is signed or unsigned, when the
1606 declaration does not use either @code{signed} or @code{unsigned}. By
1607 default, such a bit-field is signed, because this is consistent: the
1608 basic integer types such as @code{int} are signed types.
1611 @node C++ Dialect Options
1612 @section Options Controlling C++ Dialect
1614 @cindex compiler options, C++
1615 @cindex C++ options, command line
1616 @cindex options, C++
1617 This section describes the command-line options that are only meaningful
1618 for C++ programs; but you can also use most of the GNU compiler options
1619 regardless of what language your program is in. For example, you
1620 might compile a file @code{firstClass.C} like this:
1623 g++ -g -frepo -O -c firstClass.C
1627 In this example, only @option{-frepo} is an option meant
1628 only for C++ programs; you can use the other options with any
1629 language supported by GCC@.
1631 Here is a list of options that are @emph{only} for compiling C++ programs:
1635 @item -fabi-version=@var{n}
1636 @opindex fabi-version
1637 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1638 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1639 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1640 the version that conforms most closely to the C++ ABI specification.
1641 Therefore, the ABI obtained using version 0 will change as ABI bugs
1644 The default is version 2.
1646 @item -fno-access-control
1647 @opindex fno-access-control
1648 Turn off all access checking. This switch is mainly useful for working
1649 around bugs in the access control code.
1653 Check that the pointer returned by @code{operator new} is non-null
1654 before attempting to modify the storage allocated. This check is
1655 normally unnecessary because the C++ standard specifies that
1656 @code{operator new} will only return @code{0} if it is declared
1657 @samp{throw()}, in which case the compiler will always check the
1658 return value even without this option. In all other cases, when
1659 @code{operator new} has a non-empty exception specification, memory
1660 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1661 @samp{new (nothrow)}.
1663 @item -fconserve-space
1664 @opindex fconserve-space
1665 Put uninitialized or runtime-initialized global variables into the
1666 common segment, as C does. This saves space in the executable at the
1667 cost of not diagnosing duplicate definitions. If you compile with this
1668 flag and your program mysteriously crashes after @code{main()} has
1669 completed, you may have an object that is being destroyed twice because
1670 two definitions were merged.
1672 This option is no longer useful on most targets, now that support has
1673 been added for putting variables into BSS without making them common.
1675 @item -ffriend-injection
1676 @opindex ffriend-injection
1677 Inject friend functions into the enclosing namespace, so that they are
1678 visible outside the scope of the class in which they are declared.
1679 Friend functions were documented to work this way in the old Annotated
1680 C++ Reference Manual, and versions of G++ before 4.1 always worked
1681 that way. However, in ISO C++ a friend function which is not declared
1682 in an enclosing scope can only be found using argument dependent
1683 lookup. This option causes friends to be injected as they were in
1686 This option is for compatibility, and may be removed in a future
1689 @item -fno-elide-constructors
1690 @opindex fno-elide-constructors
1691 The C++ standard allows an implementation to omit creating a temporary
1692 which is only used to initialize another object of the same type.
1693 Specifying this option disables that optimization, and forces G++ to
1694 call the copy constructor in all cases.
1696 @item -fno-enforce-eh-specs
1697 @opindex fno-enforce-eh-specs
1698 Don't generate code to check for violation of exception specifications
1699 at runtime. This option violates the C++ standard, but may be useful
1700 for reducing code size in production builds, much like defining
1701 @samp{NDEBUG}. This does not give user code permission to throw
1702 exceptions in violation of the exception specifications; the compiler
1703 will still optimize based on the specifications, so throwing an
1704 unexpected exception will result in undefined behavior.
1707 @itemx -fno-for-scope
1709 @opindex fno-for-scope
1710 If @option{-ffor-scope} is specified, the scope of variables declared in
1711 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1712 as specified by the C++ standard.
1713 If @option{-fno-for-scope} is specified, the scope of variables declared in
1714 a @i{for-init-statement} extends to the end of the enclosing scope,
1715 as was the case in old versions of G++, and other (traditional)
1716 implementations of C++.
1718 The default if neither flag is given to follow the standard,
1719 but to allow and give a warning for old-style code that would
1720 otherwise be invalid, or have different behavior.
1722 @item -fno-gnu-keywords
1723 @opindex fno-gnu-keywords
1724 Do not recognize @code{typeof} as a keyword, so that code can use this
1725 word as an identifier. You can use the keyword @code{__typeof__} instead.
1726 @option{-ansi} implies @option{-fno-gnu-keywords}.
1728 @item -fno-implicit-templates
1729 @opindex fno-implicit-templates
1730 Never emit code for non-inline templates which are instantiated
1731 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1732 @xref{Template Instantiation}, for more information.
1734 @item -fno-implicit-inline-templates
1735 @opindex fno-implicit-inline-templates
1736 Don't emit code for implicit instantiations of inline templates, either.
1737 The default is to handle inlines differently so that compiles with and
1738 without optimization will need the same set of explicit instantiations.
1740 @item -fno-implement-inlines
1741 @opindex fno-implement-inlines
1742 To save space, do not emit out-of-line copies of inline functions
1743 controlled by @samp{#pragma implementation}. This will cause linker
1744 errors if these functions are not inlined everywhere they are called.
1746 @item -fms-extensions
1747 @opindex fms-extensions
1748 Disable pedantic warnings about constructs used in MFC, such as implicit
1749 int and getting a pointer to member function via non-standard syntax.
1751 @item -fno-nonansi-builtins
1752 @opindex fno-nonansi-builtins
1753 Disable built-in declarations of functions that are not mandated by
1754 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1755 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1757 @item -fno-operator-names
1758 @opindex fno-operator-names
1759 Do not treat the operator name keywords @code{and}, @code{bitand},
1760 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1761 synonyms as keywords.
1763 @item -fno-optional-diags
1764 @opindex fno-optional-diags
1765 Disable diagnostics that the standard says a compiler does not need to
1766 issue. Currently, the only such diagnostic issued by G++ is the one for
1767 a name having multiple meanings within a class.
1770 @opindex fpermissive
1771 Downgrade some diagnostics about nonconformant code from errors to
1772 warnings. Thus, using @option{-fpermissive} will allow some
1773 nonconforming code to compile.
1777 Enable automatic template instantiation at link time. This option also
1778 implies @option{-fno-implicit-templates}. @xref{Template
1779 Instantiation}, for more information.
1783 Disable generation of information about every class with virtual
1784 functions for use by the C++ runtime type identification features
1785 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1786 of the language, you can save some space by using this flag. Note that
1787 exception handling uses the same information, but it will generate it as
1788 needed. The @samp{dynamic_cast} operator can still be used for casts that
1789 do not require runtime type information, i.e. casts to @code{void *} or to
1790 unambiguous base classes.
1794 Emit statistics about front-end processing at the end of the compilation.
1795 This information is generally only useful to the G++ development team.
1797 @item -ftemplate-depth-@var{n}
1798 @opindex ftemplate-depth
1799 Set the maximum instantiation depth for template classes to @var{n}.
1800 A limit on the template instantiation depth is needed to detect
1801 endless recursions during template class instantiation. ANSI/ISO C++
1802 conforming programs must not rely on a maximum depth greater than 17.
1804 @item -fno-threadsafe-statics
1805 @opindex fno-threadsafe-statics
1806 Do not emit the extra code to use the routines specified in the C++
1807 ABI for thread-safe initialization of local statics. You can use this
1808 option to reduce code size slightly in code that doesn't need to be
1811 @item -fuse-cxa-atexit
1812 @opindex fuse-cxa-atexit
1813 Register destructors for objects with static storage duration with the
1814 @code{__cxa_atexit} function rather than the @code{atexit} function.
1815 This option is required for fully standards-compliant handling of static
1816 destructors, but will only work if your C library supports
1817 @code{__cxa_atexit}.
1819 @item -fno-use-cxa-get-exception-ptr
1820 @opindex fno-use-cxa-get-exception-ptr
1821 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1822 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1823 if the runtime routine is not available.
1825 @item -fvisibility-inlines-hidden
1826 @opindex fvisibility-inlines-hidden
1827 This switch declares that the user does not attempt to compare
1828 pointers to inline methods where the addresses of the two functions
1829 were taken in different shared objects.
1831 The effect of this is that GCC may, effectively, mark inline methods with
1832 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1833 appear in the export table of a DSO and do not require a PLT indirection
1834 when used within the DSO@. Enabling this option can have a dramatic effect
1835 on load and link times of a DSO as it massively reduces the size of the
1836 dynamic export table when the library makes heavy use of templates.
1838 The behavior of this switch is not quite the same as marking the
1839 methods as hidden directly, because it does not affect static variables
1840 local to the function or cause the compiler to deduce that
1841 the function is defined in only one shared object.
1843 You may mark a method as having a visibility explicitly to negate the
1844 effect of the switch for that method. For example, if you do want to
1845 compare pointers to a particular inline method, you might mark it as
1846 having default visibility. Marking the enclosing class with explicit
1847 visibility will have no effect.
1849 Explicitly instantiated inline methods are unaffected by this option
1850 as their linkage might otherwise cross a shared library boundary.
1851 @xref{Template Instantiation}.
1853 @item -fvisibility-ms-compat
1854 @opindex fvisibility-ms-compat
1855 This flag attempts to use visibility settings to make GCC's C++
1856 linkage model compatible with that of Microsoft Visual Studio.
1858 The flag makes these changes to GCC's linkage model:
1862 It sets the default visibility to @code{hidden}, like
1863 @option{-fvisibility=hidden}.
1866 Types, but not their members, are not hidden by default.
1869 The One Definition Rule is relaxed for types without explicit
1870 visibility specifications which are defined in more than one different
1871 shared object: those declarations are permitted if they would have
1872 been permitted when this option was not used.
1875 In new code it is better to use @option{-fvisibility=hidden} and
1876 export those classes which are intended to be externally visible.
1877 Unfortunately it is possible for code to rely, perhaps accidentally,
1878 on the Visual Studio behavior.
1880 Among the consequences of these changes are that static data members
1881 of the same type with the same name but defined in different shared
1882 objects will be different, so changing one will not change the other;
1883 and that pointers to function members defined in different shared
1884 objects may not compare equal. When this flag is given, it is a
1885 violation of the ODR to define types with the same name differently.
1889 Do not use weak symbol support, even if it is provided by the linker.
1890 By default, G++ will use weak symbols if they are available. This
1891 option exists only for testing, and should not be used by end-users;
1892 it will result in inferior code and has no benefits. This option may
1893 be removed in a future release of G++.
1897 Do not search for header files in the standard directories specific to
1898 C++, but do still search the other standard directories. (This option
1899 is used when building the C++ library.)
1902 In addition, these optimization, warning, and code generation options
1903 have meanings only for C++ programs:
1906 @item -fno-default-inline
1907 @opindex fno-default-inline
1908 Do not assume @samp{inline} for functions defined inside a class scope.
1909 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1910 functions will have linkage like inline functions; they just won't be
1913 @item -Wabi @r{(C++ only)}
1915 Warn when G++ generates code that is probably not compatible with the
1916 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1917 all such cases, there are probably some cases that are not warned about,
1918 even though G++ is generating incompatible code. There may also be
1919 cases where warnings are emitted even though the code that is generated
1922 You should rewrite your code to avoid these warnings if you are
1923 concerned about the fact that code generated by G++ may not be binary
1924 compatible with code generated by other compilers.
1926 The known incompatibilities at this point include:
1931 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1932 pack data into the same byte as a base class. For example:
1935 struct A @{ virtual void f(); int f1 : 1; @};
1936 struct B : public A @{ int f2 : 1; @};
1940 In this case, G++ will place @code{B::f2} into the same byte
1941 as@code{A::f1}; other compilers will not. You can avoid this problem
1942 by explicitly padding @code{A} so that its size is a multiple of the
1943 byte size on your platform; that will cause G++ and other compilers to
1944 layout @code{B} identically.
1947 Incorrect handling of tail-padding for virtual bases. G++ does not use
1948 tail padding when laying out virtual bases. For example:
1951 struct A @{ virtual void f(); char c1; @};
1952 struct B @{ B(); char c2; @};
1953 struct C : public A, public virtual B @{@};
1957 In this case, G++ will not place @code{B} into the tail-padding for
1958 @code{A}; other compilers will. You can avoid this problem by
1959 explicitly padding @code{A} so that its size is a multiple of its
1960 alignment (ignoring virtual base classes); that will cause G++ and other
1961 compilers to layout @code{C} identically.
1964 Incorrect handling of bit-fields with declared widths greater than that
1965 of their underlying types, when the bit-fields appear in a union. For
1969 union U @{ int i : 4096; @};
1973 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1974 union too small by the number of bits in an @code{int}.
1977 Empty classes can be placed at incorrect offsets. For example:
1987 struct C : public B, public A @{@};
1991 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1992 it should be placed at offset zero. G++ mistakenly believes that the
1993 @code{A} data member of @code{B} is already at offset zero.
1996 Names of template functions whose types involve @code{typename} or
1997 template template parameters can be mangled incorrectly.
2000 template <typename Q>
2001 void f(typename Q::X) @{@}
2003 template <template <typename> class Q>
2004 void f(typename Q<int>::X) @{@}
2008 Instantiations of these templates may be mangled incorrectly.
2012 @item -Wctor-dtor-privacy @r{(C++ only)}
2013 @opindex Wctor-dtor-privacy
2014 Warn when a class seems unusable because all the constructors or
2015 destructors in that class are private, and it has neither friends nor
2016 public static member functions.
2018 @item -Wnon-virtual-dtor @r{(C++ only)}
2019 @opindex Wnon-virtual-dtor
2020 Warn when a class has virtual functions and accessible non-virtual
2021 destructor, in which case it would be possible but unsafe to delete
2022 an instance of a derived class through a pointer to the base class.
2023 This warning is also enabled if -Weffc++ is specified.
2025 @item -Wreorder @r{(C++ only)}
2027 @cindex reordering, warning
2028 @cindex warning for reordering of member initializers
2029 Warn when the order of member initializers given in the code does not
2030 match the order in which they must be executed. For instance:
2036 A(): j (0), i (1) @{ @}
2040 The compiler will rearrange the member initializers for @samp{i}
2041 and @samp{j} to match the declaration order of the members, emitting
2042 a warning to that effect. This warning is enabled by @option{-Wall}.
2045 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2048 @item -Weffc++ @r{(C++ only)}
2050 Warn about violations of the following style guidelines from Scott Meyers'
2051 @cite{Effective C++} book:
2055 Item 11: Define a copy constructor and an assignment operator for classes
2056 with dynamically allocated memory.
2059 Item 12: Prefer initialization to assignment in constructors.
2062 Item 14: Make destructors virtual in base classes.
2065 Item 15: Have @code{operator=} return a reference to @code{*this}.
2068 Item 23: Don't try to return a reference when you must return an object.
2072 Also warn about violations of the following style guidelines from
2073 Scott Meyers' @cite{More Effective C++} book:
2077 Item 6: Distinguish between prefix and postfix forms of increment and
2078 decrement operators.
2081 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2085 When selecting this option, be aware that the standard library
2086 headers do not obey all of these guidelines; use @samp{grep -v}
2087 to filter out those warnings.
2089 @item -Wno-deprecated @r{(C++ only)}
2090 @opindex Wno-deprecated
2091 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2093 @item -Wstrict-null-sentinel @r{(C++ only)}
2094 @opindex Wstrict-null-sentinel
2095 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2096 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2097 to @code{__null}. Although it is a null pointer constant not a null pointer,
2098 it is guaranteed to of the same size as a pointer. But this use is
2099 not portable across different compilers.
2101 @item -Wno-non-template-friend @r{(C++ only)}
2102 @opindex Wno-non-template-friend
2103 Disable warnings when non-templatized friend functions are declared
2104 within a template. Since the advent of explicit template specification
2105 support in G++, if the name of the friend is an unqualified-id (i.e.,
2106 @samp{friend foo(int)}), the C++ language specification demands that the
2107 friend declare or define an ordinary, nontemplate function. (Section
2108 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2109 could be interpreted as a particular specialization of a templatized
2110 function. Because this non-conforming behavior is no longer the default
2111 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2112 check existing code for potential trouble spots and is on by default.
2113 This new compiler behavior can be turned off with
2114 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2115 but disables the helpful warning.
2117 @item -Wold-style-cast @r{(C++ only)}
2118 @opindex Wold-style-cast
2119 Warn if an old-style (C-style) cast to a non-void type is used within
2120 a C++ program. The new-style casts (@samp{dynamic_cast},
2121 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2122 less vulnerable to unintended effects and much easier to search for.
2124 @item -Woverloaded-virtual @r{(C++ only)}
2125 @opindex Woverloaded-virtual
2126 @cindex overloaded virtual fn, warning
2127 @cindex warning for overloaded virtual fn
2128 Warn when a function declaration hides virtual functions from a
2129 base class. For example, in:
2136 struct B: public A @{
2141 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2149 will fail to compile.
2151 @item -Wno-pmf-conversions @r{(C++ only)}
2152 @opindex Wno-pmf-conversions
2153 Disable the diagnostic for converting a bound pointer to member function
2156 @item -Wsign-promo @r{(C++ only)}
2157 @opindex Wsign-promo
2158 Warn when overload resolution chooses a promotion from unsigned or
2159 enumerated type to a signed type, over a conversion to an unsigned type of
2160 the same size. Previous versions of G++ would try to preserve
2161 unsignedness, but the standard mandates the current behavior.
2166 A& operator = (int);
2176 In this example, G++ will synthesize a default @samp{A& operator =
2177 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2180 @node Objective-C and Objective-C++ Dialect Options
2181 @section Options Controlling Objective-C and Objective-C++ Dialects
2183 @cindex compiler options, Objective-C and Objective-C++
2184 @cindex Objective-C and Objective-C++ options, command line
2185 @cindex options, Objective-C and Objective-C++
2186 (NOTE: This manual does not describe the Objective-C and Objective-C++
2187 languages themselves. See @xref{Standards,,Language Standards
2188 Supported by GCC}, for references.)
2190 This section describes the command-line options that are only meaningful
2191 for Objective-C and Objective-C++ programs, but you can also use most of
2192 the language-independent GNU compiler options.
2193 For example, you might compile a file @code{some_class.m} like this:
2196 gcc -g -fgnu-runtime -O -c some_class.m
2200 In this example, @option{-fgnu-runtime} is an option meant only for
2201 Objective-C and Objective-C++ programs; you can use the other options with
2202 any language supported by GCC@.
2204 Note that since Objective-C is an extension of the C language, Objective-C
2205 compilations may also use options specific to the C front-end (e.g.,
2206 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2207 C++-specific options (e.g., @option{-Wabi}).
2209 Here is a list of options that are @emph{only} for compiling Objective-C
2210 and Objective-C++ programs:
2213 @item -fconstant-string-class=@var{class-name}
2214 @opindex fconstant-string-class
2215 Use @var{class-name} as the name of the class to instantiate for each
2216 literal string specified with the syntax @code{@@"@dots{}"}. The default
2217 class name is @code{NXConstantString} if the GNU runtime is being used, and
2218 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2219 @option{-fconstant-cfstrings} option, if also present, will override the
2220 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2221 to be laid out as constant CoreFoundation strings.
2224 @opindex fgnu-runtime
2225 Generate object code compatible with the standard GNU Objective-C
2226 runtime. This is the default for most types of systems.
2228 @item -fnext-runtime
2229 @opindex fnext-runtime
2230 Generate output compatible with the NeXT runtime. This is the default
2231 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2232 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2235 @item -fno-nil-receivers
2236 @opindex fno-nil-receivers
2237 Assume that all Objective-C message dispatches (e.g.,
2238 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2239 is not @code{nil}. This allows for more efficient entry points in the runtime
2240 to be used. Currently, this option is only available in conjunction with
2241 the NeXT runtime on Mac OS X 10.3 and later.
2243 @item -fobjc-call-cxx-cdtors
2244 @opindex fobjc-call-cxx-cdtors
2245 For each Objective-C class, check if any of its instance variables is a
2246 C++ object with a non-trivial default constructor. If so, synthesize a
2247 special @code{- (id) .cxx_construct} instance method that will run
2248 non-trivial default constructors on any such instance variables, in order,
2249 and then return @code{self}. Similarly, check if any instance variable
2250 is a C++ object with a non-trivial destructor, and if so, synthesize a
2251 special @code{- (void) .cxx_destruct} method that will run
2252 all such default destructors, in reverse order.
2254 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2255 thusly generated will only operate on instance variables declared in the
2256 current Objective-C class, and not those inherited from superclasses. It
2257 is the responsibility of the Objective-C runtime to invoke all such methods
2258 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2259 will be invoked by the runtime immediately after a new object
2260 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2261 be invoked immediately before the runtime deallocates an object instance.
2263 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2264 support for invoking the @code{- (id) .cxx_construct} and
2265 @code{- (void) .cxx_destruct} methods.
2267 @item -fobjc-direct-dispatch
2268 @opindex fobjc-direct-dispatch
2269 Allow fast jumps to the message dispatcher. On Darwin this is
2270 accomplished via the comm page.
2272 @item -fobjc-exceptions
2273 @opindex fobjc-exceptions
2274 Enable syntactic support for structured exception handling in Objective-C,
2275 similar to what is offered by C++ and Java. This option is
2276 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2285 @@catch (AnObjCClass *exc) @{
2292 @@catch (AnotherClass *exc) @{
2295 @@catch (id allOthers) @{
2305 The @code{@@throw} statement may appear anywhere in an Objective-C or
2306 Objective-C++ program; when used inside of a @code{@@catch} block, the
2307 @code{@@throw} may appear without an argument (as shown above), in which case
2308 the object caught by the @code{@@catch} will be rethrown.
2310 Note that only (pointers to) Objective-C objects may be thrown and
2311 caught using this scheme. When an object is thrown, it will be caught
2312 by the nearest @code{@@catch} clause capable of handling objects of that type,
2313 analogously to how @code{catch} blocks work in C++ and Java. A
2314 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2315 any and all Objective-C exceptions not caught by previous @code{@@catch}
2318 The @code{@@finally} clause, if present, will be executed upon exit from the
2319 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2320 regardless of whether any exceptions are thrown, caught or rethrown
2321 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2322 of the @code{finally} clause in Java.
2324 There are several caveats to using the new exception mechanism:
2328 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2329 idioms provided by the @code{NSException} class, the new
2330 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2331 systems, due to additional functionality needed in the (NeXT) Objective-C
2335 As mentioned above, the new exceptions do not support handling
2336 types other than Objective-C objects. Furthermore, when used from
2337 Objective-C++, the Objective-C exception model does not interoperate with C++
2338 exceptions at this time. This means you cannot @code{@@throw} an exception
2339 from Objective-C and @code{catch} it in C++, or vice versa
2340 (i.e., @code{throw @dots{} @@catch}).
2343 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2344 blocks for thread-safe execution:
2347 @@synchronized (ObjCClass *guard) @{
2352 Upon entering the @code{@@synchronized} block, a thread of execution shall
2353 first check whether a lock has been placed on the corresponding @code{guard}
2354 object by another thread. If it has, the current thread shall wait until
2355 the other thread relinquishes its lock. Once @code{guard} becomes available,
2356 the current thread will place its own lock on it, execute the code contained in
2357 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2358 making @code{guard} available to other threads).
2360 Unlike Java, Objective-C does not allow for entire methods to be marked
2361 @code{@@synchronized}. Note that throwing exceptions out of
2362 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2363 to be unlocked properly.
2367 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2369 @item -freplace-objc-classes
2370 @opindex freplace-objc-classes
2371 Emit a special marker instructing @command{ld(1)} not to statically link in
2372 the resulting object file, and allow @command{dyld(1)} to load it in at
2373 run time instead. This is used in conjunction with the Fix-and-Continue
2374 debugging mode, where the object file in question may be recompiled and
2375 dynamically reloaded in the course of program execution, without the need
2376 to restart the program itself. Currently, Fix-and-Continue functionality
2377 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2382 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2383 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2384 compile time) with static class references that get initialized at load time,
2385 which improves run-time performance. Specifying the @option{-fzero-link} flag
2386 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2387 to be retained. This is useful in Zero-Link debugging mode, since it allows
2388 for individual class implementations to be modified during program execution.
2392 Dump interface declarations for all classes seen in the source file to a
2393 file named @file{@var{sourcename}.decl}.
2395 @item -Wassign-intercept
2396 @opindex Wassign-intercept
2397 Warn whenever an Objective-C assignment is being intercepted by the
2401 @opindex Wno-protocol
2402 If a class is declared to implement a protocol, a warning is issued for
2403 every method in the protocol that is not implemented by the class. The
2404 default behavior is to issue a warning for every method not explicitly
2405 implemented in the class, even if a method implementation is inherited
2406 from the superclass. If you use the @option{-Wno-protocol} option, then
2407 methods inherited from the superclass are considered to be implemented,
2408 and no warning is issued for them.
2412 Warn if multiple methods of different types for the same selector are
2413 found during compilation. The check is performed on the list of methods
2414 in the final stage of compilation. Additionally, a check is performed
2415 for each selector appearing in a @code{@@selector(@dots{})}
2416 expression, and a corresponding method for that selector has been found
2417 during compilation. Because these checks scan the method table only at
2418 the end of compilation, these warnings are not produced if the final
2419 stage of compilation is not reached, for example because an error is
2420 found during compilation, or because the @option{-fsyntax-only} option is
2423 @item -Wstrict-selector-match
2424 @opindex Wstrict-selector-match
2425 Warn if multiple methods with differing argument and/or return types are
2426 found for a given selector when attempting to send a message using this
2427 selector to a receiver of type @code{id} or @code{Class}. When this flag
2428 is off (which is the default behavior), the compiler will omit such warnings
2429 if any differences found are confined to types which share the same size
2432 @item -Wundeclared-selector
2433 @opindex Wundeclared-selector
2434 Warn if a @code{@@selector(@dots{})} expression referring to an
2435 undeclared selector is found. A selector is considered undeclared if no
2436 method with that name has been declared before the
2437 @code{@@selector(@dots{})} expression, either explicitly in an
2438 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2439 an @code{@@implementation} section. This option always performs its
2440 checks as soon as a @code{@@selector(@dots{})} expression is found,
2441 while @option{-Wselector} only performs its checks in the final stage of
2442 compilation. This also enforces the coding style convention
2443 that methods and selectors must be declared before being used.
2445 @item -print-objc-runtime-info
2446 @opindex print-objc-runtime-info
2447 Generate C header describing the largest structure that is passed by
2452 @node Language Independent Options
2453 @section Options to Control Diagnostic Messages Formatting
2454 @cindex options to control diagnostics formatting
2455 @cindex diagnostic messages
2456 @cindex message formatting
2458 Traditionally, diagnostic messages have been formatted irrespective of
2459 the output device's aspect (e.g.@: its width, @dots{}). The options described
2460 below can be used to control the diagnostic messages formatting
2461 algorithm, e.g.@: how many characters per line, how often source location
2462 information should be reported. Right now, only the C++ front end can
2463 honor these options. However it is expected, in the near future, that
2464 the remaining front ends would be able to digest them correctly.
2467 @item -fmessage-length=@var{n}
2468 @opindex fmessage-length
2469 Try to format error messages so that they fit on lines of about @var{n}
2470 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2471 the front ends supported by GCC@. If @var{n} is zero, then no
2472 line-wrapping will be done; each error message will appear on a single
2475 @opindex fdiagnostics-show-location
2476 @item -fdiagnostics-show-location=once
2477 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2478 reporter to emit @emph{once} source location information; that is, in
2479 case the message is too long to fit on a single physical line and has to
2480 be wrapped, the source location won't be emitted (as prefix) again,
2481 over and over, in subsequent continuation lines. This is the default
2484 @item -fdiagnostics-show-location=every-line
2485 Only meaningful in line-wrapping mode. Instructs the diagnostic
2486 messages reporter to emit the same source location information (as
2487 prefix) for physical lines that result from the process of breaking
2488 a message which is too long to fit on a single line.
2490 @item -fdiagnostics-show-option
2491 @opindex fdiagnostics-show-option
2492 This option instructs the diagnostic machinery to add text to each
2493 diagnostic emitted, which indicates which command line option directly
2494 controls that diagnostic, when such an option is known to the
2495 diagnostic machinery.
2497 @item -Wcoverage-mismatch
2498 @opindex Wcoverage-mismatch
2499 Warn if feedback profiles do not match when using the
2500 @option{-fprofile-use} option.
2501 If a source file was changed between @option{-fprofile-gen} and
2502 @option{-fprofile-use}, the files with the profile feedback can fail
2503 to match the source file and GCC can not use the profile feedback
2504 information. By default, GCC emits an error message in this case.
2505 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2506 error. GCC does not use appropriate feedback profiles, so using this
2507 option can result in poorly optimized code. This option is useful
2508 only in the case of very minor changes such as bug fixes to an
2513 @node Warning Options
2514 @section Options to Request or Suppress Warnings
2515 @cindex options to control warnings
2516 @cindex warning messages
2517 @cindex messages, warning
2518 @cindex suppressing warnings
2520 Warnings are diagnostic messages that report constructions which
2521 are not inherently erroneous but which are risky or suggest there
2522 may have been an error.
2524 The following language-independent options do not enable specific
2525 warnings but control the kinds of diagnostics produced by GCC.
2528 @cindex syntax checking
2530 @opindex fsyntax-only
2531 Check the code for syntax errors, but don't do anything beyond that.
2535 Inhibit all warning messages.
2539 Make all warnings into errors.
2543 Make the specified warning into an error. The specifier for a warning
2544 is appended, for example @option{-Werror=switch} turns the warnings
2545 controlled by @option{-Wswitch} into errors. This switch takes a
2546 negative form, to be used to negate @option{-Werror} for specific
2547 warnings, for example @option{-Wno-error=switch} makes
2548 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2549 is in effect. You can use the @option{-fdiagnostics-show-option}
2550 option to have each controllable warning amended with the option which
2551 controls it, to determine what to use with this option.
2553 Note that specifying @option{-Werror=}@var{foo} automatically implies
2554 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2557 @item -Wfatal-errors
2558 @opindex Wfatal-errors
2559 This option causes the compiler to abort compilation on the first error
2560 occurred rather than trying to keep going and printing further error
2565 You can request many specific warnings with options beginning
2566 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2567 implicit declarations. Each of these specific warning options also
2568 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2569 example, @option{-Wno-implicit}. This manual lists only one of the
2570 two forms, whichever is not the default. For further,
2571 language-specific options also refer to @ref{C++ Dialect Options} and
2572 @ref{Objective-C and Objective-C++ Dialect Options}.
2577 Issue all the warnings demanded by strict ISO C and ISO C++;
2578 reject all programs that use forbidden extensions, and some other
2579 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2580 version of the ISO C standard specified by any @option{-std} option used.
2582 Valid ISO C and ISO C++ programs should compile properly with or without
2583 this option (though a rare few will require @option{-ansi} or a
2584 @option{-std} option specifying the required version of ISO C)@. However,
2585 without this option, certain GNU extensions and traditional C and C++
2586 features are supported as well. With this option, they are rejected.
2588 @option{-pedantic} does not cause warning messages for use of the
2589 alternate keywords whose names begin and end with @samp{__}. Pedantic
2590 warnings are also disabled in the expression that follows
2591 @code{__extension__}. However, only system header files should use
2592 these escape routes; application programs should avoid them.
2593 @xref{Alternate Keywords}.
2595 Some users try to use @option{-pedantic} to check programs for strict ISO
2596 C conformance. They soon find that it does not do quite what they want:
2597 it finds some non-ISO practices, but not all---only those for which
2598 ISO C @emph{requires} a diagnostic, and some others for which
2599 diagnostics have been added.
2601 A feature to report any failure to conform to ISO C might be useful in
2602 some instances, but would require considerable additional work and would
2603 be quite different from @option{-pedantic}. We don't have plans to
2604 support such a feature in the near future.
2606 Where the standard specified with @option{-std} represents a GNU
2607 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2608 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2609 extended dialect is based. Warnings from @option{-pedantic} are given
2610 where they are required by the base standard. (It would not make sense
2611 for such warnings to be given only for features not in the specified GNU
2612 C dialect, since by definition the GNU dialects of C include all
2613 features the compiler supports with the given option, and there would be
2614 nothing to warn about.)
2616 @item -pedantic-errors
2617 @opindex pedantic-errors
2618 Like @option{-pedantic}, except that errors are produced rather than
2623 This enables all the warnings about constructions that some users
2624 consider questionable, and that are easy to avoid (or modify to
2625 prevent the warning), even in conjunction with macros. This also
2626 enables some language-specific warnings described in @ref{C++ Dialect
2627 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2629 @option{-Wall} turns on the following warning flags:
2631 @gccoptlist{-Waddress @gol
2632 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2634 -Wchar-subscripts @gol
2636 -Wimplicit-function-declaration @gol
2639 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2640 -Wmissing-braces @gol
2646 -Wsequence-point @gol
2647 -Wsign-compare @r{(only in C++)} @gol
2648 -Wstrict-aliasing @gol
2649 -Wstrict-overflow @gol
2652 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2653 -Wunknown-pragmas @gol
2654 -Wunused-function @gol
2657 -Wunused-variable @gol
2660 Note that some warning flags are not implied by @option{-Wall}. Some of
2661 them warn about constructions that users generally do not consider
2662 questionable, but which occasionally you might wish to check for;
2663 others warn about constructions that are necessary or hard to avoid in
2664 some cases, and there is no simple way to modify the code to suppress
2665 the warning. Some of them are enabled by @option{-Wextra} but many of
2666 them must be enabled individually.
2671 This enables some extra warning flags that are not enabled by
2672 @option{-Wall}. (This option used to be called @option{-W}. The older
2673 name is still supported, but the newer name is more descriptive.)
2675 @gccoptlist{-Wclobbered @gol
2677 -Wmissing-field-initializers @gol
2678 -Wmissing-parameter-type @r{(C only)} @gol
2679 -Wold-style-declaration @r{(C only)} @gol
2680 -Woverride-init @gol
2683 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2684 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2687 The option @option{-Wextra} also prints warning messages for the
2693 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2694 @samp{>}, or @samp{>=}.
2697 (C++ only) An enumerator and a non-enumerator both appear in a
2698 conditional expression.
2701 (C++ only) A non-static reference or non-static @samp{const} member
2702 appears in a class without constructors.
2705 (C++ only) Ambiguous virtual bases.
2708 (C++ only) Subscripting an array which has been declared @samp{register}.
2711 (C++ only) Taking the address of a variable which has been declared
2715 (C++ only) A base class is not initialized in a derived class' copy
2722 Inhibit warning messages about the use of @samp{#import}.
2724 @item -Wchar-subscripts
2725 @opindex Wchar-subscripts
2726 Warn if an array subscript has type @code{char}. This is a common cause
2727 of error, as programmers often forget that this type is signed on some
2729 This warning is enabled by @option{-Wall}.
2733 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2734 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2735 This warning is enabled by @option{-Wall}.
2739 @opindex ffreestanding
2740 @opindex fno-builtin
2741 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2742 the arguments supplied have types appropriate to the format string
2743 specified, and that the conversions specified in the format string make
2744 sense. This includes standard functions, and others specified by format
2745 attributes (@pxref{Function Attributes}), in the @code{printf},
2746 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2747 not in the C standard) families (or other target-specific families).
2748 Which functions are checked without format attributes having been
2749 specified depends on the standard version selected, and such checks of
2750 functions without the attribute specified are disabled by
2751 @option{-ffreestanding} or @option{-fno-builtin}.
2753 The formats are checked against the format features supported by GNU
2754 libc version 2.2. These include all ISO C90 and C99 features, as well
2755 as features from the Single Unix Specification and some BSD and GNU
2756 extensions. Other library implementations may not support all these
2757 features; GCC does not support warning about features that go beyond a
2758 particular library's limitations. However, if @option{-pedantic} is used
2759 with @option{-Wformat}, warnings will be given about format features not
2760 in the selected standard version (but not for @code{strfmon} formats,
2761 since those are not in any version of the C standard). @xref{C Dialect
2762 Options,,Options Controlling C Dialect}.
2764 Since @option{-Wformat} also checks for null format arguments for
2765 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2767 @option{-Wformat} is included in @option{-Wall}. For more control over some
2768 aspects of format checking, the options @option{-Wformat-y2k},
2769 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2770 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2771 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2774 @opindex Wformat-y2k
2775 If @option{-Wformat} is specified, also warn about @code{strftime}
2776 formats which may yield only a two-digit year.
2778 @item -Wno-format-extra-args
2779 @opindex Wno-format-extra-args
2780 If @option{-Wformat} is specified, do not warn about excess arguments to a
2781 @code{printf} or @code{scanf} format function. The C standard specifies
2782 that such arguments are ignored.
2784 Where the unused arguments lie between used arguments that are
2785 specified with @samp{$} operand number specifications, normally
2786 warnings are still given, since the implementation could not know what
2787 type to pass to @code{va_arg} to skip the unused arguments. However,
2788 in the case of @code{scanf} formats, this option will suppress the
2789 warning if the unused arguments are all pointers, since the Single
2790 Unix Specification says that such unused arguments are allowed.
2792 @item -Wno-format-zero-length
2793 @opindex Wno-format-zero-length
2794 If @option{-Wformat} is specified, do not warn about zero-length formats.
2795 The C standard specifies that zero-length formats are allowed.
2797 @item -Wformat-nonliteral
2798 @opindex Wformat-nonliteral
2799 If @option{-Wformat} is specified, also warn if the format string is not a
2800 string literal and so cannot be checked, unless the format function
2801 takes its format arguments as a @code{va_list}.
2803 @item -Wformat-security
2804 @opindex Wformat-security
2805 If @option{-Wformat} is specified, also warn about uses of format
2806 functions that represent possible security problems. At present, this
2807 warns about calls to @code{printf} and @code{scanf} functions where the
2808 format string is not a string literal and there are no format arguments,
2809 as in @code{printf (foo);}. This may be a security hole if the format
2810 string came from untrusted input and contains @samp{%n}. (This is
2811 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2812 in future warnings may be added to @option{-Wformat-security} that are not
2813 included in @option{-Wformat-nonliteral}.)
2817 Enable @option{-Wformat} plus format checks not included in
2818 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2819 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2823 Warn about passing a null pointer for arguments marked as
2824 requiring a non-null value by the @code{nonnull} function attribute.
2826 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2827 can be disabled with the @option{-Wno-nonnull} option.
2829 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2831 Warn about uninitialized variables which are initialized with themselves.
2832 Note this option can only be used with the @option{-Wuninitialized} option,
2833 which in turn only works with @option{-O1} and above.
2835 For example, GCC will warn about @code{i} being uninitialized in the
2836 following snippet only when @option{-Winit-self} has been specified:
2847 @item -Wimplicit-int
2848 @opindex Wimplicit-int
2849 Warn when a declaration does not specify a type.
2850 This warning is enabled by @option{-Wall}.
2852 @item -Wimplicit-function-declaration
2853 @opindex Wimplicit-function-declaration
2854 @opindex Wno-implicit-function-declaration
2855 Give a warning whenever a function is used before being declared. In
2856 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2857 enabled by default and it is made into an error by
2858 @option{-pedantic-errors}. This warning is also enabled by
2863 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2864 This warning is enabled by @option{-Wall}.
2868 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2869 function with external linkage, returning int, taking either zero
2870 arguments, two, or three arguments of appropriate types.
2871 This warning is enabled by @option{-Wall}.
2873 @item -Wmissing-braces
2874 @opindex Wmissing-braces
2875 Warn if an aggregate or union initializer is not fully bracketed. In
2876 the following example, the initializer for @samp{a} is not fully
2877 bracketed, but that for @samp{b} is fully bracketed.
2880 int a[2][2] = @{ 0, 1, 2, 3 @};
2881 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2884 This warning is enabled by @option{-Wall}.
2886 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2887 @opindex Wmissing-include-dirs
2888 Warn if a user-supplied include directory does not exist.
2891 @opindex Wparentheses
2892 Warn if parentheses are omitted in certain contexts, such
2893 as when there is an assignment in a context where a truth value
2894 is expected, or when operators are nested whose precedence people
2895 often get confused about.
2897 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2898 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2899 interpretation from that of ordinary mathematical notation.
2901 Also warn about constructions where there may be confusion to which
2902 @code{if} statement an @code{else} branch belongs. Here is an example of
2917 In C/C++, every @code{else} branch belongs to the innermost possible
2918 @code{if} statement, which in this example is @code{if (b)}. This is
2919 often not what the programmer expected, as illustrated in the above
2920 example by indentation the programmer chose. When there is the
2921 potential for this confusion, GCC will issue a warning when this flag
2922 is specified. To eliminate the warning, add explicit braces around
2923 the innermost @code{if} statement so there is no way the @code{else}
2924 could belong to the enclosing @code{if}. The resulting code would
2941 This warning is enabled by @option{-Wall}.
2943 @item -Wsequence-point
2944 @opindex Wsequence-point
2945 Warn about code that may have undefined semantics because of violations
2946 of sequence point rules in the C and C++ standards.
2948 The C and C++ standards defines the order in which expressions in a C/C++
2949 program are evaluated in terms of @dfn{sequence points}, which represent
2950 a partial ordering between the execution of parts of the program: those
2951 executed before the sequence point, and those executed after it. These
2952 occur after the evaluation of a full expression (one which is not part
2953 of a larger expression), after the evaluation of the first operand of a
2954 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2955 function is called (but after the evaluation of its arguments and the
2956 expression denoting the called function), and in certain other places.
2957 Other than as expressed by the sequence point rules, the order of
2958 evaluation of subexpressions of an expression is not specified. All
2959 these rules describe only a partial order rather than a total order,
2960 since, for example, if two functions are called within one expression
2961 with no sequence point between them, the order in which the functions
2962 are called is not specified. However, the standards committee have
2963 ruled that function calls do not overlap.
2965 It is not specified when between sequence points modifications to the
2966 values of objects take effect. Programs whose behavior depends on this
2967 have undefined behavior; the C and C++ standards specify that ``Between
2968 the previous and next sequence point an object shall have its stored
2969 value modified at most once by the evaluation of an expression.
2970 Furthermore, the prior value shall be read only to determine the value
2971 to be stored.''. If a program breaks these rules, the results on any
2972 particular implementation are entirely unpredictable.
2974 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2975 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2976 diagnosed by this option, and it may give an occasional false positive
2977 result, but in general it has been found fairly effective at detecting
2978 this sort of problem in programs.
2980 The standard is worded confusingly, therefore there is some debate
2981 over the precise meaning of the sequence point rules in subtle cases.
2982 Links to discussions of the problem, including proposed formal
2983 definitions, may be found on the GCC readings page, at
2984 @w{@uref{http://gcc.gnu.org/readings.html}}.
2986 This warning is enabled by @option{-Wall} for C and C++.
2989 @opindex Wreturn-type
2990 @opindex Wno-return-type
2991 Warn whenever a function is defined with a return-type that defaults
2992 to @code{int}. Also warn about any @code{return} statement with no
2993 return-value in a function whose return-type is not @code{void}
2994 (falling off the end of the function body is considered returning
2995 without a value), and about a @code{return} statement with a
2996 expression in a function whose return-type is @code{void}.
2998 Also warn if the return type of a function has a type qualifier
2999 such as @code{const}. For ISO C such a type qualifier has no effect,
3000 since the value returned by a function is not an lvalue.
3001 For C++, the warning is only emitted for scalar types or @code{void}.
3002 ISO C prohibits qualified @code{void} return types on function
3003 definitions, so such return types always receive a warning
3004 even without this option.
3006 For C++, a function without return type always produces a diagnostic
3007 message, even when @option{-Wno-return-type} is specified. The only
3008 exceptions are @samp{main} and functions defined in system headers.
3010 This warning is enabled by @option{-Wall}.
3014 Warn whenever a @code{switch} statement has an index of enumerated type
3015 and lacks a @code{case} for one or more of the named codes of that
3016 enumeration. (The presence of a @code{default} label prevents this
3017 warning.) @code{case} labels outside the enumeration range also
3018 provoke warnings when this option is used.
3019 This warning is enabled by @option{-Wall}.
3021 @item -Wswitch-default
3022 @opindex Wswitch-switch
3023 Warn whenever a @code{switch} statement does not have a @code{default}
3027 @opindex Wswitch-enum
3028 Warn whenever a @code{switch} statement has an index of enumerated type
3029 and lacks a @code{case} for one or more of the named codes of that
3030 enumeration. @code{case} labels outside the enumeration range also
3031 provoke warnings when this option is used.
3035 Warn if any trigraphs are encountered that might change the meaning of
3036 the program (trigraphs within comments are not warned about).
3037 This warning is enabled by @option{-Wall}.
3039 @item -Wunused-function
3040 @opindex Wunused-function
3041 Warn whenever a static function is declared but not defined or a
3042 non-inline static function is unused.
3043 This warning is enabled by @option{-Wall}.
3045 @item -Wunused-label
3046 @opindex Wunused-label
3047 Warn whenever a label is declared but not used.
3048 This warning is enabled by @option{-Wall}.
3050 To suppress this warning use the @samp{unused} attribute
3051 (@pxref{Variable Attributes}).
3053 @item -Wunused-parameter
3054 @opindex Wunused-parameter
3055 Warn whenever a function parameter is unused aside from its declaration.
3057 To suppress this warning use the @samp{unused} attribute
3058 (@pxref{Variable Attributes}).
3060 @item -Wunused-variable
3061 @opindex Wunused-variable
3062 Warn whenever a local variable or non-constant static variable is unused
3063 aside from its declaration.
3064 This warning is enabled by @option{-Wall}.
3066 To suppress this warning use the @samp{unused} attribute
3067 (@pxref{Variable Attributes}).
3069 @item -Wunused-value
3070 @opindex Wunused-value
3071 Warn whenever a statement computes a result that is explicitly not
3072 used. To suppress this warning cast the unused expression to
3073 @samp{void}. This includes an expression-statement or the left-hand
3074 side of a comma expression that contains no side effects. For example,
3075 an expression such as @samp{x[i,j]} will cause a warning, while
3076 @samp{x[(void)i,j]} will not.
3078 This warning is enabled by @option{-Wall}.
3082 All the above @option{-Wunused} options combined.
3084 In order to get a warning about an unused function parameter, you must
3085 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3086 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3088 @item -Wuninitialized
3089 @opindex Wuninitialized
3090 Warn if an automatic variable is used without first being initialized or
3091 if a variable may be clobbered by a @code{setjmp} call.
3093 These warnings are possible only in optimizing compilation,
3094 because they require data flow information that is computed only
3095 when optimizing. If you do not specify @option{-O}, you will not get
3096 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
3097 requiring @option{-O}.
3099 If you want to warn about code which uses the uninitialized value of the
3100 variable in its own initializer, use the @option{-Winit-self} option.
3102 These warnings occur for individual uninitialized or clobbered
3103 elements of structure, union or array variables as well as for
3104 variables which are uninitialized or clobbered as a whole. They do
3105 not occur for variables or elements declared @code{volatile}. Because
3106 these warnings depend on optimization, the exact variables or elements
3107 for which there are warnings will depend on the precise optimization
3108 options and version of GCC used.
3110 Note that there may be no warning about a variable that is used only
3111 to compute a value that itself is never used, because such
3112 computations may be deleted by data flow analysis before the warnings
3115 These warnings are made optional because GCC is not smart
3116 enough to see all the reasons why the code might be correct
3117 despite appearing to have an error. Here is one example of how
3138 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3139 always initialized, but GCC doesn't know this. Here is
3140 another common case:
3145 if (change_y) save_y = y, y = new_y;
3147 if (change_y) y = save_y;
3152 This has no bug because @code{save_y} is used only if it is set.
3154 @cindex @code{longjmp} warnings
3155 This option also warns when a non-volatile automatic variable might be
3156 changed by a call to @code{longjmp}. These warnings as well are possible
3157 only in optimizing compilation.
3159 The compiler sees only the calls to @code{setjmp}. It cannot know
3160 where @code{longjmp} will be called; in fact, a signal handler could
3161 call it at any point in the code. As a result, you may get a warning
3162 even when there is in fact no problem because @code{longjmp} cannot
3163 in fact be called at the place which would cause a problem.
3165 Some spurious warnings can be avoided if you declare all the functions
3166 you use that never return as @code{noreturn}. @xref{Function
3169 This warning is enabled by @option{-Wall} or @option{-Wextra} in
3170 optimizing compilations (@option{-O1} and above).
3172 @item -Wunknown-pragmas
3173 @opindex Wunknown-pragmas
3174 @cindex warning for unknown pragmas
3175 @cindex unknown pragmas, warning
3176 @cindex pragmas, warning of unknown
3177 Warn when a #pragma directive is encountered which is not understood by
3178 GCC@. If this command line option is used, warnings will even be issued
3179 for unknown pragmas in system header files. This is not the case if
3180 the warnings were only enabled by the @option{-Wall} command line option.
3183 @opindex Wno-pragmas
3185 Do not warn about misuses of pragmas, such as incorrect parameters,
3186 invalid syntax, or conflicts between pragmas. See also
3187 @samp{-Wunknown-pragmas}.
3189 @item -Wstrict-aliasing
3190 @opindex Wstrict-aliasing
3191 This option is only active when @option{-fstrict-aliasing} is active.
3192 It warns about code which might break the strict aliasing rules that the
3193 compiler is using for optimization. The warning does not catch all
3194 cases, but does attempt to catch the more common pitfalls. It is
3195 included in @option{-Wall}.
3196 It is equivalent to @option{-Wstrict-aliasing=3}
3198 @item -Wstrict-aliasing=n
3199 @opindex Wstrict-aliasing=n
3200 This option is only active when @option{-fstrict-aliasing} is active.
3201 It warns about code which might break the strict aliasing rules that the
3202 compiler is using for optimization.
3203 Higher levels correspond to higher accuracy (fewer false positives).
3204 Higher levels also correspond to more effort, similar to the way -O works.
3205 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3208 Level 1: Most aggressive, quick, least accurate.
3209 Possibly useful when higher levels
3210 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3211 false negatives. However, it has many false positives.
3212 Warns for all pointer conversions between possibly incompatible types,
3213 even if never dereferenced. Runs in the frontend only.
3215 Level 2: Aggressive, quick, not too precise.
3216 May still have many false positives (not as many as level 1 though),
3217 and few false negatives (but possibly more than level 1).
3218 Unlike level 1, it only warns when an address is taken. Warns about
3219 incomplete types. Runs in the frontend only.
3221 Level 3 (default for @option{-Wstrict-aliasing}):
3222 Should have very few false positives and few false
3223 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3224 Takes care of the common punn+dereference pattern in the frontend:
3225 @code{*(int*)&some_float}.
3226 If optimization is enabled, it also runs in the backend, where it deals
3227 with multiple statement cases using flow-sensitive points-to information.
3228 Only warns when the converted pointer is dereferenced.
3229 Does not warn about incomplete types.
3231 @item -Wstrict-overflow
3232 @item -Wstrict-overflow=@var{n}
3233 @opindex Wstrict-overflow
3234 This option is only active when @option{-fstrict-overflow} is active.
3235 It warns about cases where the compiler optimizes based on the
3236 assumption that signed overflow does not occur. Note that it does not
3237 warn about all cases where the code might overflow: it only warns
3238 about cases where the compiler implements some optimization. Thus
3239 this warning depends on the optimization level.
3241 An optimization which assumes that signed overflow does not occur is
3242 perfectly safe if the values of the variables involved are such that
3243 overflow never does, in fact, occur. Therefore this warning can
3244 easily give a false positive: a warning about code which is not
3245 actually a problem. To help focus on important issues, several
3246 warning levels are defined. No warnings are issued for the use of
3247 undefined signed overflow when estimating how many iterations a loop
3248 will require, in particular when determining whether a loop will be
3252 @item -Wstrict-overflow=1
3253 Warn about cases which are both questionable and easy to avoid. For
3254 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3255 compiler will simplify this to @code{1}. This level of
3256 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3257 are not, and must be explicitly requested.
3259 @item -Wstrict-overflow=2
3260 Also warn about other cases where a comparison is simplified to a
3261 constant. For example: @code{abs (x) >= 0}. This can only be
3262 simplified when @option{-fstrict-overflow} is in effect, because
3263 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3264 zero. @option{-Wstrict-overflow} (with no level) is the same as
3265 @option{-Wstrict-overflow=2}.
3267 @item -Wstrict-overflow=3
3268 Also warn about other cases where a comparison is simplified. For
3269 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3271 @item -Wstrict-overflow=4
3272 Also warn about other simplifications not covered by the above cases.
3273 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3275 @item -Wstrict-overflow=5
3276 Also warn about cases where the compiler reduces the magnitude of a
3277 constant involved in a comparison. For example: @code{x + 2 > y} will
3278 be simplified to @code{x + 1 >= y}. This is reported only at the
3279 highest warning level because this simplification applies to many
3280 comparisons, so this warning level will give a very large number of
3284 @item -Warray-bounds
3285 @opindex Wno-array-bounds
3286 @opindex Warray-bounds
3287 This option is only active when @option{-ftree-vrp} is active
3288 (default for -O2 and above). It warns about subscripts to arrays
3289 that are always out of bounds. This warning is enabled by @option{-Wall}.
3291 @item -Wno-div-by-zero
3292 @opindex Wno-div-by-zero
3293 @opindex Wdiv-by-zero
3294 Do not warn about compile-time integer division by zero. Floating point
3295 division by zero is not warned about, as it can be a legitimate way of
3296 obtaining infinities and NaNs.
3298 @item -Wsystem-headers
3299 @opindex Wsystem-headers
3300 @cindex warnings from system headers
3301 @cindex system headers, warnings from
3302 Print warning messages for constructs found in system header files.
3303 Warnings from system headers are normally suppressed, on the assumption
3304 that they usually do not indicate real problems and would only make the
3305 compiler output harder to read. Using this command line option tells
3306 GCC to emit warnings from system headers as if they occurred in user
3307 code. However, note that using @option{-Wall} in conjunction with this
3308 option will @emph{not} warn about unknown pragmas in system
3309 headers---for that, @option{-Wunknown-pragmas} must also be used.
3312 @opindex Wfloat-equal
3313 Warn if floating point values are used in equality comparisons.
3315 The idea behind this is that sometimes it is convenient (for the
3316 programmer) to consider floating-point values as approximations to
3317 infinitely precise real numbers. If you are doing this, then you need
3318 to compute (by analyzing the code, or in some other way) the maximum or
3319 likely maximum error that the computation introduces, and allow for it
3320 when performing comparisons (and when producing output, but that's a
3321 different problem). In particular, instead of testing for equality, you
3322 would check to see whether the two values have ranges that overlap; and
3323 this is done with the relational operators, so equality comparisons are
3326 @item -Wtraditional @r{(C only)}
3327 @opindex Wtraditional
3328 Warn about certain constructs that behave differently in traditional and
3329 ISO C@. Also warn about ISO C constructs that have no traditional C
3330 equivalent, and/or problematic constructs which should be avoided.
3334 Macro parameters that appear within string literals in the macro body.
3335 In traditional C macro replacement takes place within string literals,
3336 but does not in ISO C@.
3339 In traditional C, some preprocessor directives did not exist.
3340 Traditional preprocessors would only consider a line to be a directive
3341 if the @samp{#} appeared in column 1 on the line. Therefore
3342 @option{-Wtraditional} warns about directives that traditional C
3343 understands but would ignore because the @samp{#} does not appear as the
3344 first character on the line. It also suggests you hide directives like
3345 @samp{#pragma} not understood by traditional C by indenting them. Some
3346 traditional implementations would not recognize @samp{#elif}, so it
3347 suggests avoiding it altogether.
3350 A function-like macro that appears without arguments.
3353 The unary plus operator.
3356 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3357 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3358 constants.) Note, these suffixes appear in macros defined in the system
3359 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3360 Use of these macros in user code might normally lead to spurious
3361 warnings, however GCC's integrated preprocessor has enough context to
3362 avoid warning in these cases.
3365 A function declared external in one block and then used after the end of
3369 A @code{switch} statement has an operand of type @code{long}.
3372 A non-@code{static} function declaration follows a @code{static} one.
3373 This construct is not accepted by some traditional C compilers.
3376 The ISO type of an integer constant has a different width or
3377 signedness from its traditional type. This warning is only issued if
3378 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3379 typically represent bit patterns, are not warned about.
3382 Usage of ISO string concatenation is detected.
3385 Initialization of automatic aggregates.
3388 Identifier conflicts with labels. Traditional C lacks a separate
3389 namespace for labels.
3392 Initialization of unions. If the initializer is zero, the warning is
3393 omitted. This is done under the assumption that the zero initializer in
3394 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3395 initializer warnings and relies on default initialization to zero in the
3399 Conversions by prototypes between fixed/floating point values and vice
3400 versa. The absence of these prototypes when compiling with traditional
3401 C would cause serious problems. This is a subset of the possible
3402 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3405 Use of ISO C style function definitions. This warning intentionally is
3406 @emph{not} issued for prototype declarations or variadic functions
3407 because these ISO C features will appear in your code when using
3408 libiberty's traditional C compatibility macros, @code{PARAMS} and
3409 @code{VPARAMS}. This warning is also bypassed for nested functions
3410 because that feature is already a GCC extension and thus not relevant to
3411 traditional C compatibility.
3414 @item -Wtraditional-conversion @r{(C only)}
3415 @opindex Wtraditional-conversion
3416 Warn if a prototype causes a type conversion that is different from what
3417 would happen to the same argument in the absence of a prototype. This
3418 includes conversions of fixed point to floating and vice versa, and
3419 conversions changing the width or signedness of a fixed point argument
3420 except when the same as the default promotion.
3422 @item -Wdeclaration-after-statement @r{(C only)}
3423 @opindex Wdeclaration-after-statement
3424 Warn when a declaration is found after a statement in a block. This
3425 construct, known from C++, was introduced with ISO C99 and is by default
3426 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3427 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3431 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3433 @item -Wno-endif-labels
3434 @opindex Wno-endif-labels
3435 @opindex Wendif-labels
3436 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3440 Warn whenever a local variable shadows another local variable, parameter or
3441 global variable or whenever a built-in function is shadowed.
3443 @item -Wlarger-than-@var{len}
3444 @opindex Wlarger-than
3445 Warn whenever an object of larger than @var{len} bytes is defined.
3447 @item -Wunsafe-loop-optimizations
3448 @opindex Wunsafe-loop-optimizations
3449 Warn if the loop cannot be optimized because the compiler could not
3450 assume anything on the bounds of the loop indices. With
3451 @option{-funsafe-loop-optimizations} warn if the compiler made
3454 @item -Wpointer-arith
3455 @opindex Wpointer-arith
3456 Warn about anything that depends on the ``size of'' a function type or
3457 of @code{void}. GNU C assigns these types a size of 1, for
3458 convenience in calculations with @code{void *} pointers and pointers
3459 to functions. In C++, warn also when an arithmetic operation involves
3460 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3463 @opindex Wtype-limits
3464 @opindex Wno-type-limits
3465 Warn if a comparison is always true or always false due to the limited
3466 range of the data type, but do not warn for constant expressions. For
3467 example, warn if an unsigned variable is compared against zero with
3468 @samp{<} or @samp{>=}. This warning is also enabled by
3471 @item -Wbad-function-cast @r{(C only)}
3472 @opindex Wbad-function-cast
3473 Warn whenever a function call is cast to a non-matching type.
3474 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3477 Warn about ISO C constructs that are outside of the common subset of
3478 ISO C and ISO C++, e.g.@: request for implicit conversion from
3479 @code{void *} to a pointer to non-@code{void} type.
3481 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3482 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3483 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3484 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3488 Warn whenever a pointer is cast so as to remove a type qualifier from
3489 the target type. For example, warn if a @code{const char *} is cast
3490 to an ordinary @code{char *}.
3493 @opindex Wcast-align
3494 Warn whenever a pointer is cast such that the required alignment of the
3495 target is increased. For example, warn if a @code{char *} is cast to
3496 an @code{int *} on machines where integers can only be accessed at
3497 two- or four-byte boundaries.
3499 @item -Wwrite-strings
3500 @opindex Wwrite-strings
3501 When compiling C, give string constants the type @code{const
3502 char[@var{length}]} so that
3503 copying the address of one into a non-@code{const} @code{char *}
3504 pointer will get a warning; when compiling C++, warn about the
3505 deprecated conversion from string literals to @code{char *}. This
3506 warning, by default, is enabled for C++ programs.
3507 These warnings will help you find at
3508 compile time code that can try to write into a string constant, but
3509 only if you have been very careful about using @code{const} in
3510 declarations and prototypes. Otherwise, it will just be a nuisance;
3511 this is why we did not make @option{-Wall} request these warnings.
3515 Warn for variables that might be changed by @samp{longjmp} or
3516 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3519 @opindex Wconversion
3520 @opindex Wno-conversion
3521 Warn for implicit conversions that may alter a value. This includes
3522 conversions between real and integer, like @code{abs (x)} when
3523 @code{x} is @code{double}; conversions between signed and unsigned,
3524 like @code{unsigned ui = -1}; and conversions to smaller types, like
3525 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3526 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3527 changed by the conversion like in @code{abs (2.0)}. Warnings about
3528 conversions between signed and unsigned integers can be disabled by
3529 using @option{-Wno-sign-conversion}.
3531 For C++, also warn for conversions between @code{NULL} and non-pointer
3532 types; confusing overload resolution for user-defined conversions; and
3533 conversions that will never use a type conversion operator:
3534 conversions to @code{void}, the same type, a base class or a reference
3535 to them. Warnings about conversions between signed and unsigned
3536 integers are disabled by default in C++ unless
3537 @option{-Wsign-conversion} is explicitly enabled.
3540 @opindex Wempty-body
3541 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3542 while} statement. Additionally, in C++, warn when an empty body occurs
3543 in a @samp{while} or @samp{for} statement with no whitespacing before
3544 the semicolon. This warning is also enabled by @option{-Wextra}.
3546 @item -Wsign-compare
3547 @opindex Wsign-compare
3548 @cindex warning for comparison of signed and unsigned values
3549 @cindex comparison of signed and unsigned values, warning
3550 @cindex signed and unsigned values, comparison warning
3551 Warn when a comparison between signed and unsigned values could produce
3552 an incorrect result when the signed value is converted to unsigned.
3553 This warning is also enabled by @option{-Wextra}; to get the other warnings
3554 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3556 @item -Wsign-conversion
3557 @opindex Wsign-conversion
3558 @opindex Wno-sign-conversion
3559 Warn for implicit conversions that may change the sign of an integer
3560 value, like assigning a signed integer expression to an unsigned
3561 integer variable. An explicit cast silences the warning. In C, this
3562 option is enabled also by @option{-Wconversion}.
3566 @opindex Wno-address
3567 Warn about suspicious uses of memory addresses. These include using
3568 the address of a function in a conditional expression, such as
3569 @code{void func(void); if (func)}, and comparisons against the memory
3570 address of a string literal, such as @code{if (x == "abc")}. Such
3571 uses typically indicate a programmer error: the address of a function
3572 always evaluates to true, so their use in a conditional usually
3573 indicate that the programmer forgot the parentheses in a function
3574 call; and comparisons against string literals result in unspecified
3575 behavior and are not portable in C, so they usually indicate that the
3576 programmer intended to use @code{strcmp}. This warning is enabled by
3580 @opindex Wlogical-op
3581 @opindex Wno-logical-op
3582 Warn about suspicious uses of logical operators in expressions.
3583 This includes using logical operators in contexts where a
3584 bit-wise operator is likely to be expected.
3586 @item -Waggregate-return
3587 @opindex Waggregate-return
3588 Warn if any functions that return structures or unions are defined or
3589 called. (In languages where you can return an array, this also elicits
3592 @item -Wno-attributes
3593 @opindex Wno-attributes
3594 @opindex Wattributes
3595 Do not warn if an unexpected @code{__attribute__} is used, such as
3596 unrecognized attributes, function attributes applied to variables,
3597 etc. This will not stop errors for incorrect use of supported
3600 @item -Wstrict-prototypes @r{(C only)}
3601 @opindex Wstrict-prototypes
3602 Warn if a function is declared or defined without specifying the
3603 argument types. (An old-style function definition is permitted without
3604 a warning if preceded by a declaration which specifies the argument
3607 @item -Wold-style-declaration @r{(C only)}
3608 @opindex Wold-style-declaration
3609 Warn for obsolescent usages, according to the C Standard, in a
3610 declaration. For example, warn if storage-class specifiers like
3611 @code{static} are not the first things in a declaration. This warning
3612 is also enabled by @option{-Wextra}.
3614 @item -Wold-style-definition @r{(C only)}
3615 @opindex Wold-style-definition
3616 Warn if an old-style function definition is used. A warning is given
3617 even if there is a previous prototype.
3619 @item -Wmissing-parameter-type @r{(C only)}
3620 @opindex Wmissing-parameter-type
3621 A function parameter is declared without a type specifier in K&R-style
3628 This warning is also enabled by @option{-Wextra}.
3630 @item -Wmissing-prototypes @r{(C only)}
3631 @opindex Wmissing-prototypes
3632 Warn if a global function is defined without a previous prototype
3633 declaration. This warning is issued even if the definition itself
3634 provides a prototype. The aim is to detect global functions that fail
3635 to be declared in header files.
3637 @item -Wmissing-declarations @r{(C and C++ only)}
3638 @opindex Wmissing-declarations
3639 Warn if a global function is defined without a previous declaration.
3640 Do so even if the definition itself provides a prototype.
3641 Use this option to detect global functions that are not declared in
3642 header files. In C++, no warnings are issued for function templates,
3643 or for inline functions, or for functions in anonymous namespaces.
3645 @item -Wmissing-field-initializers
3646 @opindex Wmissing-field-initializers
3649 Warn if a structure's initializer has some fields missing. For
3650 example, the following code would cause such a warning, because
3651 @code{x.h} is implicitly zero:
3654 struct s @{ int f, g, h; @};
3655 struct s x = @{ 3, 4 @};
3658 This option does not warn about designated initializers, so the following
3659 modification would not trigger a warning:
3662 struct s @{ int f, g, h; @};
3663 struct s x = @{ .f = 3, .g = 4 @};
3666 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3667 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3669 @item -Wmissing-noreturn
3670 @opindex Wmissing-noreturn
3671 Warn about functions which might be candidates for attribute @code{noreturn}.
3672 Note these are only possible candidates, not absolute ones. Care should
3673 be taken to manually verify functions actually do not ever return before
3674 adding the @code{noreturn} attribute, otherwise subtle code generation
3675 bugs could be introduced. You will not get a warning for @code{main} in
3676 hosted C environments.
3678 @item -Wmissing-format-attribute
3679 @opindex Wmissing-format-attribute
3681 Warn about function pointers which might be candidates for @code{format}
3682 attributes. Note these are only possible candidates, not absolute ones.
3683 GCC will guess that function pointers with @code{format} attributes that
3684 are used in assignment, initialization, parameter passing or return
3685 statements should have a corresponding @code{format} attribute in the
3686 resulting type. I.e.@: the left-hand side of the assignment or
3687 initialization, the type of the parameter variable, or the return type
3688 of the containing function respectively should also have a @code{format}
3689 attribute to avoid the warning.
3691 GCC will also warn about function definitions which might be
3692 candidates for @code{format} attributes. Again, these are only
3693 possible candidates. GCC will guess that @code{format} attributes
3694 might be appropriate for any function that calls a function like
3695 @code{vprintf} or @code{vscanf}, but this might not always be the
3696 case, and some functions for which @code{format} attributes are
3697 appropriate may not be detected.
3699 @item -Wno-multichar
3700 @opindex Wno-multichar
3702 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3703 Usually they indicate a typo in the user's code, as they have
3704 implementation-defined values, and should not be used in portable code.
3706 @item -Wnormalized=<none|id|nfc|nfkc>
3707 @opindex Wnormalized
3710 @cindex character set, input normalization
3711 In ISO C and ISO C++, two identifiers are different if they are
3712 different sequences of characters. However, sometimes when characters
3713 outside the basic ASCII character set are used, you can have two
3714 different character sequences that look the same. To avoid confusion,
3715 the ISO 10646 standard sets out some @dfn{normalization rules} which
3716 when applied ensure that two sequences that look the same are turned into
3717 the same sequence. GCC can warn you if you are using identifiers which
3718 have not been normalized; this option controls that warning.
3720 There are four levels of warning that GCC supports. The default is
3721 @option{-Wnormalized=nfc}, which warns about any identifier which is
3722 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3723 recommended form for most uses.
3725 Unfortunately, there are some characters which ISO C and ISO C++ allow
3726 in identifiers that when turned into NFC aren't allowable as
3727 identifiers. That is, there's no way to use these symbols in portable
3728 ISO C or C++ and have all your identifiers in NFC.
3729 @option{-Wnormalized=id} suppresses the warning for these characters.
3730 It is hoped that future versions of the standards involved will correct
3731 this, which is why this option is not the default.
3733 You can switch the warning off for all characters by writing
3734 @option{-Wnormalized=none}. You would only want to do this if you
3735 were using some other normalization scheme (like ``D''), because
3736 otherwise you can easily create bugs that are literally impossible to see.
3738 Some characters in ISO 10646 have distinct meanings but look identical
3739 in some fonts or display methodologies, especially once formatting has
3740 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3741 LETTER N'', will display just like a regular @code{n} which has been
3742 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3743 normalization scheme to convert all these into a standard form as
3744 well, and GCC will warn if your code is not in NFKC if you use
3745 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3746 about every identifier that contains the letter O because it might be
3747 confused with the digit 0, and so is not the default, but may be
3748 useful as a local coding convention if the programming environment is
3749 unable to be fixed to display these characters distinctly.
3751 @item -Wno-deprecated-declarations
3752 @opindex Wno-deprecated-declarations
3753 Do not warn about uses of functions (@pxref{Function Attributes}),
3754 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3755 Attributes}) marked as deprecated by using the @code{deprecated}
3759 @opindex Wno-overflow
3760 Do not warn about compile-time overflow in constant expressions.
3762 @item -Woverride-init
3763 @opindex Woverride-init
3766 Warn if an initialized field without side effects is overridden when
3767 using designated initializers (@pxref{Designated Inits, , Designated
3770 This warning is included in @option{-Wextra}. To get other
3771 @option{-Wextra} warnings without this one, use @samp{-Wextra
3772 -Wno-override-init}.
3776 Warn if a structure is given the packed attribute, but the packed
3777 attribute has no effect on the layout or size of the structure.
3778 Such structures may be mis-aligned for little benefit. For
3779 instance, in this code, the variable @code{f.x} in @code{struct bar}
3780 will be misaligned even though @code{struct bar} does not itself
3781 have the packed attribute:
3788 @} __attribute__((packed));
3798 Warn if padding is included in a structure, either to align an element
3799 of the structure or to align the whole structure. Sometimes when this
3800 happens it is possible to rearrange the fields of the structure to
3801 reduce the padding and so make the structure smaller.
3803 @item -Wredundant-decls
3804 @opindex Wredundant-decls
3805 Warn if anything is declared more than once in the same scope, even in
3806 cases where multiple declaration is valid and changes nothing.
3808 @item -Wnested-externs @r{(C only)}
3809 @opindex Wnested-externs
3810 Warn if an @code{extern} declaration is encountered within a function.
3812 @item -Wunreachable-code
3813 @opindex Wunreachable-code
3814 Warn if the compiler detects that code will never be executed.
3816 This option is intended to warn when the compiler detects that at
3817 least a whole line of source code will never be executed, because
3818 some condition is never satisfied or because it is after a
3819 procedure that never returns.
3821 It is possible for this option to produce a warning even though there
3822 are circumstances under which part of the affected line can be executed,
3823 so care should be taken when removing apparently-unreachable code.
3825 For instance, when a function is inlined, a warning may mean that the
3826 line is unreachable in only one inlined copy of the function.
3828 This option is not made part of @option{-Wall} because in a debugging
3829 version of a program there is often substantial code which checks
3830 correct functioning of the program and is, hopefully, unreachable
3831 because the program does work. Another common use of unreachable
3832 code is to provide behavior which is selectable at compile-time.
3836 Warn if a function can not be inlined and it was declared as inline.
3837 Even with this option, the compiler will not warn about failures to
3838 inline functions declared in system headers.
3840 The compiler uses a variety of heuristics to determine whether or not
3841 to inline a function. For example, the compiler takes into account
3842 the size of the function being inlined and the amount of inlining
3843 that has already been done in the current function. Therefore,
3844 seemingly insignificant changes in the source program can cause the
3845 warnings produced by @option{-Winline} to appear or disappear.
3847 @item -Wno-invalid-offsetof @r{(C++ only)}
3848 @opindex Wno-invalid-offsetof
3849 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3850 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3851 to a non-POD type is undefined. In existing C++ implementations,
3852 however, @samp{offsetof} typically gives meaningful results even when
3853 applied to certain kinds of non-POD types. (Such as a simple
3854 @samp{struct} that fails to be a POD type only by virtue of having a
3855 constructor.) This flag is for users who are aware that they are
3856 writing nonportable code and who have deliberately chosen to ignore the
3859 The restrictions on @samp{offsetof} may be relaxed in a future version
3860 of the C++ standard.
3862 @item -Wno-int-to-pointer-cast @r{(C only)}
3863 @opindex Wno-int-to-pointer-cast
3864 Suppress warnings from casts to pointer type of an integer of a
3867 @item -Wno-pointer-to-int-cast @r{(C only)}
3868 @opindex Wno-pointer-to-int-cast
3869 Suppress warnings from casts from a pointer to an integer type of a
3873 @opindex Winvalid-pch
3874 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3875 the search path but can't be used.
3879 @opindex Wno-long-long
3880 Warn if @samp{long long} type is used. This is default. To inhibit
3881 the warning messages, use @option{-Wno-long-long}. Flags
3882 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3883 only when @option{-pedantic} flag is used.
3885 @item -Wvariadic-macros
3886 @opindex Wvariadic-macros
3887 @opindex Wno-variadic-macros
3888 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3889 alternate syntax when in pedantic ISO C99 mode. This is default.
3890 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3895 Warn if variable length array is used in the code.
3896 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3897 the variable length array.
3899 @item -Wvolatile-register-var
3900 @opindex Wvolatile-register-var
3901 @opindex Wno-volatile-register-var
3902 Warn if a register variable is declared volatile. The volatile
3903 modifier does not inhibit all optimizations that may eliminate reads
3904 and/or writes to register variables.
3906 @item -Wdisabled-optimization
3907 @opindex Wdisabled-optimization
3908 Warn if a requested optimization pass is disabled. This warning does
3909 not generally indicate that there is anything wrong with your code; it
3910 merely indicates that GCC's optimizers were unable to handle the code
3911 effectively. Often, the problem is that your code is too big or too
3912 complex; GCC will refuse to optimize programs when the optimization
3913 itself is likely to take inordinate amounts of time.
3915 @item -Wpointer-sign
3916 @opindex Wpointer-sign
3917 @opindex Wno-pointer-sign
3918 Warn for pointer argument passing or assignment with different signedness.
3919 This option is only supported for C and Objective-C@. It is implied by
3920 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3921 @option{-Wno-pointer-sign}.
3923 @item -Wstack-protector
3924 @opindex Wstack-protector
3925 This option is only active when @option{-fstack-protector} is active. It
3926 warns about functions that will not be protected against stack smashing.
3928 @item -Woverlength-strings
3929 @opindex Woverlength-strings
3930 Warn about string constants which are longer than the ``minimum
3931 maximum'' length specified in the C standard. Modern compilers
3932 generally allow string constants which are much longer than the
3933 standard's minimum limit, but very portable programs should avoid
3934 using longer strings.
3936 The limit applies @emph{after} string constant concatenation, and does
3937 not count the trailing NUL@. In C89, the limit was 509 characters; in
3938 C99, it was raised to 4095. C++98 does not specify a normative
3939 minimum maximum, so we do not diagnose overlength strings in C++@.
3941 This option is implied by @option{-pedantic}, and can be disabled with
3942 @option{-Wno-overlength-strings}.
3945 @node Debugging Options
3946 @section Options for Debugging Your Program or GCC
3947 @cindex options, debugging
3948 @cindex debugging information options
3950 GCC has various special options that are used for debugging
3951 either your program or GCC:
3956 Produce debugging information in the operating system's native format
3957 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3960 On most systems that use stabs format, @option{-g} enables use of extra
3961 debugging information that only GDB can use; this extra information
3962 makes debugging work better in GDB but will probably make other debuggers
3964 refuse to read the program. If you want to control for certain whether
3965 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3966 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3968 GCC allows you to use @option{-g} with
3969 @option{-O}. The shortcuts taken by optimized code may occasionally
3970 produce surprising results: some variables you declared may not exist
3971 at all; flow of control may briefly move where you did not expect it;
3972 some statements may not be executed because they compute constant
3973 results or their values were already at hand; some statements may
3974 execute in different places because they were moved out of loops.
3976 Nevertheless it proves possible to debug optimized output. This makes
3977 it reasonable to use the optimizer for programs that might have bugs.
3979 The following options are useful when GCC is generated with the
3980 capability for more than one debugging format.
3984 Produce debugging information for use by GDB@. This means to use the
3985 most expressive format available (DWARF 2, stabs, or the native format
3986 if neither of those are supported), including GDB extensions if at all
3991 Produce debugging information in stabs format (if that is supported),
3992 without GDB extensions. This is the format used by DBX on most BSD
3993 systems. On MIPS, Alpha and System V Release 4 systems this option
3994 produces stabs debugging output which is not understood by DBX or SDB@.
3995 On System V Release 4 systems this option requires the GNU assembler.
3997 @item -feliminate-unused-debug-symbols
3998 @opindex feliminate-unused-debug-symbols
3999 Produce debugging information in stabs format (if that is supported),
4000 for only symbols that are actually used.
4002 @item -femit-class-debug-always
4003 Instead of emitting debugging information for a C++ class in only one
4004 object file, emit it in all object files using the class. This option
4005 should be used only with debuggers that are unable to handle the way GCC
4006 normally emits debugging information for classes because using this
4007 option will increase the size of debugging information by as much as a
4012 Produce debugging information in stabs format (if that is supported),
4013 using GNU extensions understood only by the GNU debugger (GDB)@. The
4014 use of these extensions is likely to make other debuggers crash or
4015 refuse to read the program.
4019 Produce debugging information in COFF format (if that is supported).
4020 This is the format used by SDB on most System V systems prior to
4025 Produce debugging information in XCOFF format (if that is supported).
4026 This is the format used by the DBX debugger on IBM RS/6000 systems.
4030 Produce debugging information in XCOFF format (if that is supported),
4031 using GNU extensions understood only by the GNU debugger (GDB)@. The
4032 use of these extensions is likely to make other debuggers crash or
4033 refuse to read the program, and may cause assemblers other than the GNU
4034 assembler (GAS) to fail with an error.
4038 Produce debugging information in DWARF version 2 format (if that is
4039 supported). This is the format used by DBX on IRIX 6. With this
4040 option, GCC uses features of DWARF version 3 when they are useful;
4041 version 3 is upward compatible with version 2, but may still cause
4042 problems for older debuggers.
4046 Produce debugging information in VMS debug format (if that is
4047 supported). This is the format used by DEBUG on VMS systems.
4050 @itemx -ggdb@var{level}
4051 @itemx -gstabs@var{level}
4052 @itemx -gcoff@var{level}
4053 @itemx -gxcoff@var{level}
4054 @itemx -gvms@var{level}
4055 Request debugging information and also use @var{level} to specify how
4056 much information. The default level is 2.
4058 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4061 Level 1 produces minimal information, enough for making backtraces in
4062 parts of the program that you don't plan to debug. This includes
4063 descriptions of functions and external variables, but no information
4064 about local variables and no line numbers.
4066 Level 3 includes extra information, such as all the macro definitions
4067 present in the program. Some debuggers support macro expansion when
4068 you use @option{-g3}.
4070 @option{-gdwarf-2} does not accept a concatenated debug level, because
4071 GCC used to support an option @option{-gdwarf} that meant to generate
4072 debug information in version 1 of the DWARF format (which is very
4073 different from version 2), and it would have been too confusing. That
4074 debug format is long obsolete, but the option cannot be changed now.
4075 Instead use an additional @option{-g@var{level}} option to change the
4076 debug level for DWARF2.
4078 @item -feliminate-dwarf2-dups
4079 @opindex feliminate-dwarf2-dups
4080 Compress DWARF2 debugging information by eliminating duplicated
4081 information about each symbol. This option only makes sense when
4082 generating DWARF2 debugging information with @option{-gdwarf-2}.
4084 @item -femit-struct-debug-baseonly
4085 Emit debug information for struct-like types
4086 only when the base name of the compilation source file
4087 matches the base name of file in which the struct was defined.
4089 This option substantially reduces the size of debugging information,
4090 but at significant potential loss in type information to the debugger.
4091 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4092 See @option{-femit-struct-debug-detailed} for more detailed control.
4094 This option works only with DWARF 2.
4096 @item -femit-struct-debug-reduced
4097 Emit debug information for struct-like types
4098 only when the base name of the compilation source file
4099 matches the base name of file in which the type was defined,
4100 unless the struct is a template or defined in a system header.
4102 This option significantly reduces the size of debugging information,
4103 with some potential loss in type information to the debugger.
4104 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4105 See @option{-femit-struct-debug-detailed} for more detailed control.
4107 This option works only with DWARF 2.
4109 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4110 Specify the struct-like types
4111 for which the compiler will generate debug information.
4112 The intent is to reduce duplicate struct debug information
4113 between different object files within the same program.
4115 This option is a detailed version of
4116 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4117 which will serve for most needs.
4119 A specification has the syntax
4120 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4122 The optional first word limits the specification to
4123 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4124 A struct type is used directly when it is the type of a variable, member.
4125 Indirect uses arise through pointers to structs.
4126 That is, when use of an incomplete struct would be legal, the use is indirect.
4128 @samp{struct one direct; struct two * indirect;}.
4130 The optional second word limits the specification to
4131 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4132 Generic structs are a bit complicated to explain.
4133 For C++, these are non-explicit specializations of template classes,
4134 or non-template classes within the above.
4135 Other programming languages have generics,
4136 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4138 The third word specifies the source files for those
4139 structs for which the compiler will emit debug information.
4140 The values @samp{none} and @samp{any} have the normal meaning.
4141 The value @samp{base} means that
4142 the base of name of the file in which the type declaration appears
4143 must match the base of the name of the main compilation file.
4144 In practice, this means that
4145 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4146 but types declared in other header will not.
4147 The value @samp{sys} means those types satisfying @samp{base}
4148 or declared in system or compiler headers.
4150 You may need to experiment to determine the best settings for your application.
4152 The default is @samp{-femit-struct-debug-detailed=all}.
4154 This option works only with DWARF 2.
4156 @item -fdebug-prefix-map=@var{old}=@var{new}
4157 @opindex fdebug-prefix-map
4158 When compiling files in directory @file{@var{old}}, record debugging
4159 information describing them as in @file{@var{new}} instead.
4161 @cindex @command{prof}
4164 Generate extra code to write profile information suitable for the
4165 analysis program @command{prof}. You must use this option when compiling
4166 the source files you want data about, and you must also use it when
4169 @cindex @command{gprof}
4172 Generate extra code to write profile information suitable for the
4173 analysis program @command{gprof}. You must use this option when compiling
4174 the source files you want data about, and you must also use it when
4179 Makes the compiler print out each function name as it is compiled, and
4180 print some statistics about each pass when it finishes.
4183 @opindex ftime-report
4184 Makes the compiler print some statistics about the time consumed by each
4185 pass when it finishes.
4188 @opindex fmem-report
4189 Makes the compiler print some statistics about permanent memory
4190 allocation when it finishes.
4192 @item -fpre-ipa-mem-report
4193 @opindex fpre-ipa-mem-report
4194 @item -fpost-ipa-mem-report
4195 @opindex fpost-ipa-mem-report
4196 Makes the compiler print some statistics about permanent memory
4197 allocation before or after interprocedural optimization.
4199 @item -fprofile-arcs
4200 @opindex fprofile-arcs
4201 Add code so that program flow @dfn{arcs} are instrumented. During
4202 execution the program records how many times each branch and call is
4203 executed and how many times it is taken or returns. When the compiled
4204 program exits it saves this data to a file called
4205 @file{@var{auxname}.gcda} for each source file. The data may be used for
4206 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4207 test coverage analysis (@option{-ftest-coverage}). Each object file's
4208 @var{auxname} is generated from the name of the output file, if
4209 explicitly specified and it is not the final executable, otherwise it is
4210 the basename of the source file. In both cases any suffix is removed
4211 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4212 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4213 @xref{Cross-profiling}.
4215 @cindex @command{gcov}
4219 This option is used to compile and link code instrumented for coverage
4220 analysis. The option is a synonym for @option{-fprofile-arcs}
4221 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4222 linking). See the documentation for those options for more details.
4227 Compile the source files with @option{-fprofile-arcs} plus optimization
4228 and code generation options. For test coverage analysis, use the
4229 additional @option{-ftest-coverage} option. You do not need to profile
4230 every source file in a program.
4233 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4234 (the latter implies the former).
4237 Run the program on a representative workload to generate the arc profile
4238 information. This may be repeated any number of times. You can run
4239 concurrent instances of your program, and provided that the file system
4240 supports locking, the data files will be correctly updated. Also
4241 @code{fork} calls are detected and correctly handled (double counting
4245 For profile-directed optimizations, compile the source files again with
4246 the same optimization and code generation options plus
4247 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4248 Control Optimization}).
4251 For test coverage analysis, use @command{gcov} to produce human readable
4252 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4253 @command{gcov} documentation for further information.
4257 With @option{-fprofile-arcs}, for each function of your program GCC
4258 creates a program flow graph, then finds a spanning tree for the graph.
4259 Only arcs that are not on the spanning tree have to be instrumented: the
4260 compiler adds code to count the number of times that these arcs are
4261 executed. When an arc is the only exit or only entrance to a block, the
4262 instrumentation code can be added to the block; otherwise, a new basic
4263 block must be created to hold the instrumentation code.
4266 @item -ftest-coverage
4267 @opindex ftest-coverage
4268 Produce a notes file that the @command{gcov} code-coverage utility
4269 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4270 show program coverage. Each source file's note file is called
4271 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4272 above for a description of @var{auxname} and instructions on how to
4273 generate test coverage data. Coverage data will match the source files
4274 more closely, if you do not optimize.
4276 @item -fdbg-cnt-list
4277 @opindex fdbg-cnt-list
4278 Print the name and the counter upperbound for all debug counters.
4280 @item -fdbg-cnt=@var{counter-value-list}
4282 Set the internal debug counter upperbound. @var{counter-value-list}
4283 is a comma-separated list of @var{name}:@var{value} pairs
4284 which sets the upperbound of each debug counter @var{name} to @var{value}.
4285 All debug counters have the initial upperbound of @var{UINT_MAX},
4286 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4287 e.g. With -fdbg-cnt=dce:10,tail_call:0
4288 dbg_cnt(dce) will return true only for first 10 invocations
4289 and dbg_cnt(tail_call) will return false always.
4291 @item -d@var{letters}
4292 @item -fdump-rtl-@var{pass}
4294 Says to make debugging dumps during compilation at times specified by
4295 @var{letters}. This is used for debugging the RTL-based passes of the
4296 compiler. The file names for most of the dumps are made by appending a
4297 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4298 from the name of the output file, if explicitly specified and it is not
4299 an executable, otherwise it is the basename of the source file. These
4300 switches may have different effects when @option{-E} is used for
4303 Most debug dumps can be enabled either passing a letter to the @option{-d}
4304 option, or with a long @option{-fdump-rtl} switch; here are the possible
4305 letters for use in @var{letters} and @var{pass}, and their meanings:
4310 Annotate the assembler output with miscellaneous debugging information.
4313 @itemx -fdump-rtl-bbro
4315 @opindex fdump-rtl-bbro
4316 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4319 @itemx -fdump-rtl-combine
4321 @opindex fdump-rtl-combine
4322 Dump after the RTL instruction combination pass, to the file
4323 @file{@var{file}.129r.combine}.
4326 @itemx -fdump-rtl-ce1
4327 @itemx -fdump-rtl-ce2
4329 @opindex fdump-rtl-ce1
4330 @opindex fdump-rtl-ce2
4331 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4332 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4333 and @option{-fdump-rtl-ce2} enable dumping after the second if
4334 conversion, to the file @file{@var{file}.130r.ce2}.
4337 @itemx -fdump-rtl-btl
4338 @itemx -fdump-rtl-dbr
4340 @opindex fdump-rtl-btl
4341 @opindex fdump-rtl-dbr
4342 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4343 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4344 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4345 scheduling, to @file{@var{file}.36.dbr}.
4349 Dump all macro definitions, at the end of preprocessing, in addition to
4353 @itemx -fdump-rtl-ce3
4355 @opindex fdump-rtl-ce3
4356 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4359 @itemx -fdump-rtl-cfg
4360 @itemx -fdump-rtl-life
4362 @opindex fdump-rtl-cfg
4363 @opindex fdump-rtl-life
4364 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4365 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4366 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4367 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4370 @itemx -fdump-rtl-greg
4372 @opindex fdump-rtl-greg
4373 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4376 @itemx -fdump-rtl-gcse
4377 @itemx -fdump-rtl-bypass
4379 @opindex fdump-rtl-gcse
4380 @opindex fdump-rtl-bypass
4381 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4382 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4383 enable dumping after jump bypassing and control flow optimizations, to
4384 @file{@var{file}.115r.bypass}.
4387 @itemx -fdump-rtl-eh
4389 @opindex fdump-rtl-eh
4390 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4393 @itemx -fdump-rtl-sibling
4395 @opindex fdump-rtl-sibling
4396 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4399 @itemx -fdump-rtl-jump
4401 @opindex fdump-rtl-jump
4402 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4405 @itemx -fdump-rtl-stack
4407 @opindex fdump-rtl-stack
4408 Dump after conversion from GCC's "flat register file" registers to the
4409 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4412 @itemx -fdump-rtl-lreg
4414 @opindex fdump-rtl-lreg
4415 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4418 @itemx -fdump-rtl-loop2
4420 @opindex fdump-rtl-loop2
4421 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4422 loop optimization pass, to @file{@var{file}.119r.loop2},
4423 @file{@var{file}.120r.loop2_init},
4424 @file{@var{file}.121r.loop2_invariant}, and
4425 @file{@var{file}.125r.loop2_done}.
4428 @itemx -fdump-rtl-sms
4430 @opindex fdump-rtl-sms
4431 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4434 @itemx -fdump-rtl-mach
4436 @opindex fdump-rtl-mach
4437 Dump after performing the machine dependent reorganization pass, to
4438 @file{@var{file}.155r.mach} if that pass exists.
4441 @itemx -fdump-rtl-rnreg
4443 @opindex fdump-rtl-rnreg
4444 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4447 @itemx -fdump-rtl-regmove
4449 @opindex fdump-rtl-regmove
4450 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4453 @itemx -fdump-rtl-postreload
4455 @opindex fdump-rtl-postreload
4456 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4459 @itemx -fdump-rtl-expand
4461 @opindex fdump-rtl-expand
4462 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4465 @itemx -fdump-rtl-sched2
4467 @opindex fdump-rtl-sched2
4468 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4471 @itemx -fdump-rtl-cse
4473 @opindex fdump-rtl-cse
4474 Dump after CSE (including the jump optimization that sometimes follows
4475 CSE), to @file{@var{file}.113r.cse}.
4478 @itemx -fdump-rtl-sched1
4480 @opindex fdump-rtl-sched1
4481 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4484 @itemx -fdump-rtl-cse2
4486 @opindex fdump-rtl-cse2
4487 Dump after the second CSE pass (including the jump optimization that
4488 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4491 @itemx -fdump-rtl-tracer
4493 @opindex fdump-rtl-tracer
4494 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4497 @itemx -fdump-rtl-vpt
4498 @itemx -fdump-rtl-vartrack
4500 @opindex fdump-rtl-vpt
4501 @opindex fdump-rtl-vartrack
4502 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4503 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4504 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4505 to @file{@var{file}.154r.vartrack}.
4508 @itemx -fdump-rtl-flow2
4510 @opindex fdump-rtl-flow2
4511 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4514 @itemx -fdump-rtl-peephole2
4516 @opindex fdump-rtl-peephole2
4517 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4520 @itemx -fdump-rtl-web
4522 @opindex fdump-rtl-web
4523 Dump after live range splitting, to @file{@var{file}.126r.web}.
4526 @itemx -fdump-rtl-all
4528 @opindex fdump-rtl-all
4529 Produce all the dumps listed above.
4533 Produce a core dump whenever an error occurs.
4537 Print statistics on memory usage, at the end of the run, to
4542 Annotate the assembler output with a comment indicating which
4543 pattern and alternative was used. The length of each instruction is
4548 Dump the RTL in the assembler output as a comment before each instruction.
4549 Also turns on @option{-dp} annotation.
4553 For each of the other indicated dump files (either with @option{-d} or
4554 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4555 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4559 Just generate RTL for a function instead of compiling it. Usually used
4560 with @samp{r} (@option{-fdump-rtl-expand}).
4564 Dump debugging information during parsing, to standard error.
4568 @opindex fdump-noaddr
4569 When doing debugging dumps (see @option{-d} option above), suppress
4570 address output. This makes it more feasible to use diff on debugging
4571 dumps for compiler invocations with different compiler binaries and/or
4572 different text / bss / data / heap / stack / dso start locations.
4574 @item -fdump-unnumbered
4575 @opindex fdump-unnumbered
4576 When doing debugging dumps (see @option{-d} option above), suppress instruction
4577 numbers and address output. This makes it more feasible to
4578 use diff on debugging dumps for compiler invocations with different
4579 options, in particular with and without @option{-g}.
4581 @item -fdump-translation-unit @r{(C++ only)}
4582 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4583 @opindex fdump-translation-unit
4584 Dump a representation of the tree structure for the entire translation
4585 unit to a file. The file name is made by appending @file{.tu} to the
4586 source file name. If the @samp{-@var{options}} form is used, @var{options}
4587 controls the details of the dump as described for the
4588 @option{-fdump-tree} options.
4590 @item -fdump-class-hierarchy @r{(C++ only)}
4591 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4592 @opindex fdump-class-hierarchy
4593 Dump a representation of each class's hierarchy and virtual function
4594 table layout to a file. The file name is made by appending @file{.class}
4595 to the source file name. If the @samp{-@var{options}} form is used,
4596 @var{options} controls the details of the dump as described for the
4597 @option{-fdump-tree} options.
4599 @item -fdump-ipa-@var{switch}
4601 Control the dumping at various stages of inter-procedural analysis
4602 language tree to a file. The file name is generated by appending a switch
4603 specific suffix to the source file name. The following dumps are possible:
4607 Enables all inter-procedural analysis dumps; currently the only produced
4608 dump is the @samp{cgraph} dump.
4611 Dumps information about call-graph optimization, unused function removal,
4612 and inlining decisions.
4615 @item -fdump-tree-@var{switch}
4616 @itemx -fdump-tree-@var{switch}-@var{options}
4618 Control the dumping at various stages of processing the intermediate
4619 language tree to a file. The file name is generated by appending a switch
4620 specific suffix to the source file name. If the @samp{-@var{options}}
4621 form is used, @var{options} is a list of @samp{-} separated options that
4622 control the details of the dump. Not all options are applicable to all
4623 dumps, those which are not meaningful will be ignored. The following
4624 options are available
4628 Print the address of each node. Usually this is not meaningful as it
4629 changes according to the environment and source file. Its primary use
4630 is for tying up a dump file with a debug environment.
4632 Inhibit dumping of members of a scope or body of a function merely
4633 because that scope has been reached. Only dump such items when they
4634 are directly reachable by some other path. When dumping pretty-printed
4635 trees, this option inhibits dumping the bodies of control structures.
4637 Print a raw representation of the tree. By default, trees are
4638 pretty-printed into a C-like representation.
4640 Enable more detailed dumps (not honored by every dump option).
4642 Enable dumping various statistics about the pass (not honored by every dump
4645 Enable showing basic block boundaries (disabled in raw dumps).
4647 Enable showing virtual operands for every statement.
4649 Enable showing line numbers for statements.
4651 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4653 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4656 The following tree dumps are possible:
4660 Dump before any tree based optimization, to @file{@var{file}.original}.
4663 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4666 Dump after function inlining, to @file{@var{file}.inlined}.
4669 @opindex fdump-tree-gimple
4670 Dump each function before and after the gimplification pass to a file. The
4671 file name is made by appending @file{.gimple} to the source file name.
4674 @opindex fdump-tree-cfg
4675 Dump the control flow graph of each function to a file. The file name is
4676 made by appending @file{.cfg} to the source file name.
4679 @opindex fdump-tree-vcg
4680 Dump the control flow graph of each function to a file in VCG format. The
4681 file name is made by appending @file{.vcg} to the source file name. Note
4682 that if the file contains more than one function, the generated file cannot
4683 be used directly by VCG@. You will need to cut and paste each function's
4684 graph into its own separate file first.
4687 @opindex fdump-tree-ch
4688 Dump each function after copying loop headers. The file name is made by
4689 appending @file{.ch} to the source file name.
4692 @opindex fdump-tree-ssa
4693 Dump SSA related information to a file. The file name is made by appending
4694 @file{.ssa} to the source file name.
4697 @opindex fdump-tree-salias
4698 Dump structure aliasing variable information to a file. This file name
4699 is made by appending @file{.salias} to the source file name.
4702 @opindex fdump-tree-alias
4703 Dump aliasing information for each function. The file name is made by
4704 appending @file{.alias} to the source file name.
4707 @opindex fdump-tree-ccp
4708 Dump each function after CCP@. The file name is made by appending
4709 @file{.ccp} to the source file name.
4712 @opindex fdump-tree-storeccp
4713 Dump each function after STORE-CCP. The file name is made by appending
4714 @file{.storeccp} to the source file name.
4717 @opindex fdump-tree-pre
4718 Dump trees after partial redundancy elimination. The file name is made
4719 by appending @file{.pre} to the source file name.
4722 @opindex fdump-tree-fre
4723 Dump trees after full redundancy elimination. The file name is made
4724 by appending @file{.fre} to the source file name.
4727 @opindex fdump-tree-copyprop
4728 Dump trees after copy propagation. The file name is made
4729 by appending @file{.copyprop} to the source file name.
4731 @item store_copyprop
4732 @opindex fdump-tree-store_copyprop
4733 Dump trees after store copy-propagation. The file name is made
4734 by appending @file{.store_copyprop} to the source file name.
4737 @opindex fdump-tree-dce
4738 Dump each function after dead code elimination. The file name is made by
4739 appending @file{.dce} to the source file name.
4742 @opindex fdump-tree-mudflap
4743 Dump each function after adding mudflap instrumentation. The file name is
4744 made by appending @file{.mudflap} to the source file name.
4747 @opindex fdump-tree-sra
4748 Dump each function after performing scalar replacement of aggregates. The
4749 file name is made by appending @file{.sra} to the source file name.
4752 @opindex fdump-tree-sink
4753 Dump each function after performing code sinking. The file name is made
4754 by appending @file{.sink} to the source file name.
4757 @opindex fdump-tree-dom
4758 Dump each function after applying dominator tree optimizations. The file
4759 name is made by appending @file{.dom} to the source file name.
4762 @opindex fdump-tree-dse
4763 Dump each function after applying dead store elimination. The file
4764 name is made by appending @file{.dse} to the source file name.
4767 @opindex fdump-tree-phiopt
4768 Dump each function after optimizing PHI nodes into straightline code. The file
4769 name is made by appending @file{.phiopt} to the source file name.
4772 @opindex fdump-tree-forwprop
4773 Dump each function after forward propagating single use variables. The file
4774 name is made by appending @file{.forwprop} to the source file name.
4777 @opindex fdump-tree-copyrename
4778 Dump each function after applying the copy rename optimization. The file
4779 name is made by appending @file{.copyrename} to the source file name.
4782 @opindex fdump-tree-nrv
4783 Dump each function after applying the named return value optimization on
4784 generic trees. The file name is made by appending @file{.nrv} to the source
4788 @opindex fdump-tree-vect
4789 Dump each function after applying vectorization of loops. The file name is
4790 made by appending @file{.vect} to the source file name.
4793 @opindex fdump-tree-vrp
4794 Dump each function after Value Range Propagation (VRP). The file name
4795 is made by appending @file{.vrp} to the source file name.
4798 @opindex fdump-tree-all
4799 Enable all the available tree dumps with the flags provided in this option.
4802 @item -ftree-vectorizer-verbose=@var{n}
4803 @opindex ftree-vectorizer-verbose
4804 This option controls the amount of debugging output the vectorizer prints.
4805 This information is written to standard error, unless
4806 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4807 in which case it is output to the usual dump listing file, @file{.vect}.
4808 For @var{n}=0 no diagnostic information is reported.
4809 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4810 and the total number of loops that got vectorized.
4811 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4812 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4813 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4814 level that @option{-fdump-tree-vect-stats} uses.
4815 Higher verbosity levels mean either more information dumped for each
4816 reported loop, or same amount of information reported for more loops:
4817 If @var{n}=3, alignment related information is added to the reports.
4818 If @var{n}=4, data-references related information (e.g. memory dependences,
4819 memory access-patterns) is added to the reports.
4820 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4821 that did not pass the first analysis phase (i.e. may not be countable, or
4822 may have complicated control-flow).
4823 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4824 For @var{n}=7, all the information the vectorizer generates during its
4825 analysis and transformation is reported. This is the same verbosity level
4826 that @option{-fdump-tree-vect-details} uses.
4828 @item -frandom-seed=@var{string}
4829 @opindex frandom-string
4830 This option provides a seed that GCC uses when it would otherwise use
4831 random numbers. It is used to generate certain symbol names
4832 that have to be different in every compiled file. It is also used to
4833 place unique stamps in coverage data files and the object files that
4834 produce them. You can use the @option{-frandom-seed} option to produce
4835 reproducibly identical object files.
4837 The @var{string} should be different for every file you compile.
4839 @item -fsched-verbose=@var{n}
4840 @opindex fsched-verbose
4841 On targets that use instruction scheduling, this option controls the
4842 amount of debugging output the scheduler prints. This information is
4843 written to standard error, unless @option{-dS} or @option{-dR} is
4844 specified, in which case it is output to the usual dump
4845 listing file, @file{.sched} or @file{.sched2} respectively. However
4846 for @var{n} greater than nine, the output is always printed to standard
4849 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4850 same information as @option{-dRS}. For @var{n} greater than one, it
4851 also output basic block probabilities, detailed ready list information
4852 and unit/insn info. For @var{n} greater than two, it includes RTL
4853 at abort point, control-flow and regions info. And for @var{n} over
4854 four, @option{-fsched-verbose} also includes dependence info.
4858 Store the usual ``temporary'' intermediate files permanently; place them
4859 in the current directory and name them based on the source file. Thus,
4860 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4861 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4862 preprocessed @file{foo.i} output file even though the compiler now
4863 normally uses an integrated preprocessor.
4865 When used in combination with the @option{-x} command line option,
4866 @option{-save-temps} is sensible enough to avoid over writing an
4867 input source file with the same extension as an intermediate file.
4868 The corresponding intermediate file may be obtained by renaming the
4869 source file before using @option{-save-temps}.
4873 Report the CPU time taken by each subprocess in the compilation
4874 sequence. For C source files, this is the compiler proper and assembler
4875 (plus the linker if linking is done). The output looks like this:
4882 The first number on each line is the ``user time'', that is time spent
4883 executing the program itself. The second number is ``system time'',
4884 time spent executing operating system routines on behalf of the program.
4885 Both numbers are in seconds.
4887 @item -fvar-tracking
4888 @opindex fvar-tracking
4889 Run variable tracking pass. It computes where variables are stored at each
4890 position in code. Better debugging information is then generated
4891 (if the debugging information format supports this information).
4893 It is enabled by default when compiling with optimization (@option{-Os},
4894 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4895 the debug info format supports it.
4897 @item -print-file-name=@var{library}
4898 @opindex print-file-name
4899 Print the full absolute name of the library file @var{library} that
4900 would be used when linking---and don't do anything else. With this
4901 option, GCC does not compile or link anything; it just prints the
4904 @item -print-multi-directory
4905 @opindex print-multi-directory
4906 Print the directory name corresponding to the multilib selected by any
4907 other switches present in the command line. This directory is supposed
4908 to exist in @env{GCC_EXEC_PREFIX}.
4910 @item -print-multi-lib
4911 @opindex print-multi-lib
4912 Print the mapping from multilib directory names to compiler switches
4913 that enable them. The directory name is separated from the switches by
4914 @samp{;}, and each switch starts with an @samp{@@} instead of the
4915 @samp{-}, without spaces between multiple switches. This is supposed to
4916 ease shell-processing.
4918 @item -print-prog-name=@var{program}
4919 @opindex print-prog-name
4920 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4922 @item -print-libgcc-file-name
4923 @opindex print-libgcc-file-name
4924 Same as @option{-print-file-name=libgcc.a}.
4926 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4927 but you do want to link with @file{libgcc.a}. You can do
4930 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4933 @item -print-search-dirs
4934 @opindex print-search-dirs
4935 Print the name of the configured installation directory and a list of
4936 program and library directories @command{gcc} will search---and don't do anything else.
4938 This is useful when @command{gcc} prints the error message
4939 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4940 To resolve this you either need to put @file{cpp0} and the other compiler
4941 components where @command{gcc} expects to find them, or you can set the environment
4942 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4943 Don't forget the trailing @samp{/}.
4944 @xref{Environment Variables}.
4946 @item -print-sysroot-headers-suffix
4947 @opindex print-sysroot-headers-suffix
4948 Print the suffix added to the target sysroot when searching for
4949 headers, or give an error if the compiler is not configured with such
4950 a suffix---and don't do anything else.
4953 @opindex dumpmachine
4954 Print the compiler's target machine (for example,
4955 @samp{i686-pc-linux-gnu})---and don't do anything else.
4958 @opindex dumpversion
4959 Print the compiler version (for example, @samp{3.0})---and don't do
4964 Print the compiler's built-in specs---and don't do anything else. (This
4965 is used when GCC itself is being built.) @xref{Spec Files}.
4967 @item -feliminate-unused-debug-types
4968 @opindex feliminate-unused-debug-types
4969 Normally, when producing DWARF2 output, GCC will emit debugging
4970 information for all types declared in a compilation
4971 unit, regardless of whether or not they are actually used
4972 in that compilation unit. Sometimes this is useful, such as
4973 if, in the debugger, you want to cast a value to a type that is
4974 not actually used in your program (but is declared). More often,
4975 however, this results in a significant amount of wasted space.
4976 With this option, GCC will avoid producing debug symbol output
4977 for types that are nowhere used in the source file being compiled.
4980 @node Optimize Options
4981 @section Options That Control Optimization
4982 @cindex optimize options
4983 @cindex options, optimization
4985 These options control various sorts of optimizations.
4987 Without any optimization option, the compiler's goal is to reduce the
4988 cost of compilation and to make debugging produce the expected
4989 results. Statements are independent: if you stop the program with a
4990 breakpoint between statements, you can then assign a new value to any
4991 variable or change the program counter to any other statement in the
4992 function and get exactly the results you would expect from the source
4995 Turning on optimization flags makes the compiler attempt to improve
4996 the performance and/or code size at the expense of compilation time
4997 and possibly the ability to debug the program.
4999 The compiler performs optimization based on the knowledge it has of
5000 the program. Optimization levels @option{-O} and above, in
5001 particular, enable @emph{unit-at-a-time} mode, which allows the
5002 compiler to consider information gained from later functions in
5003 the file when compiling a function. Compiling multiple files at
5004 once to a single output file in @emph{unit-at-a-time} mode allows
5005 the compiler to use information gained from all of the files when
5006 compiling each of them.
5008 Not all optimizations are controlled directly by a flag. Only
5009 optimizations that have a flag are listed.
5016 Optimize. Optimizing compilation takes somewhat more time, and a lot
5017 more memory for a large function.
5019 With @option{-O}, the compiler tries to reduce code size and execution
5020 time, without performing any optimizations that take a great deal of
5023 @option{-O} turns on the following optimization flags:
5026 -fcprop-registers @gol
5029 -fdelayed-branch @gol
5031 -fguess-branch-probability @gol
5032 -fif-conversion2 @gol
5033 -fif-conversion @gol
5034 -finline-small-functions @gol
5035 -fipa-pure-const @gol
5036 -fipa-reference @gol
5038 -fsplit-wide-types @gol
5041 -ftree-copyrename @gol
5043 -ftree-dominator-opts @gol
5050 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5051 where doing so does not interfere with debugging.
5055 Optimize even more. GCC performs nearly all supported optimizations
5056 that do not involve a space-speed tradeoff. The compiler does not
5057 perform loop unrolling or function inlining when you specify @option{-O2}.
5058 As compared to @option{-O}, this option increases both compilation time
5059 and the performance of the generated code.
5061 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5062 also turns on the following optimization flags:
5063 @gccoptlist{-fthread-jumps @gol
5064 -falign-functions -falign-jumps @gol
5065 -falign-loops -falign-labels @gol
5068 -fcse-follow-jumps -fcse-skip-blocks @gol
5069 -fdelete-null-pointer-checks @gol
5070 -fexpensive-optimizations @gol
5071 -fgcse -fgcse-lm @gol
5072 -foptimize-sibling-calls @gol
5075 -freorder-blocks -freorder-functions @gol
5076 -frerun-cse-after-loop @gol
5077 -fsched-interblock -fsched-spec @gol
5078 -fschedule-insns -fschedule-insns2 @gol
5079 -fstrict-aliasing -fstrict-overflow @gol
5083 Please note the warning under @option{-fgcse} about
5084 invoking @option{-O2} on programs that use computed gotos.
5088 Optimize yet more. @option{-O3} turns on all optimizations specified by
5089 @option{-O2} and also turns on the @option{-finline-functions},
5090 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5091 @option{-fgcse-after-reload} options.
5095 Reduce compilation time and make debugging produce the expected
5096 results. This is the default.
5100 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5101 do not typically increase code size. It also performs further
5102 optimizations designed to reduce code size.
5104 @option{-Os} disables the following optimization flags:
5105 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5106 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5107 -fprefetch-loop-arrays -ftree-vect-loop-version}
5109 If you use multiple @option{-O} options, with or without level numbers,
5110 the last such option is the one that is effective.
5113 Options of the form @option{-f@var{flag}} specify machine-independent
5114 flags. Most flags have both positive and negative forms; the negative
5115 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5116 below, only one of the forms is listed---the one you typically will
5117 use. You can figure out the other form by either removing @samp{no-}
5120 The following options control specific optimizations. They are either
5121 activated by @option{-O} options or are related to ones that are. You
5122 can use the following flags in the rare cases when ``fine-tuning'' of
5123 optimizations to be performed is desired.
5126 @item -fno-default-inline
5127 @opindex fno-default-inline
5128 Do not make member functions inline by default merely because they are
5129 defined inside the class scope (C++ only). Otherwise, when you specify
5130 @w{@option{-O}}, member functions defined inside class scope are compiled
5131 inline by default; i.e., you don't need to add @samp{inline} in front of
5132 the member function name.
5134 @item -fno-defer-pop
5135 @opindex fno-defer-pop
5136 Always pop the arguments to each function call as soon as that function
5137 returns. For machines which must pop arguments after a function call,
5138 the compiler normally lets arguments accumulate on the stack for several
5139 function calls and pops them all at once.
5141 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5144 @opindex fforce-addr
5145 Force memory address constants to be copied into registers before
5146 doing arithmetic on them.
5148 @item -fforward-propagate
5149 @opindex fforward-propagate
5150 Perform a forward propagation pass on RTL. The pass tries to combine two
5151 instructions and checks if the result can be simplified. If loop unrolling
5152 is active, two passes are performed and the second is scheduled after
5155 This option is enabled by default at optimization levels @option{-O2},
5156 @option{-O3}, @option{-Os}.
5158 @item -fomit-frame-pointer
5159 @opindex fomit-frame-pointer
5160 Don't keep the frame pointer in a register for functions that
5161 don't need one. This avoids the instructions to save, set up and
5162 restore frame pointers; it also makes an extra register available
5163 in many functions. @strong{It also makes debugging impossible on
5166 On some machines, such as the VAX, this flag has no effect, because
5167 the standard calling sequence automatically handles the frame pointer
5168 and nothing is saved by pretending it doesn't exist. The
5169 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5170 whether a target machine supports this flag. @xref{Registers,,Register
5171 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5173 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5175 @item -foptimize-sibling-calls
5176 @opindex foptimize-sibling-calls
5177 Optimize sibling and tail recursive calls.
5179 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5183 Don't pay attention to the @code{inline} keyword. Normally this option
5184 is used to keep the compiler from expanding any functions inline.
5185 Note that if you are not optimizing, no functions can be expanded inline.
5187 @item -finline-small-functions
5188 @opindex finline-small-functions
5189 Integrate functions into their callers when their body is smaller than expected
5190 function call code (so overall size of program gets smaller). The compiler
5191 heuristically decides which functions are simple enough to be worth integrating
5194 Enabled at level @option{-O2}.
5196 @item -finline-functions
5197 @opindex finline-functions
5198 Integrate all simple functions into their callers. The compiler
5199 heuristically decides which functions are simple enough to be worth
5200 integrating in this way.
5202 If all calls to a given function are integrated, and the function is
5203 declared @code{static}, then the function is normally not output as
5204 assembler code in its own right.
5206 Enabled at level @option{-O3}.
5208 @item -finline-functions-called-once
5209 @opindex finline-functions-called-once
5210 Consider all @code{static} functions called once for inlining into their
5211 caller even if they are not marked @code{inline}. If a call to a given
5212 function is integrated, then the function is not output as assembler code
5215 Enabled if @option{-funit-at-a-time} is enabled.
5217 @item -fearly-inlining
5218 @opindex fearly-inlining
5219 Inline functions marked by @code{always_inline} and functions whose body seems
5220 smaller than the function call overhead early before doing
5221 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5222 makes profiling significantly cheaper and usually inlining faster on programs
5223 having large chains of nested wrapper functions.
5227 @item -finline-limit=@var{n}
5228 @opindex finline-limit
5229 By default, GCC limits the size of functions that can be inlined. This flag
5230 allows the control of this limit for functions that are explicitly marked as
5231 inline (i.e., marked with the inline keyword or defined within the class
5232 definition in c++). @var{n} is the size of functions that can be inlined in
5233 number of pseudo instructions (not counting parameter handling). The default
5234 value of @var{n} is 600.
5235 Increasing this value can result in more inlined code at
5236 the cost of compilation time and memory consumption. Decreasing usually makes
5237 the compilation faster and less code will be inlined (which presumably
5238 means slower programs). This option is particularly useful for programs that
5239 use inlining heavily such as those based on recursive templates with C++.
5241 Inlining is actually controlled by a number of parameters, which may be
5242 specified individually by using @option{--param @var{name}=@var{value}}.
5243 The @option{-finline-limit=@var{n}} option sets some of these parameters
5247 @item max-inline-insns-single
5248 is set to @var{n}/2.
5249 @item max-inline-insns-auto
5250 is set to @var{n}/2.
5251 @item min-inline-insns
5252 is set to 130 or @var{n}/4, whichever is smaller.
5253 @item max-inline-insns-rtl
5257 See below for a documentation of the individual
5258 parameters controlling inlining.
5260 @emph{Note:} pseudo instruction represents, in this particular context, an
5261 abstract measurement of function's size. In no way does it represent a count
5262 of assembly instructions and as such its exact meaning might change from one
5263 release to an another.
5265 @item -fkeep-inline-functions
5266 @opindex fkeep-inline-functions
5267 In C, emit @code{static} functions that are declared @code{inline}
5268 into the object file, even if the function has been inlined into all
5269 of its callers. This switch does not affect functions using the
5270 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5271 inline functions into the object file.
5273 @item -fkeep-static-consts
5274 @opindex fkeep-static-consts
5275 Emit variables declared @code{static const} when optimization isn't turned
5276 on, even if the variables aren't referenced.
5278 GCC enables this option by default. If you want to force the compiler to
5279 check if the variable was referenced, regardless of whether or not
5280 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5282 @item -fmerge-constants
5283 @opindex fmerge-constants
5284 Attempt to merge identical constants (string constants and floating point
5285 constants) across compilation units.
5287 This option is the default for optimized compilation if the assembler and
5288 linker support it. Use @option{-fno-merge-constants} to inhibit this
5291 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5293 @item -fmerge-all-constants
5294 @opindex fmerge-all-constants
5295 Attempt to merge identical constants and identical variables.
5297 This option implies @option{-fmerge-constants}. In addition to
5298 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5299 arrays or initialized constant variables with integral or floating point
5300 types. Languages like C or C++ require each non-automatic variable to
5301 have distinct location, so using this option will result in non-conforming
5304 @item -fmodulo-sched
5305 @opindex fmodulo-sched
5306 Perform swing modulo scheduling immediately before the first scheduling
5307 pass. This pass looks at innermost loops and reorders their
5308 instructions by overlapping different iterations.
5310 @item -fmodulo-sched-allow-regmoves
5311 @opindex fmodulo-sched-allow-regmoves
5312 Perform more aggressive SMS based modulo scheduling with register moves
5313 allowed. By setting this flag certain anti-dependences edges will be
5314 deleted which will trigger the generation of reg-moves based on the
5315 life-range analysis. This option is effective only with
5316 @option{-fmodulo-sched} enabled.
5318 @item -fno-branch-count-reg
5319 @opindex fno-branch-count-reg
5320 Do not use ``decrement and branch'' instructions on a count register,
5321 but instead generate a sequence of instructions that decrement a
5322 register, compare it against zero, then branch based upon the result.
5323 This option is only meaningful on architectures that support such
5324 instructions, which include x86, PowerPC, IA-64 and S/390.
5326 The default is @option{-fbranch-count-reg}.
5328 @item -fno-function-cse
5329 @opindex fno-function-cse
5330 Do not put function addresses in registers; make each instruction that
5331 calls a constant function contain the function's address explicitly.
5333 This option results in less efficient code, but some strange hacks
5334 that alter the assembler output may be confused by the optimizations
5335 performed when this option is not used.
5337 The default is @option{-ffunction-cse}
5339 @item -fno-zero-initialized-in-bss
5340 @opindex fno-zero-initialized-in-bss
5341 If the target supports a BSS section, GCC by default puts variables that
5342 are initialized to zero into BSS@. This can save space in the resulting
5345 This option turns off this behavior because some programs explicitly
5346 rely on variables going to the data section. E.g., so that the
5347 resulting executable can find the beginning of that section and/or make
5348 assumptions based on that.
5350 The default is @option{-fzero-initialized-in-bss}.
5352 @item -fmudflap -fmudflapth -fmudflapir
5356 @cindex bounds checking
5358 For front-ends that support it (C and C++), instrument all risky
5359 pointer/array dereferencing operations, some standard library
5360 string/heap functions, and some other associated constructs with
5361 range/validity tests. Modules so instrumented should be immune to
5362 buffer overflows, invalid heap use, and some other classes of C/C++
5363 programming errors. The instrumentation relies on a separate runtime
5364 library (@file{libmudflap}), which will be linked into a program if
5365 @option{-fmudflap} is given at link time. Run-time behavior of the
5366 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5367 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5370 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5371 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5372 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5373 instrumentation should ignore pointer reads. This produces less
5374 instrumentation (and therefore faster execution) and still provides
5375 some protection against outright memory corrupting writes, but allows
5376 erroneously read data to propagate within a program.
5378 @item -fthread-jumps
5379 @opindex fthread-jumps
5380 Perform optimizations where we check to see if a jump branches to a
5381 location where another comparison subsumed by the first is found. If
5382 so, the first branch is redirected to either the destination of the
5383 second branch or a point immediately following it, depending on whether
5384 the condition is known to be true or false.
5386 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5388 @item -fsplit-wide-types
5389 @opindex fsplit-wide-types
5390 When using a type that occupies multiple registers, such as @code{long
5391 long} on a 32-bit system, split the registers apart and allocate them
5392 independently. This normally generates better code for those types,
5393 but may make debugging more difficult.
5395 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5398 @item -fcse-follow-jumps
5399 @opindex fcse-follow-jumps
5400 In common subexpression elimination (CSE), scan through jump instructions
5401 when the target of the jump is not reached by any other path. For
5402 example, when CSE encounters an @code{if} statement with an
5403 @code{else} clause, CSE will follow the jump when the condition
5406 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5408 @item -fcse-skip-blocks
5409 @opindex fcse-skip-blocks
5410 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5411 follow jumps which conditionally skip over blocks. When CSE
5412 encounters a simple @code{if} statement with no else clause,
5413 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5414 body of the @code{if}.
5416 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5418 @item -frerun-cse-after-loop
5419 @opindex frerun-cse-after-loop
5420 Re-run common subexpression elimination after loop optimizations has been
5423 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5427 Perform a global common subexpression elimination pass.
5428 This pass also performs global constant and copy propagation.
5430 @emph{Note:} When compiling a program using computed gotos, a GCC
5431 extension, you may get better runtime performance if you disable
5432 the global common subexpression elimination pass by adding
5433 @option{-fno-gcse} to the command line.
5435 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5439 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5440 attempt to move loads which are only killed by stores into themselves. This
5441 allows a loop containing a load/store sequence to be changed to a load outside
5442 the loop, and a copy/store within the loop.
5444 Enabled by default when gcse is enabled.
5448 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5449 global common subexpression elimination. This pass will attempt to move
5450 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5451 loops containing a load/store sequence can be changed to a load before
5452 the loop and a store after the loop.
5454 Not enabled at any optimization level.
5458 When @option{-fgcse-las} is enabled, the global common subexpression
5459 elimination pass eliminates redundant loads that come after stores to the
5460 same memory location (both partial and full redundancies).
5462 Not enabled at any optimization level.
5464 @item -fgcse-after-reload
5465 @opindex fgcse-after-reload
5466 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5467 pass is performed after reload. The purpose of this pass is to cleanup
5470 @item -funsafe-loop-optimizations
5471 @opindex funsafe-loop-optimizations
5472 If given, the loop optimizer will assume that loop indices do not
5473 overflow, and that the loops with nontrivial exit condition are not
5474 infinite. This enables a wider range of loop optimizations even if
5475 the loop optimizer itself cannot prove that these assumptions are valid.
5476 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5477 if it finds this kind of loop.
5479 @item -fcrossjumping
5480 @opindex fcrossjumping
5481 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5482 resulting code may or may not perform better than without cross-jumping.
5484 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5486 @item -fauto-inc-dec
5487 @opindex fauto-inc-dec
5488 Combine increments or decrements of addresses with memory accesses.
5489 This pass is always skipped on architectures that do not have
5490 instructions to support this. Enabled by default at @option{-O} and
5491 higher on architectures that support this.
5495 Perform dead code elimination (DCE) on RTL.
5496 Enabled by default at @option{-O} and higher.
5500 Perform dead store elimination (DSE) on RTL.
5501 Enabled by default at @option{-O} and higher.
5503 @item -fif-conversion
5504 @opindex fif-conversion
5505 Attempt to transform conditional jumps into branch-less equivalents. This
5506 include use of conditional moves, min, max, set flags and abs instructions, and
5507 some tricks doable by standard arithmetics. The use of conditional execution
5508 on chips where it is available is controlled by @code{if-conversion2}.
5510 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5512 @item -fif-conversion2
5513 @opindex fif-conversion2
5514 Use conditional execution (where available) to transform conditional jumps into
5515 branch-less equivalents.
5517 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5519 @item -fdelete-null-pointer-checks
5520 @opindex fdelete-null-pointer-checks
5521 Use global dataflow analysis to identify and eliminate useless checks
5522 for null pointers. The compiler assumes that dereferencing a null
5523 pointer would have halted the program. If a pointer is checked after
5524 it has already been dereferenced, it cannot be null.
5526 In some environments, this assumption is not true, and programs can
5527 safely dereference null pointers. Use
5528 @option{-fno-delete-null-pointer-checks} to disable this optimization
5529 for programs which depend on that behavior.
5531 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5533 @item -fexpensive-optimizations
5534 @opindex fexpensive-optimizations
5535 Perform a number of minor optimizations that are relatively expensive.
5537 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5539 @item -foptimize-register-move
5541 @opindex foptimize-register-move
5543 Attempt to reassign register numbers in move instructions and as
5544 operands of other simple instructions in order to maximize the amount of
5545 register tying. This is especially helpful on machines with two-operand
5548 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5551 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5553 @item -fdelayed-branch
5554 @opindex fdelayed-branch
5555 If supported for the target machine, attempt to reorder instructions
5556 to exploit instruction slots available after delayed branch
5559 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5561 @item -fschedule-insns
5562 @opindex fschedule-insns
5563 If supported for the target machine, attempt to reorder instructions to
5564 eliminate execution stalls due to required data being unavailable. This
5565 helps machines that have slow floating point or memory load instructions
5566 by allowing other instructions to be issued until the result of the load
5567 or floating point instruction is required.
5569 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5571 @item -fschedule-insns2
5572 @opindex fschedule-insns2
5573 Similar to @option{-fschedule-insns}, but requests an additional pass of
5574 instruction scheduling after register allocation has been done. This is
5575 especially useful on machines with a relatively small number of
5576 registers and where memory load instructions take more than one cycle.
5578 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5580 @item -fno-sched-interblock
5581 @opindex fno-sched-interblock
5582 Don't schedule instructions across basic blocks. This is normally
5583 enabled by default when scheduling before register allocation, i.e.@:
5584 with @option{-fschedule-insns} or at @option{-O2} or higher.
5586 @item -fno-sched-spec
5587 @opindex fno-sched-spec
5588 Don't allow speculative motion of non-load instructions. This is normally
5589 enabled by default when scheduling before register allocation, i.e.@:
5590 with @option{-fschedule-insns} or at @option{-O2} or higher.
5592 @item -fsched-spec-load
5593 @opindex fsched-spec-load
5594 Allow speculative motion of some load instructions. This only makes
5595 sense when scheduling before register allocation, i.e.@: with
5596 @option{-fschedule-insns} or at @option{-O2} or higher.
5598 @item -fsched-spec-load-dangerous
5599 @opindex fsched-spec-load-dangerous
5600 Allow speculative motion of more load instructions. This only makes
5601 sense when scheduling before register allocation, i.e.@: with
5602 @option{-fschedule-insns} or at @option{-O2} or higher.
5604 @item -fsched-stalled-insns
5605 @itemx -fsched-stalled-insns=@var{n}
5606 @opindex fsched-stalled-insns
5607 Define how many insns (if any) can be moved prematurely from the queue
5608 of stalled insns into the ready list, during the second scheduling pass.
5609 @option{-fno-sched-stalled-insns} means that no insns will be moved
5610 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5611 on how many queued insns can be moved prematurely.
5612 @option{-fsched-stalled-insns} without a value is equivalent to
5613 @option{-fsched-stalled-insns=1}.
5615 @item -fsched-stalled-insns-dep
5616 @itemx -fsched-stalled-insns-dep=@var{n}
5617 @opindex fsched-stalled-insns-dep
5618 Define how many insn groups (cycles) will be examined for a dependency
5619 on a stalled insn that is candidate for premature removal from the queue
5620 of stalled insns. This has an effect only during the second scheduling pass,
5621 and only if @option{-fsched-stalled-insns} is used.
5622 @option{-fno-sched-stalled-insns-dep} is equivalent to
5623 @option{-fsched-stalled-insns-dep=0}.
5624 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5625 @option{-fsched-stalled-insns-dep=1}.
5627 @item -fsched2-use-superblocks
5628 @opindex fsched2-use-superblocks
5629 When scheduling after register allocation, do use superblock scheduling
5630 algorithm. Superblock scheduling allows motion across basic block boundaries
5631 resulting on faster schedules. This option is experimental, as not all machine
5632 descriptions used by GCC model the CPU closely enough to avoid unreliable
5633 results from the algorithm.
5635 This only makes sense when scheduling after register allocation, i.e.@: with
5636 @option{-fschedule-insns2} or at @option{-O2} or higher.
5638 @item -fsched2-use-traces
5639 @opindex fsched2-use-traces
5640 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5641 allocation and additionally perform code duplication in order to increase the
5642 size of superblocks using tracer pass. See @option{-ftracer} for details on
5645 This mode should produce faster but significantly longer programs. Also
5646 without @option{-fbranch-probabilities} the traces constructed may not
5647 match the reality and hurt the performance. This only makes
5648 sense when scheduling after register allocation, i.e.@: with
5649 @option{-fschedule-insns2} or at @option{-O2} or higher.
5653 Eliminate redundant sign extension instructions and move the non-redundant
5654 ones to optimal placement using lazy code motion (LCM).
5656 @item -freschedule-modulo-scheduled-loops
5657 @opindex freschedule-modulo-scheduled-loops
5658 The modulo scheduling comes before the traditional scheduling, if a loop
5659 was modulo scheduled we may want to prevent the later scheduling passes
5660 from changing its schedule, we use this option to control that.
5662 @item -fcaller-saves
5663 @opindex fcaller-saves
5664 Enable values to be allocated in registers that will be clobbered by
5665 function calls, by emitting extra instructions to save and restore the
5666 registers around such calls. Such allocation is done only when it
5667 seems to result in better code than would otherwise be produced.
5669 This option is always enabled by default on certain machines, usually
5670 those which have no call-preserved registers to use instead.
5672 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5674 @item -ftree-reassoc
5675 @opindex ftree-reassoc
5676 Perform reassociation on trees. This flag is enabled by default
5677 at @option{-O} and higher.
5681 Perform partial redundancy elimination (PRE) on trees. This flag is
5682 enabled by default at @option{-O2} and @option{-O3}.
5686 Perform full redundancy elimination (FRE) on trees. The difference
5687 between FRE and PRE is that FRE only considers expressions
5688 that are computed on all paths leading to the redundant computation.
5689 This analysis is faster than PRE, though it exposes fewer redundancies.
5690 This flag is enabled by default at @option{-O} and higher.
5692 @item -ftree-copy-prop
5693 @opindex ftree-copy-prop
5694 Perform copy propagation on trees. This pass eliminates unnecessary
5695 copy operations. This flag is enabled by default at @option{-O} and
5699 @opindex ftree-salias
5700 Perform structural alias analysis on trees. This flag
5701 is enabled by default at @option{-O} and higher.
5703 @item -fipa-pure-const
5704 @opindex fipa-pure-const
5705 Discover which functions are pure or constant.
5706 Enabled by default at @option{-O} and higher.
5708 @item -fipa-reference
5709 @opindex fipa-reference
5710 Discover which static variables do not escape cannot escape the
5712 Enabled by default at @option{-O} and higher.
5716 Perform interprocedural pointer analysis.
5720 Perform forward store motion on trees. This flag is
5721 enabled by default at @option{-O} and higher.
5725 Perform sparse conditional constant propagation (CCP) on trees. This
5726 pass only operates on local scalar variables and is enabled by default
5727 at @option{-O} and higher.
5729 @item -ftree-store-ccp
5730 @opindex ftree-store-ccp
5731 Perform sparse conditional constant propagation (CCP) on trees. This
5732 pass operates on both local scalar variables and memory stores and
5733 loads (global variables, structures, arrays, etc). This flag is
5734 enabled by default at @option{-O2} and higher.
5738 Perform dead code elimination (DCE) on trees. This flag is enabled by
5739 default at @option{-O} and higher.
5741 @item -ftree-dominator-opts
5742 @opindex ftree-dominator-opts
5743 Perform a variety of simple scalar cleanups (constant/copy
5744 propagation, redundancy elimination, range propagation and expression
5745 simplification) based on a dominator tree traversal. This also
5746 performs jump threading (to reduce jumps to jumps). This flag is
5747 enabled by default at @option{-O} and higher.
5751 Perform dead store elimination (DSE) on trees. A dead store is a store into
5752 a memory location which will later be overwritten by another store without
5753 any intervening loads. In this case the earlier store can be deleted. This
5754 flag is enabled by default at @option{-O} and higher.
5758 Perform loop header copying on trees. This is beneficial since it increases
5759 effectiveness of code motion optimizations. It also saves one jump. This flag
5760 is enabled by default at @option{-O} and higher. It is not enabled
5761 for @option{-Os}, since it usually increases code size.
5763 @item -ftree-loop-optimize
5764 @opindex ftree-loop-optimize
5765 Perform loop optimizations on trees. This flag is enabled by default
5766 at @option{-O} and higher.
5768 @item -ftree-loop-linear
5769 @opindex ftree-loop-linear
5770 Perform linear loop transformations on tree. This flag can improve cache
5771 performance and allow further loop optimizations to take place.
5773 @item -fcheck-data-deps
5774 @opindex fcheck-data-deps
5775 Compare the results of several data dependence analyzers. This option
5776 is used for debugging the data dependence analyzers.
5778 @item -ftree-loop-im
5779 @opindex ftree-loop-im
5780 Perform loop invariant motion on trees. This pass moves only invariants that
5781 would be hard to handle at RTL level (function calls, operations that expand to
5782 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5783 operands of conditions that are invariant out of the loop, so that we can use
5784 just trivial invariantness analysis in loop unswitching. The pass also includes
5787 @item -ftree-loop-ivcanon
5788 @opindex ftree-loop-ivcanon
5789 Create a canonical counter for number of iterations in the loop for that
5790 determining number of iterations requires complicated analysis. Later
5791 optimizations then may determine the number easily. Useful especially
5792 in connection with unrolling.
5796 Perform induction variable optimizations (strength reduction, induction
5797 variable merging and induction variable elimination) on trees.
5799 @item -ftree-parallelize-loops=n
5800 @opindex ftree-parallelize-loops
5801 Parallelize loops, i.e., split their iteration space to run in n threads.
5802 This is only possible for loops whose iterations are independent
5803 and can be arbitrarily reordered. The optimization is only
5804 profitable on multiprocessor machines, for loops that are CPU-intensive,
5805 rather than constrained e.g. by memory bandwidth.
5809 Perform scalar replacement of aggregates. This pass replaces structure
5810 references with scalars to prevent committing structures to memory too
5811 early. This flag is enabled by default at @option{-O} and higher.
5813 @item -ftree-copyrename
5814 @opindex ftree-copyrename
5815 Perform copy renaming on trees. This pass attempts to rename compiler
5816 temporaries to other variables at copy locations, usually resulting in
5817 variable names which more closely resemble the original variables. This flag
5818 is enabled by default at @option{-O} and higher.
5822 Perform temporary expression replacement during the SSA->normal phase. Single
5823 use/single def temporaries are replaced at their use location with their
5824 defining expression. This results in non-GIMPLE code, but gives the expanders
5825 much more complex trees to work on resulting in better RTL generation. This is
5826 enabled by default at @option{-O} and higher.
5828 @item -ftree-vectorize
5829 @opindex ftree-vectorize
5830 Perform loop vectorization on trees.
5832 @item -ftree-vect-loop-version
5833 @opindex ftree-vect-loop-version
5834 Perform loop versioning when doing loop vectorization on trees. When a loop
5835 appears to be vectorizable except that data alignment or data dependence cannot
5836 be determined at compile time then vectorized and non-vectorized versions of
5837 the loop are generated along with runtime checks for alignment or dependence
5838 to control which version is executed. This option is enabled by default
5839 except at level @option{-Os} where it is disabled.
5841 @item -fvect-cost-model
5842 @opindex fvect-cost-model
5843 Enable cost model for vectorization.
5847 Perform Value Range Propagation on trees. This is similar to the
5848 constant propagation pass, but instead of values, ranges of values are
5849 propagated. This allows the optimizers to remove unnecessary range
5850 checks like array bound checks and null pointer checks. This is
5851 enabled by default at @option{-O2} and higher. Null pointer check
5852 elimination is only done if @option{-fdelete-null-pointer-checks} is
5857 Perform tail duplication to enlarge superblock size. This transformation
5858 simplifies the control flow of the function allowing other optimizations to do
5861 @item -funroll-loops
5862 @opindex funroll-loops
5863 Unroll loops whose number of iterations can be determined at compile
5864 time or upon entry to the loop. @option{-funroll-loops} implies
5865 @option{-frerun-cse-after-loop}. This option makes code larger,
5866 and may or may not make it run faster.
5868 @item -funroll-all-loops
5869 @opindex funroll-all-loops
5870 Unroll all loops, even if their number of iterations is uncertain when
5871 the loop is entered. This usually makes programs run more slowly.
5872 @option{-funroll-all-loops} implies the same options as
5873 @option{-funroll-loops},
5875 @item -fsplit-ivs-in-unroller
5876 @opindex fsplit-ivs-in-unroller
5877 Enables expressing of values of induction variables in later iterations
5878 of the unrolled loop using the value in the first iteration. This breaks
5879 long dependency chains, thus improving efficiency of the scheduling passes.
5881 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5882 same effect. However in cases the loop body is more complicated than
5883 a single basic block, this is not reliable. It also does not work at all
5884 on some of the architectures due to restrictions in the CSE pass.
5886 This optimization is enabled by default.
5888 @item -fvariable-expansion-in-unroller
5889 @opindex fvariable-expansion-in-unroller
5890 With this option, the compiler will create multiple copies of some
5891 local variables when unrolling a loop which can result in superior code.
5893 @item -fpredictive-commoning
5894 @opindex fpredictive-commoning
5895 Perform predictive commoning optimization, i.e., reusing computations
5896 (especially memory loads and stores) performed in previous
5897 iterations of loops.
5899 This option is enabled at level @option{-O3}.
5901 @item -fprefetch-loop-arrays
5902 @opindex fprefetch-loop-arrays
5903 If supported by the target machine, generate instructions to prefetch
5904 memory to improve the performance of loops that access large arrays.
5906 This option may generate better or worse code; results are highly
5907 dependent on the structure of loops within the source code.
5909 Disabled at level @option{-Os}.
5912 @itemx -fno-peephole2
5913 @opindex fno-peephole
5914 @opindex fno-peephole2
5915 Disable any machine-specific peephole optimizations. The difference
5916 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5917 are implemented in the compiler; some targets use one, some use the
5918 other, a few use both.
5920 @option{-fpeephole} is enabled by default.
5921 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5923 @item -fno-guess-branch-probability
5924 @opindex fno-guess-branch-probability
5925 Do not guess branch probabilities using heuristics.
5927 GCC will use heuristics to guess branch probabilities if they are
5928 not provided by profiling feedback (@option{-fprofile-arcs}). These
5929 heuristics are based on the control flow graph. If some branch probabilities
5930 are specified by @samp{__builtin_expect}, then the heuristics will be
5931 used to guess branch probabilities for the rest of the control flow graph,
5932 taking the @samp{__builtin_expect} info into account. The interactions
5933 between the heuristics and @samp{__builtin_expect} can be complex, and in
5934 some cases, it may be useful to disable the heuristics so that the effects
5935 of @samp{__builtin_expect} are easier to understand.
5937 The default is @option{-fguess-branch-probability} at levels
5938 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5940 @item -freorder-blocks
5941 @opindex freorder-blocks
5942 Reorder basic blocks in the compiled function in order to reduce number of
5943 taken branches and improve code locality.
5945 Enabled at levels @option{-O2}, @option{-O3}.
5947 @item -freorder-blocks-and-partition
5948 @opindex freorder-blocks-and-partition
5949 In addition to reordering basic blocks in the compiled function, in order
5950 to reduce number of taken branches, partitions hot and cold basic blocks
5951 into separate sections of the assembly and .o files, to improve
5952 paging and cache locality performance.
5954 This optimization is automatically turned off in the presence of
5955 exception handling, for linkonce sections, for functions with a user-defined
5956 section attribute and on any architecture that does not support named
5959 @item -freorder-functions
5960 @opindex freorder-functions
5961 Reorder functions in the object file in order to
5962 improve code locality. This is implemented by using special
5963 subsections @code{.text.hot} for most frequently executed functions and
5964 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5965 the linker so object file format must support named sections and linker must
5966 place them in a reasonable way.
5968 Also profile feedback must be available in to make this option effective. See
5969 @option{-fprofile-arcs} for details.
5971 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5973 @item -fstrict-aliasing
5974 @opindex fstrict-aliasing
5975 Allows the compiler to assume the strictest aliasing rules applicable to
5976 the language being compiled. For C (and C++), this activates
5977 optimizations based on the type of expressions. In particular, an
5978 object of one type is assumed never to reside at the same address as an
5979 object of a different type, unless the types are almost the same. For
5980 example, an @code{unsigned int} can alias an @code{int}, but not a
5981 @code{void*} or a @code{double}. A character type may alias any other
5984 Pay special attention to code like this:
5997 The practice of reading from a different union member than the one most
5998 recently written to (called ``type-punning'') is common. Even with
5999 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6000 is accessed through the union type. So, the code above will work as
6001 expected. However, this code might not:
6012 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6014 @item -fstrict-overflow
6015 @opindex fstrict-overflow
6016 Allow the compiler to assume strict signed overflow rules, depending
6017 on the language being compiled. For C (and C++) this means that
6018 overflow when doing arithmetic with signed numbers is undefined, which
6019 means that the compiler may assume that it will not happen. This
6020 permits various optimizations. For example, the compiler will assume
6021 that an expression like @code{i + 10 > i} will always be true for
6022 signed @code{i}. This assumption is only valid if signed overflow is
6023 undefined, as the expression is false if @code{i + 10} overflows when
6024 using twos complement arithmetic. When this option is in effect any
6025 attempt to determine whether an operation on signed numbers will
6026 overflow must be written carefully to not actually involve overflow.
6028 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6029 that signed overflow is fully defined: it wraps. When
6030 @option{-fwrapv} is used, there is no difference between
6031 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
6032 @option{-fwrapv} certain types of overflow are permitted. For
6033 example, if the compiler gets an overflow when doing arithmetic on
6034 constants, the overflowed value can still be used with
6035 @option{-fwrapv}, but not otherwise.
6037 The @option{-fstrict-overflow} option is enabled at levels
6038 @option{-O2}, @option{-O3}, @option{-Os}.
6040 @item -falign-functions
6041 @itemx -falign-functions=@var{n}
6042 @opindex falign-functions
6043 Align the start of functions to the next power-of-two greater than
6044 @var{n}, skipping up to @var{n} bytes. For instance,
6045 @option{-falign-functions=32} aligns functions to the next 32-byte
6046 boundary, but @option{-falign-functions=24} would align to the next
6047 32-byte boundary only if this can be done by skipping 23 bytes or less.
6049 @option{-fno-align-functions} and @option{-falign-functions=1} are
6050 equivalent and mean that functions will not be aligned.
6052 Some assemblers only support this flag when @var{n} is a power of two;
6053 in that case, it is rounded up.
6055 If @var{n} is not specified or is zero, use a machine-dependent default.
6057 Enabled at levels @option{-O2}, @option{-O3}.
6059 @item -falign-labels
6060 @itemx -falign-labels=@var{n}
6061 @opindex falign-labels
6062 Align all branch targets to a power-of-two boundary, skipping up to
6063 @var{n} bytes like @option{-falign-functions}. This option can easily
6064 make code slower, because it must insert dummy operations for when the
6065 branch target is reached in the usual flow of the code.
6067 @option{-fno-align-labels} and @option{-falign-labels=1} are
6068 equivalent and mean that labels will not be aligned.
6070 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6071 are greater than this value, then their values are used instead.
6073 If @var{n} is not specified or is zero, use a machine-dependent default
6074 which is very likely to be @samp{1}, meaning no alignment.
6076 Enabled at levels @option{-O2}, @option{-O3}.
6079 @itemx -falign-loops=@var{n}
6080 @opindex falign-loops
6081 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6082 like @option{-falign-functions}. The hope is that the loop will be
6083 executed many times, which will make up for any execution of the dummy
6086 @option{-fno-align-loops} and @option{-falign-loops=1} are
6087 equivalent and mean that loops will not be aligned.
6089 If @var{n} is not specified or is zero, use a machine-dependent default.
6091 Enabled at levels @option{-O2}, @option{-O3}.
6094 @itemx -falign-jumps=@var{n}
6095 @opindex falign-jumps
6096 Align branch targets to a power-of-two boundary, for branch targets
6097 where the targets can only be reached by jumping, skipping up to @var{n}
6098 bytes like @option{-falign-functions}. In this case, no dummy operations
6101 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6102 equivalent and mean that loops will not be aligned.
6104 If @var{n} is not specified or is zero, use a machine-dependent default.
6106 Enabled at levels @option{-O2}, @option{-O3}.
6108 @item -funit-at-a-time
6109 @opindex funit-at-a-time
6110 Parse the whole compilation unit before starting to produce code.
6111 This allows some extra optimizations to take place but consumes
6112 more memory (in general). There are some compatibility issues
6113 with @emph{unit-at-a-time} mode:
6116 enabling @emph{unit-at-a-time} mode may change the order
6117 in which functions, variables, and top-level @code{asm} statements
6118 are emitted, and will likely break code relying on some particular
6119 ordering. The majority of such top-level @code{asm} statements,
6120 though, can be replaced by @code{section} attributes. The
6121 @option{fno-toplevel-reorder} option may be used to keep the ordering
6122 used in the input file, at the cost of some optimizations.
6125 @emph{unit-at-a-time} mode removes unreferenced static variables
6126 and functions. This may result in undefined references
6127 when an @code{asm} statement refers directly to variables or functions
6128 that are otherwise unused. In that case either the variable/function
6129 shall be listed as an operand of the @code{asm} statement operand or,
6130 in the case of top-level @code{asm} statements the attribute @code{used}
6131 shall be used on the declaration.
6134 Static functions now can use non-standard passing conventions that
6135 may break @code{asm} statements calling functions directly. Again,
6136 attribute @code{used} will prevent this behavior.
6139 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6140 but this scheme may not be supported by future releases of GCC@.
6142 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6144 @item -fno-toplevel-reorder
6145 Do not reorder top-level functions, variables, and @code{asm}
6146 statements. Output them in the same order that they appear in the
6147 input file. When this option is used, unreferenced static variables
6148 will not be removed. This option is intended to support existing code
6149 which relies on a particular ordering. For new code, it is better to
6154 Constructs webs as commonly used for register allocation purposes and assign
6155 each web individual pseudo register. This allows the register allocation pass
6156 to operate on pseudos directly, but also strengthens several other optimization
6157 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6158 however, make debugging impossible, since variables will no longer stay in a
6161 Enabled by default with @option{-funroll-loops}.
6163 @item -fwhole-program
6164 @opindex fwhole-program
6165 Assume that the current compilation unit represents whole program being
6166 compiled. All public functions and variables with the exception of @code{main}
6167 and those merged by attribute @code{externally_visible} become static functions
6168 and in a affect gets more aggressively optimized by interprocedural optimizers.
6169 While this option is equivalent to proper use of @code{static} keyword for
6170 programs consisting of single file, in combination with option
6171 @option{--combine} this flag can be used to compile most of smaller scale C
6172 programs since the functions and variables become local for the whole combined
6173 compilation unit, not for the single source file itself.
6175 This option is not supported for Fortran programs.
6177 @item -fcprop-registers
6178 @opindex fcprop-registers
6179 After register allocation and post-register allocation instruction splitting,
6180 we perform a copy-propagation pass to try to reduce scheduling dependencies
6181 and occasionally eliminate the copy.
6183 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6185 @item -fprofile-generate
6186 @opindex fprofile-generate
6188 Enable options usually used for instrumenting application to produce
6189 profile useful for later recompilation with profile feedback based
6190 optimization. You must use @option{-fprofile-generate} both when
6191 compiling and when linking your program.
6193 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6196 @opindex fprofile-use
6197 Enable profile feedback directed optimizations, and optimizations
6198 generally profitable only with profile feedback available.
6200 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6201 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6203 By default, GCC emits an error message if the feedback profiles do not
6204 match the source code. This error can be turned into a warning by using
6205 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6209 The following options control compiler behavior regarding floating
6210 point arithmetic. These options trade off between speed and
6211 correctness. All must be specifically enabled.
6215 @opindex ffloat-store
6216 Do not store floating point variables in registers, and inhibit other
6217 options that might change whether a floating point value is taken from a
6220 @cindex floating point precision
6221 This option prevents undesirable excess precision on machines such as
6222 the 68000 where the floating registers (of the 68881) keep more
6223 precision than a @code{double} is supposed to have. Similarly for the
6224 x86 architecture. For most programs, the excess precision does only
6225 good, but a few programs rely on the precise definition of IEEE floating
6226 point. Use @option{-ffloat-store} for such programs, after modifying
6227 them to store all pertinent intermediate computations into variables.
6231 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6232 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6233 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6235 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6237 This option is not turned on by any @option{-O} option since
6238 it can result in incorrect output for programs which depend on
6239 an exact implementation of IEEE or ISO rules/specifications for
6240 math functions. It may, however, yield faster code for programs
6241 that do not require the guarantees of these specifications.
6243 @item -fno-math-errno
6244 @opindex fno-math-errno
6245 Do not set ERRNO after calling math functions that are executed
6246 with a single instruction, e.g., sqrt. A program that relies on
6247 IEEE exceptions for math error handling may want to use this flag
6248 for speed while maintaining IEEE arithmetic compatibility.
6250 This option is not turned on by any @option{-O} option since
6251 it can result in incorrect output for programs which depend on
6252 an exact implementation of IEEE or ISO rules/specifications for
6253 math functions. It may, however, yield faster code for programs
6254 that do not require the guarantees of these specifications.
6256 The default is @option{-fmath-errno}.
6258 On Darwin systems, the math library never sets @code{errno}. There is
6259 therefore no reason for the compiler to consider the possibility that
6260 it might, and @option{-fno-math-errno} is the default.
6262 @item -funsafe-math-optimizations
6263 @opindex funsafe-math-optimizations
6265 Allow optimizations for floating-point arithmetic that (a) assume
6266 that arguments and results are valid and (b) may violate IEEE or
6267 ANSI standards. When used at link-time, it may include libraries
6268 or startup files that change the default FPU control word or other
6269 similar optimizations.
6271 This option is not turned on by any @option{-O} option since
6272 it can result in incorrect output for programs which depend on
6273 an exact implementation of IEEE or ISO rules/specifications for
6274 math functions. It may, however, yield faster code for programs
6275 that do not require the guarantees of these specifications.
6276 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6277 @option{-fassociative-math} and @option{-freciprocal-math}.
6279 The default is @option{-fno-unsafe-math-optimizations}.
6281 @item -fassociative-math
6282 @opindex fassociative-math
6284 Allow re-association of operands in series of floating-point operations.
6285 This violates the ISO C and C++ language standard by possibly changing
6286 computation result. NOTE: re-ordering may change the sign of zero as
6287 well as ignore NaNs and inhibit or create underflow or overflow (and
6288 thus cannot be used on a code which relies on rounding behavior like
6289 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6290 and thus may not be used when ordered comparisons are required.
6291 This option requires that both @option{-fno-signed-zeros} and
6292 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6293 much sense with @option{-frounding-math}.
6295 The default is @option{-fno-associative-math}.
6297 @item -freciprocal-math
6298 @opindex freciprocal-math
6300 Allow the reciprocal of a value to be used instead of dividing by
6301 the value if this enables optimizations. For example @code{x / y}
6302 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6303 is subject to common subexpression elimination. Note that this loses
6304 precision and increases the number of flops operating on the value.
6306 The default is @option{-fno-reciprocal-math}.
6308 @item -ffinite-math-only
6309 @opindex ffinite-math-only
6310 Allow optimizations for floating-point arithmetic that assume
6311 that arguments and results are not NaNs or +-Infs.
6313 This option is not turned on by any @option{-O} option since
6314 it can result in incorrect output for programs which depend on
6315 an exact implementation of IEEE or ISO rules/specifications for
6316 math functions. It may, however, yield faster code for programs
6317 that do not require the guarantees of these specifications.
6319 The default is @option{-fno-finite-math-only}.
6321 @item -fno-signed-zeros
6322 @opindex fno-signed-zeros
6323 Allow optimizations for floating point arithmetic that ignore the
6324 signedness of zero. IEEE arithmetic specifies the behavior of
6325 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6326 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6327 This option implies that the sign of a zero result isn't significant.
6329 The default is @option{-fsigned-zeros}.
6331 @item -fno-trapping-math
6332 @opindex fno-trapping-math
6333 Compile code assuming that floating-point operations cannot generate
6334 user-visible traps. These traps include division by zero, overflow,
6335 underflow, inexact result and invalid operation. This option requires
6336 that @option{-fno-signaling-nans} be in effect. Setting this option may
6337 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6339 This option should never be turned on by any @option{-O} option since
6340 it can result in incorrect output for programs which depend on
6341 an exact implementation of IEEE or ISO rules/specifications for
6344 The default is @option{-ftrapping-math}.
6346 @item -frounding-math
6347 @opindex frounding-math
6348 Disable transformations and optimizations that assume default floating
6349 point rounding behavior. This is round-to-zero for all floating point
6350 to integer conversions, and round-to-nearest for all other arithmetic
6351 truncations. This option should be specified for programs that change
6352 the FP rounding mode dynamically, or that may be executed with a
6353 non-default rounding mode. This option disables constant folding of
6354 floating point expressions at compile-time (which may be affected by
6355 rounding mode) and arithmetic transformations that are unsafe in the
6356 presence of sign-dependent rounding modes.
6358 The default is @option{-fno-rounding-math}.
6360 This option is experimental and does not currently guarantee to
6361 disable all GCC optimizations that are affected by rounding mode.
6362 Future versions of GCC may provide finer control of this setting
6363 using C99's @code{FENV_ACCESS} pragma. This command line option
6364 will be used to specify the default state for @code{FENV_ACCESS}.
6366 @item -frtl-abstract-sequences
6367 @opindex frtl-abstract-sequences
6368 It is a size optimization method. This option is to find identical
6369 sequences of code, which can be turned into pseudo-procedures and
6370 then replace all occurrences with calls to the newly created
6371 subroutine. It is kind of an opposite of @option{-finline-functions}.
6372 This optimization runs at RTL level.
6374 @item -fsignaling-nans
6375 @opindex fsignaling-nans
6376 Compile code assuming that IEEE signaling NaNs may generate user-visible
6377 traps during floating-point operations. Setting this option disables
6378 optimizations that may change the number of exceptions visible with
6379 signaling NaNs. This option implies @option{-ftrapping-math}.
6381 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6384 The default is @option{-fno-signaling-nans}.
6386 This option is experimental and does not currently guarantee to
6387 disable all GCC optimizations that affect signaling NaN behavior.
6389 @item -fsingle-precision-constant
6390 @opindex fsingle-precision-constant
6391 Treat floating point constant as single precision constant instead of
6392 implicitly converting it to double precision constant.
6394 @item -fcx-limited-range
6395 @opindex fcx-limited-range
6396 When enabled, this option states that a range reduction step is not
6397 needed when performing complex division. The default is
6398 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6400 This option controls the default setting of the ISO C99
6401 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6406 The following options control optimizations that may improve
6407 performance, but are not enabled by any @option{-O} options. This
6408 section includes experimental options that may produce broken code.
6411 @item -fbranch-probabilities
6412 @opindex fbranch-probabilities
6413 After running a program compiled with @option{-fprofile-arcs}
6414 (@pxref{Debugging Options,, Options for Debugging Your Program or
6415 @command{gcc}}), you can compile it a second time using
6416 @option{-fbranch-probabilities}, to improve optimizations based on
6417 the number of times each branch was taken. When the program
6418 compiled with @option{-fprofile-arcs} exits it saves arc execution
6419 counts to a file called @file{@var{sourcename}.gcda} for each source
6420 file. The information in this data file is very dependent on the
6421 structure of the generated code, so you must use the same source code
6422 and the same optimization options for both compilations.
6424 With @option{-fbranch-probabilities}, GCC puts a
6425 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6426 These can be used to improve optimization. Currently, they are only
6427 used in one place: in @file{reorg.c}, instead of guessing which path a
6428 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6429 exactly determine which path is taken more often.
6431 @item -fprofile-values
6432 @opindex fprofile-values
6433 If combined with @option{-fprofile-arcs}, it adds code so that some
6434 data about values of expressions in the program is gathered.
6436 With @option{-fbranch-probabilities}, it reads back the data gathered
6437 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6438 notes to instructions for their later usage in optimizations.
6440 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6444 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6445 a code to gather information about values of expressions.
6447 With @option{-fbranch-probabilities}, it reads back the data gathered
6448 and actually performs the optimizations based on them.
6449 Currently the optimizations include specialization of division operation
6450 using the knowledge about the value of the denominator.
6452 @item -frename-registers
6453 @opindex frename-registers
6454 Attempt to avoid false dependencies in scheduled code by making use
6455 of registers left over after register allocation. This optimization
6456 will most benefit processors with lots of registers. Depending on the
6457 debug information format adopted by the target, however, it can
6458 make debugging impossible, since variables will no longer stay in
6459 a ``home register''.
6461 Enabled by default with @option{-funroll-loops}.
6465 Perform tail duplication to enlarge superblock size. This transformation
6466 simplifies the control flow of the function allowing other optimizations to do
6469 Enabled with @option{-fprofile-use}.
6471 @item -funroll-loops
6472 @opindex funroll-loops
6473 Unroll loops whose number of iterations can be determined at compile time or
6474 upon entry to the loop. @option{-funroll-loops} implies
6475 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6476 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6477 small constant number of iterations). This option makes code larger, and may
6478 or may not make it run faster.
6480 Enabled with @option{-fprofile-use}.
6482 @item -funroll-all-loops
6483 @opindex funroll-all-loops
6484 Unroll all loops, even if their number of iterations is uncertain when
6485 the loop is entered. This usually makes programs run more slowly.
6486 @option{-funroll-all-loops} implies the same options as
6487 @option{-funroll-loops}.
6490 @opindex fpeel-loops
6491 Peels the loops for that there is enough information that they do not
6492 roll much (from profile feedback). It also turns on complete loop peeling
6493 (i.e.@: complete removal of loops with small constant number of iterations).
6495 Enabled with @option{-fprofile-use}.
6497 @item -fmove-loop-invariants
6498 @opindex fmove-loop-invariants
6499 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6500 at level @option{-O1}
6502 @item -funswitch-loops
6503 @opindex funswitch-loops
6504 Move branches with loop invariant conditions out of the loop, with duplicates
6505 of the loop on both branches (modified according to result of the condition).
6507 @item -ffunction-sections
6508 @itemx -fdata-sections
6509 @opindex ffunction-sections
6510 @opindex fdata-sections
6511 Place each function or data item into its own section in the output
6512 file if the target supports arbitrary sections. The name of the
6513 function or the name of the data item determines the section's name
6516 Use these options on systems where the linker can perform optimizations
6517 to improve locality of reference in the instruction space. Most systems
6518 using the ELF object format and SPARC processors running Solaris 2 have
6519 linkers with such optimizations. AIX may have these optimizations in
6522 Only use these options when there are significant benefits from doing
6523 so. When you specify these options, the assembler and linker will
6524 create larger object and executable files and will also be slower.
6525 You will not be able to use @code{gprof} on all systems if you
6526 specify this option and you may have problems with debugging if
6527 you specify both this option and @option{-g}.
6529 @item -fbranch-target-load-optimize
6530 @opindex fbranch-target-load-optimize
6531 Perform branch target register load optimization before prologue / epilogue
6533 The use of target registers can typically be exposed only during reload,
6534 thus hoisting loads out of loops and doing inter-block scheduling needs
6535 a separate optimization pass.
6537 @item -fbranch-target-load-optimize2
6538 @opindex fbranch-target-load-optimize2
6539 Perform branch target register load optimization after prologue / epilogue
6542 @item -fbtr-bb-exclusive
6543 @opindex fbtr-bb-exclusive
6544 When performing branch target register load optimization, don't reuse
6545 branch target registers in within any basic block.
6547 @item -fstack-protector
6548 @opindex fstack-protector
6549 Emit extra code to check for buffer overflows, such as stack smashing
6550 attacks. This is done by adding a guard variable to functions with
6551 vulnerable objects. This includes functions that call alloca, and
6552 functions with buffers larger than 8 bytes. The guards are initialized
6553 when a function is entered and then checked when the function exits.
6554 If a guard check fails, an error message is printed and the program exits.
6556 @item -fstack-protector-all
6557 @opindex fstack-protector-all
6558 Like @option{-fstack-protector} except that all functions are protected.
6560 @item -fsection-anchors
6561 @opindex fsection-anchors
6562 Try to reduce the number of symbolic address calculations by using
6563 shared ``anchor'' symbols to address nearby objects. This transformation
6564 can help to reduce the number of GOT entries and GOT accesses on some
6567 For example, the implementation of the following function @code{foo}:
6571 int foo (void) @{ return a + b + c; @}
6574 would usually calculate the addresses of all three variables, but if you
6575 compile it with @option{-fsection-anchors}, it will access the variables
6576 from a common anchor point instead. The effect is similar to the
6577 following pseudocode (which isn't valid C):
6582 register int *xr = &x;
6583 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6587 Not all targets support this option.
6589 @item --param @var{name}=@var{value}
6591 In some places, GCC uses various constants to control the amount of
6592 optimization that is done. For example, GCC will not inline functions
6593 that contain more that a certain number of instructions. You can
6594 control some of these constants on the command-line using the
6595 @option{--param} option.
6597 The names of specific parameters, and the meaning of the values, are
6598 tied to the internals of the compiler, and are subject to change
6599 without notice in future releases.
6601 In each case, the @var{value} is an integer. The allowable choices for
6602 @var{name} are given in the following table:
6605 @item salias-max-implicit-fields
6606 The maximum number of fields in a variable without direct
6607 structure accesses for which structure aliasing will consider trying
6608 to track each field. The default is 5
6610 @item salias-max-array-elements
6611 The maximum number of elements an array can have and its elements
6612 still be tracked individually by structure aliasing. The default is 4
6614 @item sra-max-structure-size
6615 The maximum structure size, in bytes, at which the scalar replacement
6616 of aggregates (SRA) optimization will perform block copies. The
6617 default value, 0, implies that GCC will select the most appropriate
6620 @item sra-field-structure-ratio
6621 The threshold ratio (as a percentage) between instantiated fields and
6622 the complete structure size. We say that if the ratio of the number
6623 of bytes in instantiated fields to the number of bytes in the complete
6624 structure exceeds this parameter, then block copies are not used. The
6627 @item max-crossjump-edges
6628 The maximum number of incoming edges to consider for crossjumping.
6629 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6630 the number of edges incoming to each block. Increasing values mean
6631 more aggressive optimization, making the compile time increase with
6632 probably small improvement in executable size.
6634 @item min-crossjump-insns
6635 The minimum number of instructions which must be matched at the end
6636 of two blocks before crossjumping will be performed on them. This
6637 value is ignored in the case where all instructions in the block being
6638 crossjumped from are matched. The default value is 5.
6640 @item max-grow-copy-bb-insns
6641 The maximum code size expansion factor when copying basic blocks
6642 instead of jumping. The expansion is relative to a jump instruction.
6643 The default value is 8.
6645 @item max-goto-duplication-insns
6646 The maximum number of instructions to duplicate to a block that jumps
6647 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6648 passes, GCC factors computed gotos early in the compilation process,
6649 and unfactors them as late as possible. Only computed jumps at the
6650 end of a basic blocks with no more than max-goto-duplication-insns are
6651 unfactored. The default value is 8.
6653 @item max-delay-slot-insn-search
6654 The maximum number of instructions to consider when looking for an
6655 instruction to fill a delay slot. If more than this arbitrary number of
6656 instructions is searched, the time savings from filling the delay slot
6657 will be minimal so stop searching. Increasing values mean more
6658 aggressive optimization, making the compile time increase with probably
6659 small improvement in executable run time.
6661 @item max-delay-slot-live-search
6662 When trying to fill delay slots, the maximum number of instructions to
6663 consider when searching for a block with valid live register
6664 information. Increasing this arbitrarily chosen value means more
6665 aggressive optimization, increasing the compile time. This parameter
6666 should be removed when the delay slot code is rewritten to maintain the
6669 @item max-gcse-memory
6670 The approximate maximum amount of memory that will be allocated in
6671 order to perform the global common subexpression elimination
6672 optimization. If more memory than specified is required, the
6673 optimization will not be done.
6675 @item max-gcse-passes
6676 The maximum number of passes of GCSE to run. The default is 1.
6678 @item max-pending-list-length
6679 The maximum number of pending dependencies scheduling will allow
6680 before flushing the current state and starting over. Large functions
6681 with few branches or calls can create excessively large lists which
6682 needlessly consume memory and resources.
6684 @item max-inline-insns-single
6685 Several parameters control the tree inliner used in gcc.
6686 This number sets the maximum number of instructions (counted in GCC's
6687 internal representation) in a single function that the tree inliner
6688 will consider for inlining. This only affects functions declared
6689 inline and methods implemented in a class declaration (C++).
6690 The default value is 450.
6692 @item max-inline-insns-auto
6693 When you use @option{-finline-functions} (included in @option{-O3}),
6694 a lot of functions that would otherwise not be considered for inlining
6695 by the compiler will be investigated. To those functions, a different
6696 (more restrictive) limit compared to functions declared inline can
6698 The default value is 90.
6700 @item large-function-insns
6701 The limit specifying really large functions. For functions larger than this
6702 limit after inlining inlining is constrained by
6703 @option{--param large-function-growth}. This parameter is useful primarily
6704 to avoid extreme compilation time caused by non-linear algorithms used by the
6706 This parameter is ignored when @option{-funit-at-a-time} is not used.
6707 The default value is 2700.
6709 @item large-function-growth
6710 Specifies maximal growth of large function caused by inlining in percents.
6711 This parameter is ignored when @option{-funit-at-a-time} is not used.
6712 The default value is 100 which limits large function growth to 2.0 times
6715 @item large-unit-insns
6716 The limit specifying large translation unit. Growth caused by inlining of
6717 units larger than this limit is limited by @option{--param inline-unit-growth}.
6718 For small units this might be too tight (consider unit consisting of function A
6719 that is inline and B that just calls A three time. If B is small relative to
6720 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6721 large units consisting of small inlininable functions however the overall unit
6722 growth limit is needed to avoid exponential explosion of code size. Thus for
6723 smaller units, the size is increased to @option{--param large-unit-insns}
6724 before applying @option{--param inline-unit-growth}. The default is 10000
6726 @item inline-unit-growth
6727 Specifies maximal overall growth of the compilation unit caused by inlining.
6728 This parameter is ignored when @option{-funit-at-a-time} is not used.
6729 The default value is 30 which limits unit growth to 1.3 times the original
6732 @item large-stack-frame
6733 The limit specifying large stack frames. While inlining the algorithm is trying
6734 to not grow past this limit too much. Default value is 256 bytes.
6736 @item large-stack-frame-growth
6737 Specifies maximal growth of large stack frames caused by inlining in percents.
6738 The default value is 1000 which limits large stack frame growth to 11 times
6741 @item max-inline-insns-recursive
6742 @itemx max-inline-insns-recursive-auto
6743 Specifies maximum number of instructions out-of-line copy of self recursive inline
6744 function can grow into by performing recursive inlining.
6746 For functions declared inline @option{--param max-inline-insns-recursive} is
6747 taken into account. For function not declared inline, recursive inlining
6748 happens only when @option{-finline-functions} (included in @option{-O3}) is
6749 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6750 default value is 450.
6752 @item max-inline-recursive-depth
6753 @itemx max-inline-recursive-depth-auto
6754 Specifies maximum recursion depth used by the recursive inlining.
6756 For functions declared inline @option{--param max-inline-recursive-depth} is
6757 taken into account. For function not declared inline, recursive inlining
6758 happens only when @option{-finline-functions} (included in @option{-O3}) is
6759 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6760 default value is 450.
6762 @item min-inline-recursive-probability
6763 Recursive inlining is profitable only for function having deep recursion
6764 in average and can hurt for function having little recursion depth by
6765 increasing the prologue size or complexity of function body to other
6768 When profile feedback is available (see @option{-fprofile-generate}) the actual
6769 recursion depth can be guessed from probability that function will recurse via
6770 given call expression. This parameter limits inlining only to call expression
6771 whose probability exceeds given threshold (in percents). The default value is
6774 @item inline-call-cost
6775 Specify cost of call instruction relative to simple arithmetics operations
6776 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6777 functions and at the same time increases size of leaf function that is believed to
6778 reduce function size by being inlined. In effect it increases amount of
6779 inlining for code having large abstraction penalty (many functions that just
6780 pass the arguments to other functions) and decrease inlining for code with low
6781 abstraction penalty. The default value is 12.
6783 @item min-vect-loop-bound
6784 The minimum number of iterations under which a loop will not get vectorized
6785 when @option{-ftree-vectorize} is used. The number of iterations after
6786 vectorization needs to be greater than the value specified by this option
6787 to allow vectorization. The default value is 0.
6789 @item max-unrolled-insns
6790 The maximum number of instructions that a loop should have if that loop
6791 is unrolled, and if the loop is unrolled, it determines how many times
6792 the loop code is unrolled.
6794 @item max-average-unrolled-insns
6795 The maximum number of instructions biased by probabilities of their execution
6796 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6797 it determines how many times the loop code is unrolled.
6799 @item max-unroll-times
6800 The maximum number of unrollings of a single loop.
6802 @item max-peeled-insns
6803 The maximum number of instructions that a loop should have if that loop
6804 is peeled, and if the loop is peeled, it determines how many times
6805 the loop code is peeled.
6807 @item max-peel-times
6808 The maximum number of peelings of a single loop.
6810 @item max-completely-peeled-insns
6811 The maximum number of insns of a completely peeled loop.
6813 @item max-completely-peel-times
6814 The maximum number of iterations of a loop to be suitable for complete peeling.
6816 @item max-unswitch-insns
6817 The maximum number of insns of an unswitched loop.
6819 @item max-unswitch-level
6820 The maximum number of branches unswitched in a single loop.
6823 The minimum cost of an expensive expression in the loop invariant motion.
6825 @item iv-consider-all-candidates-bound
6826 Bound on number of candidates for induction variables below that
6827 all candidates are considered for each use in induction variable
6828 optimizations. Only the most relevant candidates are considered
6829 if there are more candidates, to avoid quadratic time complexity.
6831 @item iv-max-considered-uses
6832 The induction variable optimizations give up on loops that contain more
6833 induction variable uses.
6835 @item iv-always-prune-cand-set-bound
6836 If number of candidates in the set is smaller than this value,
6837 we always try to remove unnecessary ivs from the set during its
6838 optimization when a new iv is added to the set.
6840 @item scev-max-expr-size
6841 Bound on size of expressions used in the scalar evolutions analyzer.
6842 Large expressions slow the analyzer.
6844 @item omega-max-vars
6845 The maximum number of variables in an Omega constraint system.
6846 The default value is 128.
6848 @item omega-max-geqs
6849 The maximum number of inequalities in an Omega constraint system.
6850 The default value is 256.
6853 The maximum number of equalities in an Omega constraint system.
6854 The default value is 128.
6856 @item omega-max-wild-cards
6857 The maximum number of wildcard variables that the Omega solver will
6858 be able to insert. The default value is 18.
6860 @item omega-hash-table-size
6861 The size of the hash table in the Omega solver. The default value is
6864 @item omega-max-keys
6865 The maximal number of keys used by the Omega solver. The default
6868 @item omega-eliminate-redundant-constraints
6869 When set to 1, use expensive methods to eliminate all redundant
6870 constraints. The default value is 0.
6872 @item vect-max-version-for-alignment-checks
6873 The maximum number of runtime checks that can be performed when
6874 doing loop versioning for alignment in the vectorizer. See option
6875 ftree-vect-loop-version for more information.
6877 @item vect-max-version-for-alias-checks
6878 The maximum number of runtime checks that can be performed when
6879 doing loop versioning for alias in the vectorizer. See option
6880 ftree-vect-loop-version for more information.
6882 @item max-iterations-to-track
6884 The maximum number of iterations of a loop the brute force algorithm
6885 for analysis of # of iterations of the loop tries to evaluate.
6887 @item hot-bb-count-fraction
6888 Select fraction of the maximal count of repetitions of basic block in program
6889 given basic block needs to have to be considered hot.
6891 @item hot-bb-frequency-fraction
6892 Select fraction of the maximal frequency of executions of basic block in
6893 function given basic block needs to have to be considered hot
6895 @item max-predicted-iterations
6896 The maximum number of loop iterations we predict statically. This is useful
6897 in cases where function contain single loop with known bound and other loop
6898 with unknown. We predict the known number of iterations correctly, while
6899 the unknown number of iterations average to roughly 10. This means that the
6900 loop without bounds would appear artificially cold relative to the other one.
6902 @item align-threshold
6904 Select fraction of the maximal frequency of executions of basic block in
6905 function given basic block will get aligned.
6907 @item align-loop-iterations
6909 A loop expected to iterate at lest the selected number of iterations will get
6912 @item tracer-dynamic-coverage
6913 @itemx tracer-dynamic-coverage-feedback
6915 This value is used to limit superblock formation once the given percentage of
6916 executed instructions is covered. This limits unnecessary code size
6919 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6920 feedback is available. The real profiles (as opposed to statically estimated
6921 ones) are much less balanced allowing the threshold to be larger value.
6923 @item tracer-max-code-growth
6924 Stop tail duplication once code growth has reached given percentage. This is
6925 rather hokey argument, as most of the duplicates will be eliminated later in
6926 cross jumping, so it may be set to much higher values than is the desired code
6929 @item tracer-min-branch-ratio
6931 Stop reverse growth when the reverse probability of best edge is less than this
6932 threshold (in percent).
6934 @item tracer-min-branch-ratio
6935 @itemx tracer-min-branch-ratio-feedback
6937 Stop forward growth if the best edge do have probability lower than this
6940 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6941 compilation for profile feedback and one for compilation without. The value
6942 for compilation with profile feedback needs to be more conservative (higher) in
6943 order to make tracer effective.
6945 @item max-cse-path-length
6947 Maximum number of basic blocks on path that cse considers. The default is 10.
6950 The maximum instructions CSE process before flushing. The default is 1000.
6952 @item max-aliased-vops
6954 Maximum number of virtual operands per function allowed to represent
6955 aliases before triggering the alias partitioning heuristic. Alias
6956 partitioning reduces compile times and memory consumption needed for
6957 aliasing at the expense of precision loss in alias information. The
6958 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
6961 Notice that if a function contains more memory statements than the
6962 value of this parameter, it is not really possible to achieve this
6963 reduction. In this case, the compiler will use the number of memory
6964 statements as the value for @option{max-aliased-vops}.
6966 @item avg-aliased-vops
6968 Average number of virtual operands per statement allowed to represent
6969 aliases before triggering the alias partitioning heuristic. This
6970 works in conjunction with @option{max-aliased-vops}. If a function
6971 contains more than @option{max-aliased-vops} virtual operators, then
6972 memory symbols will be grouped into memory partitions until either the
6973 total number of virtual operators is below @option{max-aliased-vops}
6974 or the average number of virtual operators per memory statement is
6975 below @option{avg-aliased-vops}. The default value for this parameter
6976 is 1 for -O1 and -O2, and 3 for -O3.
6978 @item ggc-min-expand
6980 GCC uses a garbage collector to manage its own memory allocation. This
6981 parameter specifies the minimum percentage by which the garbage
6982 collector's heap should be allowed to expand between collections.
6983 Tuning this may improve compilation speed; it has no effect on code
6986 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6987 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6988 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6989 GCC is not able to calculate RAM on a particular platform, the lower
6990 bound of 30% is used. Setting this parameter and
6991 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6992 every opportunity. This is extremely slow, but can be useful for
6995 @item ggc-min-heapsize
6997 Minimum size of the garbage collector's heap before it begins bothering
6998 to collect garbage. The first collection occurs after the heap expands
6999 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7000 tuning this may improve compilation speed, and has no effect on code
7003 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7004 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7005 with a lower bound of 4096 (four megabytes) and an upper bound of
7006 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7007 particular platform, the lower bound is used. Setting this parameter
7008 very large effectively disables garbage collection. Setting this
7009 parameter and @option{ggc-min-expand} to zero causes a full collection
7010 to occur at every opportunity.
7012 @item max-reload-search-insns
7013 The maximum number of instruction reload should look backward for equivalent
7014 register. Increasing values mean more aggressive optimization, making the
7015 compile time increase with probably slightly better performance. The default
7018 @item max-cselib-memory-locations
7019 The maximum number of memory locations cselib should take into account.
7020 Increasing values mean more aggressive optimization, making the compile time
7021 increase with probably slightly better performance. The default value is 500.
7023 @item max-flow-memory-locations
7024 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
7025 The default value is 100.
7027 @item reorder-blocks-duplicate
7028 @itemx reorder-blocks-duplicate-feedback
7030 Used by basic block reordering pass to decide whether to use unconditional
7031 branch or duplicate the code on its destination. Code is duplicated when its
7032 estimated size is smaller than this value multiplied by the estimated size of
7033 unconditional jump in the hot spots of the program.
7035 The @option{reorder-block-duplicate-feedback} is used only when profile
7036 feedback is available and may be set to higher values than
7037 @option{reorder-block-duplicate} since information about the hot spots is more
7040 @item max-sched-ready-insns
7041 The maximum number of instructions ready to be issued the scheduler should
7042 consider at any given time during the first scheduling pass. Increasing
7043 values mean more thorough searches, making the compilation time increase
7044 with probably little benefit. The default value is 100.
7046 @item max-sched-region-blocks
7047 The maximum number of blocks in a region to be considered for
7048 interblock scheduling. The default value is 10.
7050 @item max-sched-region-insns
7051 The maximum number of insns in a region to be considered for
7052 interblock scheduling. The default value is 100.
7055 The minimum probability (in percents) of reaching a source block
7056 for interblock speculative scheduling. The default value is 40.
7058 @item max-sched-extend-regions-iters
7059 The maximum number of iterations through CFG to extend regions.
7060 0 - disable region extension,
7061 N - do at most N iterations.
7062 The default value is 0.
7064 @item max-sched-insn-conflict-delay
7065 The maximum conflict delay for an insn to be considered for speculative motion.
7066 The default value is 3.
7068 @item sched-spec-prob-cutoff
7069 The minimal probability of speculation success (in percents), so that
7070 speculative insn will be scheduled.
7071 The default value is 40.
7073 @item max-last-value-rtl
7075 The maximum size measured as number of RTLs that can be recorded in an expression
7076 in combiner for a pseudo register as last known value of that register. The default
7079 @item integer-share-limit
7080 Small integer constants can use a shared data structure, reducing the
7081 compiler's memory usage and increasing its speed. This sets the maximum
7082 value of a shared integer constant's. The default value is 256.
7084 @item min-virtual-mappings
7085 Specifies the minimum number of virtual mappings in the incremental
7086 SSA updater that should be registered to trigger the virtual mappings
7087 heuristic defined by virtual-mappings-ratio. The default value is
7090 @item virtual-mappings-ratio
7091 If the number of virtual mappings is virtual-mappings-ratio bigger
7092 than the number of virtual symbols to be updated, then the incremental
7093 SSA updater switches to a full update for those symbols. The default
7096 @item ssp-buffer-size
7097 The minimum size of buffers (i.e. arrays) that will receive stack smashing
7098 protection when @option{-fstack-protection} is used.
7100 @item max-jump-thread-duplication-stmts
7101 Maximum number of statements allowed in a block that needs to be
7102 duplicated when threading jumps.
7104 @item max-fields-for-field-sensitive
7105 Maximum number of fields in a structure we will treat in
7106 a field sensitive manner during pointer analysis.
7108 @item prefetch-latency
7109 Estimate on average number of instructions that are executed before
7110 prefetch finishes. The distance we prefetch ahead is proportional
7111 to this constant. Increasing this number may also lead to less
7112 streams being prefetched (see @option{simultaneous-prefetches}).
7114 @item simultaneous-prefetches
7115 Maximum number of prefetches that can run at the same time.
7117 @item l1-cache-line-size
7118 The size of cache line in L1 cache, in bytes.
7121 The size of L1 cache, in kilobytes.
7124 The size of L2 cache, in kilobytes.
7126 @item use-canonical-types
7127 Whether the compiler should use the ``canonical'' type system. By
7128 default, this should always be 1, which uses a more efficient internal
7129 mechanism for comparing types in C++ and Objective-C++. However, if
7130 bugs in the canonical type system are causing compilation failures,
7131 set this value to 0 to disable canonical types.
7133 @item max-partial-antic-length
7134 Maximum length of the partial antic set computed during the tree
7135 partial redundancy elimination optimization (@option{-ftree-pre}) when
7136 optimizing at @option{-O3} and above. For some sorts of source code
7137 the enhanced partial redundancy elimination optimization can run away,
7138 consuming all of the memory available on the host machine. This
7139 parameter sets a limit on the length of the sets that are computed,
7140 which prevents the runaway behaviour. Setting a value of 0 for
7141 this paramter will allow an unlimited set length.
7146 @node Preprocessor Options
7147 @section Options Controlling the Preprocessor
7148 @cindex preprocessor options
7149 @cindex options, preprocessor
7151 These options control the C preprocessor, which is run on each C source
7152 file before actual compilation.
7154 If you use the @option{-E} option, nothing is done except preprocessing.
7155 Some of these options make sense only together with @option{-E} because
7156 they cause the preprocessor output to be unsuitable for actual
7161 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7162 and pass @var{option} directly through to the preprocessor. If
7163 @var{option} contains commas, it is split into multiple options at the
7164 commas. However, many options are modified, translated or interpreted
7165 by the compiler driver before being passed to the preprocessor, and
7166 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7167 interface is undocumented and subject to change, so whenever possible
7168 you should avoid using @option{-Wp} and let the driver handle the
7171 @item -Xpreprocessor @var{option}
7172 @opindex preprocessor
7173 Pass @var{option} as an option to the preprocessor. You can use this to
7174 supply system-specific preprocessor options which GCC does not know how to
7177 If you want to pass an option that takes an argument, you must use
7178 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7181 @include cppopts.texi
7183 @node Assembler Options
7184 @section Passing Options to the Assembler
7186 @c prevent bad page break with this line
7187 You can pass options to the assembler.
7190 @item -Wa,@var{option}
7192 Pass @var{option} as an option to the assembler. If @var{option}
7193 contains commas, it is split into multiple options at the commas.
7195 @item -Xassembler @var{option}
7197 Pass @var{option} as an option to the assembler. You can use this to
7198 supply system-specific assembler options which GCC does not know how to
7201 If you want to pass an option that takes an argument, you must use
7202 @option{-Xassembler} twice, once for the option and once for the argument.
7207 @section Options for Linking
7208 @cindex link options
7209 @cindex options, linking
7211 These options come into play when the compiler links object files into
7212 an executable output file. They are meaningless if the compiler is
7213 not doing a link step.
7217 @item @var{object-file-name}
7218 A file name that does not end in a special recognized suffix is
7219 considered to name an object file or library. (Object files are
7220 distinguished from libraries by the linker according to the file
7221 contents.) If linking is done, these object files are used as input
7230 If any of these options is used, then the linker is not run, and
7231 object file names should not be used as arguments. @xref{Overall
7235 @item -l@var{library}
7236 @itemx -l @var{library}
7238 Search the library named @var{library} when linking. (The second
7239 alternative with the library as a separate argument is only for
7240 POSIX compliance and is not recommended.)
7242 It makes a difference where in the command you write this option; the
7243 linker searches and processes libraries and object files in the order they
7244 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7245 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7246 to functions in @samp{z}, those functions may not be loaded.
7248 The linker searches a standard list of directories for the library,
7249 which is actually a file named @file{lib@var{library}.a}. The linker
7250 then uses this file as if it had been specified precisely by name.
7252 The directories searched include several standard system directories
7253 plus any that you specify with @option{-L}.
7255 Normally the files found this way are library files---archive files
7256 whose members are object files. The linker handles an archive file by
7257 scanning through it for members which define symbols that have so far
7258 been referenced but not defined. But if the file that is found is an
7259 ordinary object file, it is linked in the usual fashion. The only
7260 difference between using an @option{-l} option and specifying a file name
7261 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7262 and searches several directories.
7266 You need this special case of the @option{-l} option in order to
7267 link an Objective-C or Objective-C++ program.
7270 @opindex nostartfiles
7271 Do not use the standard system startup files when linking.
7272 The standard system libraries are used normally, unless @option{-nostdlib}
7273 or @option{-nodefaultlibs} is used.
7275 @item -nodefaultlibs
7276 @opindex nodefaultlibs
7277 Do not use the standard system libraries when linking.
7278 Only the libraries you specify will be passed to the linker.
7279 The standard startup files are used normally, unless @option{-nostartfiles}
7280 is used. The compiler may generate calls to @code{memcmp},
7281 @code{memset}, @code{memcpy} and @code{memmove}.
7282 These entries are usually resolved by entries in
7283 libc. These entry points should be supplied through some other
7284 mechanism when this option is specified.
7288 Do not use the standard system startup files or libraries when linking.
7289 No startup files and only the libraries you specify will be passed to
7290 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7291 @code{memcpy} and @code{memmove}.
7292 These entries are usually resolved by entries in
7293 libc. These entry points should be supplied through some other
7294 mechanism when this option is specified.
7296 @cindex @option{-lgcc}, use with @option{-nostdlib}
7297 @cindex @option{-nostdlib} and unresolved references
7298 @cindex unresolved references and @option{-nostdlib}
7299 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7300 @cindex @option{-nodefaultlibs} and unresolved references
7301 @cindex unresolved references and @option{-nodefaultlibs}
7302 One of the standard libraries bypassed by @option{-nostdlib} and
7303 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7304 that GCC uses to overcome shortcomings of particular machines, or special
7305 needs for some languages.
7306 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7307 Collection (GCC) Internals},
7308 for more discussion of @file{libgcc.a}.)
7309 In most cases, you need @file{libgcc.a} even when you want to avoid
7310 other standard libraries. In other words, when you specify @option{-nostdlib}
7311 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7312 This ensures that you have no unresolved references to internal GCC
7313 library subroutines. (For example, @samp{__main}, used to ensure C++
7314 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7315 GNU Compiler Collection (GCC) Internals}.)
7319 Produce a position independent executable on targets which support it.
7320 For predictable results, you must also specify the same set of options
7321 that were used to generate code (@option{-fpie}, @option{-fPIE},
7322 or model suboptions) when you specify this option.
7326 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7327 that support it. This instructs the linker to add all symbols, not
7328 only used ones, to the dynamic symbol table. This option is needed
7329 for some uses of @code{dlopen} or to allow obtaining backtraces
7330 from within a program.
7334 Remove all symbol table and relocation information from the executable.
7338 On systems that support dynamic linking, this prevents linking with the shared
7339 libraries. On other systems, this option has no effect.
7343 Produce a shared object which can then be linked with other objects to
7344 form an executable. Not all systems support this option. For predictable
7345 results, you must also specify the same set of options that were used to
7346 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7347 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7348 needs to build supplementary stub code for constructors to work. On
7349 multi-libbed systems, @samp{gcc -shared} must select the correct support
7350 libraries to link against. Failing to supply the correct flags may lead
7351 to subtle defects. Supplying them in cases where they are not necessary
7354 @item -shared-libgcc
7355 @itemx -static-libgcc
7356 @opindex shared-libgcc
7357 @opindex static-libgcc
7358 On systems that provide @file{libgcc} as a shared library, these options
7359 force the use of either the shared or static version respectively.
7360 If no shared version of @file{libgcc} was built when the compiler was
7361 configured, these options have no effect.
7363 There are several situations in which an application should use the
7364 shared @file{libgcc} instead of the static version. The most common
7365 of these is when the application wishes to throw and catch exceptions
7366 across different shared libraries. In that case, each of the libraries
7367 as well as the application itself should use the shared @file{libgcc}.
7369 Therefore, the G++ and GCJ drivers automatically add
7370 @option{-shared-libgcc} whenever you build a shared library or a main
7371 executable, because C++ and Java programs typically use exceptions, so
7372 this is the right thing to do.
7374 If, instead, you use the GCC driver to create shared libraries, you may
7375 find that they will not always be linked with the shared @file{libgcc}.
7376 If GCC finds, at its configuration time, that you have a non-GNU linker
7377 or a GNU linker that does not support option @option{--eh-frame-hdr},
7378 it will link the shared version of @file{libgcc} into shared libraries
7379 by default. Otherwise, it will take advantage of the linker and optimize
7380 away the linking with the shared version of @file{libgcc}, linking with
7381 the static version of libgcc by default. This allows exceptions to
7382 propagate through such shared libraries, without incurring relocation
7383 costs at library load time.
7385 However, if a library or main executable is supposed to throw or catch
7386 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7387 for the languages used in the program, or using the option
7388 @option{-shared-libgcc}, such that it is linked with the shared
7393 Bind references to global symbols when building a shared object. Warn
7394 about any unresolved references (unless overridden by the link editor
7395 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7398 @item -Xlinker @var{option}
7400 Pass @var{option} as an option to the linker. You can use this to
7401 supply system-specific linker options which GCC does not know how to
7404 If you want to pass an option that takes an argument, you must use
7405 @option{-Xlinker} twice, once for the option and once for the argument.
7406 For example, to pass @option{-assert definitions}, you must write
7407 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7408 @option{-Xlinker "-assert definitions"}, because this passes the entire
7409 string as a single argument, which is not what the linker expects.
7411 @item -Wl,@var{option}
7413 Pass @var{option} as an option to the linker. If @var{option} contains
7414 commas, it is split into multiple options at the commas.
7416 @item -u @var{symbol}
7418 Pretend the symbol @var{symbol} is undefined, to force linking of
7419 library modules to define it. You can use @option{-u} multiple times with
7420 different symbols to force loading of additional library modules.
7423 @node Directory Options
7424 @section Options for Directory Search
7425 @cindex directory options
7426 @cindex options, directory search
7429 These options specify directories to search for header files, for
7430 libraries and for parts of the compiler:
7435 Add the directory @var{dir} to the head of the list of directories to be
7436 searched for header files. This can be used to override a system header
7437 file, substituting your own version, since these directories are
7438 searched before the system header file directories. However, you should
7439 not use this option to add directories that contain vendor-supplied
7440 system header files (use @option{-isystem} for that). If you use more than
7441 one @option{-I} option, the directories are scanned in left-to-right
7442 order; the standard system directories come after.
7444 If a standard system include directory, or a directory specified with
7445 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7446 option will be ignored. The directory will still be searched but as a
7447 system directory at its normal position in the system include chain.
7448 This is to ensure that GCC's procedure to fix buggy system headers and
7449 the ordering for the include_next directive are not inadvertently changed.
7450 If you really need to change the search order for system directories,
7451 use the @option{-nostdinc} and/or @option{-isystem} options.
7453 @item -iquote@var{dir}
7455 Add the directory @var{dir} to the head of the list of directories to
7456 be searched for header files only for the case of @samp{#include
7457 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7458 otherwise just like @option{-I}.
7462 Add directory @var{dir} to the list of directories to be searched
7465 @item -B@var{prefix}
7467 This option specifies where to find the executables, libraries,
7468 include files, and data files of the compiler itself.
7470 The compiler driver program runs one or more of the subprograms
7471 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7472 @var{prefix} as a prefix for each program it tries to run, both with and
7473 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7475 For each subprogram to be run, the compiler driver first tries the
7476 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7477 was not specified, the driver tries two standard prefixes, which are
7478 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7479 those results in a file name that is found, the unmodified program
7480 name is searched for using the directories specified in your
7481 @env{PATH} environment variable.
7483 The compiler will check to see if the path provided by the @option{-B}
7484 refers to a directory, and if necessary it will add a directory
7485 separator character at the end of the path.
7487 @option{-B} prefixes that effectively specify directory names also apply
7488 to libraries in the linker, because the compiler translates these
7489 options into @option{-L} options for the linker. They also apply to
7490 includes files in the preprocessor, because the compiler translates these
7491 options into @option{-isystem} options for the preprocessor. In this case,
7492 the compiler appends @samp{include} to the prefix.
7494 The run-time support file @file{libgcc.a} can also be searched for using
7495 the @option{-B} prefix, if needed. If it is not found there, the two
7496 standard prefixes above are tried, and that is all. The file is left
7497 out of the link if it is not found by those means.
7499 Another way to specify a prefix much like the @option{-B} prefix is to use
7500 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7503 As a special kludge, if the path provided by @option{-B} is
7504 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7505 9, then it will be replaced by @file{[dir/]include}. This is to help
7506 with boot-strapping the compiler.
7508 @item -specs=@var{file}
7510 Process @var{file} after the compiler reads in the standard @file{specs}
7511 file, in order to override the defaults that the @file{gcc} driver
7512 program uses when determining what switches to pass to @file{cc1},
7513 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7514 @option{-specs=@var{file}} can be specified on the command line, and they
7515 are processed in order, from left to right.
7517 @item --sysroot=@var{dir}
7519 Use @var{dir} as the logical root directory for headers and libraries.
7520 For example, if the compiler would normally search for headers in
7521 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7522 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7524 If you use both this option and the @option{-isysroot} option, then
7525 the @option{--sysroot} option will apply to libraries, but the
7526 @option{-isysroot} option will apply to header files.
7528 The GNU linker (beginning with version 2.16) has the necessary support
7529 for this option. If your linker does not support this option, the
7530 header file aspect of @option{--sysroot} will still work, but the
7531 library aspect will not.
7535 This option has been deprecated. Please use @option{-iquote} instead for
7536 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7537 Any directories you specify with @option{-I} options before the @option{-I-}
7538 option are searched only for the case of @samp{#include "@var{file}"};
7539 they are not searched for @samp{#include <@var{file}>}.
7541 If additional directories are specified with @option{-I} options after
7542 the @option{-I-}, these directories are searched for all @samp{#include}
7543 directives. (Ordinarily @emph{all} @option{-I} directories are used
7546 In addition, the @option{-I-} option inhibits the use of the current
7547 directory (where the current input file came from) as the first search
7548 directory for @samp{#include "@var{file}"}. There is no way to
7549 override this effect of @option{-I-}. With @option{-I.} you can specify
7550 searching the directory which was current when the compiler was
7551 invoked. That is not exactly the same as what the preprocessor does
7552 by default, but it is often satisfactory.
7554 @option{-I-} does not inhibit the use of the standard system directories
7555 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7562 @section Specifying subprocesses and the switches to pass to them
7565 @command{gcc} is a driver program. It performs its job by invoking a
7566 sequence of other programs to do the work of compiling, assembling and
7567 linking. GCC interprets its command-line parameters and uses these to
7568 deduce which programs it should invoke, and which command-line options
7569 it ought to place on their command lines. This behavior is controlled
7570 by @dfn{spec strings}. In most cases there is one spec string for each
7571 program that GCC can invoke, but a few programs have multiple spec
7572 strings to control their behavior. The spec strings built into GCC can
7573 be overridden by using the @option{-specs=} command-line switch to specify
7576 @dfn{Spec files} are plaintext files that are used to construct spec
7577 strings. They consist of a sequence of directives separated by blank
7578 lines. The type of directive is determined by the first non-whitespace
7579 character on the line and it can be one of the following:
7582 @item %@var{command}
7583 Issues a @var{command} to the spec file processor. The commands that can
7587 @item %include <@var{file}>
7589 Search for @var{file} and insert its text at the current point in the
7592 @item %include_noerr <@var{file}>
7593 @cindex %include_noerr
7594 Just like @samp{%include}, but do not generate an error message if the include
7595 file cannot be found.
7597 @item %rename @var{old_name} @var{new_name}
7599 Rename the spec string @var{old_name} to @var{new_name}.
7603 @item *[@var{spec_name}]:
7604 This tells the compiler to create, override or delete the named spec
7605 string. All lines after this directive up to the next directive or
7606 blank line are considered to be the text for the spec string. If this
7607 results in an empty string then the spec will be deleted. (Or, if the
7608 spec did not exist, then nothing will happened.) Otherwise, if the spec
7609 does not currently exist a new spec will be created. If the spec does
7610 exist then its contents will be overridden by the text of this
7611 directive, unless the first character of that text is the @samp{+}
7612 character, in which case the text will be appended to the spec.
7614 @item [@var{suffix}]:
7615 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7616 and up to the next directive or blank line are considered to make up the
7617 spec string for the indicated suffix. When the compiler encounters an
7618 input file with the named suffix, it will processes the spec string in
7619 order to work out how to compile that file. For example:
7626 This says that any input file whose name ends in @samp{.ZZ} should be
7627 passed to the program @samp{z-compile}, which should be invoked with the
7628 command-line switch @option{-input} and with the result of performing the
7629 @samp{%i} substitution. (See below.)
7631 As an alternative to providing a spec string, the text that follows a
7632 suffix directive can be one of the following:
7635 @item @@@var{language}
7636 This says that the suffix is an alias for a known @var{language}. This is
7637 similar to using the @option{-x} command-line switch to GCC to specify a
7638 language explicitly. For example:
7645 Says that .ZZ files are, in fact, C++ source files.
7648 This causes an error messages saying:
7651 @var{name} compiler not installed on this system.
7655 GCC already has an extensive list of suffixes built into it.
7656 This directive will add an entry to the end of the list of suffixes, but
7657 since the list is searched from the end backwards, it is effectively
7658 possible to override earlier entries using this technique.
7662 GCC has the following spec strings built into it. Spec files can
7663 override these strings or create their own. Note that individual
7664 targets can also add their own spec strings to this list.
7667 asm Options to pass to the assembler
7668 asm_final Options to pass to the assembler post-processor
7669 cpp Options to pass to the C preprocessor
7670 cc1 Options to pass to the C compiler
7671 cc1plus Options to pass to the C++ compiler
7672 endfile Object files to include at the end of the link
7673 link Options to pass to the linker
7674 lib Libraries to include on the command line to the linker
7675 libgcc Decides which GCC support library to pass to the linker
7676 linker Sets the name of the linker
7677 predefines Defines to be passed to the C preprocessor
7678 signed_char Defines to pass to CPP to say whether @code{char} is signed
7680 startfile Object files to include at the start of the link
7683 Here is a small example of a spec file:
7689 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7692 This example renames the spec called @samp{lib} to @samp{old_lib} and
7693 then overrides the previous definition of @samp{lib} with a new one.
7694 The new definition adds in some extra command-line options before
7695 including the text of the old definition.
7697 @dfn{Spec strings} are a list of command-line options to be passed to their
7698 corresponding program. In addition, the spec strings can contain
7699 @samp{%}-prefixed sequences to substitute variable text or to
7700 conditionally insert text into the command line. Using these constructs
7701 it is possible to generate quite complex command lines.
7703 Here is a table of all defined @samp{%}-sequences for spec
7704 strings. Note that spaces are not generated automatically around the
7705 results of expanding these sequences. Therefore you can concatenate them
7706 together or combine them with constant text in a single argument.
7710 Substitute one @samp{%} into the program name or argument.
7713 Substitute the name of the input file being processed.
7716 Substitute the basename of the input file being processed.
7717 This is the substring up to (and not including) the last period
7718 and not including the directory.
7721 This is the same as @samp{%b}, but include the file suffix (text after
7725 Marks the argument containing or following the @samp{%d} as a
7726 temporary file name, so that that file will be deleted if GCC exits
7727 successfully. Unlike @samp{%g}, this contributes no text to the
7730 @item %g@var{suffix}
7731 Substitute a file name that has suffix @var{suffix} and is chosen
7732 once per compilation, and mark the argument in the same way as
7733 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7734 name is now chosen in a way that is hard to predict even when previously
7735 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7736 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7737 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7738 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7739 was simply substituted with a file name chosen once per compilation,
7740 without regard to any appended suffix (which was therefore treated
7741 just like ordinary text), making such attacks more likely to succeed.
7743 @item %u@var{suffix}
7744 Like @samp{%g}, but generates a new temporary file name even if
7745 @samp{%u@var{suffix}} was already seen.
7747 @item %U@var{suffix}
7748 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7749 new one if there is no such last file name. In the absence of any
7750 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7751 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7752 would involve the generation of two distinct file names, one
7753 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7754 simply substituted with a file name chosen for the previous @samp{%u},
7755 without regard to any appended suffix.
7757 @item %j@var{suffix}
7758 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7759 writable, and if save-temps is off; otherwise, substitute the name
7760 of a temporary file, just like @samp{%u}. This temporary file is not
7761 meant for communication between processes, but rather as a junk
7764 @item %|@var{suffix}
7765 @itemx %m@var{suffix}
7766 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7767 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7768 all. These are the two most common ways to instruct a program that it
7769 should read from standard input or write to standard output. If you
7770 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7771 construct: see for example @file{f/lang-specs.h}.
7773 @item %.@var{SUFFIX}
7774 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7775 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7776 terminated by the next space or %.
7779 Marks the argument containing or following the @samp{%w} as the
7780 designated output file of this compilation. This puts the argument
7781 into the sequence of arguments that @samp{%o} will substitute later.
7784 Substitutes the names of all the output files, with spaces
7785 automatically placed around them. You should write spaces
7786 around the @samp{%o} as well or the results are undefined.
7787 @samp{%o} is for use in the specs for running the linker.
7788 Input files whose names have no recognized suffix are not compiled
7789 at all, but they are included among the output files, so they will
7793 Substitutes the suffix for object files. Note that this is
7794 handled specially when it immediately follows @samp{%g, %u, or %U},
7795 because of the need for those to form complete file names. The
7796 handling is such that @samp{%O} is treated exactly as if it had already
7797 been substituted, except that @samp{%g, %u, and %U} do not currently
7798 support additional @var{suffix} characters following @samp{%O} as they would
7799 following, for example, @samp{.o}.
7802 Substitutes the standard macro predefinitions for the
7803 current target machine. Use this when running @code{cpp}.
7806 Like @samp{%p}, but puts @samp{__} before and after the name of each
7807 predefined macro, except for macros that start with @samp{__} or with
7808 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7812 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7813 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7814 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7815 and @option{-imultilib} as necessary.
7818 Current argument is the name of a library or startup file of some sort.
7819 Search for that file in a standard list of directories and substitute
7820 the full name found.
7823 Print @var{str} as an error message. @var{str} is terminated by a newline.
7824 Use this when inconsistent options are detected.
7827 Substitute the contents of spec string @var{name} at this point.
7830 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7832 @item %x@{@var{option}@}
7833 Accumulate an option for @samp{%X}.
7836 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7840 Output the accumulated assembler options specified by @option{-Wa}.
7843 Output the accumulated preprocessor options specified by @option{-Wp}.
7846 Process the @code{asm} spec. This is used to compute the
7847 switches to be passed to the assembler.
7850 Process the @code{asm_final} spec. This is a spec string for
7851 passing switches to an assembler post-processor, if such a program is
7855 Process the @code{link} spec. This is the spec for computing the
7856 command line passed to the linker. Typically it will make use of the
7857 @samp{%L %G %S %D and %E} sequences.
7860 Dump out a @option{-L} option for each directory that GCC believes might
7861 contain startup files. If the target supports multilibs then the
7862 current multilib directory will be prepended to each of these paths.
7865 Process the @code{lib} spec. This is a spec string for deciding which
7866 libraries should be included on the command line to the linker.
7869 Process the @code{libgcc} spec. This is a spec string for deciding
7870 which GCC support library should be included on the command line to the linker.
7873 Process the @code{startfile} spec. This is a spec for deciding which
7874 object files should be the first ones passed to the linker. Typically
7875 this might be a file named @file{crt0.o}.
7878 Process the @code{endfile} spec. This is a spec string that specifies
7879 the last object files that will be passed to the linker.
7882 Process the @code{cpp} spec. This is used to construct the arguments
7883 to be passed to the C preprocessor.
7886 Process the @code{cc1} spec. This is used to construct the options to be
7887 passed to the actual C compiler (@samp{cc1}).
7890 Process the @code{cc1plus} spec. This is used to construct the options to be
7891 passed to the actual C++ compiler (@samp{cc1plus}).
7894 Substitute the variable part of a matched option. See below.
7895 Note that each comma in the substituted string is replaced by
7899 Remove all occurrences of @code{-S} from the command line. Note---this
7900 command is position dependent. @samp{%} commands in the spec string
7901 before this one will see @code{-S}, @samp{%} commands in the spec string
7902 after this one will not.
7904 @item %:@var{function}(@var{args})
7905 Call the named function @var{function}, passing it @var{args}.
7906 @var{args} is first processed as a nested spec string, then split
7907 into an argument vector in the usual fashion. The function returns
7908 a string which is processed as if it had appeared literally as part
7909 of the current spec.
7911 The following built-in spec functions are provided:
7915 The @code{getenv} spec function takes two arguments: an environment
7916 variable name and a string. If the environment variable is not
7917 defined, a fatal error is issued. Otherwise, the return value is the
7918 value of the environment variable concatenated with the string. For
7919 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7922 %:getenv(TOPDIR /include)
7925 expands to @file{/path/to/top/include}.
7927 @item @code{if-exists}
7928 The @code{if-exists} spec function takes one argument, an absolute
7929 pathname to a file. If the file exists, @code{if-exists} returns the
7930 pathname. Here is a small example of its usage:
7934 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7937 @item @code{if-exists-else}
7938 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7939 spec function, except that it takes two arguments. The first argument is
7940 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7941 returns the pathname. If it does not exist, it returns the second argument.
7942 This way, @code{if-exists-else} can be used to select one file or another,
7943 based on the existence of the first. Here is a small example of its usage:
7947 crt0%O%s %:if-exists(crti%O%s) \
7948 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7951 @item @code{replace-outfile}
7952 The @code{replace-outfile} spec function takes two arguments. It looks for the
7953 first argument in the outfiles array and replaces it with the second argument. Here
7954 is a small example of its usage:
7957 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7960 @item @code{print-asm-header}
7961 The @code{print-asm-header} function takes no arguments and simply
7962 prints a banner like:
7968 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
7971 It is used to separate compiler options from assembler options
7972 in the @option{--target-help} output.
7976 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7977 If that switch was not specified, this substitutes nothing. Note that
7978 the leading dash is omitted when specifying this option, and it is
7979 automatically inserted if the substitution is performed. Thus the spec
7980 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7981 and would output the command line option @option{-foo}.
7983 @item %W@{@code{S}@}
7984 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7987 @item %@{@code{S}*@}
7988 Substitutes all the switches specified to GCC whose names start
7989 with @code{-S}, but which also take an argument. This is used for
7990 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7991 GCC considers @option{-o foo} as being
7992 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7993 text, including the space. Thus two arguments would be generated.
7995 @item %@{@code{S}*&@code{T}*@}
7996 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7997 (the order of @code{S} and @code{T} in the spec is not significant).
7998 There can be any number of ampersand-separated variables; for each the
7999 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8001 @item %@{@code{S}:@code{X}@}
8002 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8004 @item %@{!@code{S}:@code{X}@}
8005 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8007 @item %@{@code{S}*:@code{X}@}
8008 Substitutes @code{X} if one or more switches whose names start with
8009 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8010 once, no matter how many such switches appeared. However, if @code{%*}
8011 appears somewhere in @code{X}, then @code{X} will be substituted once
8012 for each matching switch, with the @code{%*} replaced by the part of
8013 that switch that matched the @code{*}.
8015 @item %@{.@code{S}:@code{X}@}
8016 Substitutes @code{X}, if processing a file with suffix @code{S}.
8018 @item %@{!.@code{S}:@code{X}@}
8019 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8021 @item %@{,@code{S}:@code{X}@}
8022 Substitutes @code{X}, if processing a file for language @code{S}.
8024 @item %@{!,@code{S}:@code{X}@}
8025 Substitutes @code{X}, if not processing a file for language @code{S}.
8027 @item %@{@code{S}|@code{P}:@code{X}@}
8028 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8029 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8030 @code{*} sequences as well, although they have a stronger binding than
8031 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8032 alternatives must be starred, and only the first matching alternative
8035 For example, a spec string like this:
8038 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8041 will output the following command-line options from the following input
8042 command-line options:
8047 -d fred.c -foo -baz -boggle
8048 -d jim.d -bar -baz -boggle
8051 @item %@{S:X; T:Y; :D@}
8053 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8054 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8055 be as many clauses as you need. This may be combined with @code{.},
8056 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8061 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8062 construct may contain other nested @samp{%} constructs or spaces, or
8063 even newlines. They are processed as usual, as described above.
8064 Trailing white space in @code{X} is ignored. White space may also
8065 appear anywhere on the left side of the colon in these constructs,
8066 except between @code{.} or @code{*} and the corresponding word.
8068 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8069 handled specifically in these constructs. If another value of
8070 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8071 @option{-W} switch is found later in the command line, the earlier
8072 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8073 just one letter, which passes all matching options.
8075 The character @samp{|} at the beginning of the predicate text is used to
8076 indicate that a command should be piped to the following command, but
8077 only if @option{-pipe} is specified.
8079 It is built into GCC which switches take arguments and which do not.
8080 (You might think it would be useful to generalize this to allow each
8081 compiler's spec to say which switches take arguments. But this cannot
8082 be done in a consistent fashion. GCC cannot even decide which input
8083 files have been specified without knowing which switches take arguments,
8084 and it must know which input files to compile in order to tell which
8087 GCC also knows implicitly that arguments starting in @option{-l} are to be
8088 treated as compiler output files, and passed to the linker in their
8089 proper position among the other output files.
8091 @c man begin OPTIONS
8093 @node Target Options
8094 @section Specifying Target Machine and Compiler Version
8095 @cindex target options
8096 @cindex cross compiling
8097 @cindex specifying machine version
8098 @cindex specifying compiler version and target machine
8099 @cindex compiler version, specifying
8100 @cindex target machine, specifying
8102 The usual way to run GCC is to run the executable called @file{gcc}, or
8103 @file{<machine>-gcc} when cross-compiling, or
8104 @file{<machine>-gcc-<version>} to run a version other than the one that
8105 was installed last. Sometimes this is inconvenient, so GCC provides
8106 options that will switch to another cross-compiler or version.
8109 @item -b @var{machine}
8111 The argument @var{machine} specifies the target machine for compilation.
8113 The value to use for @var{machine} is the same as was specified as the
8114 machine type when configuring GCC as a cross-compiler. For
8115 example, if a cross-compiler was configured with @samp{configure
8116 arm-elf}, meaning to compile for an arm processor with elf binaries,
8117 then you would specify @option{-b arm-elf} to run that cross compiler.
8118 Because there are other options beginning with @option{-b}, the
8119 configuration must contain a hyphen.
8121 @item -V @var{version}
8123 The argument @var{version} specifies which version of GCC to run.
8124 This is useful when multiple versions are installed. For example,
8125 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8128 The @option{-V} and @option{-b} options work by running the
8129 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8130 use them if you can just run that directly.
8132 @node Submodel Options
8133 @section Hardware Models and Configurations
8134 @cindex submodel options
8135 @cindex specifying hardware config
8136 @cindex hardware models and configurations, specifying
8137 @cindex machine dependent options
8139 Earlier we discussed the standard option @option{-b} which chooses among
8140 different installed compilers for completely different target
8141 machines, such as VAX vs.@: 68000 vs.@: 80386.
8143 In addition, each of these target machine types can have its own
8144 special options, starting with @samp{-m}, to choose among various
8145 hardware models or configurations---for example, 68010 vs 68020,
8146 floating coprocessor or none. A single installed version of the
8147 compiler can compile for any model or configuration, according to the
8150 Some configurations of the compiler also support additional special
8151 options, usually for compatibility with other compilers on the same
8154 @c This list is ordered alphanumerically by subsection name.
8155 @c It should be the same order and spelling as these options are listed
8156 @c in Machine Dependent Options
8162 * Blackfin Options::
8166 * DEC Alpha Options::
8167 * DEC Alpha/VMS Options::
8169 * GNU/Linux Options::
8172 * i386 and x86-64 Options::
8185 * RS/6000 and PowerPC Options::
8186 * S/390 and zSeries Options::
8191 * System V Options::
8192 * TMS320C3x/C4x Options::
8197 * Xstormy16 Options::
8203 @subsection ARC Options
8206 These options are defined for ARC implementations:
8211 Compile code for little endian mode. This is the default.
8215 Compile code for big endian mode.
8218 @opindex mmangle-cpu
8219 Prepend the name of the cpu to all public symbol names.
8220 In multiple-processor systems, there are many ARC variants with different
8221 instruction and register set characteristics. This flag prevents code
8222 compiled for one cpu to be linked with code compiled for another.
8223 No facility exists for handling variants that are ``almost identical''.
8224 This is an all or nothing option.
8226 @item -mcpu=@var{cpu}
8228 Compile code for ARC variant @var{cpu}.
8229 Which variants are supported depend on the configuration.
8230 All variants support @option{-mcpu=base}, this is the default.
8232 @item -mtext=@var{text-section}
8233 @itemx -mdata=@var{data-section}
8234 @itemx -mrodata=@var{readonly-data-section}
8238 Put functions, data, and readonly data in @var{text-section},
8239 @var{data-section}, and @var{readonly-data-section} respectively
8240 by default. This can be overridden with the @code{section} attribute.
8241 @xref{Variable Attributes}.
8246 @subsection ARM Options
8249 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8253 @item -mabi=@var{name}
8255 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8256 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8259 @opindex mapcs-frame
8260 Generate a stack frame that is compliant with the ARM Procedure Call
8261 Standard for all functions, even if this is not strictly necessary for
8262 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8263 with this option will cause the stack frames not to be generated for
8264 leaf functions. The default is @option{-mno-apcs-frame}.
8268 This is a synonym for @option{-mapcs-frame}.
8271 @c not currently implemented
8272 @item -mapcs-stack-check
8273 @opindex mapcs-stack-check
8274 Generate code to check the amount of stack space available upon entry to
8275 every function (that actually uses some stack space). If there is
8276 insufficient space available then either the function
8277 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8278 called, depending upon the amount of stack space required. The run time
8279 system is required to provide these functions. The default is
8280 @option{-mno-apcs-stack-check}, since this produces smaller code.
8282 @c not currently implemented
8284 @opindex mapcs-float
8285 Pass floating point arguments using the float point registers. This is
8286 one of the variants of the APCS@. This option is recommended if the
8287 target hardware has a floating point unit or if a lot of floating point
8288 arithmetic is going to be performed by the code. The default is
8289 @option{-mno-apcs-float}, since integer only code is slightly increased in
8290 size if @option{-mapcs-float} is used.
8292 @c not currently implemented
8293 @item -mapcs-reentrant
8294 @opindex mapcs-reentrant
8295 Generate reentrant, position independent code. The default is
8296 @option{-mno-apcs-reentrant}.
8299 @item -mthumb-interwork
8300 @opindex mthumb-interwork
8301 Generate code which supports calling between the ARM and Thumb
8302 instruction sets. Without this option the two instruction sets cannot
8303 be reliably used inside one program. The default is
8304 @option{-mno-thumb-interwork}, since slightly larger code is generated
8305 when @option{-mthumb-interwork} is specified.
8307 @item -mno-sched-prolog
8308 @opindex mno-sched-prolog
8309 Prevent the reordering of instructions in the function prolog, or the
8310 merging of those instruction with the instructions in the function's
8311 body. This means that all functions will start with a recognizable set
8312 of instructions (or in fact one of a choice from a small set of
8313 different function prologues), and this information can be used to
8314 locate the start if functions inside an executable piece of code. The
8315 default is @option{-msched-prolog}.
8318 @opindex mhard-float
8319 Generate output containing floating point instructions. This is the
8323 @opindex msoft-float
8324 Generate output containing library calls for floating point.
8325 @strong{Warning:} the requisite libraries are not available for all ARM
8326 targets. Normally the facilities of the machine's usual C compiler are
8327 used, but this cannot be done directly in cross-compilation. You must make
8328 your own arrangements to provide suitable library functions for
8331 @option{-msoft-float} changes the calling convention in the output file;
8332 therefore, it is only useful if you compile @emph{all} of a program with
8333 this option. In particular, you need to compile @file{libgcc.a}, the
8334 library that comes with GCC, with @option{-msoft-float} in order for
8337 @item -mfloat-abi=@var{name}
8339 Specifies which ABI to use for floating point values. Permissible values
8340 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8342 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8343 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8344 of floating point instructions, but still uses the soft-float calling
8347 @item -mlittle-endian
8348 @opindex mlittle-endian
8349 Generate code for a processor running in little-endian mode. This is
8350 the default for all standard configurations.
8353 @opindex mbig-endian
8354 Generate code for a processor running in big-endian mode; the default is
8355 to compile code for a little-endian processor.
8357 @item -mwords-little-endian
8358 @opindex mwords-little-endian
8359 This option only applies when generating code for big-endian processors.
8360 Generate code for a little-endian word order but a big-endian byte
8361 order. That is, a byte order of the form @samp{32107654}. Note: this
8362 option should only be used if you require compatibility with code for
8363 big-endian ARM processors generated by versions of the compiler prior to
8366 @item -mcpu=@var{name}
8368 This specifies the name of the target ARM processor. GCC uses this name
8369 to determine what kind of instructions it can emit when generating
8370 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8371 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8372 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8373 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8374 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8375 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8376 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8377 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8378 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8379 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8380 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8381 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8382 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8383 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8384 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8385 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8387 @itemx -mtune=@var{name}
8389 This option is very similar to the @option{-mcpu=} option, except that
8390 instead of specifying the actual target processor type, and hence
8391 restricting which instructions can be used, it specifies that GCC should
8392 tune the performance of the code as if the target were of the type
8393 specified in this option, but still choosing the instructions that it
8394 will generate based on the cpu specified by a @option{-mcpu=} option.
8395 For some ARM implementations better performance can be obtained by using
8398 @item -march=@var{name}
8400 This specifies the name of the target ARM architecture. GCC uses this
8401 name to determine what kind of instructions it can emit when generating
8402 assembly code. This option can be used in conjunction with or instead
8403 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8404 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8405 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8406 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8407 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8409 @item -mfpu=@var{name}
8410 @itemx -mfpe=@var{number}
8411 @itemx -mfp=@var{number}
8415 This specifies what floating point hardware (or hardware emulation) is
8416 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8417 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8418 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8419 with older versions of GCC@.
8421 If @option{-msoft-float} is specified this specifies the format of
8422 floating point values.
8424 @item -mstructure-size-boundary=@var{n}
8425 @opindex mstructure-size-boundary
8426 The size of all structures and unions will be rounded up to a multiple
8427 of the number of bits set by this option. Permissible values are 8, 32
8428 and 64. The default value varies for different toolchains. For the COFF
8429 targeted toolchain the default value is 8. A value of 64 is only allowed
8430 if the underlying ABI supports it.
8432 Specifying the larger number can produce faster, more efficient code, but
8433 can also increase the size of the program. Different values are potentially
8434 incompatible. Code compiled with one value cannot necessarily expect to
8435 work with code or libraries compiled with another value, if they exchange
8436 information using structures or unions.
8438 @item -mabort-on-noreturn
8439 @opindex mabort-on-noreturn
8440 Generate a call to the function @code{abort} at the end of a
8441 @code{noreturn} function. It will be executed if the function tries to
8445 @itemx -mno-long-calls
8446 @opindex mlong-calls
8447 @opindex mno-long-calls
8448 Tells the compiler to perform function calls by first loading the
8449 address of the function into a register and then performing a subroutine
8450 call on this register. This switch is needed if the target function
8451 will lie outside of the 64 megabyte addressing range of the offset based
8452 version of subroutine call instruction.
8454 Even if this switch is enabled, not all function calls will be turned
8455 into long calls. The heuristic is that static functions, functions
8456 which have the @samp{short-call} attribute, functions that are inside
8457 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8458 definitions have already been compiled within the current compilation
8459 unit, will not be turned into long calls. The exception to this rule is
8460 that weak function definitions, functions with the @samp{long-call}
8461 attribute or the @samp{section} attribute, and functions that are within
8462 the scope of a @samp{#pragma long_calls} directive, will always be
8463 turned into long calls.
8465 This feature is not enabled by default. Specifying
8466 @option{-mno-long-calls} will restore the default behavior, as will
8467 placing the function calls within the scope of a @samp{#pragma
8468 long_calls_off} directive. Note these switches have no effect on how
8469 the compiler generates code to handle function calls via function
8472 @item -mnop-fun-dllimport
8473 @opindex mnop-fun-dllimport
8474 Disable support for the @code{dllimport} attribute.
8476 @item -msingle-pic-base
8477 @opindex msingle-pic-base
8478 Treat the register used for PIC addressing as read-only, rather than
8479 loading it in the prologue for each function. The run-time system is
8480 responsible for initializing this register with an appropriate value
8481 before execution begins.
8483 @item -mpic-register=@var{reg}
8484 @opindex mpic-register
8485 Specify the register to be used for PIC addressing. The default is R10
8486 unless stack-checking is enabled, when R9 is used.
8488 @item -mcirrus-fix-invalid-insns
8489 @opindex mcirrus-fix-invalid-insns
8490 @opindex mno-cirrus-fix-invalid-insns
8491 Insert NOPs into the instruction stream to in order to work around
8492 problems with invalid Maverick instruction combinations. This option
8493 is only valid if the @option{-mcpu=ep9312} option has been used to
8494 enable generation of instructions for the Cirrus Maverick floating
8495 point co-processor. This option is not enabled by default, since the
8496 problem is only present in older Maverick implementations. The default
8497 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8500 @item -mpoke-function-name
8501 @opindex mpoke-function-name
8502 Write the name of each function into the text section, directly
8503 preceding the function prologue. The generated code is similar to this:
8507 .ascii "arm_poke_function_name", 0
8510 .word 0xff000000 + (t1 - t0)
8511 arm_poke_function_name
8513 stmfd sp!, @{fp, ip, lr, pc@}
8517 When performing a stack backtrace, code can inspect the value of
8518 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8519 location @code{pc - 12} and the top 8 bits are set, then we know that
8520 there is a function name embedded immediately preceding this location
8521 and has length @code{((pc[-3]) & 0xff000000)}.
8525 Generate code for the Thumb instruction set. The default is to
8526 use the 32-bit ARM instruction set.
8527 This option automatically enables either 16-bit Thumb-1 or
8528 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8529 and @option{-march=@var{name}} options.
8532 @opindex mtpcs-frame
8533 Generate a stack frame that is compliant with the Thumb Procedure Call
8534 Standard for all non-leaf functions. (A leaf function is one that does
8535 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8537 @item -mtpcs-leaf-frame
8538 @opindex mtpcs-leaf-frame
8539 Generate a stack frame that is compliant with the Thumb Procedure Call
8540 Standard for all leaf functions. (A leaf function is one that does
8541 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8543 @item -mcallee-super-interworking
8544 @opindex mcallee-super-interworking
8545 Gives all externally visible functions in the file being compiled an ARM
8546 instruction set header which switches to Thumb mode before executing the
8547 rest of the function. This allows these functions to be called from
8548 non-interworking code.
8550 @item -mcaller-super-interworking
8551 @opindex mcaller-super-interworking
8552 Allows calls via function pointers (including virtual functions) to
8553 execute correctly regardless of whether the target code has been
8554 compiled for interworking or not. There is a small overhead in the cost
8555 of executing a function pointer if this option is enabled.
8557 @item -mtp=@var{name}
8559 Specify the access model for the thread local storage pointer. The valid
8560 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8561 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8562 (supported in the arm6k architecture), and @option{auto}, which uses the
8563 best available method for the selected processor. The default setting is
8569 @subsection AVR Options
8572 These options are defined for AVR implementations:
8575 @item -mmcu=@var{mcu}
8577 Specify ATMEL AVR instruction set or MCU type.
8579 Instruction set avr1 is for the minimal AVR core, not supported by the C
8580 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8581 attiny11, attiny12, attiny15, attiny28).
8583 Instruction set avr2 (default) is for the classic AVR core with up to
8584 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8585 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8586 at90c8534, at90s8535).
8588 Instruction set avr3 is for the classic AVR core with up to 128K program
8589 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8591 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8592 memory space (MCU types: atmega8, atmega83, atmega85).
8594 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8595 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8596 atmega64, atmega128, at43usb355, at94k).
8600 Output instruction sizes to the asm file.
8602 @item -minit-stack=@var{N}
8603 @opindex minit-stack
8604 Specify the initial stack address, which may be a symbol or numeric value,
8605 @samp{__stack} is the default.
8607 @item -mno-interrupts
8608 @opindex mno-interrupts
8609 Generated code is not compatible with hardware interrupts.
8610 Code size will be smaller.
8612 @item -mcall-prologues
8613 @opindex mcall-prologues
8614 Functions prologues/epilogues expanded as call to appropriate
8615 subroutines. Code size will be smaller.
8617 @item -mno-tablejump
8618 @opindex mno-tablejump
8619 Do not generate tablejump insns which sometimes increase code size.
8622 @opindex mtiny-stack
8623 Change only the low 8 bits of the stack pointer.
8627 Assume int to be 8 bit integer. This affects the sizes of all types: A
8628 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8629 and long long will be 4 bytes. Please note that this option does not
8630 comply to the C standards, but it will provide you with smaller code
8634 @node Blackfin Options
8635 @subsection Blackfin Options
8636 @cindex Blackfin Options
8639 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8641 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8642 can be one of @samp{bf522}, @samp{bf525}, @samp{bf527},
8643 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8644 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8645 @samp{bf542}, @samp{bf544}, @samp{bf548}, @samp{bf549},
8647 The optional @var{sirevision} specifies the silicon revision of the target
8648 Blackfin processor. Any workarounds available for the targeted silicon revision
8649 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8650 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8651 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8652 hexadecimal digits representing the major and minor numbers in the silicon
8653 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8654 is not defined. If @var{sirevision} is @samp{any}, the
8655 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8656 If this optional @var{sirevision} is not used, GCC assumes the latest known
8657 silicon revision of the targeted Blackfin processor.
8659 Support for @samp{bf561} is incomplete. For @samp{bf561},
8660 Only the processor macro is defined.
8661 Without this option, @samp{bf532} is used as the processor by default.
8662 The corresponding predefined processor macros for @var{cpu} is to
8663 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8664 provided by libgloss to be linked in if @option{-msim} is not given.
8668 Specifies that the program will be run on the simulator. This causes
8669 the simulator BSP provided by libgloss to be linked in. This option
8670 has effect only for @samp{bfin-elf} toolchain.
8672 @item -momit-leaf-frame-pointer
8673 @opindex momit-leaf-frame-pointer
8674 Don't keep the frame pointer in a register for leaf functions. This
8675 avoids the instructions to save, set up and restore frame pointers and
8676 makes an extra register available in leaf functions. The option
8677 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8678 which might make debugging harder.
8680 @item -mspecld-anomaly
8681 @opindex mspecld-anomaly
8682 When enabled, the compiler will ensure that the generated code does not
8683 contain speculative loads after jump instructions. If this option is used,
8684 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8686 @item -mno-specld-anomaly
8687 @opindex mno-specld-anomaly
8688 Don't generate extra code to prevent speculative loads from occurring.
8690 @item -mcsync-anomaly
8691 @opindex mcsync-anomaly
8692 When enabled, the compiler will ensure that the generated code does not
8693 contain CSYNC or SSYNC instructions too soon after conditional branches.
8694 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8696 @item -mno-csync-anomaly
8697 @opindex mno-csync-anomaly
8698 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8699 occurring too soon after a conditional branch.
8703 When enabled, the compiler is free to take advantage of the knowledge that
8704 the entire program fits into the low 64k of memory.
8707 @opindex mno-low-64k
8708 Assume that the program is arbitrarily large. This is the default.
8710 @item -mstack-check-l1
8711 @opindex mstack-check-l1
8712 Do stack checking using information placed into L1 scratchpad memory by the
8715 @item -mid-shared-library
8716 @opindex mid-shared-library
8717 Generate code that supports shared libraries via the library ID method.
8718 This allows for execute in place and shared libraries in an environment
8719 without virtual memory management. This option implies @option{-fPIC}.
8721 @item -mno-id-shared-library
8722 @opindex mno-id-shared-library
8723 Generate code that doesn't assume ID based shared libraries are being used.
8724 This is the default.
8726 @item -mleaf-id-shared-library
8727 @opindex mleaf-id-shared-library
8728 Generate code that supports shared libraries via the library ID method,
8729 but assumes that this library or executable won't link against any other
8730 ID shared libraries. That allows the compiler to use faster code for jumps
8733 @item -mno-leaf-id-shared-library
8734 @opindex mno-leaf-id-shared-library
8735 Do not assume that the code being compiled won't link against any ID shared
8736 libraries. Slower code will be generated for jump and call insns.
8738 @item -mshared-library-id=n
8739 @opindex mshared-library-id
8740 Specified the identification number of the ID based shared library being
8741 compiled. Specifying a value of 0 will generate more compact code, specifying
8742 other values will force the allocation of that number to the current
8743 library but is no more space or time efficient than omitting this option.
8747 Generate code that allows the data segment to be located in a different
8748 area of memory from the text segment. This allows for execute in place in
8749 an environment without virtual memory management by eliminating relocations
8750 against the text section.
8753 @opindex mno-sep-data
8754 Generate code that assumes that the data segment follows the text segment.
8755 This is the default.
8758 @itemx -mno-long-calls
8759 @opindex mlong-calls
8760 @opindex mno-long-calls
8761 Tells the compiler to perform function calls by first loading the
8762 address of the function into a register and then performing a subroutine
8763 call on this register. This switch is needed if the target function
8764 will lie outside of the 24 bit addressing range of the offset based
8765 version of subroutine call instruction.
8767 This feature is not enabled by default. Specifying
8768 @option{-mno-long-calls} will restore the default behavior. Note these
8769 switches have no effect on how the compiler generates code to handle
8770 function calls via function pointers.
8774 Link with the fast floating-point library. This library relaxes some of
8775 the IEEE floating-point standard's rules for checking inputs against
8776 Not-a-Number (NAN), in the interest of performance.
8779 @opindex minline-plt
8780 Enable inlining of PLT entries in function calls to functions that are
8781 not known to bind locally. It has no effect without @option{-mfdpic}.
8785 @subsection CRIS Options
8786 @cindex CRIS Options
8788 These options are defined specifically for the CRIS ports.
8791 @item -march=@var{architecture-type}
8792 @itemx -mcpu=@var{architecture-type}
8795 Generate code for the specified architecture. The choices for
8796 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8797 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8798 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8801 @item -mtune=@var{architecture-type}
8803 Tune to @var{architecture-type} everything applicable about the generated
8804 code, except for the ABI and the set of available instructions. The
8805 choices for @var{architecture-type} are the same as for
8806 @option{-march=@var{architecture-type}}.
8808 @item -mmax-stack-frame=@var{n}
8809 @opindex mmax-stack-frame
8810 Warn when the stack frame of a function exceeds @var{n} bytes.
8812 @item -melinux-stacksize=@var{n}
8813 @opindex melinux-stacksize
8814 Only available with the @samp{cris-axis-aout} target. Arranges for
8815 indications in the program to the kernel loader that the stack of the
8816 program should be set to @var{n} bytes.
8822 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8823 @option{-march=v3} and @option{-march=v8} respectively.
8825 @item -mmul-bug-workaround
8826 @itemx -mno-mul-bug-workaround
8827 @opindex mmul-bug-workaround
8828 @opindex mno-mul-bug-workaround
8829 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8830 models where it applies. This option is active by default.
8834 Enable CRIS-specific verbose debug-related information in the assembly
8835 code. This option also has the effect to turn off the @samp{#NO_APP}
8836 formatted-code indicator to the assembler at the beginning of the
8841 Do not use condition-code results from previous instruction; always emit
8842 compare and test instructions before use of condition codes.
8844 @item -mno-side-effects
8845 @opindex mno-side-effects
8846 Do not emit instructions with side-effects in addressing modes other than
8850 @itemx -mno-stack-align
8852 @itemx -mno-data-align
8853 @itemx -mconst-align
8854 @itemx -mno-const-align
8855 @opindex mstack-align
8856 @opindex mno-stack-align
8857 @opindex mdata-align
8858 @opindex mno-data-align
8859 @opindex mconst-align
8860 @opindex mno-const-align
8861 These options (no-options) arranges (eliminate arrangements) for the
8862 stack-frame, individual data and constants to be aligned for the maximum
8863 single data access size for the chosen CPU model. The default is to
8864 arrange for 32-bit alignment. ABI details such as structure layout are
8865 not affected by these options.
8873 Similar to the stack- data- and const-align options above, these options
8874 arrange for stack-frame, writable data and constants to all be 32-bit,
8875 16-bit or 8-bit aligned. The default is 32-bit alignment.
8877 @item -mno-prologue-epilogue
8878 @itemx -mprologue-epilogue
8879 @opindex mno-prologue-epilogue
8880 @opindex mprologue-epilogue
8881 With @option{-mno-prologue-epilogue}, the normal function prologue and
8882 epilogue that sets up the stack-frame are omitted and no return
8883 instructions or return sequences are generated in the code. Use this
8884 option only together with visual inspection of the compiled code: no
8885 warnings or errors are generated when call-saved registers must be saved,
8886 or storage for local variable needs to be allocated.
8892 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8893 instruction sequences that load addresses for functions from the PLT part
8894 of the GOT rather than (traditional on other architectures) calls to the
8895 PLT@. The default is @option{-mgotplt}.
8899 Legacy no-op option only recognized with the cris-axis-aout target.
8903 Legacy no-op option only recognized with the cris-axis-elf and
8904 cris-axis-linux-gnu targets.
8908 Only recognized with the cris-axis-aout target, where it selects a
8909 GNU/linux-like multilib, include files and instruction set for
8914 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8918 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8919 to link with input-output functions from a simulator library. Code,
8920 initialized data and zero-initialized data are allocated consecutively.
8924 Like @option{-sim}, but pass linker options to locate initialized data at
8925 0x40000000 and zero-initialized data at 0x80000000.
8929 @subsection CRX Options
8932 These options are defined specifically for the CRX ports.
8938 Enable the use of multiply-accumulate instructions. Disabled by default.
8942 Push instructions will be used to pass outgoing arguments when functions
8943 are called. Enabled by default.
8946 @node Darwin Options
8947 @subsection Darwin Options
8948 @cindex Darwin options
8950 These options are defined for all architectures running the Darwin operating
8953 FSF GCC on Darwin does not create ``fat'' object files; it will create
8954 an object file for the single architecture that it was built to
8955 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8956 @option{-arch} options are used; it does so by running the compiler or
8957 linker multiple times and joining the results together with
8960 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8961 @samp{i686}) is determined by the flags that specify the ISA
8962 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8963 @option{-force_cpusubtype_ALL} option can be used to override this.
8965 The Darwin tools vary in their behavior when presented with an ISA
8966 mismatch. The assembler, @file{as}, will only permit instructions to
8967 be used that are valid for the subtype of the file it is generating,
8968 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8969 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8970 and print an error if asked to create a shared library with a less
8971 restrictive subtype than its input files (for instance, trying to put
8972 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8973 for executables, @file{ld}, will quietly give the executable the most
8974 restrictive subtype of any of its input files.
8979 Add the framework directory @var{dir} to the head of the list of
8980 directories to be searched for header files. These directories are
8981 interleaved with those specified by @option{-I} options and are
8982 scanned in a left-to-right order.
8984 A framework directory is a directory with frameworks in it. A
8985 framework is a directory with a @samp{"Headers"} and/or
8986 @samp{"PrivateHeaders"} directory contained directly in it that ends
8987 in @samp{".framework"}. The name of a framework is the name of this
8988 directory excluding the @samp{".framework"}. Headers associated with
8989 the framework are found in one of those two directories, with
8990 @samp{"Headers"} being searched first. A subframework is a framework
8991 directory that is in a framework's @samp{"Frameworks"} directory.
8992 Includes of subframework headers can only appear in a header of a
8993 framework that contains the subframework, or in a sibling subframework
8994 header. Two subframeworks are siblings if they occur in the same
8995 framework. A subframework should not have the same name as a
8996 framework, a warning will be issued if this is violated. Currently a
8997 subframework cannot have subframeworks, in the future, the mechanism
8998 may be extended to support this. The standard frameworks can be found
8999 in @samp{"/System/Library/Frameworks"} and
9000 @samp{"/Library/Frameworks"}. An example include looks like
9001 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9002 the name of the framework and header.h is found in the
9003 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9005 @item -iframework@var{dir}
9007 Like @option{-F} except the directory is a treated as a system
9008 directory. The main difference between this @option{-iframework} and
9009 @option{-F} is that with @option{-iframework} the compiler does not
9010 warn about constructs contained within header files found via
9011 @var{dir}. This option is valid only for the C family of languages.
9015 Emit debugging information for symbols that are used. For STABS
9016 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9017 This is by default ON@.
9021 Emit debugging information for all symbols and types.
9023 @item -mmacosx-version-min=@var{version}
9024 The earliest version of MacOS X that this executable will run on
9025 is @var{version}. Typical values of @var{version} include @code{10.1},
9026 @code{10.2}, and @code{10.3.9}.
9028 If the compiler was built to use the system's headers by default,
9029 then the default for this option is the system version on which the
9030 compiler is running, otherwise the default is to make choices which
9031 are compatible with as many systems and code bases as possible.
9035 Enable kernel development mode. The @option{-mkernel} option sets
9036 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9037 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9038 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9039 applicable. This mode also sets @option{-mno-altivec},
9040 @option{-msoft-float}, @option{-fno-builtin} and
9041 @option{-mlong-branch} for PowerPC targets.
9043 @item -mone-byte-bool
9044 @opindex mone-byte-bool
9045 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9046 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9047 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9048 option has no effect on x86.
9050 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9051 to generate code that is not binary compatible with code generated
9052 without that switch. Using this switch may require recompiling all
9053 other modules in a program, including system libraries. Use this
9054 switch to conform to a non-default data model.
9056 @item -mfix-and-continue
9057 @itemx -ffix-and-continue
9058 @itemx -findirect-data
9059 @opindex mfix-and-continue
9060 @opindex ffix-and-continue
9061 @opindex findirect-data
9062 Generate code suitable for fast turn around development. Needed to
9063 enable gdb to dynamically load @code{.o} files into already running
9064 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9065 are provided for backwards compatibility.
9069 Loads all members of static archive libraries.
9070 See man ld(1) for more information.
9072 @item -arch_errors_fatal
9073 @opindex arch_errors_fatal
9074 Cause the errors having to do with files that have the wrong architecture
9078 @opindex bind_at_load
9079 Causes the output file to be marked such that the dynamic linker will
9080 bind all undefined references when the file is loaded or launched.
9084 Produce a Mach-o bundle format file.
9085 See man ld(1) for more information.
9087 @item -bundle_loader @var{executable}
9088 @opindex bundle_loader
9089 This option specifies the @var{executable} that will be loading the build
9090 output file being linked. See man ld(1) for more information.
9094 When passed this option, GCC will produce a dynamic library instead of
9095 an executable when linking, using the Darwin @file{libtool} command.
9097 @item -force_cpusubtype_ALL
9098 @opindex force_cpusubtype_ALL
9099 This causes GCC's output file to have the @var{ALL} subtype, instead of
9100 one controlled by the @option{-mcpu} or @option{-march} option.
9102 @item -allowable_client @var{client_name}
9104 @itemx -compatibility_version
9105 @itemx -current_version
9107 @itemx -dependency-file
9109 @itemx -dylinker_install_name
9111 @itemx -exported_symbols_list
9113 @itemx -flat_namespace
9114 @itemx -force_flat_namespace
9115 @itemx -headerpad_max_install_names
9118 @itemx -install_name
9119 @itemx -keep_private_externs
9120 @itemx -multi_module
9121 @itemx -multiply_defined
9122 @itemx -multiply_defined_unused
9124 @itemx -no_dead_strip_inits_and_terms
9125 @itemx -nofixprebinding
9128 @itemx -noseglinkedit
9129 @itemx -pagezero_size
9131 @itemx -prebind_all_twolevel_modules
9132 @itemx -private_bundle
9133 @itemx -read_only_relocs
9135 @itemx -sectobjectsymbols
9139 @itemx -sectobjectsymbols
9142 @itemx -segs_read_only_addr
9143 @itemx -segs_read_write_addr
9144 @itemx -seg_addr_table
9145 @itemx -seg_addr_table_filename
9148 @itemx -segs_read_only_addr
9149 @itemx -segs_read_write_addr
9150 @itemx -single_module
9153 @itemx -sub_umbrella
9154 @itemx -twolevel_namespace
9157 @itemx -unexported_symbols_list
9158 @itemx -weak_reference_mismatches
9161 @opindex allowable_client
9162 @opindex client_name
9163 @opindex compatibility_version
9164 @opindex current_version
9166 @opindex dependency-file
9168 @opindex dylinker_install_name
9170 @opindex exported_symbols_list
9172 @opindex flat_namespace
9173 @opindex force_flat_namespace
9174 @opindex headerpad_max_install_names
9177 @opindex install_name
9178 @opindex keep_private_externs
9179 @opindex multi_module
9180 @opindex multiply_defined
9181 @opindex multiply_defined_unused
9183 @opindex no_dead_strip_inits_and_terms
9184 @opindex nofixprebinding
9185 @opindex nomultidefs
9187 @opindex noseglinkedit
9188 @opindex pagezero_size
9190 @opindex prebind_all_twolevel_modules
9191 @opindex private_bundle
9192 @opindex read_only_relocs
9194 @opindex sectobjectsymbols
9198 @opindex sectobjectsymbols
9201 @opindex segs_read_only_addr
9202 @opindex segs_read_write_addr
9203 @opindex seg_addr_table
9204 @opindex seg_addr_table_filename
9205 @opindex seglinkedit
9207 @opindex segs_read_only_addr
9208 @opindex segs_read_write_addr
9209 @opindex single_module
9211 @opindex sub_library
9212 @opindex sub_umbrella
9213 @opindex twolevel_namespace
9216 @opindex unexported_symbols_list
9217 @opindex weak_reference_mismatches
9218 @opindex whatsloaded
9220 These options are passed to the Darwin linker. The Darwin linker man page
9221 describes them in detail.
9224 @node DEC Alpha Options
9225 @subsection DEC Alpha Options
9227 These @samp{-m} options are defined for the DEC Alpha implementations:
9230 @item -mno-soft-float
9232 @opindex mno-soft-float
9233 @opindex msoft-float
9234 Use (do not use) the hardware floating-point instructions for
9235 floating-point operations. When @option{-msoft-float} is specified,
9236 functions in @file{libgcc.a} will be used to perform floating-point
9237 operations. Unless they are replaced by routines that emulate the
9238 floating-point operations, or compiled in such a way as to call such
9239 emulations routines, these routines will issue floating-point
9240 operations. If you are compiling for an Alpha without floating-point
9241 operations, you must ensure that the library is built so as not to call
9244 Note that Alpha implementations without floating-point operations are
9245 required to have floating-point registers.
9250 @opindex mno-fp-regs
9251 Generate code that uses (does not use) the floating-point register set.
9252 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9253 register set is not used, floating point operands are passed in integer
9254 registers as if they were integers and floating-point results are passed
9255 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9256 so any function with a floating-point argument or return value called by code
9257 compiled with @option{-mno-fp-regs} must also be compiled with that
9260 A typical use of this option is building a kernel that does not use,
9261 and hence need not save and restore, any floating-point registers.
9265 The Alpha architecture implements floating-point hardware optimized for
9266 maximum performance. It is mostly compliant with the IEEE floating
9267 point standard. However, for full compliance, software assistance is
9268 required. This option generates code fully IEEE compliant code
9269 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9270 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9271 defined during compilation. The resulting code is less efficient but is
9272 able to correctly support denormalized numbers and exceptional IEEE
9273 values such as not-a-number and plus/minus infinity. Other Alpha
9274 compilers call this option @option{-ieee_with_no_inexact}.
9276 @item -mieee-with-inexact
9277 @opindex mieee-with-inexact
9278 This is like @option{-mieee} except the generated code also maintains
9279 the IEEE @var{inexact-flag}. Turning on this option causes the
9280 generated code to implement fully-compliant IEEE math. In addition to
9281 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9282 macro. On some Alpha implementations the resulting code may execute
9283 significantly slower than the code generated by default. Since there is
9284 very little code that depends on the @var{inexact-flag}, you should
9285 normally not specify this option. Other Alpha compilers call this
9286 option @option{-ieee_with_inexact}.
9288 @item -mfp-trap-mode=@var{trap-mode}
9289 @opindex mfp-trap-mode
9290 This option controls what floating-point related traps are enabled.
9291 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9292 The trap mode can be set to one of four values:
9296 This is the default (normal) setting. The only traps that are enabled
9297 are the ones that cannot be disabled in software (e.g., division by zero
9301 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9305 Like @samp{u}, but the instructions are marked to be safe for software
9306 completion (see Alpha architecture manual for details).
9309 Like @samp{su}, but inexact traps are enabled as well.
9312 @item -mfp-rounding-mode=@var{rounding-mode}
9313 @opindex mfp-rounding-mode
9314 Selects the IEEE rounding mode. Other Alpha compilers call this option
9315 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9320 Normal IEEE rounding mode. Floating point numbers are rounded towards
9321 the nearest machine number or towards the even machine number in case
9325 Round towards minus infinity.
9328 Chopped rounding mode. Floating point numbers are rounded towards zero.
9331 Dynamic rounding mode. A field in the floating point control register
9332 (@var{fpcr}, see Alpha architecture reference manual) controls the
9333 rounding mode in effect. The C library initializes this register for
9334 rounding towards plus infinity. Thus, unless your program modifies the
9335 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9338 @item -mtrap-precision=@var{trap-precision}
9339 @opindex mtrap-precision
9340 In the Alpha architecture, floating point traps are imprecise. This
9341 means without software assistance it is impossible to recover from a
9342 floating trap and program execution normally needs to be terminated.
9343 GCC can generate code that can assist operating system trap handlers
9344 in determining the exact location that caused a floating point trap.
9345 Depending on the requirements of an application, different levels of
9346 precisions can be selected:
9350 Program precision. This option is the default and means a trap handler
9351 can only identify which program caused a floating point exception.
9354 Function precision. The trap handler can determine the function that
9355 caused a floating point exception.
9358 Instruction precision. The trap handler can determine the exact
9359 instruction that caused a floating point exception.
9362 Other Alpha compilers provide the equivalent options called
9363 @option{-scope_safe} and @option{-resumption_safe}.
9365 @item -mieee-conformant
9366 @opindex mieee-conformant
9367 This option marks the generated code as IEEE conformant. You must not
9368 use this option unless you also specify @option{-mtrap-precision=i} and either
9369 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9370 is to emit the line @samp{.eflag 48} in the function prologue of the
9371 generated assembly file. Under DEC Unix, this has the effect that
9372 IEEE-conformant math library routines will be linked in.
9374 @item -mbuild-constants
9375 @opindex mbuild-constants
9376 Normally GCC examines a 32- or 64-bit integer constant to
9377 see if it can construct it from smaller constants in two or three
9378 instructions. If it cannot, it will output the constant as a literal and
9379 generate code to load it from the data segment at runtime.
9381 Use this option to require GCC to construct @emph{all} integer constants
9382 using code, even if it takes more instructions (the maximum is six).
9384 You would typically use this option to build a shared library dynamic
9385 loader. Itself a shared library, it must relocate itself in memory
9386 before it can find the variables and constants in its own data segment.
9392 Select whether to generate code to be assembled by the vendor-supplied
9393 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9411 Indicate whether GCC should generate code to use the optional BWX,
9412 CIX, FIX and MAX instruction sets. The default is to use the instruction
9413 sets supported by the CPU type specified via @option{-mcpu=} option or that
9414 of the CPU on which GCC was built if none was specified.
9419 @opindex mfloat-ieee
9420 Generate code that uses (does not use) VAX F and G floating point
9421 arithmetic instead of IEEE single and double precision.
9423 @item -mexplicit-relocs
9424 @itemx -mno-explicit-relocs
9425 @opindex mexplicit-relocs
9426 @opindex mno-explicit-relocs
9427 Older Alpha assemblers provided no way to generate symbol relocations
9428 except via assembler macros. Use of these macros does not allow
9429 optimal instruction scheduling. GNU binutils as of version 2.12
9430 supports a new syntax that allows the compiler to explicitly mark
9431 which relocations should apply to which instructions. This option
9432 is mostly useful for debugging, as GCC detects the capabilities of
9433 the assembler when it is built and sets the default accordingly.
9437 @opindex msmall-data
9438 @opindex mlarge-data
9439 When @option{-mexplicit-relocs} is in effect, static data is
9440 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9441 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9442 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9443 16-bit relocations off of the @code{$gp} register. This limits the
9444 size of the small data area to 64KB, but allows the variables to be
9445 directly accessed via a single instruction.
9447 The default is @option{-mlarge-data}. With this option the data area
9448 is limited to just below 2GB@. Programs that require more than 2GB of
9449 data must use @code{malloc} or @code{mmap} to allocate the data in the
9450 heap instead of in the program's data segment.
9452 When generating code for shared libraries, @option{-fpic} implies
9453 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9457 @opindex msmall-text
9458 @opindex mlarge-text
9459 When @option{-msmall-text} is used, the compiler assumes that the
9460 code of the entire program (or shared library) fits in 4MB, and is
9461 thus reachable with a branch instruction. When @option{-msmall-data}
9462 is used, the compiler can assume that all local symbols share the
9463 same @code{$gp} value, and thus reduce the number of instructions
9464 required for a function call from 4 to 1.
9466 The default is @option{-mlarge-text}.
9468 @item -mcpu=@var{cpu_type}
9470 Set the instruction set and instruction scheduling parameters for
9471 machine type @var{cpu_type}. You can specify either the @samp{EV}
9472 style name or the corresponding chip number. GCC supports scheduling
9473 parameters for the EV4, EV5 and EV6 family of processors and will
9474 choose the default values for the instruction set from the processor
9475 you specify. If you do not specify a processor type, GCC will default
9476 to the processor on which the compiler was built.
9478 Supported values for @var{cpu_type} are
9484 Schedules as an EV4 and has no instruction set extensions.
9488 Schedules as an EV5 and has no instruction set extensions.
9492 Schedules as an EV5 and supports the BWX extension.
9497 Schedules as an EV5 and supports the BWX and MAX extensions.
9501 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9505 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9508 @item -mtune=@var{cpu_type}
9510 Set only the instruction scheduling parameters for machine type
9511 @var{cpu_type}. The instruction set is not changed.
9513 @item -mmemory-latency=@var{time}
9514 @opindex mmemory-latency
9515 Sets the latency the scheduler should assume for typical memory
9516 references as seen by the application. This number is highly
9517 dependent on the memory access patterns used by the application
9518 and the size of the external cache on the machine.
9520 Valid options for @var{time} are
9524 A decimal number representing clock cycles.
9530 The compiler contains estimates of the number of clock cycles for
9531 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9532 (also called Dcache, Scache, and Bcache), as well as to main memory.
9533 Note that L3 is only valid for EV5.
9538 @node DEC Alpha/VMS Options
9539 @subsection DEC Alpha/VMS Options
9541 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9544 @item -mvms-return-codes
9545 @opindex mvms-return-codes
9546 Return VMS condition codes from main. The default is to return POSIX
9547 style condition (e.g.@: error) codes.
9551 @subsection FRV Options
9558 Only use the first 32 general purpose registers.
9563 Use all 64 general purpose registers.
9568 Use only the first 32 floating point registers.
9573 Use all 64 floating point registers
9576 @opindex mhard-float
9578 Use hardware instructions for floating point operations.
9581 @opindex msoft-float
9583 Use library routines for floating point operations.
9588 Dynamically allocate condition code registers.
9593 Do not try to dynamically allocate condition code registers, only
9594 use @code{icc0} and @code{fcc0}.
9599 Change ABI to use double word insns.
9604 Do not use double word instructions.
9609 Use floating point double instructions.
9614 Do not use floating point double instructions.
9619 Use media instructions.
9624 Do not use media instructions.
9629 Use multiply and add/subtract instructions.
9634 Do not use multiply and add/subtract instructions.
9639 Select the FDPIC ABI, that uses function descriptors to represent
9640 pointers to functions. Without any PIC/PIE-related options, it
9641 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9642 assumes GOT entries and small data are within a 12-bit range from the
9643 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9644 are computed with 32 bits.
9647 @opindex minline-plt
9649 Enable inlining of PLT entries in function calls to functions that are
9650 not known to bind locally. It has no effect without @option{-mfdpic}.
9651 It's enabled by default if optimizing for speed and compiling for
9652 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9653 optimization option such as @option{-O3} or above is present in the
9659 Assume a large TLS segment when generating thread-local code.
9664 Do not assume a large TLS segment when generating thread-local code.
9669 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9670 that is known to be in read-only sections. It's enabled by default,
9671 except for @option{-fpic} or @option{-fpie}: even though it may help
9672 make the global offset table smaller, it trades 1 instruction for 4.
9673 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9674 one of which may be shared by multiple symbols, and it avoids the need
9675 for a GOT entry for the referenced symbol, so it's more likely to be a
9676 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9678 @item -multilib-library-pic
9679 @opindex multilib-library-pic
9681 Link with the (library, not FD) pic libraries. It's implied by
9682 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9683 @option{-fpic} without @option{-mfdpic}. You should never have to use
9689 Follow the EABI requirement of always creating a frame pointer whenever
9690 a stack frame is allocated. This option is enabled by default and can
9691 be disabled with @option{-mno-linked-fp}.
9694 @opindex mlong-calls
9696 Use indirect addressing to call functions outside the current
9697 compilation unit. This allows the functions to be placed anywhere
9698 within the 32-bit address space.
9700 @item -malign-labels
9701 @opindex malign-labels
9703 Try to align labels to an 8-byte boundary by inserting nops into the
9704 previous packet. This option only has an effect when VLIW packing
9705 is enabled. It doesn't create new packets; it merely adds nops to
9709 @opindex mlibrary-pic
9711 Generate position-independent EABI code.
9716 Use only the first four media accumulator registers.
9721 Use all eight media accumulator registers.
9726 Pack VLIW instructions.
9731 Do not pack VLIW instructions.
9736 Do not mark ABI switches in e_flags.
9741 Enable the use of conditional-move instructions (default).
9743 This switch is mainly for debugging the compiler and will likely be removed
9744 in a future version.
9746 @item -mno-cond-move
9747 @opindex mno-cond-move
9749 Disable the use of conditional-move instructions.
9751 This switch is mainly for debugging the compiler and will likely be removed
9752 in a future version.
9757 Enable the use of conditional set instructions (default).
9759 This switch is mainly for debugging the compiler and will likely be removed
9760 in a future version.
9765 Disable the use of conditional set instructions.
9767 This switch is mainly for debugging the compiler and will likely be removed
9768 in a future version.
9773 Enable the use of conditional execution (default).
9775 This switch is mainly for debugging the compiler and will likely be removed
9776 in a future version.
9778 @item -mno-cond-exec
9779 @opindex mno-cond-exec
9781 Disable the use of conditional execution.
9783 This switch is mainly for debugging the compiler and will likely be removed
9784 in a future version.
9787 @opindex mvliw-branch
9789 Run a pass to pack branches into VLIW instructions (default).
9791 This switch is mainly for debugging the compiler and will likely be removed
9792 in a future version.
9794 @item -mno-vliw-branch
9795 @opindex mno-vliw-branch
9797 Do not run a pass to pack branches into VLIW instructions.
9799 This switch is mainly for debugging the compiler and will likely be removed
9800 in a future version.
9802 @item -mmulti-cond-exec
9803 @opindex mmulti-cond-exec
9805 Enable optimization of @code{&&} and @code{||} in conditional execution
9808 This switch is mainly for debugging the compiler and will likely be removed
9809 in a future version.
9811 @item -mno-multi-cond-exec
9812 @opindex mno-multi-cond-exec
9814 Disable optimization of @code{&&} and @code{||} in conditional execution.
9816 This switch is mainly for debugging the compiler and will likely be removed
9817 in a future version.
9819 @item -mnested-cond-exec
9820 @opindex mnested-cond-exec
9822 Enable nested conditional execution optimizations (default).
9824 This switch is mainly for debugging the compiler and will likely be removed
9825 in a future version.
9827 @item -mno-nested-cond-exec
9828 @opindex mno-nested-cond-exec
9830 Disable nested conditional execution optimizations.
9832 This switch is mainly for debugging the compiler and will likely be removed
9833 in a future version.
9835 @item -moptimize-membar
9836 @opindex moptimize-membar
9838 This switch removes redundant @code{membar} instructions from the
9839 compiler generated code. It is enabled by default.
9841 @item -mno-optimize-membar
9842 @opindex mno-optimize-membar
9844 This switch disables the automatic removal of redundant @code{membar}
9845 instructions from the generated code.
9847 @item -mtomcat-stats
9848 @opindex mtomcat-stats
9850 Cause gas to print out tomcat statistics.
9852 @item -mcpu=@var{cpu}
9855 Select the processor type for which to generate code. Possible values are
9856 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9857 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9861 @node GNU/Linux Options
9862 @subsection GNU/Linux Options
9864 These @samp{-m} options are defined for GNU/Linux targets:
9869 Use the GNU C library instead of uClibc. This is the default except
9870 on @samp{*-*-linux-*uclibc*} targets.
9874 Use uClibc instead of the GNU C library. This is the default on
9875 @samp{*-*-linux-*uclibc*} targets.
9878 @node H8/300 Options
9879 @subsection H8/300 Options
9881 These @samp{-m} options are defined for the H8/300 implementations:
9886 Shorten some address references at link time, when possible; uses the
9887 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9888 ld, Using ld}, for a fuller description.
9892 Generate code for the H8/300H@.
9896 Generate code for the H8S@.
9900 Generate code for the H8S and H8/300H in the normal mode. This switch
9901 must be used either with @option{-mh} or @option{-ms}.
9905 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9909 Make @code{int} data 32 bits by default.
9913 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9914 The default for the H8/300H and H8S is to align longs and floats on 4
9916 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9917 This option has no effect on the H8/300.
9921 @subsection HPPA Options
9922 @cindex HPPA Options
9924 These @samp{-m} options are defined for the HPPA family of computers:
9927 @item -march=@var{architecture-type}
9929 Generate code for the specified architecture. The choices for
9930 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9931 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9932 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9933 architecture option for your machine. Code compiled for lower numbered
9934 architectures will run on higher numbered architectures, but not the
9938 @itemx -mpa-risc-1-1
9939 @itemx -mpa-risc-2-0
9940 @opindex mpa-risc-1-0
9941 @opindex mpa-risc-1-1
9942 @opindex mpa-risc-2-0
9943 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9946 @opindex mbig-switch
9947 Generate code suitable for big switch tables. Use this option only if
9948 the assembler/linker complain about out of range branches within a switch
9951 @item -mjump-in-delay
9952 @opindex mjump-in-delay
9953 Fill delay slots of function calls with unconditional jump instructions
9954 by modifying the return pointer for the function call to be the target
9955 of the conditional jump.
9957 @item -mdisable-fpregs
9958 @opindex mdisable-fpregs
9959 Prevent floating point registers from being used in any manner. This is
9960 necessary for compiling kernels which perform lazy context switching of
9961 floating point registers. If you use this option and attempt to perform
9962 floating point operations, the compiler will abort.
9964 @item -mdisable-indexing
9965 @opindex mdisable-indexing
9966 Prevent the compiler from using indexing address modes. This avoids some
9967 rather obscure problems when compiling MIG generated code under MACH@.
9969 @item -mno-space-regs
9970 @opindex mno-space-regs
9971 Generate code that assumes the target has no space registers. This allows
9972 GCC to generate faster indirect calls and use unscaled index address modes.
9974 Such code is suitable for level 0 PA systems and kernels.
9976 @item -mfast-indirect-calls
9977 @opindex mfast-indirect-calls
9978 Generate code that assumes calls never cross space boundaries. This
9979 allows GCC to emit code which performs faster indirect calls.
9981 This option will not work in the presence of shared libraries or nested
9984 @item -mfixed-range=@var{register-range}
9985 @opindex mfixed-range
9986 Generate code treating the given register range as fixed registers.
9987 A fixed register is one that the register allocator can not use. This is
9988 useful when compiling kernel code. A register range is specified as
9989 two registers separated by a dash. Multiple register ranges can be
9990 specified separated by a comma.
9992 @item -mlong-load-store
9993 @opindex mlong-load-store
9994 Generate 3-instruction load and store sequences as sometimes required by
9995 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9998 @item -mportable-runtime
9999 @opindex mportable-runtime
10000 Use the portable calling conventions proposed by HP for ELF systems.
10004 Enable the use of assembler directives only GAS understands.
10006 @item -mschedule=@var{cpu-type}
10008 Schedule code according to the constraints for the machine type
10009 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10010 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10011 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10012 proper scheduling option for your machine. The default scheduling is
10016 @opindex mlinker-opt
10017 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10018 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10019 linkers in which they give bogus error messages when linking some programs.
10022 @opindex msoft-float
10023 Generate output containing library calls for floating point.
10024 @strong{Warning:} the requisite libraries are not available for all HPPA
10025 targets. Normally the facilities of the machine's usual C compiler are
10026 used, but this cannot be done directly in cross-compilation. You must make
10027 your own arrangements to provide suitable library functions for
10028 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
10029 does provide software floating point support.
10031 @option{-msoft-float} changes the calling convention in the output file;
10032 therefore, it is only useful if you compile @emph{all} of a program with
10033 this option. In particular, you need to compile @file{libgcc.a}, the
10034 library that comes with GCC, with @option{-msoft-float} in order for
10039 Generate the predefine, @code{_SIO}, for server IO@. The default is
10040 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10041 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10042 options are available under HP-UX and HI-UX@.
10046 Use GNU ld specific options. This passes @option{-shared} to ld when
10047 building a shared library. It is the default when GCC is configured,
10048 explicitly or implicitly, with the GNU linker. This option does not
10049 have any affect on which ld is called, it only changes what parameters
10050 are passed to that ld. The ld that is called is determined by the
10051 @option{--with-ld} configure option, GCC's program search path, and
10052 finally by the user's @env{PATH}. The linker used by GCC can be printed
10053 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10054 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
10058 Use HP ld specific options. This passes @option{-b} to ld when building
10059 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10060 links. It is the default when GCC is configured, explicitly or
10061 implicitly, with the HP linker. This option does not have any affect on
10062 which ld is called, it only changes what parameters are passed to that
10063 ld. The ld that is called is determined by the @option{--with-ld}
10064 configure option, GCC's program search path, and finally by the user's
10065 @env{PATH}. The linker used by GCC can be printed using @samp{which
10066 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10067 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
10070 @opindex mno-long-calls
10071 Generate code that uses long call sequences. This ensures that a call
10072 is always able to reach linker generated stubs. The default is to generate
10073 long calls only when the distance from the call site to the beginning
10074 of the function or translation unit, as the case may be, exceeds a
10075 predefined limit set by the branch type being used. The limits for
10076 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10077 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10080 Distances are measured from the beginning of functions when using the
10081 @option{-ffunction-sections} option, or when using the @option{-mgas}
10082 and @option{-mno-portable-runtime} options together under HP-UX with
10085 It is normally not desirable to use this option as it will degrade
10086 performance. However, it may be useful in large applications,
10087 particularly when partial linking is used to build the application.
10089 The types of long calls used depends on the capabilities of the
10090 assembler and linker, and the type of code being generated. The
10091 impact on systems that support long absolute calls, and long pic
10092 symbol-difference or pc-relative calls should be relatively small.
10093 However, an indirect call is used on 32-bit ELF systems in pic code
10094 and it is quite long.
10096 @item -munix=@var{unix-std}
10098 Generate compiler predefines and select a startfile for the specified
10099 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10100 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10101 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10102 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10103 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10106 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10107 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10108 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10109 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10110 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10111 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10113 It is @emph{important} to note that this option changes the interfaces
10114 for various library routines. It also affects the operational behavior
10115 of the C library. Thus, @emph{extreme} care is needed in using this
10118 Library code that is intended to operate with more than one UNIX
10119 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10120 as appropriate. Most GNU software doesn't provide this capability.
10124 Suppress the generation of link options to search libdld.sl when the
10125 @option{-static} option is specified on HP-UX 10 and later.
10129 The HP-UX implementation of setlocale in libc has a dependency on
10130 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10131 when the @option{-static} option is specified, special link options
10132 are needed to resolve this dependency.
10134 On HP-UX 10 and later, the GCC driver adds the necessary options to
10135 link with libdld.sl when the @option{-static} option is specified.
10136 This causes the resulting binary to be dynamic. On the 64-bit port,
10137 the linkers generate dynamic binaries by default in any case. The
10138 @option{-nolibdld} option can be used to prevent the GCC driver from
10139 adding these link options.
10143 Add support for multithreading with the @dfn{dce thread} library
10144 under HP-UX@. This option sets flags for both the preprocessor and
10148 @node i386 and x86-64 Options
10149 @subsection Intel 386 and AMD x86-64 Options
10150 @cindex i386 Options
10151 @cindex x86-64 Options
10152 @cindex Intel 386 Options
10153 @cindex AMD x86-64 Options
10155 These @samp{-m} options are defined for the i386 and x86-64 family of
10159 @item -mtune=@var{cpu-type}
10161 Tune to @var{cpu-type} everything applicable about the generated code, except
10162 for the ABI and the set of available instructions. The choices for
10163 @var{cpu-type} are:
10166 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10167 If you know the CPU on which your code will run, then you should use
10168 the corresponding @option{-mtune} option instead of
10169 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10170 of your application will have, then you should use this option.
10172 As new processors are deployed in the marketplace, the behavior of this
10173 option will change. Therefore, if you upgrade to a newer version of
10174 GCC, the code generated option will change to reflect the processors
10175 that were most common when that version of GCC was released.
10177 There is no @option{-march=generic} option because @option{-march}
10178 indicates the instruction set the compiler can use, and there is no
10179 generic instruction set applicable to all processors. In contrast,
10180 @option{-mtune} indicates the processor (or, in this case, collection of
10181 processors) for which the code is optimized.
10183 This selects the CPU to tune for at compilation time by determining
10184 the processor type of the compiling machine. Using @option{-mtune=native}
10185 will produce code optimized for the local machine under the constraints
10186 of the selected instruction set. Using @option{-march=native} will
10187 enable all instruction subsets supported by the local machine (hence
10188 the result might not run on different machines).
10190 Original Intel's i386 CPU@.
10192 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10193 @item i586, pentium
10194 Intel Pentium CPU with no MMX support.
10196 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10198 Intel PentiumPro CPU@.
10200 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10201 instruction set will be used, so the code will run on all i686 family chips.
10203 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10204 @item pentium3, pentium3m
10205 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10208 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10209 support. Used by Centrino notebooks.
10210 @item pentium4, pentium4m
10211 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10213 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10216 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10217 SSE2 and SSE3 instruction set support.
10219 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10220 instruction set support.
10222 AMD K6 CPU with MMX instruction set support.
10224 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
10225 @item athlon, athlon-tbird
10226 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
10228 @item athlon-4, athlon-xp, athlon-mp
10229 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
10230 instruction set support.
10231 @item k8, opteron, athlon64, athlon-fx
10232 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10233 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
10234 @item k8-sse3, opteron-sse3, athlon64-sse3
10235 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10236 @item amdfam10, barcelona
10237 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10238 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10239 instruction set extensions.)
10241 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10244 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
10245 instruction set support.
10247 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
10248 implemented for this chip.)
10250 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10251 implemented for this chip.)
10253 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10256 While picking a specific @var{cpu-type} will schedule things appropriately
10257 for that particular chip, the compiler will not generate any code that
10258 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10261 @item -march=@var{cpu-type}
10263 Generate instructions for the machine type @var{cpu-type}. The choices
10264 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10265 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10267 @item -mcpu=@var{cpu-type}
10269 A deprecated synonym for @option{-mtune}.
10271 @item -mfpmath=@var{unit}
10273 Generate floating point arithmetics for selected unit @var{unit}. The choices
10274 for @var{unit} are:
10278 Use the standard 387 floating point coprocessor present majority of chips and
10279 emulated otherwise. Code compiled with this option will run almost everywhere.
10280 The temporary results are computed in 80bit precision instead of precision
10281 specified by the type resulting in slightly different results compared to most
10282 of other chips. See @option{-ffloat-store} for more detailed description.
10284 This is the default choice for i386 compiler.
10287 Use scalar floating point instructions present in the SSE instruction set.
10288 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10289 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10290 instruction set supports only single precision arithmetics, thus the double and
10291 extended precision arithmetics is still done using 387. Later version, present
10292 only in Pentium4 and the future AMD x86-64 chips supports double precision
10295 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10296 or @option{-msse2} switches to enable SSE extensions and make this option
10297 effective. For the x86-64 compiler, these extensions are enabled by default.
10299 The resulting code should be considerably faster in the majority of cases and avoid
10300 the numerical instability problems of 387 code, but may break some existing
10301 code that expects temporaries to be 80bit.
10303 This is the default choice for the x86-64 compiler.
10306 Attempt to utilize both instruction sets at once. This effectively double the
10307 amount of available registers and on chips with separate execution units for
10308 387 and SSE the execution resources too. Use this option with care, as it is
10309 still experimental, because the GCC register allocator does not model separate
10310 functional units well resulting in instable performance.
10313 @item -masm=@var{dialect}
10314 @opindex masm=@var{dialect}
10315 Output asm instructions using selected @var{dialect}. Supported
10316 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10317 not support @samp{intel}.
10320 @itemx -mno-ieee-fp
10322 @opindex mno-ieee-fp
10323 Control whether or not the compiler uses IEEE floating point
10324 comparisons. These handle correctly the case where the result of a
10325 comparison is unordered.
10328 @opindex msoft-float
10329 Generate output containing library calls for floating point.
10330 @strong{Warning:} the requisite libraries are not part of GCC@.
10331 Normally the facilities of the machine's usual C compiler are used, but
10332 this can't be done directly in cross-compilation. You must make your
10333 own arrangements to provide suitable library functions for
10336 On machines where a function returns floating point results in the 80387
10337 register stack, some floating point opcodes may be emitted even if
10338 @option{-msoft-float} is used.
10340 @item -mno-fp-ret-in-387
10341 @opindex mno-fp-ret-in-387
10342 Do not use the FPU registers for return values of functions.
10344 The usual calling convention has functions return values of types
10345 @code{float} and @code{double} in an FPU register, even if there
10346 is no FPU@. The idea is that the operating system should emulate
10349 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10350 in ordinary CPU registers instead.
10352 @item -mno-fancy-math-387
10353 @opindex mno-fancy-math-387
10354 Some 387 emulators do not support the @code{sin}, @code{cos} and
10355 @code{sqrt} instructions for the 387. Specify this option to avoid
10356 generating those instructions. This option is the default on FreeBSD,
10357 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10358 indicates that the target cpu will always have an FPU and so the
10359 instruction will not need emulation. As of revision 2.6.1, these
10360 instructions are not generated unless you also use the
10361 @option{-funsafe-math-optimizations} switch.
10363 @item -malign-double
10364 @itemx -mno-align-double
10365 @opindex malign-double
10366 @opindex mno-align-double
10367 Control whether GCC aligns @code{double}, @code{long double}, and
10368 @code{long long} variables on a two word boundary or a one word
10369 boundary. Aligning @code{double} variables on a two word boundary will
10370 produce code that runs somewhat faster on a @samp{Pentium} at the
10371 expense of more memory.
10373 On x86-64, @option{-malign-double} is enabled by default.
10375 @strong{Warning:} if you use the @option{-malign-double} switch,
10376 structures containing the above types will be aligned differently than
10377 the published application binary interface specifications for the 386
10378 and will not be binary compatible with structures in code compiled
10379 without that switch.
10381 @item -m96bit-long-double
10382 @itemx -m128bit-long-double
10383 @opindex m96bit-long-double
10384 @opindex m128bit-long-double
10385 These switches control the size of @code{long double} type. The i386
10386 application binary interface specifies the size to be 96 bits,
10387 so @option{-m96bit-long-double} is the default in 32 bit mode.
10389 Modern architectures (Pentium and newer) would prefer @code{long double}
10390 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10391 conforming to the ABI, this would not be possible. So specifying a
10392 @option{-m128bit-long-double} will align @code{long double}
10393 to a 16 byte boundary by padding the @code{long double} with an additional
10396 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10397 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10399 Notice that neither of these options enable any extra precision over the x87
10400 standard of 80 bits for a @code{long double}.
10402 @strong{Warning:} if you override the default value for your target ABI, the
10403 structures and arrays containing @code{long double} variables will change
10404 their size as well as function calling convention for function taking
10405 @code{long double} will be modified. Hence they will not be binary
10406 compatible with arrays or structures in code compiled without that switch.
10408 @item -mmlarge-data-threshold=@var{number}
10409 @opindex mlarge-data-threshold=@var{number}
10410 When @option{-mcmodel=medium} is specified, the data greater than
10411 @var{threshold} are placed in large data section. This value must be the
10412 same across all object linked into the binary and defaults to 65535.
10416 Use a different function-calling convention, in which functions that
10417 take a fixed number of arguments return with the @code{ret} @var{num}
10418 instruction, which pops their arguments while returning. This saves one
10419 instruction in the caller since there is no need to pop the arguments
10422 You can specify that an individual function is called with this calling
10423 sequence with the function attribute @samp{stdcall}. You can also
10424 override the @option{-mrtd} option by using the function attribute
10425 @samp{cdecl}. @xref{Function Attributes}.
10427 @strong{Warning:} this calling convention is incompatible with the one
10428 normally used on Unix, so you cannot use it if you need to call
10429 libraries compiled with the Unix compiler.
10431 Also, you must provide function prototypes for all functions that
10432 take variable numbers of arguments (including @code{printf});
10433 otherwise incorrect code will be generated for calls to those
10436 In addition, seriously incorrect code will result if you call a
10437 function with too many arguments. (Normally, extra arguments are
10438 harmlessly ignored.)
10440 @item -mregparm=@var{num}
10442 Control how many registers are used to pass integer arguments. By
10443 default, no registers are used to pass arguments, and at most 3
10444 registers can be used. You can control this behavior for a specific
10445 function by using the function attribute @samp{regparm}.
10446 @xref{Function Attributes}.
10448 @strong{Warning:} if you use this switch, and
10449 @var{num} is nonzero, then you must build all modules with the same
10450 value, including any libraries. This includes the system libraries and
10454 @opindex msseregparm
10455 Use SSE register passing conventions for float and double arguments
10456 and return values. You can control this behavior for a specific
10457 function by using the function attribute @samp{sseregparm}.
10458 @xref{Function Attributes}.
10460 @strong{Warning:} if you use this switch then you must build all
10461 modules with the same value, including any libraries. This includes
10462 the system libraries and startup modules.
10471 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10472 is specified, the significands of results of floating-point operations are
10473 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10474 significands of results of floating-point operations to 53 bits (double
10475 precision) and @option{-mpc80} rounds the significands of results of
10476 floating-point operations to 64 bits (extended double precision), which is
10477 the default. When this option is used, floating-point operations in higher
10478 precisions are not available to the programmer without setting the FPU
10479 control word explicitly.
10481 Setting the rounding of floating-point operations to less than the default
10482 80 bits can speed some programs by 2% or more. Note that some mathematical
10483 libraries assume that extended precision (80 bit) floating-point operations
10484 are enabled by default; routines in such libraries could suffer significant
10485 loss of accuracy, typically through so-called "catastrophic cancellation",
10486 when this option is used to set the precision to less than extended precision.
10488 @item -mstackrealign
10489 @opindex mstackrealign
10490 Realign the stack at entry. On the Intel x86, the
10491 @option{-mstackrealign} option will generate an alternate prologue and
10492 epilogue that realigns the runtime stack. This supports mixing legacy
10493 codes that keep a 4-byte aligned stack with modern codes that keep a
10494 16-byte stack for SSE compatibility. The alternate prologue and
10495 epilogue are slower and bigger than the regular ones, and the
10496 alternate prologue requires an extra scratch register; this lowers the
10497 number of registers available if used in conjunction with the
10498 @code{regparm} attribute. The @option{-mstackrealign} option is
10499 incompatible with the nested function prologue; this is considered a
10500 hard error. See also the attribute @code{force_align_arg_pointer},
10501 applicable to individual functions.
10503 @item -mpreferred-stack-boundary=@var{num}
10504 @opindex mpreferred-stack-boundary
10505 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10506 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10507 the default is 4 (16 bytes or 128 bits).
10509 On Pentium and PentiumPro, @code{double} and @code{long double} values
10510 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10511 suffer significant run time performance penalties. On Pentium III, the
10512 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10513 properly if it is not 16 byte aligned.
10515 To ensure proper alignment of this values on the stack, the stack boundary
10516 must be as aligned as that required by any value stored on the stack.
10517 Further, every function must be generated such that it keeps the stack
10518 aligned. Thus calling a function compiled with a higher preferred
10519 stack boundary from a function compiled with a lower preferred stack
10520 boundary will most likely misalign the stack. It is recommended that
10521 libraries that use callbacks always use the default setting.
10523 This extra alignment does consume extra stack space, and generally
10524 increases code size. Code that is sensitive to stack space usage, such
10525 as embedded systems and operating system kernels, may want to reduce the
10526 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10560 These switches enable or disable the use of instructions in the MMX,
10561 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, SSE5, ABM or 3DNow! extended
10563 These extensions are also available as built-in functions: see
10564 @ref{X86 Built-in Functions}, for details of the functions enabled and
10565 disabled by these switches.
10567 To have SSE/SSE2 instructions generated automatically from floating-point
10568 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10570 These options will enable GCC to use these extended instructions in
10571 generated code, even without @option{-mfpmath=sse}. Applications which
10572 perform runtime CPU detection must compile separate files for each
10573 supported architecture, using the appropriate flags. In particular,
10574 the file containing the CPU detection code should be compiled without
10579 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10580 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10581 data types. This is useful for high resolution counters that could be updated
10582 by multiple processors (or cores). This instruction is generated as part of
10583 atomic built-in functions: see @ref{Atomic Builtins} for details.
10587 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10588 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10589 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10590 SAHF are load and store instructions, respectively, for certain status flags.
10591 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10592 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10596 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10597 vectorized variants RCPPS and RSQRTPS) instead of DIVSS and SQRTSS (and their
10598 vectorized variants). These instructions will be generated only when
10599 @option{-funsafe-math-optimizations} is enabled.
10601 @item -mveclibabi=@var{type}
10602 @opindex mveclibabi
10603 Specifies the ABI type to use for vectorizing intrinsics using an
10604 external library. Supported types are @code{acml} for the AMD
10605 math core library style of interfacing. GCC will currently emit
10606 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10607 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10608 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10609 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10610 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10611 is enabled. A ACML ABI compatible library will have to be specified
10615 @itemx -mno-push-args
10616 @opindex mpush-args
10617 @opindex mno-push-args
10618 Use PUSH operations to store outgoing parameters. This method is shorter
10619 and usually equally fast as method using SUB/MOV operations and is enabled
10620 by default. In some cases disabling it may improve performance because of
10621 improved scheduling and reduced dependencies.
10623 @item -maccumulate-outgoing-args
10624 @opindex maccumulate-outgoing-args
10625 If enabled, the maximum amount of space required for outgoing arguments will be
10626 computed in the function prologue. This is faster on most modern CPUs
10627 because of reduced dependencies, improved scheduling and reduced stack usage
10628 when preferred stack boundary is not equal to 2. The drawback is a notable
10629 increase in code size. This switch implies @option{-mno-push-args}.
10633 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10634 on thread-safe exception handling must compile and link all code with the
10635 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10636 @option{-D_MT}; when linking, it links in a special thread helper library
10637 @option{-lmingwthrd} which cleans up per thread exception handling data.
10639 @item -mno-align-stringops
10640 @opindex mno-align-stringops
10641 Do not align destination of inlined string operations. This switch reduces
10642 code size and improves performance in case the destination is already aligned,
10643 but GCC doesn't know about it.
10645 @item -minline-all-stringops
10646 @opindex minline-all-stringops
10647 By default GCC inlines string operations only when destination is known to be
10648 aligned at least to 4 byte boundary. This enables more inlining, increase code
10649 size, but may improve performance of code that depends on fast memcpy, strlen
10650 and memset for short lengths.
10652 @item -minline-stringops-dynamically
10653 @opindex minline-stringops-dynamically
10654 For string operation of unknown size, inline runtime checks so for small
10655 blocks inline code is used, while for large blocks library call is used.
10657 @item -mstringop-strategy=@var{alg}
10658 @opindex mstringop-strategy=@var{alg}
10659 Overwrite internal decision heuristic about particular algorithm to inline
10660 string operation with. The allowed values are @code{rep_byte},
10661 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10662 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10663 expanding inline loop, @code{libcall} for always expanding library call.
10665 @item -momit-leaf-frame-pointer
10666 @opindex momit-leaf-frame-pointer
10667 Don't keep the frame pointer in a register for leaf functions. This
10668 avoids the instructions to save, set up and restore frame pointers and
10669 makes an extra register available in leaf functions. The option
10670 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10671 which might make debugging harder.
10673 @item -mtls-direct-seg-refs
10674 @itemx -mno-tls-direct-seg-refs
10675 @opindex mtls-direct-seg-refs
10676 Controls whether TLS variables may be accessed with offsets from the
10677 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10678 or whether the thread base pointer must be added. Whether or not this
10679 is legal depends on the operating system, and whether it maps the
10680 segment to cover the entire TLS area.
10682 For systems that use GNU libc, the default is on.
10685 @itemx -mno-fused-madd
10686 @opindex mfused-madd
10687 Enable automatic generation of fused floating point multiply-add instructions
10688 if the ISA supports such instructions. The -mfused-madd option is on by
10689 default. The fused multiply-add instructions have a different
10690 rounding behavior compared to executing a multiply followed by an add.
10693 These @samp{-m} switches are supported in addition to the above
10694 on AMD x86-64 processors in 64-bit environments.
10701 Generate code for a 32-bit or 64-bit environment.
10702 The 32-bit environment sets int, long and pointer to 32 bits and
10703 generates code that runs on any i386 system.
10704 The 64-bit environment sets int to 32 bits and long and pointer
10705 to 64 bits and generates code for AMD's x86-64 architecture. For
10706 darwin only the -m64 option turns off the @option{-fno-pic} and
10707 @option{-mdynamic-no-pic} options.
10709 @item -mno-red-zone
10710 @opindex no-red-zone
10711 Do not use a so called red zone for x86-64 code. The red zone is mandated
10712 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10713 stack pointer that will not be modified by signal or interrupt handlers
10714 and therefore can be used for temporary data without adjusting the stack
10715 pointer. The flag @option{-mno-red-zone} disables this red zone.
10717 @item -mcmodel=small
10718 @opindex mcmodel=small
10719 Generate code for the small code model: the program and its symbols must
10720 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10721 Programs can be statically or dynamically linked. This is the default
10724 @item -mcmodel=kernel
10725 @opindex mcmodel=kernel
10726 Generate code for the kernel code model. The kernel runs in the
10727 negative 2 GB of the address space.
10728 This model has to be used for Linux kernel code.
10730 @item -mcmodel=medium
10731 @opindex mcmodel=medium
10732 Generate code for the medium model: The program is linked in the lower 2
10733 GB of the address space but symbols can be located anywhere in the
10734 address space. Programs can be statically or dynamically linked, but
10735 building of shared libraries are not supported with the medium model.
10737 @item -mcmodel=large
10738 @opindex mcmodel=large
10739 Generate code for the large model: This model makes no assumptions
10740 about addresses and sizes of sections.
10743 @node IA-64 Options
10744 @subsection IA-64 Options
10745 @cindex IA-64 Options
10747 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10751 @opindex mbig-endian
10752 Generate code for a big endian target. This is the default for HP-UX@.
10754 @item -mlittle-endian
10755 @opindex mlittle-endian
10756 Generate code for a little endian target. This is the default for AIX5
10762 @opindex mno-gnu-as
10763 Generate (or don't) code for the GNU assembler. This is the default.
10764 @c Also, this is the default if the configure option @option{--with-gnu-as}
10770 @opindex mno-gnu-ld
10771 Generate (or don't) code for the GNU linker. This is the default.
10772 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10777 Generate code that does not use a global pointer register. The result
10778 is not position independent code, and violates the IA-64 ABI@.
10780 @item -mvolatile-asm-stop
10781 @itemx -mno-volatile-asm-stop
10782 @opindex mvolatile-asm-stop
10783 @opindex mno-volatile-asm-stop
10784 Generate (or don't) a stop bit immediately before and after volatile asm
10787 @item -mregister-names
10788 @itemx -mno-register-names
10789 @opindex mregister-names
10790 @opindex mno-register-names
10791 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10792 the stacked registers. This may make assembler output more readable.
10798 Disable (or enable) optimizations that use the small data section. This may
10799 be useful for working around optimizer bugs.
10801 @item -mconstant-gp
10802 @opindex mconstant-gp
10803 Generate code that uses a single constant global pointer value. This is
10804 useful when compiling kernel code.
10808 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10809 This is useful when compiling firmware code.
10811 @item -minline-float-divide-min-latency
10812 @opindex minline-float-divide-min-latency
10813 Generate code for inline divides of floating point values
10814 using the minimum latency algorithm.
10816 @item -minline-float-divide-max-throughput
10817 @opindex minline-float-divide-max-throughput
10818 Generate code for inline divides of floating point values
10819 using the maximum throughput algorithm.
10821 @item -minline-int-divide-min-latency
10822 @opindex minline-int-divide-min-latency
10823 Generate code for inline divides of integer values
10824 using the minimum latency algorithm.
10826 @item -minline-int-divide-max-throughput
10827 @opindex minline-int-divide-max-throughput
10828 Generate code for inline divides of integer values
10829 using the maximum throughput algorithm.
10831 @item -minline-sqrt-min-latency
10832 @opindex minline-sqrt-min-latency
10833 Generate code for inline square roots
10834 using the minimum latency algorithm.
10836 @item -minline-sqrt-max-throughput
10837 @opindex minline-sqrt-max-throughput
10838 Generate code for inline square roots
10839 using the maximum throughput algorithm.
10841 @item -mno-dwarf2-asm
10842 @itemx -mdwarf2-asm
10843 @opindex mno-dwarf2-asm
10844 @opindex mdwarf2-asm
10845 Don't (or do) generate assembler code for the DWARF2 line number debugging
10846 info. This may be useful when not using the GNU assembler.
10848 @item -mearly-stop-bits
10849 @itemx -mno-early-stop-bits
10850 @opindex mearly-stop-bits
10851 @opindex mno-early-stop-bits
10852 Allow stop bits to be placed earlier than immediately preceding the
10853 instruction that triggered the stop bit. This can improve instruction
10854 scheduling, but does not always do so.
10856 @item -mfixed-range=@var{register-range}
10857 @opindex mfixed-range
10858 Generate code treating the given register range as fixed registers.
10859 A fixed register is one that the register allocator can not use. This is
10860 useful when compiling kernel code. A register range is specified as
10861 two registers separated by a dash. Multiple register ranges can be
10862 specified separated by a comma.
10864 @item -mtls-size=@var{tls-size}
10866 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10869 @item -mtune=@var{cpu-type}
10871 Tune the instruction scheduling for a particular CPU, Valid values are
10872 itanium, itanium1, merced, itanium2, and mckinley.
10878 Add support for multithreading using the POSIX threads library. This
10879 option sets flags for both the preprocessor and linker. It does
10880 not affect the thread safety of object code produced by the compiler or
10881 that of libraries supplied with it. These are HP-UX specific flags.
10887 Generate code for a 32-bit or 64-bit environment.
10888 The 32-bit environment sets int, long and pointer to 32 bits.
10889 The 64-bit environment sets int to 32 bits and long and pointer
10890 to 64 bits. These are HP-UX specific flags.
10892 @item -mno-sched-br-data-spec
10893 @itemx -msched-br-data-spec
10894 @opindex mno-sched-br-data-spec
10895 @opindex msched-br-data-spec
10896 (Dis/En)able data speculative scheduling before reload.
10897 This will result in generation of the ld.a instructions and
10898 the corresponding check instructions (ld.c / chk.a).
10899 The default is 'disable'.
10901 @item -msched-ar-data-spec
10902 @itemx -mno-sched-ar-data-spec
10903 @opindex msched-ar-data-spec
10904 @opindex mno-sched-ar-data-spec
10905 (En/Dis)able data speculative scheduling after reload.
10906 This will result in generation of the ld.a instructions and
10907 the corresponding check instructions (ld.c / chk.a).
10908 The default is 'enable'.
10910 @item -mno-sched-control-spec
10911 @itemx -msched-control-spec
10912 @opindex mno-sched-control-spec
10913 @opindex msched-control-spec
10914 (Dis/En)able control speculative scheduling. This feature is
10915 available only during region scheduling (i.e. before reload).
10916 This will result in generation of the ld.s instructions and
10917 the corresponding check instructions chk.s .
10918 The default is 'disable'.
10920 @item -msched-br-in-data-spec
10921 @itemx -mno-sched-br-in-data-spec
10922 @opindex msched-br-in-data-spec
10923 @opindex mno-sched-br-in-data-spec
10924 (En/Dis)able speculative scheduling of the instructions that
10925 are dependent on the data speculative loads before reload.
10926 This is effective only with @option{-msched-br-data-spec} enabled.
10927 The default is 'enable'.
10929 @item -msched-ar-in-data-spec
10930 @itemx -mno-sched-ar-in-data-spec
10931 @opindex msched-ar-in-data-spec
10932 @opindex mno-sched-ar-in-data-spec
10933 (En/Dis)able speculative scheduling of the instructions that
10934 are dependent on the data speculative loads after reload.
10935 This is effective only with @option{-msched-ar-data-spec} enabled.
10936 The default is 'enable'.
10938 @item -msched-in-control-spec
10939 @itemx -mno-sched-in-control-spec
10940 @opindex msched-in-control-spec
10941 @opindex mno-sched-in-control-spec
10942 (En/Dis)able speculative scheduling of the instructions that
10943 are dependent on the control speculative loads.
10944 This is effective only with @option{-msched-control-spec} enabled.
10945 The default is 'enable'.
10948 @itemx -mno-sched-ldc
10949 @opindex msched-ldc
10950 @opindex mno-sched-ldc
10951 (En/Dis)able use of simple data speculation checks ld.c .
10952 If disabled, only chk.a instructions will be emitted to check
10953 data speculative loads.
10954 The default is 'enable'.
10956 @item -mno-sched-control-ldc
10957 @itemx -msched-control-ldc
10958 @opindex mno-sched-control-ldc
10959 @opindex msched-control-ldc
10960 (Dis/En)able use of ld.c instructions to check control speculative loads.
10961 If enabled, in case of control speculative load with no speculatively
10962 scheduled dependent instructions this load will be emitted as ld.sa and
10963 ld.c will be used to check it.
10964 The default is 'disable'.
10966 @item -mno-sched-spec-verbose
10967 @itemx -msched-spec-verbose
10968 @opindex mno-sched-spec-verbose
10969 @opindex msched-spec-verbose
10970 (Dis/En)able printing of the information about speculative motions.
10972 @item -mno-sched-prefer-non-data-spec-insns
10973 @itemx -msched-prefer-non-data-spec-insns
10974 @opindex mno-sched-prefer-non-data-spec-insns
10975 @opindex msched-prefer-non-data-spec-insns
10976 If enabled, data speculative instructions will be chosen for schedule
10977 only if there are no other choices at the moment. This will make
10978 the use of the data speculation much more conservative.
10979 The default is 'disable'.
10981 @item -mno-sched-prefer-non-control-spec-insns
10982 @itemx -msched-prefer-non-control-spec-insns
10983 @opindex mno-sched-prefer-non-control-spec-insns
10984 @opindex msched-prefer-non-control-spec-insns
10985 If enabled, control speculative instructions will be chosen for schedule
10986 only if there are no other choices at the moment. This will make
10987 the use of the control speculation much more conservative.
10988 The default is 'disable'.
10990 @item -mno-sched-count-spec-in-critical-path
10991 @itemx -msched-count-spec-in-critical-path
10992 @opindex mno-sched-count-spec-in-critical-path
10993 @opindex msched-count-spec-in-critical-path
10994 If enabled, speculative dependencies will be considered during
10995 computation of the instructions priorities. This will make the use of the
10996 speculation a bit more conservative.
10997 The default is 'disable'.
11002 @subsection M32C Options
11003 @cindex M32C options
11006 @item -mcpu=@var{name}
11008 Select the CPU for which code is generated. @var{name} may be one of
11009 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11010 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11011 the M32C/80 series.
11015 Specifies that the program will be run on the simulator. This causes
11016 an alternate runtime library to be linked in which supports, for
11017 example, file I/O. You must not use this option when generating
11018 programs that will run on real hardware; you must provide your own
11019 runtime library for whatever I/O functions are needed.
11021 @item -memregs=@var{number}
11023 Specifies the number of memory-based pseudo-registers GCC will use
11024 during code generation. These pseudo-registers will be used like real
11025 registers, so there is a tradeoff between GCC's ability to fit the
11026 code into available registers, and the performance penalty of using
11027 memory instead of registers. Note that all modules in a program must
11028 be compiled with the same value for this option. Because of that, you
11029 must not use this option with the default runtime libraries gcc
11034 @node M32R/D Options
11035 @subsection M32R/D Options
11036 @cindex M32R/D options
11038 These @option{-m} options are defined for Renesas M32R/D architectures:
11043 Generate code for the M32R/2@.
11047 Generate code for the M32R/X@.
11051 Generate code for the M32R@. This is the default.
11053 @item -mmodel=small
11054 @opindex mmodel=small
11055 Assume all objects live in the lower 16MB of memory (so that their addresses
11056 can be loaded with the @code{ld24} instruction), and assume all subroutines
11057 are reachable with the @code{bl} instruction.
11058 This is the default.
11060 The addressability of a particular object can be set with the
11061 @code{model} attribute.
11063 @item -mmodel=medium
11064 @opindex mmodel=medium
11065 Assume objects may be anywhere in the 32-bit address space (the compiler
11066 will generate @code{seth/add3} instructions to load their addresses), and
11067 assume all subroutines are reachable with the @code{bl} instruction.
11069 @item -mmodel=large
11070 @opindex mmodel=large
11071 Assume objects may be anywhere in the 32-bit address space (the compiler
11072 will generate @code{seth/add3} instructions to load their addresses), and
11073 assume subroutines may not be reachable with the @code{bl} instruction
11074 (the compiler will generate the much slower @code{seth/add3/jl}
11075 instruction sequence).
11078 @opindex msdata=none
11079 Disable use of the small data area. Variables will be put into
11080 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11081 @code{section} attribute has been specified).
11082 This is the default.
11084 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11085 Objects may be explicitly put in the small data area with the
11086 @code{section} attribute using one of these sections.
11088 @item -msdata=sdata
11089 @opindex msdata=sdata
11090 Put small global and static data in the small data area, but do not
11091 generate special code to reference them.
11094 @opindex msdata=use
11095 Put small global and static data in the small data area, and generate
11096 special instructions to reference them.
11100 @cindex smaller data references
11101 Put global and static objects less than or equal to @var{num} bytes
11102 into the small data or bss sections instead of the normal data or bss
11103 sections. The default value of @var{num} is 8.
11104 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11105 for this option to have any effect.
11107 All modules should be compiled with the same @option{-G @var{num}} value.
11108 Compiling with different values of @var{num} may or may not work; if it
11109 doesn't the linker will give an error message---incorrect code will not be
11114 Makes the M32R specific code in the compiler display some statistics
11115 that might help in debugging programs.
11117 @item -malign-loops
11118 @opindex malign-loops
11119 Align all loops to a 32-byte boundary.
11121 @item -mno-align-loops
11122 @opindex mno-align-loops
11123 Do not enforce a 32-byte alignment for loops. This is the default.
11125 @item -missue-rate=@var{number}
11126 @opindex missue-rate=@var{number}
11127 Issue @var{number} instructions per cycle. @var{number} can only be 1
11130 @item -mbranch-cost=@var{number}
11131 @opindex mbranch-cost=@var{number}
11132 @var{number} can only be 1 or 2. If it is 1 then branches will be
11133 preferred over conditional code, if it is 2, then the opposite will
11136 @item -mflush-trap=@var{number}
11137 @opindex mflush-trap=@var{number}
11138 Specifies the trap number to use to flush the cache. The default is
11139 12. Valid numbers are between 0 and 15 inclusive.
11141 @item -mno-flush-trap
11142 @opindex mno-flush-trap
11143 Specifies that the cache cannot be flushed by using a trap.
11145 @item -mflush-func=@var{name}
11146 @opindex mflush-func=@var{name}
11147 Specifies the name of the operating system function to call to flush
11148 the cache. The default is @emph{_flush_cache}, but a function call
11149 will only be used if a trap is not available.
11151 @item -mno-flush-func
11152 @opindex mno-flush-func
11153 Indicates that there is no OS function for flushing the cache.
11157 @node M680x0 Options
11158 @subsection M680x0 Options
11159 @cindex M680x0 options
11161 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11162 The default settings depend on which architecture was selected when
11163 the compiler was configured; the defaults for the most common choices
11167 @item -march=@var{arch}
11169 Generate code for a specific M680x0 or ColdFire instruction set
11170 architecture. Permissible values of @var{arch} for M680x0
11171 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11172 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11173 architectures are selected according to Freescale's ISA classification
11174 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11175 @samp{isab} and @samp{isac}.
11177 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11178 code for a ColdFire target. The @var{arch} in this macro is one of the
11179 @option{-march} arguments given above.
11181 When used together, @option{-march} and @option{-mtune} select code
11182 that runs on a family of similar processors but that is optimized
11183 for a particular microarchitecture.
11185 @item -mcpu=@var{cpu}
11187 Generate code for a specific M680x0 or ColdFire processor.
11188 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11189 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11190 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11191 below, which also classifies the CPUs into families:
11193 @multitable @columnfractions 0.20 0.80
11194 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11195 @item @samp{51qe} @tab @samp{51qe}
11196 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11197 @item @samp{5206e} @tab @samp{5206e}
11198 @item @samp{5208} @tab @samp{5207} @samp{5208}
11199 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11200 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11201 @item @samp{5216} @tab @samp{5214} @samp{5216}
11202 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11203 @item @samp{5225} @tab @samp{5224} @samp{5225}
11204 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11205 @item @samp{5249} @tab @samp{5249}
11206 @item @samp{5250} @tab @samp{5250}
11207 @item @samp{5271} @tab @samp{5270} @samp{5271}
11208 @item @samp{5272} @tab @samp{5272}
11209 @item @samp{5275} @tab @samp{5274} @samp{5275}
11210 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11211 @item @samp{5307} @tab @samp{5307}
11212 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11213 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11214 @item @samp{5407} @tab @samp{5407}
11215 @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}
11218 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11219 @var{arch} is compatible with @var{cpu}. Other combinations of
11220 @option{-mcpu} and @option{-march} are rejected.
11222 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11223 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11224 where the value of @var{family} is given by the table above.
11226 @item -mtune=@var{tune}
11228 Tune the code for a particular microarchitecture, within the
11229 constraints set by @option{-march} and @option{-mcpu}.
11230 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11231 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11232 and @samp{cpu32}. The ColdFire microarchitectures
11233 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11235 You can also use @option{-mtune=68020-40} for code that needs
11236 to run relatively well on 68020, 68030 and 68040 targets.
11237 @option{-mtune=68020-60} is similar but includes 68060 targets
11238 as well. These two options select the same tuning decisions as
11239 @option{-m68020-40} and @option{-m68020-60} respectively.
11241 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11242 when tuning for 680x0 architecture @var{arch}. It also defines
11243 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11244 option is used. If gcc is tuning for a range of architectures,
11245 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11246 it defines the macros for every architecture in the range.
11248 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11249 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11250 of the arguments given above.
11256 Generate output for a 68000. This is the default
11257 when the compiler is configured for 68000-based systems.
11258 It is equivalent to @option{-march=68000}.
11260 Use this option for microcontrollers with a 68000 or EC000 core,
11261 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11265 Generate output for a 68010. This is the default
11266 when the compiler is configured for 68010-based systems.
11267 It is equivalent to @option{-march=68010}.
11273 Generate output for a 68020. This is the default
11274 when the compiler is configured for 68020-based systems.
11275 It is equivalent to @option{-march=68020}.
11279 Generate output for a 68030. This is the default when the compiler is
11280 configured for 68030-based systems. It is equivalent to
11281 @option{-march=68030}.
11285 Generate output for a 68040. This is the default when the compiler is
11286 configured for 68040-based systems. It is equivalent to
11287 @option{-march=68040}.
11289 This option inhibits the use of 68881/68882 instructions that have to be
11290 emulated by software on the 68040. Use this option if your 68040 does not
11291 have code to emulate those instructions.
11295 Generate output for a 68060. This is the default when the compiler is
11296 configured for 68060-based systems. It is equivalent to
11297 @option{-march=68060}.
11299 This option inhibits the use of 68020 and 68881/68882 instructions that
11300 have to be emulated by software on the 68060. Use this option if your 68060
11301 does not have code to emulate those instructions.
11305 Generate output for a CPU32. This is the default
11306 when the compiler is configured for CPU32-based systems.
11307 It is equivalent to @option{-march=cpu32}.
11309 Use this option for microcontrollers with a
11310 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11311 68336, 68340, 68341, 68349 and 68360.
11315 Generate output for a 520X ColdFire CPU. This is the default
11316 when the compiler is configured for 520X-based systems.
11317 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11318 in favor of that option.
11320 Use this option for microcontroller with a 5200 core, including
11321 the MCF5202, MCF5203, MCF5204 and MCF5206.
11325 Generate output for a 5206e ColdFire CPU. The option is now
11326 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11330 Generate output for a member of the ColdFire 528X family.
11331 The option is now deprecated in favor of the equivalent
11332 @option{-mcpu=528x}.
11336 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11337 in favor of the equivalent @option{-mcpu=5307}.
11341 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11342 in favor of the equivalent @option{-mcpu=5407}.
11346 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11347 This includes use of hardware floating point instructions.
11348 The option is equivalent to @option{-mcpu=547x}, and is now
11349 deprecated in favor of that option.
11353 Generate output for a 68040, without using any of the new instructions.
11354 This results in code which can run relatively efficiently on either a
11355 68020/68881 or a 68030 or a 68040. The generated code does use the
11356 68881 instructions that are emulated on the 68040.
11358 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11362 Generate output for a 68060, without using any of the new instructions.
11363 This results in code which can run relatively efficiently on either a
11364 68020/68881 or a 68030 or a 68040. The generated code does use the
11365 68881 instructions that are emulated on the 68060.
11367 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11371 @opindex mhard-float
11373 Generate floating-point instructions. This is the default for 68020
11374 and above, and for ColdFire devices that have an FPU. It defines the
11375 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11376 on ColdFire targets.
11379 @opindex msoft-float
11380 Do not generate floating-point instructions; use library calls instead.
11381 This is the default for 68000, 68010, and 68832 targets. It is also
11382 the default for ColdFire devices that have no FPU.
11388 Generate (do not generate) ColdFire hardware divide and remainder
11389 instructions. If @option{-march} is used without @option{-mcpu},
11390 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11391 architectures. Otherwise, the default is taken from the target CPU
11392 (either the default CPU, or the one specified by @option{-mcpu}). For
11393 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11394 @option{-mcpu=5206e}.
11396 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11400 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11401 Additionally, parameters passed on the stack are also aligned to a
11402 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11406 Do not consider type @code{int} to be 16 bits wide. This is the default.
11409 @itemx -mno-bitfield
11410 @opindex mnobitfield
11411 @opindex mno-bitfield
11412 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11413 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11417 Do use the bit-field instructions. The @option{-m68020} option implies
11418 @option{-mbitfield}. This is the default if you use a configuration
11419 designed for a 68020.
11423 Use a different function-calling convention, in which functions
11424 that take a fixed number of arguments return with the @code{rtd}
11425 instruction, which pops their arguments while returning. This
11426 saves one instruction in the caller since there is no need to pop
11427 the arguments there.
11429 This calling convention is incompatible with the one normally
11430 used on Unix, so you cannot use it if you need to call libraries
11431 compiled with the Unix compiler.
11433 Also, you must provide function prototypes for all functions that
11434 take variable numbers of arguments (including @code{printf});
11435 otherwise incorrect code will be generated for calls to those
11438 In addition, seriously incorrect code will result if you call a
11439 function with too many arguments. (Normally, extra arguments are
11440 harmlessly ignored.)
11442 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11443 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11447 Do not use the calling conventions selected by @option{-mrtd}.
11448 This is the default.
11451 @itemx -mno-align-int
11452 @opindex malign-int
11453 @opindex mno-align-int
11454 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11455 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11456 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11457 Aligning variables on 32-bit boundaries produces code that runs somewhat
11458 faster on processors with 32-bit busses at the expense of more memory.
11460 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11461 align structures containing the above types differently than
11462 most published application binary interface specifications for the m68k.
11466 Use the pc-relative addressing mode of the 68000 directly, instead of
11467 using a global offset table. At present, this option implies @option{-fpic},
11468 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11469 not presently supported with @option{-mpcrel}, though this could be supported for
11470 68020 and higher processors.
11472 @item -mno-strict-align
11473 @itemx -mstrict-align
11474 @opindex mno-strict-align
11475 @opindex mstrict-align
11476 Do not (do) assume that unaligned memory references will be handled by
11480 Generate code that allows the data segment to be located in a different
11481 area of memory from the text segment. This allows for execute in place in
11482 an environment without virtual memory management. This option implies
11485 @item -mno-sep-data
11486 Generate code that assumes that the data segment follows the text segment.
11487 This is the default.
11489 @item -mid-shared-library
11490 Generate code that supports shared libraries via the library ID method.
11491 This allows for execute in place and shared libraries in an environment
11492 without virtual memory management. This option implies @option{-fPIC}.
11494 @item -mno-id-shared-library
11495 Generate code that doesn't assume ID based shared libraries are being used.
11496 This is the default.
11498 @item -mshared-library-id=n
11499 Specified the identification number of the ID based shared library being
11500 compiled. Specifying a value of 0 will generate more compact code, specifying
11501 other values will force the allocation of that number to the current
11502 library but is no more space or time efficient than omitting this option.
11506 @node M68hc1x Options
11507 @subsection M68hc1x Options
11508 @cindex M68hc1x options
11510 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11511 microcontrollers. The default values for these options depends on
11512 which style of microcontroller was selected when the compiler was configured;
11513 the defaults for the most common choices are given below.
11520 Generate output for a 68HC11. This is the default
11521 when the compiler is configured for 68HC11-based systems.
11527 Generate output for a 68HC12. This is the default
11528 when the compiler is configured for 68HC12-based systems.
11534 Generate output for a 68HCS12.
11536 @item -mauto-incdec
11537 @opindex mauto-incdec
11538 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11545 Enable the use of 68HC12 min and max instructions.
11548 @itemx -mno-long-calls
11549 @opindex mlong-calls
11550 @opindex mno-long-calls
11551 Treat all calls as being far away (near). If calls are assumed to be
11552 far away, the compiler will use the @code{call} instruction to
11553 call a function and the @code{rtc} instruction for returning.
11557 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11559 @item -msoft-reg-count=@var{count}
11560 @opindex msoft-reg-count
11561 Specify the number of pseudo-soft registers which are used for the
11562 code generation. The maximum number is 32. Using more pseudo-soft
11563 register may or may not result in better code depending on the program.
11564 The default is 4 for 68HC11 and 2 for 68HC12.
11568 @node MCore Options
11569 @subsection MCore Options
11570 @cindex MCore options
11572 These are the @samp{-m} options defined for the Motorola M*Core
11578 @itemx -mno-hardlit
11580 @opindex mno-hardlit
11581 Inline constants into the code stream if it can be done in two
11582 instructions or less.
11588 Use the divide instruction. (Enabled by default).
11590 @item -mrelax-immediate
11591 @itemx -mno-relax-immediate
11592 @opindex mrelax-immediate
11593 @opindex mno-relax-immediate
11594 Allow arbitrary sized immediates in bit operations.
11596 @item -mwide-bitfields
11597 @itemx -mno-wide-bitfields
11598 @opindex mwide-bitfields
11599 @opindex mno-wide-bitfields
11600 Always treat bit-fields as int-sized.
11602 @item -m4byte-functions
11603 @itemx -mno-4byte-functions
11604 @opindex m4byte-functions
11605 @opindex mno-4byte-functions
11606 Force all functions to be aligned to a four byte boundary.
11608 @item -mcallgraph-data
11609 @itemx -mno-callgraph-data
11610 @opindex mcallgraph-data
11611 @opindex mno-callgraph-data
11612 Emit callgraph information.
11615 @itemx -mno-slow-bytes
11616 @opindex mslow-bytes
11617 @opindex mno-slow-bytes
11618 Prefer word access when reading byte quantities.
11620 @item -mlittle-endian
11621 @itemx -mbig-endian
11622 @opindex mlittle-endian
11623 @opindex mbig-endian
11624 Generate code for a little endian target.
11630 Generate code for the 210 processor.
11634 @subsection MIPS Options
11635 @cindex MIPS options
11641 Generate big-endian code.
11645 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11648 @item -march=@var{arch}
11650 Generate code that will run on @var{arch}, which can be the name of a
11651 generic MIPS ISA, or the name of a particular processor.
11653 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11654 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11655 The processor names are:
11656 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11657 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11658 @samp{5kc}, @samp{5kf},
11660 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11661 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11662 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11663 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11666 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11667 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11668 @samp{rm7000}, @samp{rm9000},
11671 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11672 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11673 The special value @samp{from-abi} selects the
11674 most compatible architecture for the selected ABI (that is,
11675 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11677 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11678 (for example, @samp{-march=r2k}). Prefixes are optional, and
11679 @samp{vr} may be written @samp{r}.
11681 Names of the form @samp{@var{n}f2_1} refer to processors with
11682 FPUs clocked at half the rate of the core, names of the form
11683 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11684 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11685 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11686 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11687 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11688 accepted as synonyms for @samp{@var{n}f1_1}.
11690 GCC defines two macros based on the value of this option. The first
11691 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11692 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11693 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11694 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11695 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11697 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11698 above. In other words, it will have the full prefix and will not
11699 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11700 the macro names the resolved architecture (either @samp{"mips1"} or
11701 @samp{"mips3"}). It names the default architecture when no
11702 @option{-march} option is given.
11704 @item -mtune=@var{arch}
11706 Optimize for @var{arch}. Among other things, this option controls
11707 the way instructions are scheduled, and the perceived cost of arithmetic
11708 operations. The list of @var{arch} values is the same as for
11711 When this option is not used, GCC will optimize for the processor
11712 specified by @option{-march}. By using @option{-march} and
11713 @option{-mtune} together, it is possible to generate code that will
11714 run on a family of processors, but optimize the code for one
11715 particular member of that family.
11717 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11718 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11719 @samp{-march} ones described above.
11723 Equivalent to @samp{-march=mips1}.
11727 Equivalent to @samp{-march=mips2}.
11731 Equivalent to @samp{-march=mips3}.
11735 Equivalent to @samp{-march=mips4}.
11739 Equivalent to @samp{-march=mips32}.
11743 Equivalent to @samp{-march=mips32r2}.
11747 Equivalent to @samp{-march=mips64}.
11752 @opindex mno-mips16
11753 Generate (do not generate) MIPS16 code. If GCC is targetting a
11754 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11756 MIPS16 code generation can also be controlled on a per-function basis
11757 by means of @code{mips16} and @code{nomips16} attributes.
11758 @xref{Function Attributes}, for more information.
11760 @item -mflip-mips16
11761 @opindex mflip-mips16
11762 Generate MIPS16 code on alternating functions. This option is provided
11763 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11764 not intended for ordinary use in compiling user code.
11766 @item -minterlink-mips16
11767 @itemx -mno-interlink-mips16
11768 @opindex minterlink-mips16
11769 @opindex mno-interlink-mips16
11770 Require (do not require) that non-MIPS16 code be link-compatible with
11773 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
11774 it must either use a call or an indirect jump. @option{-minterlink-mips16}
11775 therefore disables direct jumps unless GCC knows that the target of the
11776 jump is not MIPS16.
11788 Generate code for the given ABI@.
11790 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11791 generates 64-bit code when you select a 64-bit architecture, but you
11792 can use @option{-mgp32} to get 32-bit code instead.
11794 For information about the O64 ABI, see
11795 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11797 GCC supports a variant of the o32 ABI in which floating-point registers
11798 are 64 rather than 32 bits wide. You can select this combination with
11799 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11800 and @samp{mfhc1} instructions and is therefore only supported for
11801 MIPS32R2 processors.
11803 The register assignments for arguments and return values remain the
11804 same, but each scalar value is passed in a single 64-bit register
11805 rather than a pair of 32-bit registers. For example, scalar
11806 floating-point values are returned in @samp{$f0} only, not a
11807 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11808 remains the same, but all 64 bits are saved.
11811 @itemx -mno-abicalls
11813 @opindex mno-abicalls
11814 Generate (do not generate) code that is suitable for SVR4-style
11815 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11820 Generate (do not generate) code that is fully position-independent,
11821 and that can therefore be linked into shared libraries. This option
11822 only affects @option{-mabicalls}.
11824 All @option{-mabicalls} code has traditionally been position-independent,
11825 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11826 as an extension, the GNU toolchain allows executables to use absolute
11827 accesses for locally-binding symbols. It can also use shorter GP
11828 initialization sequences and generate direct calls to locally-defined
11829 functions. This mode is selected by @option{-mno-shared}.
11831 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11832 objects that can only be linked by the GNU linker. However, the option
11833 does not affect the ABI of the final executable; it only affects the ABI
11834 of relocatable objects. Using @option{-mno-shared} will generally make
11835 executables both smaller and quicker.
11837 @option{-mshared} is the default.
11843 Lift (do not lift) the usual restrictions on the size of the global
11846 GCC normally uses a single instruction to load values from the GOT@.
11847 While this is relatively efficient, it will only work if the GOT
11848 is smaller than about 64k. Anything larger will cause the linker
11849 to report an error such as:
11851 @cindex relocation truncated to fit (MIPS)
11853 relocation truncated to fit: R_MIPS_GOT16 foobar
11856 If this happens, you should recompile your code with @option{-mxgot}.
11857 It should then work with very large GOTs, although it will also be
11858 less efficient, since it will take three instructions to fetch the
11859 value of a global symbol.
11861 Note that some linkers can create multiple GOTs. If you have such a
11862 linker, you should only need to use @option{-mxgot} when a single object
11863 file accesses more than 64k's worth of GOT entries. Very few do.
11865 These options have no effect unless GCC is generating position
11870 Assume that general-purpose registers are 32 bits wide.
11874 Assume that general-purpose registers are 64 bits wide.
11878 Assume that floating-point registers are 32 bits wide.
11882 Assume that floating-point registers are 64 bits wide.
11885 @opindex mhard-float
11886 Use floating-point coprocessor instructions.
11889 @opindex msoft-float
11890 Do not use floating-point coprocessor instructions. Implement
11891 floating-point calculations using library calls instead.
11893 @item -msingle-float
11894 @opindex msingle-float
11895 Assume that the floating-point coprocessor only supports single-precision
11898 @item -mdouble-float
11899 @opindex mdouble-float
11900 Assume that the floating-point coprocessor supports double-precision
11901 operations. This is the default.
11907 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
11908 implement atomic memory built-in functions. When neither option is
11909 specified, GCC will use the instructions if the target architecture
11912 @option{-mllsc} is useful if the runtime environment can emulate the
11913 instructions and @option{-mno-llsc} can be useful when compiling for
11914 nonstandard ISAs. You can make either option the default by
11915 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
11916 respectively. @option{--with-llsc} is the default for some
11917 configurations; see the installation documentation for details.
11923 Use (do not use) revision 1 of the MIPS DSP ASE.
11924 @xref{MIPS DSP Built-in Functions}. This option defines the
11925 preprocessor macro @samp{__mips_dsp}. It also defines
11926 @samp{__mips_dsp_rev} to 1.
11932 Use (do not use) revision 2 of the MIPS DSP ASE.
11933 @xref{MIPS DSP Built-in Functions}. This option defines the
11934 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
11935 It also defines @samp{__mips_dsp_rev} to 2.
11938 @itemx -mno-smartmips
11939 @opindex msmartmips
11940 @opindex mno-smartmips
11941 Use (do not use) the MIPS SmartMIPS ASE.
11943 @item -mpaired-single
11944 @itemx -mno-paired-single
11945 @opindex mpaired-single
11946 @opindex mno-paired-single
11947 Use (do not use) paired-single floating-point instructions.
11948 @xref{MIPS Paired-Single Support}. This option requires
11949 hardware floating-point support to be enabled.
11955 Use (do not use) MIPS Digital Media Extension instructions.
11956 This option can only be used when generating 64-bit code and requires
11957 hardware floating-point support to be enabled.
11962 @opindex mno-mips3d
11963 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11964 The option @option{-mips3d} implies @option{-mpaired-single}.
11970 Use (do not use) MT Multithreading instructions.
11974 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11975 an explanation of the default and the way that the pointer size is
11980 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11982 The default size of @code{int}s, @code{long}s and pointers depends on
11983 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11984 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11985 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11986 or the same size as integer registers, whichever is smaller.
11992 Assume (do not assume) that all symbols have 32-bit values, regardless
11993 of the selected ABI@. This option is useful in combination with
11994 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11995 to generate shorter and faster references to symbolic addresses.
11999 Put definitions of externally-visible data in a small data section
12000 if that data is no bigger than @var{num} bytes. GCC can then access
12001 the data more efficiently; see @option{-mgpopt} for details.
12003 The default @option{-G} option depends on the configuration.
12005 @item -mlocal-sdata
12006 @itemx -mno-local-sdata
12007 @opindex mlocal-sdata
12008 @opindex mno-local-sdata
12009 Extend (do not extend) the @option{-G} behavior to local data too,
12010 such as to static variables in C. @option{-mlocal-sdata} is the
12011 default for all configurations.
12013 If the linker complains that an application is using too much small data,
12014 you might want to try rebuilding the less performance-critical parts with
12015 @option{-mno-local-sdata}. You might also want to build large
12016 libraries with @option{-mno-local-sdata}, so that the libraries leave
12017 more room for the main program.
12019 @item -mextern-sdata
12020 @itemx -mno-extern-sdata
12021 @opindex mextern-sdata
12022 @opindex mno-extern-sdata
12023 Assume (do not assume) that externally-defined data will be in
12024 a small data section if that data is within the @option{-G} limit.
12025 @option{-mextern-sdata} is the default for all configurations.
12027 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12028 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12029 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12030 is placed in a small data section. If @var{Var} is defined by another
12031 module, you must either compile that module with a high-enough
12032 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12033 definition. If @var{Var} is common, you must link the application
12034 with a high-enough @option{-G} setting.
12036 The easiest way of satisfying these restrictions is to compile
12037 and link every module with the same @option{-G} option. However,
12038 you may wish to build a library that supports several different
12039 small data limits. You can do this by compiling the library with
12040 the highest supported @option{-G} setting and additionally using
12041 @option{-mno-extern-sdata} to stop the library from making assumptions
12042 about externally-defined data.
12048 Use (do not use) GP-relative accesses for symbols that are known to be
12049 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12050 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12053 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12054 might not hold the value of @code{_gp}. For example, if the code is
12055 part of a library that might be used in a boot monitor, programs that
12056 call boot monitor routines will pass an unknown value in @code{$gp}.
12057 (In such situations, the boot monitor itself would usually be compiled
12058 with @option{-G0}.)
12060 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12061 @option{-mno-extern-sdata}.
12063 @item -membedded-data
12064 @itemx -mno-embedded-data
12065 @opindex membedded-data
12066 @opindex mno-embedded-data
12067 Allocate variables to the read-only data section first if possible, then
12068 next in the small data section if possible, otherwise in data. This gives
12069 slightly slower code than the default, but reduces the amount of RAM required
12070 when executing, and thus may be preferred for some embedded systems.
12072 @item -muninit-const-in-rodata
12073 @itemx -mno-uninit-const-in-rodata
12074 @opindex muninit-const-in-rodata
12075 @opindex mno-uninit-const-in-rodata
12076 Put uninitialized @code{const} variables in the read-only data section.
12077 This option is only meaningful in conjunction with @option{-membedded-data}.
12079 @item -mcode-readable=@var{setting}
12080 @opindex mcode-readable
12081 Specify whether GCC may generate code that reads from executable sections.
12082 There are three possible settings:
12085 @item -mcode-readable=yes
12086 Instructions may freely access executable sections. This is the
12089 @item -mcode-readable=pcrel
12090 MIPS16 PC-relative load instructions can access executable sections,
12091 but other instructions must not do so. This option is useful on 4KSc
12092 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12093 It is also useful on processors that can be configured to have a dual
12094 instruction/data SRAM interface and that, like the M4K, automatically
12095 redirect PC-relative loads to the instruction RAM.
12097 @item -mcode-readable=no
12098 Instructions must not access executable sections. This option can be
12099 useful on targets that are configured to have a dual instruction/data
12100 SRAM interface but that (unlike the M4K) do not automatically redirect
12101 PC-relative loads to the instruction RAM.
12104 @item -msplit-addresses
12105 @itemx -mno-split-addresses
12106 @opindex msplit-addresses
12107 @opindex mno-split-addresses
12108 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12109 relocation operators. This option has been superseded by
12110 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12112 @item -mexplicit-relocs
12113 @itemx -mno-explicit-relocs
12114 @opindex mexplicit-relocs
12115 @opindex mno-explicit-relocs
12116 Use (do not use) assembler relocation operators when dealing with symbolic
12117 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12118 is to use assembler macros instead.
12120 @option{-mexplicit-relocs} is the default if GCC was configured
12121 to use an assembler that supports relocation operators.
12123 @item -mcheck-zero-division
12124 @itemx -mno-check-zero-division
12125 @opindex mcheck-zero-division
12126 @opindex mno-check-zero-division
12127 Trap (do not trap) on integer division by zero.
12129 The default is @option{-mcheck-zero-division}.
12131 @item -mdivide-traps
12132 @itemx -mdivide-breaks
12133 @opindex mdivide-traps
12134 @opindex mdivide-breaks
12135 MIPS systems check for division by zero by generating either a
12136 conditional trap or a break instruction. Using traps results in
12137 smaller code, but is only supported on MIPS II and later. Also, some
12138 versions of the Linux kernel have a bug that prevents trap from
12139 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12140 allow conditional traps on architectures that support them and
12141 @option{-mdivide-breaks} to force the use of breaks.
12143 The default is usually @option{-mdivide-traps}, but this can be
12144 overridden at configure time using @option{--with-divide=breaks}.
12145 Divide-by-zero checks can be completely disabled using
12146 @option{-mno-check-zero-division}.
12151 @opindex mno-memcpy
12152 Force (do not force) the use of @code{memcpy()} for non-trivial block
12153 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12154 most constant-sized copies.
12157 @itemx -mno-long-calls
12158 @opindex mlong-calls
12159 @opindex mno-long-calls
12160 Disable (do not disable) use of the @code{jal} instruction. Calling
12161 functions using @code{jal} is more efficient but requires the caller
12162 and callee to be in the same 256 megabyte segment.
12164 This option has no effect on abicalls code. The default is
12165 @option{-mno-long-calls}.
12171 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12172 instructions, as provided by the R4650 ISA@.
12175 @itemx -mno-fused-madd
12176 @opindex mfused-madd
12177 @opindex mno-fused-madd
12178 Enable (disable) use of the floating point multiply-accumulate
12179 instructions, when they are available. The default is
12180 @option{-mfused-madd}.
12182 When multiply-accumulate instructions are used, the intermediate
12183 product is calculated to infinite precision and is not subject to
12184 the FCSR Flush to Zero bit. This may be undesirable in some
12189 Tell the MIPS assembler to not run its preprocessor over user
12190 assembler files (with a @samp{.s} suffix) when assembling them.
12193 @itemx -mno-fix-r4000
12194 @opindex mfix-r4000
12195 @opindex mno-fix-r4000
12196 Work around certain R4000 CPU errata:
12199 A double-word or a variable shift may give an incorrect result if executed
12200 immediately after starting an integer division.
12202 A double-word or a variable shift may give an incorrect result if executed
12203 while an integer multiplication is in progress.
12205 An integer division may give an incorrect result if started in a delay slot
12206 of a taken branch or a jump.
12210 @itemx -mno-fix-r4400
12211 @opindex mfix-r4400
12212 @opindex mno-fix-r4400
12213 Work around certain R4400 CPU errata:
12216 A double-word or a variable shift may give an incorrect result if executed
12217 immediately after starting an integer division.
12221 @itemx -mno-fix-vr4120
12222 @opindex mfix-vr4120
12223 Work around certain VR4120 errata:
12226 @code{dmultu} does not always produce the correct result.
12228 @code{div} and @code{ddiv} do not always produce the correct result if one
12229 of the operands is negative.
12231 The workarounds for the division errata rely on special functions in
12232 @file{libgcc.a}. At present, these functions are only provided by
12233 the @code{mips64vr*-elf} configurations.
12235 Other VR4120 errata require a nop to be inserted between certain pairs of
12236 instructions. These errata are handled by the assembler, not by GCC itself.
12239 @opindex mfix-vr4130
12240 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12241 workarounds are implemented by the assembler rather than by GCC,
12242 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12243 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12244 instructions are available instead.
12247 @itemx -mno-fix-sb1
12249 Work around certain SB-1 CPU core errata.
12250 (This flag currently works around the SB-1 revision 2
12251 ``F1'' and ``F2'' floating point errata.)
12253 @item -mflush-func=@var{func}
12254 @itemx -mno-flush-func
12255 @opindex mflush-func
12256 Specifies the function to call to flush the I and D caches, or to not
12257 call any such function. If called, the function must take the same
12258 arguments as the common @code{_flush_func()}, that is, the address of the
12259 memory range for which the cache is being flushed, the size of the
12260 memory range, and the number 3 (to flush both caches). The default
12261 depends on the target GCC was configured for, but commonly is either
12262 @samp{_flush_func} or @samp{__cpu_flush}.
12264 @item mbranch-cost=@var{num}
12265 @opindex mbranch-cost
12266 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12267 This cost is only a heuristic and is not guaranteed to produce
12268 consistent results across releases. A zero cost redundantly selects
12269 the default, which is based on the @option{-mtune} setting.
12271 @item -mbranch-likely
12272 @itemx -mno-branch-likely
12273 @opindex mbranch-likely
12274 @opindex mno-branch-likely
12275 Enable or disable use of Branch Likely instructions, regardless of the
12276 default for the selected architecture. By default, Branch Likely
12277 instructions may be generated if they are supported by the selected
12278 architecture. An exception is for the MIPS32 and MIPS64 architectures
12279 and processors which implement those architectures; for those, Branch
12280 Likely instructions will not be generated by default because the MIPS32
12281 and MIPS64 architectures specifically deprecate their use.
12283 @item -mfp-exceptions
12284 @itemx -mno-fp-exceptions
12285 @opindex mfp-exceptions
12286 Specifies whether FP exceptions are enabled. This affects how we schedule
12287 FP instructions for some processors. The default is that FP exceptions are
12290 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12291 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12294 @item -mvr4130-align
12295 @itemx -mno-vr4130-align
12296 @opindex mvr4130-align
12297 The VR4130 pipeline is two-way superscalar, but can only issue two
12298 instructions together if the first one is 8-byte aligned. When this
12299 option is enabled, GCC will align pairs of instructions that it
12300 thinks should execute in parallel.
12302 This option only has an effect when optimizing for the VR4130.
12303 It normally makes code faster, but at the expense of making it bigger.
12304 It is enabled by default at optimization level @option{-O3}.
12308 @subsection MMIX Options
12309 @cindex MMIX Options
12311 These options are defined for the MMIX:
12315 @itemx -mno-libfuncs
12317 @opindex mno-libfuncs
12318 Specify that intrinsic library functions are being compiled, passing all
12319 values in registers, no matter the size.
12322 @itemx -mno-epsilon
12324 @opindex mno-epsilon
12325 Generate floating-point comparison instructions that compare with respect
12326 to the @code{rE} epsilon register.
12328 @item -mabi=mmixware
12330 @opindex mabi-mmixware
12332 Generate code that passes function parameters and return values that (in
12333 the called function) are seen as registers @code{$0} and up, as opposed to
12334 the GNU ABI which uses global registers @code{$231} and up.
12336 @item -mzero-extend
12337 @itemx -mno-zero-extend
12338 @opindex mzero-extend
12339 @opindex mno-zero-extend
12340 When reading data from memory in sizes shorter than 64 bits, use (do not
12341 use) zero-extending load instructions by default, rather than
12342 sign-extending ones.
12345 @itemx -mno-knuthdiv
12347 @opindex mno-knuthdiv
12348 Make the result of a division yielding a remainder have the same sign as
12349 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12350 remainder follows the sign of the dividend. Both methods are
12351 arithmetically valid, the latter being almost exclusively used.
12353 @item -mtoplevel-symbols
12354 @itemx -mno-toplevel-symbols
12355 @opindex mtoplevel-symbols
12356 @opindex mno-toplevel-symbols
12357 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12358 code can be used with the @code{PREFIX} assembly directive.
12362 Generate an executable in the ELF format, rather than the default
12363 @samp{mmo} format used by the @command{mmix} simulator.
12365 @item -mbranch-predict
12366 @itemx -mno-branch-predict
12367 @opindex mbranch-predict
12368 @opindex mno-branch-predict
12369 Use (do not use) the probable-branch instructions, when static branch
12370 prediction indicates a probable branch.
12372 @item -mbase-addresses
12373 @itemx -mno-base-addresses
12374 @opindex mbase-addresses
12375 @opindex mno-base-addresses
12376 Generate (do not generate) code that uses @emph{base addresses}. Using a
12377 base address automatically generates a request (handled by the assembler
12378 and the linker) for a constant to be set up in a global register. The
12379 register is used for one or more base address requests within the range 0
12380 to 255 from the value held in the register. The generally leads to short
12381 and fast code, but the number of different data items that can be
12382 addressed is limited. This means that a program that uses lots of static
12383 data may require @option{-mno-base-addresses}.
12385 @item -msingle-exit
12386 @itemx -mno-single-exit
12387 @opindex msingle-exit
12388 @opindex mno-single-exit
12389 Force (do not force) generated code to have a single exit point in each
12393 @node MN10300 Options
12394 @subsection MN10300 Options
12395 @cindex MN10300 options
12397 These @option{-m} options are defined for Matsushita MN10300 architectures:
12402 Generate code to avoid bugs in the multiply instructions for the MN10300
12403 processors. This is the default.
12405 @item -mno-mult-bug
12406 @opindex mno-mult-bug
12407 Do not generate code to avoid bugs in the multiply instructions for the
12408 MN10300 processors.
12412 Generate code which uses features specific to the AM33 processor.
12416 Do not generate code which uses features specific to the AM33 processor. This
12419 @item -mreturn-pointer-on-d0
12420 @opindex mreturn-pointer-on-d0
12421 When generating a function which returns a pointer, return the pointer
12422 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12423 only in a0, and attempts to call such functions without a prototype
12424 would result in errors. Note that this option is on by default; use
12425 @option{-mno-return-pointer-on-d0} to disable it.
12429 Do not link in the C run-time initialization object file.
12433 Indicate to the linker that it should perform a relaxation optimization pass
12434 to shorten branches, calls and absolute memory addresses. This option only
12435 has an effect when used on the command line for the final link step.
12437 This option makes symbolic debugging impossible.
12441 @subsection MT Options
12444 These @option{-m} options are defined for Morpho MT architectures:
12448 @item -march=@var{cpu-type}
12450 Generate code that will run on @var{cpu-type}, which is the name of a system
12451 representing a certain processor type. Possible values for
12452 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12453 @samp{ms1-16-003} and @samp{ms2}.
12455 When this option is not used, the default is @option{-march=ms1-16-002}.
12459 Use byte loads and stores when generating code.
12463 Do not use byte loads and stores when generating code.
12467 Use simulator runtime
12471 Do not link in the C run-time initialization object file
12472 @file{crti.o}. Other run-time initialization and termination files
12473 such as @file{startup.o} and @file{exit.o} are still included on the
12474 linker command line.
12478 @node PDP-11 Options
12479 @subsection PDP-11 Options
12480 @cindex PDP-11 Options
12482 These options are defined for the PDP-11:
12487 Use hardware FPP floating point. This is the default. (FIS floating
12488 point on the PDP-11/40 is not supported.)
12491 @opindex msoft-float
12492 Do not use hardware floating point.
12496 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12500 Return floating-point results in memory. This is the default.
12504 Generate code for a PDP-11/40.
12508 Generate code for a PDP-11/45. This is the default.
12512 Generate code for a PDP-11/10.
12514 @item -mbcopy-builtin
12515 @opindex bcopy-builtin
12516 Use inline @code{movmemhi} patterns for copying memory. This is the
12521 Do not use inline @code{movmemhi} patterns for copying memory.
12527 Use 16-bit @code{int}. This is the default.
12533 Use 32-bit @code{int}.
12536 @itemx -mno-float32
12538 @opindex mno-float32
12539 Use 64-bit @code{float}. This is the default.
12542 @itemx -mno-float64
12544 @opindex mno-float64
12545 Use 32-bit @code{float}.
12549 Use @code{abshi2} pattern. This is the default.
12553 Do not use @code{abshi2} pattern.
12555 @item -mbranch-expensive
12556 @opindex mbranch-expensive
12557 Pretend that branches are expensive. This is for experimenting with
12558 code generation only.
12560 @item -mbranch-cheap
12561 @opindex mbranch-cheap
12562 Do not pretend that branches are expensive. This is the default.
12566 Generate code for a system with split I&D@.
12570 Generate code for a system without split I&D@. This is the default.
12574 Use Unix assembler syntax. This is the default when configured for
12575 @samp{pdp11-*-bsd}.
12579 Use DEC assembler syntax. This is the default when configured for any
12580 PDP-11 target other than @samp{pdp11-*-bsd}.
12583 @node PowerPC Options
12584 @subsection PowerPC Options
12585 @cindex PowerPC options
12587 These are listed under @xref{RS/6000 and PowerPC Options}.
12589 @node RS/6000 and PowerPC Options
12590 @subsection IBM RS/6000 and PowerPC Options
12591 @cindex RS/6000 and PowerPC Options
12592 @cindex IBM RS/6000 and PowerPC Options
12594 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12601 @itemx -mno-powerpc
12602 @itemx -mpowerpc-gpopt
12603 @itemx -mno-powerpc-gpopt
12604 @itemx -mpowerpc-gfxopt
12605 @itemx -mno-powerpc-gfxopt
12607 @itemx -mno-powerpc64
12611 @itemx -mno-popcntb
12619 @itemx -mno-hard-dfp
12623 @opindex mno-power2
12625 @opindex mno-powerpc
12626 @opindex mpowerpc-gpopt
12627 @opindex mno-powerpc-gpopt
12628 @opindex mpowerpc-gfxopt
12629 @opindex mno-powerpc-gfxopt
12630 @opindex mpowerpc64
12631 @opindex mno-powerpc64
12635 @opindex mno-popcntb
12641 @opindex mno-mfpgpr
12643 @opindex mno-hard-dfp
12644 GCC supports two related instruction set architectures for the
12645 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12646 instructions supported by the @samp{rios} chip set used in the original
12647 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12648 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12649 the IBM 4xx, 6xx, and follow-on microprocessors.
12651 Neither architecture is a subset of the other. However there is a
12652 large common subset of instructions supported by both. An MQ
12653 register is included in processors supporting the POWER architecture.
12655 You use these options to specify which instructions are available on the
12656 processor you are using. The default value of these options is
12657 determined when configuring GCC@. Specifying the
12658 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12659 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12660 rather than the options listed above.
12662 The @option{-mpower} option allows GCC to generate instructions that
12663 are found only in the POWER architecture and to use the MQ register.
12664 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12665 to generate instructions that are present in the POWER2 architecture but
12666 not the original POWER architecture.
12668 The @option{-mpowerpc} option allows GCC to generate instructions that
12669 are found only in the 32-bit subset of the PowerPC architecture.
12670 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12671 GCC to use the optional PowerPC architecture instructions in the
12672 General Purpose group, including floating-point square root. Specifying
12673 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12674 use the optional PowerPC architecture instructions in the Graphics
12675 group, including floating-point select.
12677 The @option{-mmfcrf} option allows GCC to generate the move from
12678 condition register field instruction implemented on the POWER4
12679 processor and other processors that support the PowerPC V2.01
12681 The @option{-mpopcntb} option allows GCC to generate the popcount and
12682 double precision FP reciprocal estimate instruction implemented on the
12683 POWER5 processor and other processors that support the PowerPC V2.02
12685 The @option{-mfprnd} option allows GCC to generate the FP round to
12686 integer instructions implemented on the POWER5+ processor and other
12687 processors that support the PowerPC V2.03 architecture.
12688 The @option{-mcmpb} option allows GCC to generate the compare bytes
12689 instruction implemented on the POWER6 processor and other processors
12690 that support the PowerPC V2.05 architecture.
12691 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12692 general purpose register instructions implemented on the POWER6X
12693 processor and other processors that support the extended PowerPC V2.05
12695 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12696 point instructions implemented on some POWER processors.
12698 The @option{-mpowerpc64} option allows GCC to generate the additional
12699 64-bit instructions that are found in the full PowerPC64 architecture
12700 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12701 @option{-mno-powerpc64}.
12703 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12704 will use only the instructions in the common subset of both
12705 architectures plus some special AIX common-mode calls, and will not use
12706 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12707 permits GCC to use any instruction from either architecture and to
12708 allow use of the MQ register; specify this for the Motorola MPC601.
12710 @item -mnew-mnemonics
12711 @itemx -mold-mnemonics
12712 @opindex mnew-mnemonics
12713 @opindex mold-mnemonics
12714 Select which mnemonics to use in the generated assembler code. With
12715 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12716 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12717 assembler mnemonics defined for the POWER architecture. Instructions
12718 defined in only one architecture have only one mnemonic; GCC uses that
12719 mnemonic irrespective of which of these options is specified.
12721 GCC defaults to the mnemonics appropriate for the architecture in
12722 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12723 value of these option. Unless you are building a cross-compiler, you
12724 should normally not specify either @option{-mnew-mnemonics} or
12725 @option{-mold-mnemonics}, but should instead accept the default.
12727 @item -mcpu=@var{cpu_type}
12729 Set architecture type, register usage, choice of mnemonics, and
12730 instruction scheduling parameters for machine type @var{cpu_type}.
12731 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12732 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12733 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12734 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12735 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12736 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12737 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12738 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12739 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12740 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12742 @option{-mcpu=common} selects a completely generic processor. Code
12743 generated under this option will run on any POWER or PowerPC processor.
12744 GCC will use only the instructions in the common subset of both
12745 architectures, and will not use the MQ register. GCC assumes a generic
12746 processor model for scheduling purposes.
12748 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12749 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12750 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12751 types, with an appropriate, generic processor model assumed for
12752 scheduling purposes.
12754 The other options specify a specific processor. Code generated under
12755 those options will run best on that processor, and may not run at all on
12758 The @option{-mcpu} options automatically enable or disable the
12761 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12762 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12763 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12765 The particular options set for any particular CPU will vary between
12766 compiler versions, depending on what setting seems to produce optimal
12767 code for that CPU; it doesn't necessarily reflect the actual hardware's
12768 capabilities. If you wish to set an individual option to a particular
12769 value, you may specify it after the @option{-mcpu} option, like
12770 @samp{-mcpu=970 -mno-altivec}.
12772 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12773 not enabled or disabled by the @option{-mcpu} option at present because
12774 AIX does not have full support for these options. You may still
12775 enable or disable them individually if you're sure it'll work in your
12778 @item -mtune=@var{cpu_type}
12780 Set the instruction scheduling parameters for machine type
12781 @var{cpu_type}, but do not set the architecture type, register usage, or
12782 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12783 values for @var{cpu_type} are used for @option{-mtune} as for
12784 @option{-mcpu}. If both are specified, the code generated will use the
12785 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12786 scheduling parameters set by @option{-mtune}.
12792 Generate code to compute division as reciprocal estimate and iterative
12793 refinement, creating opportunities for increased throughput. This
12794 feature requires: optional PowerPC Graphics instruction set for single
12795 precision and FRE instruction for double precision, assuming divides
12796 cannot generate user-visible traps, and the domain values not include
12797 Infinities, denormals or zero denominator.
12800 @itemx -mno-altivec
12802 @opindex mno-altivec
12803 Generate code that uses (does not use) AltiVec instructions, and also
12804 enable the use of built-in functions that allow more direct access to
12805 the AltiVec instruction set. You may also need to set
12806 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12812 @opindex mno-vrsave
12813 Generate VRSAVE instructions when generating AltiVec code.
12816 @opindex msecure-plt
12817 Generate code that allows ld and ld.so to build executables and shared
12818 libraries with non-exec .plt and .got sections. This is a PowerPC
12819 32-bit SYSV ABI option.
12823 Generate code that uses a BSS .plt section that ld.so fills in, and
12824 requires .plt and .got sections that are both writable and executable.
12825 This is a PowerPC 32-bit SYSV ABI option.
12831 This switch enables or disables the generation of ISEL instructions.
12833 @item -misel=@var{yes/no}
12834 This switch has been deprecated. Use @option{-misel} and
12835 @option{-mno-isel} instead.
12841 This switch enables or disables the generation of SPE simd
12847 @opindex mno-paired
12848 This switch enables or disables the generation of PAIRED simd
12851 @item -mspe=@var{yes/no}
12852 This option has been deprecated. Use @option{-mspe} and
12853 @option{-mno-spe} instead.
12855 @item -mfloat-gprs=@var{yes/single/double/no}
12856 @itemx -mfloat-gprs
12857 @opindex mfloat-gprs
12858 This switch enables or disables the generation of floating point
12859 operations on the general purpose registers for architectures that
12862 The argument @var{yes} or @var{single} enables the use of
12863 single-precision floating point operations.
12865 The argument @var{double} enables the use of single and
12866 double-precision floating point operations.
12868 The argument @var{no} disables floating point operations on the
12869 general purpose registers.
12871 This option is currently only available on the MPC854x.
12877 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12878 targets (including GNU/Linux). The 32-bit environment sets int, long
12879 and pointer to 32 bits and generates code that runs on any PowerPC
12880 variant. The 64-bit environment sets int to 32 bits and long and
12881 pointer to 64 bits, and generates code for PowerPC64, as for
12882 @option{-mpowerpc64}.
12885 @itemx -mno-fp-in-toc
12886 @itemx -mno-sum-in-toc
12887 @itemx -mminimal-toc
12889 @opindex mno-fp-in-toc
12890 @opindex mno-sum-in-toc
12891 @opindex mminimal-toc
12892 Modify generation of the TOC (Table Of Contents), which is created for
12893 every executable file. The @option{-mfull-toc} option is selected by
12894 default. In that case, GCC will allocate at least one TOC entry for
12895 each unique non-automatic variable reference in your program. GCC
12896 will also place floating-point constants in the TOC@. However, only
12897 16,384 entries are available in the TOC@.
12899 If you receive a linker error message that saying you have overflowed
12900 the available TOC space, you can reduce the amount of TOC space used
12901 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12902 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12903 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12904 generate code to calculate the sum of an address and a constant at
12905 run-time instead of putting that sum into the TOC@. You may specify one
12906 or both of these options. Each causes GCC to produce very slightly
12907 slower and larger code at the expense of conserving TOC space.
12909 If you still run out of space in the TOC even when you specify both of
12910 these options, specify @option{-mminimal-toc} instead. This option causes
12911 GCC to make only one TOC entry for every file. When you specify this
12912 option, GCC will produce code that is slower and larger but which
12913 uses extremely little TOC space. You may wish to use this option
12914 only on files that contain less frequently executed code.
12920 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12921 @code{long} type, and the infrastructure needed to support them.
12922 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12923 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12924 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12927 @itemx -mno-xl-compat
12928 @opindex mxl-compat
12929 @opindex mno-xl-compat
12930 Produce code that conforms more closely to IBM XL compiler semantics
12931 when using AIX-compatible ABI. Pass floating-point arguments to
12932 prototyped functions beyond the register save area (RSA) on the stack
12933 in addition to argument FPRs. Do not assume that most significant
12934 double in 128-bit long double value is properly rounded when comparing
12935 values and converting to double. Use XL symbol names for long double
12938 The AIX calling convention was extended but not initially documented to
12939 handle an obscure K&R C case of calling a function that takes the
12940 address of its arguments with fewer arguments than declared. IBM XL
12941 compilers access floating point arguments which do not fit in the
12942 RSA from the stack when a subroutine is compiled without
12943 optimization. Because always storing floating-point arguments on the
12944 stack is inefficient and rarely needed, this option is not enabled by
12945 default and only is necessary when calling subroutines compiled by IBM
12946 XL compilers without optimization.
12950 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12951 application written to use message passing with special startup code to
12952 enable the application to run. The system must have PE installed in the
12953 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12954 must be overridden with the @option{-specs=} option to specify the
12955 appropriate directory location. The Parallel Environment does not
12956 support threads, so the @option{-mpe} option and the @option{-pthread}
12957 option are incompatible.
12959 @item -malign-natural
12960 @itemx -malign-power
12961 @opindex malign-natural
12962 @opindex malign-power
12963 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12964 @option{-malign-natural} overrides the ABI-defined alignment of larger
12965 types, such as floating-point doubles, on their natural size-based boundary.
12966 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12967 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12969 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12973 @itemx -mhard-float
12974 @opindex msoft-float
12975 @opindex mhard-float
12976 Generate code that does not use (uses) the floating-point register set.
12977 Software floating point emulation is provided if you use the
12978 @option{-msoft-float} option, and pass the option to GCC when linking.
12981 @itemx -mno-multiple
12983 @opindex mno-multiple
12984 Generate code that uses (does not use) the load multiple word
12985 instructions and the store multiple word instructions. These
12986 instructions are generated by default on POWER systems, and not
12987 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12988 endian PowerPC systems, since those instructions do not work when the
12989 processor is in little endian mode. The exceptions are PPC740 and
12990 PPC750 which permit the instructions usage in little endian mode.
12995 @opindex mno-string
12996 Generate code that uses (does not use) the load string instructions
12997 and the store string word instructions to save multiple registers and
12998 do small block moves. These instructions are generated by default on
12999 POWER systems, and not generated on PowerPC systems. Do not use
13000 @option{-mstring} on little endian PowerPC systems, since those
13001 instructions do not work when the processor is in little endian mode.
13002 The exceptions are PPC740 and PPC750 which permit the instructions
13003 usage in little endian mode.
13008 @opindex mno-update
13009 Generate code that uses (does not use) the load or store instructions
13010 that update the base register to the address of the calculated memory
13011 location. These instructions are generated by default. If you use
13012 @option{-mno-update}, there is a small window between the time that the
13013 stack pointer is updated and the address of the previous frame is
13014 stored, which means code that walks the stack frame across interrupts or
13015 signals may get corrupted data.
13018 @itemx -mno-fused-madd
13019 @opindex mfused-madd
13020 @opindex mno-fused-madd
13021 Generate code that uses (does not use) the floating point multiply and
13022 accumulate instructions. These instructions are generated by default if
13023 hardware floating is used.
13029 Generate code that uses (does not use) the half-word multiply and
13030 multiply-accumulate instructions on the IBM 405 and 440 processors.
13031 These instructions are generated by default when targetting those
13038 Generate code that uses (does not use) the string-search @samp{dlmzb}
13039 instruction on the IBM 405 and 440 processors. This instruction is
13040 generated by default when targetting those processors.
13042 @item -mno-bit-align
13044 @opindex mno-bit-align
13045 @opindex mbit-align
13046 On System V.4 and embedded PowerPC systems do not (do) force structures
13047 and unions that contain bit-fields to be aligned to the base type of the
13050 For example, by default a structure containing nothing but 8
13051 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13052 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13053 the structure would be aligned to a 1 byte boundary and be one byte in
13056 @item -mno-strict-align
13057 @itemx -mstrict-align
13058 @opindex mno-strict-align
13059 @opindex mstrict-align
13060 On System V.4 and embedded PowerPC systems do not (do) assume that
13061 unaligned memory references will be handled by the system.
13063 @item -mrelocatable
13064 @itemx -mno-relocatable
13065 @opindex mrelocatable
13066 @opindex mno-relocatable
13067 On embedded PowerPC systems generate code that allows (does not allow)
13068 the program to be relocated to a different address at runtime. If you
13069 use @option{-mrelocatable} on any module, all objects linked together must
13070 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13072 @item -mrelocatable-lib
13073 @itemx -mno-relocatable-lib
13074 @opindex mrelocatable-lib
13075 @opindex mno-relocatable-lib
13076 On embedded PowerPC systems generate code that allows (does not allow)
13077 the program to be relocated to a different address at runtime. Modules
13078 compiled with @option{-mrelocatable-lib} can be linked with either modules
13079 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13080 with modules compiled with the @option{-mrelocatable} options.
13086 On System V.4 and embedded PowerPC systems do not (do) assume that
13087 register 2 contains a pointer to a global area pointing to the addresses
13088 used in the program.
13091 @itemx -mlittle-endian
13093 @opindex mlittle-endian
13094 On System V.4 and embedded PowerPC systems compile code for the
13095 processor in little endian mode. The @option{-mlittle-endian} option is
13096 the same as @option{-mlittle}.
13099 @itemx -mbig-endian
13101 @opindex mbig-endian
13102 On System V.4 and embedded PowerPC systems compile code for the
13103 processor in big endian mode. The @option{-mbig-endian} option is
13104 the same as @option{-mbig}.
13106 @item -mdynamic-no-pic
13107 @opindex mdynamic-no-pic
13108 On Darwin and Mac OS X systems, compile code so that it is not
13109 relocatable, but that its external references are relocatable. The
13110 resulting code is suitable for applications, but not shared
13113 @item -mprioritize-restricted-insns=@var{priority}
13114 @opindex mprioritize-restricted-insns
13115 This option controls the priority that is assigned to
13116 dispatch-slot restricted instructions during the second scheduling
13117 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13118 @var{no/highest/second-highest} priority to dispatch slot restricted
13121 @item -msched-costly-dep=@var{dependence_type}
13122 @opindex msched-costly-dep
13123 This option controls which dependences are considered costly
13124 by the target during instruction scheduling. The argument
13125 @var{dependence_type} takes one of the following values:
13126 @var{no}: no dependence is costly,
13127 @var{all}: all dependences are costly,
13128 @var{true_store_to_load}: a true dependence from store to load is costly,
13129 @var{store_to_load}: any dependence from store to load is costly,
13130 @var{number}: any dependence which latency >= @var{number} is costly.
13132 @item -minsert-sched-nops=@var{scheme}
13133 @opindex minsert-sched-nops
13134 This option controls which nop insertion scheme will be used during
13135 the second scheduling pass. The argument @var{scheme} takes one of the
13137 @var{no}: Don't insert nops.
13138 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13139 according to the scheduler's grouping.
13140 @var{regroup_exact}: Insert nops to force costly dependent insns into
13141 separate groups. Insert exactly as many nops as needed to force an insn
13142 to a new group, according to the estimated processor grouping.
13143 @var{number}: Insert nops to force costly dependent insns into
13144 separate groups. Insert @var{number} nops to force an insn to a new group.
13147 @opindex mcall-sysv
13148 On System V.4 and embedded PowerPC systems compile code using calling
13149 conventions that adheres to the March 1995 draft of the System V
13150 Application Binary Interface, PowerPC processor supplement. This is the
13151 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13153 @item -mcall-sysv-eabi
13154 @opindex mcall-sysv-eabi
13155 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13157 @item -mcall-sysv-noeabi
13158 @opindex mcall-sysv-noeabi
13159 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13161 @item -mcall-solaris
13162 @opindex mcall-solaris
13163 On System V.4 and embedded PowerPC systems compile code for the Solaris
13167 @opindex mcall-linux
13168 On System V.4 and embedded PowerPC systems compile code for the
13169 Linux-based GNU system.
13173 On System V.4 and embedded PowerPC systems compile code for the
13174 Hurd-based GNU system.
13176 @item -mcall-netbsd
13177 @opindex mcall-netbsd
13178 On System V.4 and embedded PowerPC systems compile code for the
13179 NetBSD operating system.
13181 @item -maix-struct-return
13182 @opindex maix-struct-return
13183 Return all structures in memory (as specified by the AIX ABI)@.
13185 @item -msvr4-struct-return
13186 @opindex msvr4-struct-return
13187 Return structures smaller than 8 bytes in registers (as specified by the
13190 @item -mabi=@var{abi-type}
13192 Extend the current ABI with a particular extension, or remove such extension.
13193 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13194 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13198 Extend the current ABI with SPE ABI extensions. This does not change
13199 the default ABI, instead it adds the SPE ABI extensions to the current
13203 @opindex mabi=no-spe
13204 Disable Booke SPE ABI extensions for the current ABI@.
13206 @item -mabi=ibmlongdouble
13207 @opindex mabi=ibmlongdouble
13208 Change the current ABI to use IBM extended precision long double.
13209 This is a PowerPC 32-bit SYSV ABI option.
13211 @item -mabi=ieeelongdouble
13212 @opindex mabi=ieeelongdouble
13213 Change the current ABI to use IEEE extended precision long double.
13214 This is a PowerPC 32-bit Linux ABI option.
13217 @itemx -mno-prototype
13218 @opindex mprototype
13219 @opindex mno-prototype
13220 On System V.4 and embedded PowerPC systems assume that all calls to
13221 variable argument functions are properly prototyped. Otherwise, the
13222 compiler must insert an instruction before every non prototyped call to
13223 set or clear bit 6 of the condition code register (@var{CR}) to
13224 indicate whether floating point values were passed in the floating point
13225 registers in case the function takes a variable arguments. With
13226 @option{-mprototype}, only calls to prototyped variable argument functions
13227 will set or clear the bit.
13231 On embedded PowerPC systems, assume that the startup module is called
13232 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13233 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
13238 On embedded PowerPC systems, assume that the startup module is called
13239 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13244 On embedded PowerPC systems, assume that the startup module is called
13245 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13248 @item -myellowknife
13249 @opindex myellowknife
13250 On embedded PowerPC systems, assume that the startup module is called
13251 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13256 On System V.4 and embedded PowerPC systems, specify that you are
13257 compiling for a VxWorks system.
13261 Specify that you are compiling for the WindISS simulation environment.
13265 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13266 header to indicate that @samp{eabi} extended relocations are used.
13272 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13273 Embedded Applications Binary Interface (eabi) which is a set of
13274 modifications to the System V.4 specifications. Selecting @option{-meabi}
13275 means that the stack is aligned to an 8 byte boundary, a function
13276 @code{__eabi} is called to from @code{main} to set up the eabi
13277 environment, and the @option{-msdata} option can use both @code{r2} and
13278 @code{r13} to point to two separate small data areas. Selecting
13279 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13280 do not call an initialization function from @code{main}, and the
13281 @option{-msdata} option will only use @code{r13} to point to a single
13282 small data area. The @option{-meabi} option is on by default if you
13283 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13286 @opindex msdata=eabi
13287 On System V.4 and embedded PowerPC systems, put small initialized
13288 @code{const} global and static data in the @samp{.sdata2} section, which
13289 is pointed to by register @code{r2}. Put small initialized
13290 non-@code{const} global and static data in the @samp{.sdata} section,
13291 which is pointed to by register @code{r13}. Put small uninitialized
13292 global and static data in the @samp{.sbss} section, which is adjacent to
13293 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13294 incompatible with the @option{-mrelocatable} option. The
13295 @option{-msdata=eabi} option also sets the @option{-memb} option.
13298 @opindex msdata=sysv
13299 On System V.4 and embedded PowerPC systems, put small global and static
13300 data in the @samp{.sdata} section, which is pointed to by register
13301 @code{r13}. Put small uninitialized global and static data in the
13302 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13303 The @option{-msdata=sysv} option is incompatible with the
13304 @option{-mrelocatable} option.
13306 @item -msdata=default
13308 @opindex msdata=default
13310 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13311 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13312 same as @option{-msdata=sysv}.
13315 @opindex msdata-data
13316 On System V.4 and embedded PowerPC systems, put small global
13317 data in the @samp{.sdata} section. Put small uninitialized global
13318 data in the @samp{.sbss} section. Do not use register @code{r13}
13319 to address small data however. This is the default behavior unless
13320 other @option{-msdata} options are used.
13324 @opindex msdata=none
13326 On embedded PowerPC systems, put all initialized global and static data
13327 in the @samp{.data} section, and all uninitialized data in the
13328 @samp{.bss} section.
13332 @cindex smaller data references (PowerPC)
13333 @cindex .sdata/.sdata2 references (PowerPC)
13334 On embedded PowerPC systems, put global and static items less than or
13335 equal to @var{num} bytes into the small data or bss sections instead of
13336 the normal data or bss section. By default, @var{num} is 8. The
13337 @option{-G @var{num}} switch is also passed to the linker.
13338 All modules should be compiled with the same @option{-G @var{num}} value.
13341 @itemx -mno-regnames
13343 @opindex mno-regnames
13344 On System V.4 and embedded PowerPC systems do (do not) emit register
13345 names in the assembly language output using symbolic forms.
13348 @itemx -mno-longcall
13350 @opindex mno-longcall
13351 By default assume that all calls are far away so that a longer more
13352 expensive calling sequence is required. This is required for calls
13353 further than 32 megabytes (33,554,432 bytes) from the current location.
13354 A short call will be generated if the compiler knows
13355 the call cannot be that far away. This setting can be overridden by
13356 the @code{shortcall} function attribute, or by @code{#pragma
13359 Some linkers are capable of detecting out-of-range calls and generating
13360 glue code on the fly. On these systems, long calls are unnecessary and
13361 generate slower code. As of this writing, the AIX linker can do this,
13362 as can the GNU linker for PowerPC/64. It is planned to add this feature
13363 to the GNU linker for 32-bit PowerPC systems as well.
13365 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13366 callee, L42'', plus a ``branch island'' (glue code). The two target
13367 addresses represent the callee and the ``branch island''. The
13368 Darwin/PPC linker will prefer the first address and generate a ``bl
13369 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13370 otherwise, the linker will generate ``bl L42'' to call the ``branch
13371 island''. The ``branch island'' is appended to the body of the
13372 calling function; it computes the full 32-bit address of the callee
13375 On Mach-O (Darwin) systems, this option directs the compiler emit to
13376 the glue for every direct call, and the Darwin linker decides whether
13377 to use or discard it.
13379 In the future, we may cause GCC to ignore all longcall specifications
13380 when the linker is known to generate glue.
13384 Adds support for multithreading with the @dfn{pthreads} library.
13385 This option sets flags for both the preprocessor and linker.
13389 @node S/390 and zSeries Options
13390 @subsection S/390 and zSeries Options
13391 @cindex S/390 and zSeries Options
13393 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13397 @itemx -msoft-float
13398 @opindex mhard-float
13399 @opindex msoft-float
13400 Use (do not use) the hardware floating-point instructions and registers
13401 for floating-point operations. When @option{-msoft-float} is specified,
13402 functions in @file{libgcc.a} will be used to perform floating-point
13403 operations. When @option{-mhard-float} is specified, the compiler
13404 generates IEEE floating-point instructions. This is the default.
13406 @item -mlong-double-64
13407 @itemx -mlong-double-128
13408 @opindex mlong-double-64
13409 @opindex mlong-double-128
13410 These switches control the size of @code{long double} type. A size
13411 of 64bit makes the @code{long double} type equivalent to the @code{double}
13412 type. This is the default.
13415 @itemx -mno-backchain
13416 @opindex mbackchain
13417 @opindex mno-backchain
13418 Store (do not store) the address of the caller's frame as backchain pointer
13419 into the callee's stack frame.
13420 A backchain may be needed to allow debugging using tools that do not understand
13421 DWARF-2 call frame information.
13422 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13423 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13424 the backchain is placed into the topmost word of the 96/160 byte register
13427 In general, code compiled with @option{-mbackchain} is call-compatible with
13428 code compiled with @option{-mmo-backchain}; however, use of the backchain
13429 for debugging purposes usually requires that the whole binary is built with
13430 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13431 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13432 to build a linux kernel use @option{-msoft-float}.
13434 The default is to not maintain the backchain.
13436 @item -mpacked-stack
13437 @item -mno-packed-stack
13438 @opindex mpacked-stack
13439 @opindex mno-packed-stack
13440 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13441 specified, the compiler uses the all fields of the 96/160 byte register save
13442 area only for their default purpose; unused fields still take up stack space.
13443 When @option{-mpacked-stack} is specified, register save slots are densely
13444 packed at the top of the register save area; unused space is reused for other
13445 purposes, allowing for more efficient use of the available stack space.
13446 However, when @option{-mbackchain} is also in effect, the topmost word of
13447 the save area is always used to store the backchain, and the return address
13448 register is always saved two words below the backchain.
13450 As long as the stack frame backchain is not used, code generated with
13451 @option{-mpacked-stack} is call-compatible with code generated with
13452 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13453 S/390 or zSeries generated code that uses the stack frame backchain at run
13454 time, not just for debugging purposes. Such code is not call-compatible
13455 with code compiled with @option{-mpacked-stack}. Also, note that the
13456 combination of @option{-mbackchain},
13457 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13458 to build a linux kernel use @option{-msoft-float}.
13460 The default is to not use the packed stack layout.
13463 @itemx -mno-small-exec
13464 @opindex msmall-exec
13465 @opindex mno-small-exec
13466 Generate (or do not generate) code using the @code{bras} instruction
13467 to do subroutine calls.
13468 This only works reliably if the total executable size does not
13469 exceed 64k. The default is to use the @code{basr} instruction instead,
13470 which does not have this limitation.
13476 When @option{-m31} is specified, generate code compliant to the
13477 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13478 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13479 particular to generate 64-bit instructions. For the @samp{s390}
13480 targets, the default is @option{-m31}, while the @samp{s390x}
13481 targets default to @option{-m64}.
13487 When @option{-mzarch} is specified, generate code using the
13488 instructions available on z/Architecture.
13489 When @option{-mesa} is specified, generate code using the
13490 instructions available on ESA/390. Note that @option{-mesa} is
13491 not possible with @option{-m64}.
13492 When generating code compliant to the GNU/Linux for S/390 ABI,
13493 the default is @option{-mesa}. When generating code compliant
13494 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13500 Generate (or do not generate) code using the @code{mvcle} instruction
13501 to perform block moves. When @option{-mno-mvcle} is specified,
13502 use a @code{mvc} loop instead. This is the default unless optimizing for
13509 Print (or do not print) additional debug information when compiling.
13510 The default is to not print debug information.
13512 @item -march=@var{cpu-type}
13514 Generate code that will run on @var{cpu-type}, which is the name of a system
13515 representing a certain processor type. Possible values for
13516 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13517 When generating code using the instructions available on z/Architecture,
13518 the default is @option{-march=z900}. Otherwise, the default is
13519 @option{-march=g5}.
13521 @item -mtune=@var{cpu-type}
13523 Tune to @var{cpu-type} everything applicable about the generated code,
13524 except for the ABI and the set of available instructions.
13525 The list of @var{cpu-type} values is the same as for @option{-march}.
13526 The default is the value used for @option{-march}.
13529 @itemx -mno-tpf-trace
13530 @opindex mtpf-trace
13531 @opindex mno-tpf-trace
13532 Generate code that adds (does not add) in TPF OS specific branches to trace
13533 routines in the operating system. This option is off by default, even
13534 when compiling for the TPF OS@.
13537 @itemx -mno-fused-madd
13538 @opindex mfused-madd
13539 @opindex mno-fused-madd
13540 Generate code that uses (does not use) the floating point multiply and
13541 accumulate instructions. These instructions are generated by default if
13542 hardware floating point is used.
13544 @item -mwarn-framesize=@var{framesize}
13545 @opindex mwarn-framesize
13546 Emit a warning if the current function exceeds the given frame size. Because
13547 this is a compile time check it doesn't need to be a real problem when the program
13548 runs. It is intended to identify functions which most probably cause
13549 a stack overflow. It is useful to be used in an environment with limited stack
13550 size e.g.@: the linux kernel.
13552 @item -mwarn-dynamicstack
13553 @opindex mwarn-dynamicstack
13554 Emit a warning if the function calls alloca or uses dynamically
13555 sized arrays. This is generally a bad idea with a limited stack size.
13557 @item -mstack-guard=@var{stack-guard}
13558 @item -mstack-size=@var{stack-size}
13559 @opindex mstack-guard
13560 @opindex mstack-size
13561 If these options are provided the s390 back end emits additional instructions in
13562 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13563 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13564 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13565 the frame size of the compiled function is chosen.
13566 These options are intended to be used to help debugging stack overflow problems.
13567 The additionally emitted code causes only little overhead and hence can also be
13568 used in production like systems without greater performance degradation. The given
13569 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13570 @var{stack-guard} without exceeding 64k.
13571 In order to be efficient the extra code makes the assumption that the stack starts
13572 at an address aligned to the value given by @var{stack-size}.
13573 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13576 @node Score Options
13577 @subsection Score Options
13578 @cindex Score Options
13580 These options are defined for Score implementations:
13585 Compile code for big endian mode. This is the default.
13589 Compile code for little endian mode.
13593 Disable generate bcnz instruction.
13597 Enable generate unaligned load and store instruction.
13601 Enable the use of multiply-accumulate instructions. Disabled by default.
13605 Specify the SCORE5 as the target architecture.
13609 Specify the SCORE5U of the target architecture.
13613 Specify the SCORE7 as the target architecture. This is the default.
13617 Specify the SCORE7D as the target architecture.
13621 @subsection SH Options
13623 These @samp{-m} options are defined for the SH implementations:
13628 Generate code for the SH1.
13632 Generate code for the SH2.
13635 Generate code for the SH2e.
13639 Generate code for the SH3.
13643 Generate code for the SH3e.
13647 Generate code for the SH4 without a floating-point unit.
13649 @item -m4-single-only
13650 @opindex m4-single-only
13651 Generate code for the SH4 with a floating-point unit that only
13652 supports single-precision arithmetic.
13656 Generate code for the SH4 assuming the floating-point unit is in
13657 single-precision mode by default.
13661 Generate code for the SH4.
13665 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13666 floating-point unit is not used.
13668 @item -m4a-single-only
13669 @opindex m4a-single-only
13670 Generate code for the SH4a, in such a way that no double-precision
13671 floating point operations are used.
13674 @opindex m4a-single
13675 Generate code for the SH4a assuming the floating-point unit is in
13676 single-precision mode by default.
13680 Generate code for the SH4a.
13684 Same as @option{-m4a-nofpu}, except that it implicitly passes
13685 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13686 instructions at the moment.
13690 Compile code for the processor in big endian mode.
13694 Compile code for the processor in little endian mode.
13698 Align doubles at 64-bit boundaries. Note that this changes the calling
13699 conventions, and thus some functions from the standard C library will
13700 not work unless you recompile it first with @option{-mdalign}.
13704 Shorten some address references at link time, when possible; uses the
13705 linker option @option{-relax}.
13709 Use 32-bit offsets in @code{switch} tables. The default is to use
13714 Enable the use of the instruction @code{fmovd}.
13718 Comply with the calling conventions defined by Renesas.
13722 Comply with the calling conventions defined by Renesas.
13726 Comply with the calling conventions defined for GCC before the Renesas
13727 conventions were available. This option is the default for all
13728 targets of the SH toolchain except for @samp{sh-symbianelf}.
13731 @opindex mnomacsave
13732 Mark the @code{MAC} register as call-clobbered, even if
13733 @option{-mhitachi} is given.
13737 Increase IEEE-compliance of floating-point code.
13738 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13739 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13740 comparisons of NANs / infinities incurs extra overhead in every
13741 floating point comparison, therefore the default is set to
13742 @option{-ffinite-math-only}.
13744 @item -minline-ic_invalidate
13745 @opindex minline-ic_invalidate
13746 Inline code to invalidate instruction cache entries after setting up
13747 nested function trampolines.
13748 This option has no effect if -musermode is in effect and the selected
13749 code generation option (e.g. -m4) does not allow the use of the icbi
13751 If the selected code generation option does not allow the use of the icbi
13752 instruction, and -musermode is not in effect, the inlined code will
13753 manipulate the instruction cache address array directly with an associative
13754 write. This not only requires privileged mode, but it will also
13755 fail if the cache line had been mapped via the TLB and has become unmapped.
13759 Dump instruction size and location in the assembly code.
13762 @opindex mpadstruct
13763 This option is deprecated. It pads structures to multiple of 4 bytes,
13764 which is incompatible with the SH ABI@.
13768 Optimize for space instead of speed. Implied by @option{-Os}.
13771 @opindex mprefergot
13772 When generating position-independent code, emit function calls using
13773 the Global Offset Table instead of the Procedure Linkage Table.
13777 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13778 if the inlined code would not work in user mode.
13779 This is the default when the target is @code{sh-*-linux*}.
13781 @item -multcost=@var{number}
13782 @opindex multcost=@var{number}
13783 Set the cost to assume for a multiply insn.
13785 @item -mdiv=@var{strategy}
13786 @opindex mdiv=@var{strategy}
13787 Set the division strategy to use for SHmedia code. @var{strategy} must be
13788 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13789 inv:call2, inv:fp .
13790 "fp" performs the operation in floating point. This has a very high latency,
13791 but needs only a few instructions, so it might be a good choice if
13792 your code has enough easily exploitable ILP to allow the compiler to
13793 schedule the floating point instructions together with other instructions.
13794 Division by zero causes a floating point exception.
13795 "inv" uses integer operations to calculate the inverse of the divisor,
13796 and then multiplies the dividend with the inverse. This strategy allows
13797 cse and hoisting of the inverse calculation. Division by zero calculates
13798 an unspecified result, but does not trap.
13799 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13800 have been found, or if the entire operation has been hoisted to the same
13801 place, the last stages of the inverse calculation are intertwined with the
13802 final multiply to reduce the overall latency, at the expense of using a few
13803 more instructions, and thus offering fewer scheduling opportunities with
13805 "call" calls a library function that usually implements the inv:minlat
13807 This gives high code density for m5-*media-nofpu compilations.
13808 "call2" uses a different entry point of the same library function, where it
13809 assumes that a pointer to a lookup table has already been set up, which
13810 exposes the pointer load to cse / code hoisting optimizations.
13811 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13812 code generation, but if the code stays unoptimized, revert to the "call",
13813 "call2", or "fp" strategies, respectively. Note that the
13814 potentially-trapping side effect of division by zero is carried by a
13815 separate instruction, so it is possible that all the integer instructions
13816 are hoisted out, but the marker for the side effect stays where it is.
13817 A recombination to fp operations or a call is not possible in that case.
13818 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13819 that the inverse calculation was nor separated from the multiply, they speed
13820 up division where the dividend fits into 20 bits (plus sign where applicable),
13821 by inserting a test to skip a number of operations in this case; this test
13822 slows down the case of larger dividends. inv20u assumes the case of a such
13823 a small dividend to be unlikely, and inv20l assumes it to be likely.
13825 @item -mdivsi3_libfunc=@var{name}
13826 @opindex mdivsi3_libfunc=@var{name}
13827 Set the name of the library function used for 32 bit signed division to
13828 @var{name}. This only affect the name used in the call and inv:call
13829 division strategies, and the compiler will still expect the same
13830 sets of input/output/clobbered registers as if this option was not present.
13832 @item -madjust-unroll
13833 @opindex madjust-unroll
13834 Throttle unrolling to avoid thrashing target registers.
13835 This option only has an effect if the gcc code base supports the
13836 TARGET_ADJUST_UNROLL_MAX target hook.
13838 @item -mindexed-addressing
13839 @opindex mindexed-addressing
13840 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13841 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13842 semantics for the indexed addressing mode. The architecture allows the
13843 implementation of processors with 64 bit MMU, which the OS could use to
13844 get 32 bit addressing, but since no current hardware implementation supports
13845 this or any other way to make the indexed addressing mode safe to use in
13846 the 32 bit ABI, the default is -mno-indexed-addressing.
13848 @item -mgettrcost=@var{number}
13849 @opindex mgettrcost=@var{number}
13850 Set the cost assumed for the gettr instruction to @var{number}.
13851 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13855 Assume pt* instructions won't trap. This will generally generate better
13856 scheduled code, but is unsafe on current hardware. The current architecture
13857 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13858 This has the unintentional effect of making it unsafe to schedule ptabs /
13859 ptrel before a branch, or hoist it out of a loop. For example,
13860 __do_global_ctors, a part of libgcc that runs constructors at program
13861 startup, calls functions in a list which is delimited by @minus{}1. With the
13862 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13863 That means that all the constructors will be run a bit quicker, but when
13864 the loop comes to the end of the list, the program crashes because ptabs
13865 loads @minus{}1 into a target register. Since this option is unsafe for any
13866 hardware implementing the current architecture specification, the default
13867 is -mno-pt-fixed. Unless the user specifies a specific cost with
13868 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13869 this deters register allocation using target registers for storing
13872 @item -minvalid-symbols
13873 @opindex minvalid-symbols
13874 Assume symbols might be invalid. Ordinary function symbols generated by
13875 the compiler will always be valid to load with movi/shori/ptabs or
13876 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13877 to generate symbols that will cause ptabs / ptrel to trap.
13878 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13879 It will then prevent cross-basic-block cse, hoisting and most scheduling
13880 of symbol loads. The default is @option{-mno-invalid-symbols}.
13883 @node SPARC Options
13884 @subsection SPARC Options
13885 @cindex SPARC options
13887 These @samp{-m} options are supported on the SPARC:
13890 @item -mno-app-regs
13892 @opindex mno-app-regs
13894 Specify @option{-mapp-regs} to generate output using the global registers
13895 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13898 To be fully SVR4 ABI compliant at the cost of some performance loss,
13899 specify @option{-mno-app-regs}. You should compile libraries and system
13900 software with this option.
13903 @itemx -mhard-float
13905 @opindex mhard-float
13906 Generate output containing floating point instructions. This is the
13910 @itemx -msoft-float
13912 @opindex msoft-float
13913 Generate output containing library calls for floating point.
13914 @strong{Warning:} the requisite libraries are not available for all SPARC
13915 targets. Normally the facilities of the machine's usual C compiler are
13916 used, but this cannot be done directly in cross-compilation. You must make
13917 your own arrangements to provide suitable library functions for
13918 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13919 @samp{sparclite-*-*} do provide software floating point support.
13921 @option{-msoft-float} changes the calling convention in the output file;
13922 therefore, it is only useful if you compile @emph{all} of a program with
13923 this option. In particular, you need to compile @file{libgcc.a}, the
13924 library that comes with GCC, with @option{-msoft-float} in order for
13927 @item -mhard-quad-float
13928 @opindex mhard-quad-float
13929 Generate output containing quad-word (long double) floating point
13932 @item -msoft-quad-float
13933 @opindex msoft-quad-float
13934 Generate output containing library calls for quad-word (long double)
13935 floating point instructions. The functions called are those specified
13936 in the SPARC ABI@. This is the default.
13938 As of this writing, there are no SPARC implementations that have hardware
13939 support for the quad-word floating point instructions. They all invoke
13940 a trap handler for one of these instructions, and then the trap handler
13941 emulates the effect of the instruction. Because of the trap handler overhead,
13942 this is much slower than calling the ABI library routines. Thus the
13943 @option{-msoft-quad-float} option is the default.
13945 @item -mno-unaligned-doubles
13946 @itemx -munaligned-doubles
13947 @opindex mno-unaligned-doubles
13948 @opindex munaligned-doubles
13949 Assume that doubles have 8 byte alignment. This is the default.
13951 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13952 alignment only if they are contained in another type, or if they have an
13953 absolute address. Otherwise, it assumes they have 4 byte alignment.
13954 Specifying this option avoids some rare compatibility problems with code
13955 generated by other compilers. It is not the default because it results
13956 in a performance loss, especially for floating point code.
13958 @item -mno-faster-structs
13959 @itemx -mfaster-structs
13960 @opindex mno-faster-structs
13961 @opindex mfaster-structs
13962 With @option{-mfaster-structs}, the compiler assumes that structures
13963 should have 8 byte alignment. This enables the use of pairs of
13964 @code{ldd} and @code{std} instructions for copies in structure
13965 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13966 However, the use of this changed alignment directly violates the SPARC
13967 ABI@. Thus, it's intended only for use on targets where the developer
13968 acknowledges that their resulting code will not be directly in line with
13969 the rules of the ABI@.
13971 @item -mimpure-text
13972 @opindex mimpure-text
13973 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13974 the compiler to not pass @option{-z text} to the linker when linking a
13975 shared object. Using this option, you can link position-dependent
13976 code into a shared object.
13978 @option{-mimpure-text} suppresses the ``relocations remain against
13979 allocatable but non-writable sections'' linker error message.
13980 However, the necessary relocations will trigger copy-on-write, and the
13981 shared object is not actually shared across processes. Instead of
13982 using @option{-mimpure-text}, you should compile all source code with
13983 @option{-fpic} or @option{-fPIC}.
13985 This option is only available on SunOS and Solaris.
13987 @item -mcpu=@var{cpu_type}
13989 Set the instruction set, register set, and instruction scheduling parameters
13990 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13991 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13992 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13993 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13994 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
13996 Default instruction scheduling parameters are used for values that select
13997 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13998 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14000 Here is a list of each supported architecture and their supported
14005 v8: supersparc, hypersparc
14006 sparclite: f930, f934, sparclite86x
14008 v9: ultrasparc, ultrasparc3, niagara, niagara2
14011 By default (unless configured otherwise), GCC generates code for the V7
14012 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14013 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14014 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14015 SPARCStation 1, 2, IPX etc.
14017 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14018 architecture. The only difference from V7 code is that the compiler emits
14019 the integer multiply and integer divide instructions which exist in SPARC-V8
14020 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14021 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14024 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14025 the SPARC architecture. This adds the integer multiply, integer divide step
14026 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14027 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14028 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14029 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14030 MB86934 chip, which is the more recent SPARClite with FPU@.
14032 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14033 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14034 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14035 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14036 optimizes it for the TEMIC SPARClet chip.
14038 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14039 architecture. This adds 64-bit integer and floating-point move instructions,
14040 3 additional floating-point condition code registers and conditional move
14041 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14042 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14043 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14044 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14045 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14046 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14047 additionally optimizes it for Sun UltraSPARC T2 chips.
14049 @item -mtune=@var{cpu_type}
14051 Set the instruction scheduling parameters for machine type
14052 @var{cpu_type}, but do not set the instruction set or register set that the
14053 option @option{-mcpu=@var{cpu_type}} would.
14055 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14056 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14057 that select a particular cpu implementation. Those are @samp{cypress},
14058 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14059 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14060 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14065 @opindex mno-v8plus
14066 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14067 difference from the V8 ABI is that the global and out registers are
14068 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14069 mode for all SPARC-V9 processors.
14075 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14076 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14079 These @samp{-m} options are supported in addition to the above
14080 on SPARC-V9 processors in 64-bit environments:
14083 @item -mlittle-endian
14084 @opindex mlittle-endian
14085 Generate code for a processor running in little-endian mode. It is only
14086 available for a few configurations and most notably not on Solaris and Linux.
14092 Generate code for a 32-bit or 64-bit environment.
14093 The 32-bit environment sets int, long and pointer to 32 bits.
14094 The 64-bit environment sets int to 32 bits and long and pointer
14097 @item -mcmodel=medlow
14098 @opindex mcmodel=medlow
14099 Generate code for the Medium/Low code model: 64-bit addresses, programs
14100 must be linked in the low 32 bits of memory. Programs can be statically
14101 or dynamically linked.
14103 @item -mcmodel=medmid
14104 @opindex mcmodel=medmid
14105 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14106 must be linked in the low 44 bits of memory, the text and data segments must
14107 be less than 2GB in size and the data segment must be located within 2GB of
14110 @item -mcmodel=medany
14111 @opindex mcmodel=medany
14112 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14113 may be linked anywhere in memory, the text and data segments must be less
14114 than 2GB in size and the data segment must be located within 2GB of the
14117 @item -mcmodel=embmedany
14118 @opindex mcmodel=embmedany
14119 Generate code for the Medium/Anywhere code model for embedded systems:
14120 64-bit addresses, the text and data segments must be less than 2GB in
14121 size, both starting anywhere in memory (determined at link time). The
14122 global register %g4 points to the base of the data segment. Programs
14123 are statically linked and PIC is not supported.
14126 @itemx -mno-stack-bias
14127 @opindex mstack-bias
14128 @opindex mno-stack-bias
14129 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14130 frame pointer if present, are offset by @minus{}2047 which must be added back
14131 when making stack frame references. This is the default in 64-bit mode.
14132 Otherwise, assume no such offset is present.
14135 These switches are supported in addition to the above on Solaris:
14140 Add support for multithreading using the Solaris threads library. This
14141 option sets flags for both the preprocessor and linker. This option does
14142 not affect the thread safety of object code produced by the compiler or
14143 that of libraries supplied with it.
14147 Add support for multithreading using the POSIX threads library. This
14148 option sets flags for both the preprocessor and linker. This option does
14149 not affect the thread safety of object code produced by the compiler or
14150 that of libraries supplied with it.
14154 This is a synonym for @option{-pthreads}.
14158 @subsection SPU Options
14159 @cindex SPU options
14161 These @samp{-m} options are supported on the SPU:
14165 @itemx -merror-reloc
14166 @opindex mwarn-reloc
14167 @opindex merror-reloc
14169 The loader for SPU does not handle dynamic relocations. By default, GCC
14170 will give an error when it generates code that requires a dynamic
14171 relocation. @option{-mno-error-reloc} disables the error,
14172 @option{-mwarn-reloc} will generate a warning instead.
14175 @itemx -munsafe-dma
14177 @opindex munsafe-dma
14179 Instructions which initiate or test completion of DMA must not be
14180 reordered with respect to loads and stores of the memory which is being
14181 accessed. Users typically address this problem using the volatile
14182 keyword, but that can lead to inefficient code in places where the
14183 memory is known to not change. Rather than mark the memory as volatile
14184 we treat the DMA instructions as potentially effecting all memory. With
14185 @option{-munsafe-dma} users must use the volatile keyword to protect
14188 @item -mbranch-hints
14189 @opindex mbranch-hints
14191 By default, GCC will generate a branch hint instruction to avoid
14192 pipeline stalls for always taken or probably taken branches. A hint
14193 will not be generated closer than 8 instructions away from its branch.
14194 There is little reason to disable them, except for debugging purposes,
14195 or to make an object a little bit smaller.
14199 @opindex msmall-mem
14200 @opindex mlarge-mem
14202 By default, GCC generates code assuming that addresses are never larger
14203 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14204 a full 32 bit address.
14209 By default, GCC links against startup code that assumes the SPU-style
14210 main function interface (which has an unconventional parameter list).
14211 With @option{-mstdmain}, GCC will link your program against startup
14212 code that assumes a C99-style interface to @code{main}, including a
14213 local copy of @code{argv} strings.
14215 @item -mfixed-range=@var{register-range}
14216 @opindex mfixed-range
14217 Generate code treating the given register range as fixed registers.
14218 A fixed register is one that the register allocator can not use. This is
14219 useful when compiling kernel code. A register range is specified as
14220 two registers separated by a dash. Multiple register ranges can be
14221 specified separated by a comma.
14225 @node System V Options
14226 @subsection Options for System V
14228 These additional options are available on System V Release 4 for
14229 compatibility with other compilers on those systems:
14234 Create a shared object.
14235 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14239 Identify the versions of each tool used by the compiler, in a
14240 @code{.ident} assembler directive in the output.
14244 Refrain from adding @code{.ident} directives to the output file (this is
14247 @item -YP,@var{dirs}
14249 Search the directories @var{dirs}, and no others, for libraries
14250 specified with @option{-l}.
14252 @item -Ym,@var{dir}
14254 Look in the directory @var{dir} to find the M4 preprocessor.
14255 The assembler uses this option.
14256 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14257 @c the generic assembler that comes with Solaris takes just -Ym.
14260 @node TMS320C3x/C4x Options
14261 @subsection TMS320C3x/C4x Options
14262 @cindex TMS320C3x/C4x Options
14264 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
14268 @item -mcpu=@var{cpu_type}
14270 Set the instruction set, register set, and instruction scheduling
14271 parameters for machine type @var{cpu_type}. Supported values for
14272 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
14273 @samp{c44}. The default is @samp{c40} to generate code for the
14278 @itemx -msmall-memory
14280 @opindex mbig-memory
14282 @opindex msmall-memory
14284 Generates code for the big or small memory model. The small memory
14285 model assumed that all data fits into one 64K word page. At run-time
14286 the data page (DP) register must be set to point to the 64K page
14287 containing the .bss and .data program sections. The big memory model is
14288 the default and requires reloading of the DP register for every direct
14295 Allow (disallow) allocation of general integer operands into the block
14296 count register BK@.
14302 Enable (disable) generation of code using decrement and branch,
14303 DBcond(D), instructions. This is enabled by default for the C4x. To be
14304 on the safe side, this is disabled for the C3x, since the maximum
14305 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
14306 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
14307 that it can utilize the decrement and branch instruction, but will give
14308 up if there is more than one memory reference in the loop. Thus a loop
14309 where the loop counter is decremented can generate slightly more
14310 efficient code, in cases where the RPTB instruction cannot be utilized.
14312 @item -mdp-isr-reload
14314 @opindex mdp-isr-reload
14316 Force the DP register to be saved on entry to an interrupt service
14317 routine (ISR), reloaded to point to the data section, and restored on
14318 exit from the ISR@. This should not be required unless someone has
14319 violated the small memory model by modifying the DP register, say within
14326 For the C3x use the 24-bit MPYI instruction for integer multiplies
14327 instead of a library call to guarantee 32-bit results. Note that if one
14328 of the operands is a constant, then the multiplication will be performed
14329 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
14330 then squaring operations are performed inline instead of a library call.
14333 @itemx -mno-fast-fix
14335 @opindex mno-fast-fix
14336 The C3x/C4x FIX instruction to convert a floating point value to an
14337 integer value chooses the nearest integer less than or equal to the
14338 floating point value rather than to the nearest integer. Thus if the
14339 floating point number is negative, the result will be incorrectly
14340 truncated an additional code is necessary to detect and correct this
14341 case. This option can be used to disable generation of the additional
14342 code required to correct the result.
14348 Enable (disable) generation of repeat block sequences using the RPTB
14349 instruction for zero overhead looping. The RPTB construct is only used
14350 for innermost loops that do not call functions or jump across the loop
14351 boundaries. There is no advantage having nested RPTB loops due to the
14352 overhead required to save and restore the RC, RS, and RE registers.
14353 This is enabled by default with @option{-O2}.
14355 @item -mrpts=@var{count}
14359 Enable (disable) the use of the single instruction repeat instruction
14360 RPTS@. If a repeat block contains a single instruction, and the loop
14361 count can be guaranteed to be less than the value @var{count}, GCC will
14362 emit a RPTS instruction instead of a RPTB@. If no value is specified,
14363 then a RPTS will be emitted even if the loop count cannot be determined
14364 at compile time. Note that the repeated instruction following RPTS does
14365 not have to be reloaded from memory each iteration, thus freeing up the
14366 CPU buses for operands. However, since interrupts are blocked by this
14367 instruction, it is disabled by default.
14369 @item -mloop-unsigned
14370 @itemx -mno-loop-unsigned
14371 @opindex mloop-unsigned
14372 @opindex mno-loop-unsigned
14373 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
14374 is @math{2^{31} + 1} since these instructions test if the iteration count is
14375 negative to terminate the loop. If the iteration count is unsigned
14376 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
14377 exceeded. This switch allows an unsigned iteration count.
14381 Try to emit an assembler syntax that the TI assembler (asm30) is happy
14382 with. This also enforces compatibility with the API employed by the TI
14383 C3x C compiler. For example, long doubles are passed as structures
14384 rather than in floating point registers.
14390 Generate code that uses registers (stack) for passing arguments to functions.
14391 By default, arguments are passed in registers where possible rather
14392 than by pushing arguments on to the stack.
14394 @item -mparallel-insns
14395 @itemx -mno-parallel-insns
14396 @opindex mparallel-insns
14397 @opindex mno-parallel-insns
14398 Allow the generation of parallel instructions. This is enabled by
14399 default with @option{-O2}.
14401 @item -mparallel-mpy
14402 @itemx -mno-parallel-mpy
14403 @opindex mparallel-mpy
14404 @opindex mno-parallel-mpy
14405 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
14406 provided @option{-mparallel-insns} is also specified. These instructions have
14407 tight register constraints which can pessimize the code generation
14408 of large functions.
14413 @subsection V850 Options
14414 @cindex V850 Options
14416 These @samp{-m} options are defined for V850 implementations:
14420 @itemx -mno-long-calls
14421 @opindex mlong-calls
14422 @opindex mno-long-calls
14423 Treat all calls as being far away (near). If calls are assumed to be
14424 far away, the compiler will always load the functions address up into a
14425 register, and call indirect through the pointer.
14431 Do not optimize (do optimize) basic blocks that use the same index
14432 pointer 4 or more times to copy pointer into the @code{ep} register, and
14433 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14434 option is on by default if you optimize.
14436 @item -mno-prolog-function
14437 @itemx -mprolog-function
14438 @opindex mno-prolog-function
14439 @opindex mprolog-function
14440 Do not use (do use) external functions to save and restore registers
14441 at the prologue and epilogue of a function. The external functions
14442 are slower, but use less code space if more than one function saves
14443 the same number of registers. The @option{-mprolog-function} option
14444 is on by default if you optimize.
14448 Try to make the code as small as possible. At present, this just turns
14449 on the @option{-mep} and @option{-mprolog-function} options.
14451 @item -mtda=@var{n}
14453 Put static or global variables whose size is @var{n} bytes or less into
14454 the tiny data area that register @code{ep} points to. The tiny data
14455 area can hold up to 256 bytes in total (128 bytes for byte references).
14457 @item -msda=@var{n}
14459 Put static or global variables whose size is @var{n} bytes or less into
14460 the small data area that register @code{gp} points to. The small data
14461 area can hold up to 64 kilobytes.
14463 @item -mzda=@var{n}
14465 Put static or global variables whose size is @var{n} bytes or less into
14466 the first 32 kilobytes of memory.
14470 Specify that the target processor is the V850.
14473 @opindex mbig-switch
14474 Generate code suitable for big switch tables. Use this option only if
14475 the assembler/linker complain about out of range branches within a switch
14480 This option will cause r2 and r5 to be used in the code generated by
14481 the compiler. This setting is the default.
14483 @item -mno-app-regs
14484 @opindex mno-app-regs
14485 This option will cause r2 and r5 to be treated as fixed registers.
14489 Specify that the target processor is the V850E1. The preprocessor
14490 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14491 this option is used.
14495 Specify that the target processor is the V850E@. The preprocessor
14496 constant @samp{__v850e__} will be defined if this option is used.
14498 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14499 are defined then a default target processor will be chosen and the
14500 relevant @samp{__v850*__} preprocessor constant will be defined.
14502 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14503 defined, regardless of which processor variant is the target.
14505 @item -mdisable-callt
14506 @opindex mdisable-callt
14507 This option will suppress generation of the CALLT instruction for the
14508 v850e and v850e1 flavors of the v850 architecture. The default is
14509 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14514 @subsection VAX Options
14515 @cindex VAX options
14517 These @samp{-m} options are defined for the VAX:
14522 Do not output certain jump instructions (@code{aobleq} and so on)
14523 that the Unix assembler for the VAX cannot handle across long
14528 Do output those jump instructions, on the assumption that you
14529 will assemble with the GNU assembler.
14533 Output code for g-format floating point numbers instead of d-format.
14536 @node VxWorks Options
14537 @subsection VxWorks Options
14538 @cindex VxWorks Options
14540 The options in this section are defined for all VxWorks targets.
14541 Options specific to the target hardware are listed with the other
14542 options for that target.
14547 GCC can generate code for both VxWorks kernels and real time processes
14548 (RTPs). This option switches from the former to the latter. It also
14549 defines the preprocessor macro @code{__RTP__}.
14552 @opindex non-static
14553 Link an RTP executable against shared libraries rather than static
14554 libraries. The options @option{-static} and @option{-shared} can
14555 also be used for RTPs (@pxref{Link Options}); @option{-static}
14562 These options are passed down to the linker. They are defined for
14563 compatibility with Diab.
14566 @opindex Xbind-lazy
14567 Enable lazy binding of function calls. This option is equivalent to
14568 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14572 Disable lazy binding of function calls. This option is the default and
14573 is defined for compatibility with Diab.
14576 @node x86-64 Options
14577 @subsection x86-64 Options
14578 @cindex x86-64 options
14580 These are listed under @xref{i386 and x86-64 Options}.
14582 @node Xstormy16 Options
14583 @subsection Xstormy16 Options
14584 @cindex Xstormy16 Options
14586 These options are defined for Xstormy16:
14591 Choose startup files and linker script suitable for the simulator.
14594 @node Xtensa Options
14595 @subsection Xtensa Options
14596 @cindex Xtensa Options
14598 These options are supported for Xtensa targets:
14602 @itemx -mno-const16
14604 @opindex mno-const16
14605 Enable or disable use of @code{CONST16} instructions for loading
14606 constant values. The @code{CONST16} instruction is currently not a
14607 standard option from Tensilica. When enabled, @code{CONST16}
14608 instructions are always used in place of the standard @code{L32R}
14609 instructions. The use of @code{CONST16} is enabled by default only if
14610 the @code{L32R} instruction is not available.
14613 @itemx -mno-fused-madd
14614 @opindex mfused-madd
14615 @opindex mno-fused-madd
14616 Enable or disable use of fused multiply/add and multiply/subtract
14617 instructions in the floating-point option. This has no effect if the
14618 floating-point option is not also enabled. Disabling fused multiply/add
14619 and multiply/subtract instructions forces the compiler to use separate
14620 instructions for the multiply and add/subtract operations. This may be
14621 desirable in some cases where strict IEEE 754-compliant results are
14622 required: the fused multiply add/subtract instructions do not round the
14623 intermediate result, thereby producing results with @emph{more} bits of
14624 precision than specified by the IEEE standard. Disabling fused multiply
14625 add/subtract instructions also ensures that the program output is not
14626 sensitive to the compiler's ability to combine multiply and add/subtract
14629 @item -mtext-section-literals
14630 @itemx -mno-text-section-literals
14631 @opindex mtext-section-literals
14632 @opindex mno-text-section-literals
14633 Control the treatment of literal pools. The default is
14634 @option{-mno-text-section-literals}, which places literals in a separate
14635 section in the output file. This allows the literal pool to be placed
14636 in a data RAM/ROM, and it also allows the linker to combine literal
14637 pools from separate object files to remove redundant literals and
14638 improve code size. With @option{-mtext-section-literals}, the literals
14639 are interspersed in the text section in order to keep them as close as
14640 possible to their references. This may be necessary for large assembly
14643 @item -mtarget-align
14644 @itemx -mno-target-align
14645 @opindex mtarget-align
14646 @opindex mno-target-align
14647 When this option is enabled, GCC instructs the assembler to
14648 automatically align instructions to reduce branch penalties at the
14649 expense of some code density. The assembler attempts to widen density
14650 instructions to align branch targets and the instructions following call
14651 instructions. If there are not enough preceding safe density
14652 instructions to align a target, no widening will be performed. The
14653 default is @option{-mtarget-align}. These options do not affect the
14654 treatment of auto-aligned instructions like @code{LOOP}, which the
14655 assembler will always align, either by widening density instructions or
14656 by inserting no-op instructions.
14659 @itemx -mno-longcalls
14660 @opindex mlongcalls
14661 @opindex mno-longcalls
14662 When this option is enabled, GCC instructs the assembler to translate
14663 direct calls to indirect calls unless it can determine that the target
14664 of a direct call is in the range allowed by the call instruction. This
14665 translation typically occurs for calls to functions in other source
14666 files. Specifically, the assembler translates a direct @code{CALL}
14667 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14668 The default is @option{-mno-longcalls}. This option should be used in
14669 programs where the call target can potentially be out of range. This
14670 option is implemented in the assembler, not the compiler, so the
14671 assembly code generated by GCC will still show direct call
14672 instructions---look at the disassembled object code to see the actual
14673 instructions. Note that the assembler will use an indirect call for
14674 every cross-file call, not just those that really will be out of range.
14677 @node zSeries Options
14678 @subsection zSeries Options
14679 @cindex zSeries options
14681 These are listed under @xref{S/390 and zSeries Options}.
14683 @node Code Gen Options
14684 @section Options for Code Generation Conventions
14685 @cindex code generation conventions
14686 @cindex options, code generation
14687 @cindex run-time options
14689 These machine-independent options control the interface conventions
14690 used in code generation.
14692 Most of them have both positive and negative forms; the negative form
14693 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14694 one of the forms is listed---the one which is not the default. You
14695 can figure out the other form by either removing @samp{no-} or adding
14699 @item -fbounds-check
14700 @opindex fbounds-check
14701 For front-ends that support it, generate additional code to check that
14702 indices used to access arrays are within the declared range. This is
14703 currently only supported by the Java and Fortran front-ends, where
14704 this option defaults to true and false respectively.
14708 This option generates traps for signed overflow on addition, subtraction,
14709 multiplication operations.
14713 This option instructs the compiler to assume that signed arithmetic
14714 overflow of addition, subtraction and multiplication wraps around
14715 using twos-complement representation. This flag enables some optimizations
14716 and disables others. This option is enabled by default for the Java
14717 front-end, as required by the Java language specification.
14720 @opindex fexceptions
14721 Enable exception handling. Generates extra code needed to propagate
14722 exceptions. For some targets, this implies GCC will generate frame
14723 unwind information for all functions, which can produce significant data
14724 size overhead, although it does not affect execution. If you do not
14725 specify this option, GCC will enable it by default for languages like
14726 C++ which normally require exception handling, and disable it for
14727 languages like C that do not normally require it. However, you may need
14728 to enable this option when compiling C code that needs to interoperate
14729 properly with exception handlers written in C++. You may also wish to
14730 disable this option if you are compiling older C++ programs that don't
14731 use exception handling.
14733 @item -fnon-call-exceptions
14734 @opindex fnon-call-exceptions
14735 Generate code that allows trapping instructions to throw exceptions.
14736 Note that this requires platform-specific runtime support that does
14737 not exist everywhere. Moreover, it only allows @emph{trapping}
14738 instructions to throw exceptions, i.e.@: memory references or floating
14739 point instructions. It does not allow exceptions to be thrown from
14740 arbitrary signal handlers such as @code{SIGALRM}.
14742 @item -funwind-tables
14743 @opindex funwind-tables
14744 Similar to @option{-fexceptions}, except that it will just generate any needed
14745 static data, but will not affect the generated code in any other way.
14746 You will normally not enable this option; instead, a language processor
14747 that needs this handling would enable it on your behalf.
14749 @item -fasynchronous-unwind-tables
14750 @opindex fasynchronous-unwind-tables
14751 Generate unwind table in dwarf2 format, if supported by target machine. The
14752 table is exact at each instruction boundary, so it can be used for stack
14753 unwinding from asynchronous events (such as debugger or garbage collector).
14755 @item -fpcc-struct-return
14756 @opindex fpcc-struct-return
14757 Return ``short'' @code{struct} and @code{union} values in memory like
14758 longer ones, rather than in registers. This convention is less
14759 efficient, but it has the advantage of allowing intercallability between
14760 GCC-compiled files and files compiled with other compilers, particularly
14761 the Portable C Compiler (pcc).
14763 The precise convention for returning structures in memory depends
14764 on the target configuration macros.
14766 Short structures and unions are those whose size and alignment match
14767 that of some integer type.
14769 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14770 switch is not binary compatible with code compiled with the
14771 @option{-freg-struct-return} switch.
14772 Use it to conform to a non-default application binary interface.
14774 @item -freg-struct-return
14775 @opindex freg-struct-return
14776 Return @code{struct} and @code{union} values in registers when possible.
14777 This is more efficient for small structures than
14778 @option{-fpcc-struct-return}.
14780 If you specify neither @option{-fpcc-struct-return} nor
14781 @option{-freg-struct-return}, GCC defaults to whichever convention is
14782 standard for the target. If there is no standard convention, GCC
14783 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14784 the principal compiler. In those cases, we can choose the standard, and
14785 we chose the more efficient register return alternative.
14787 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14788 switch is not binary compatible with code compiled with the
14789 @option{-fpcc-struct-return} switch.
14790 Use it to conform to a non-default application binary interface.
14792 @item -fshort-enums
14793 @opindex fshort-enums
14794 Allocate to an @code{enum} type only as many bytes as it needs for the
14795 declared range of possible values. Specifically, the @code{enum} type
14796 will be equivalent to the smallest integer type which has enough room.
14798 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14799 code that is not binary compatible with code generated without that switch.
14800 Use it to conform to a non-default application binary interface.
14802 @item -fshort-double
14803 @opindex fshort-double
14804 Use the same size for @code{double} as for @code{float}.
14806 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14807 code that is not binary compatible with code generated without that switch.
14808 Use it to conform to a non-default application binary interface.
14810 @item -fshort-wchar
14811 @opindex fshort-wchar
14812 Override the underlying type for @samp{wchar_t} to be @samp{short
14813 unsigned int} instead of the default for the target. This option is
14814 useful for building programs to run under WINE@.
14816 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14817 code that is not binary compatible with code generated without that switch.
14818 Use it to conform to a non-default application binary interface.
14821 @opindex fno-common
14822 In C, allocate even uninitialized global variables in the data section of the
14823 object file, rather than generating them as common blocks. This has the
14824 effect that if the same variable is declared (without @code{extern}) in
14825 two different compilations, you will get an error when you link them.
14826 The only reason this might be useful is if you wish to verify that the
14827 program will work on other systems which always work this way.
14831 Ignore the @samp{#ident} directive.
14833 @item -finhibit-size-directive
14834 @opindex finhibit-size-directive
14835 Don't output a @code{.size} assembler directive, or anything else that
14836 would cause trouble if the function is split in the middle, and the
14837 two halves are placed at locations far apart in memory. This option is
14838 used when compiling @file{crtstuff.c}; you should not need to use it
14841 @item -fverbose-asm
14842 @opindex fverbose-asm
14843 Put extra commentary information in the generated assembly code to
14844 make it more readable. This option is generally only of use to those
14845 who actually need to read the generated assembly code (perhaps while
14846 debugging the compiler itself).
14848 @option{-fno-verbose-asm}, the default, causes the
14849 extra information to be omitted and is useful when comparing two assembler
14852 @item -frecord-gcc-switches
14853 @opindex frecord-gcc-switches
14854 This switch causes the command line that was used to invoke the
14855 compiler to be recorded into the object file that is being created.
14856 This switch is only implemented on some targets and the exact format
14857 of the recording is target and binary file format dependent, but it
14858 usually takes the form of a section containing ASCII text. This
14859 switch is related to the @option{-fverbose-asm} switch, but that
14860 switch only records information in the assembler output file as
14861 comments, so it never reaches the object file.
14865 @cindex global offset table
14867 Generate position-independent code (PIC) suitable for use in a shared
14868 library, if supported for the target machine. Such code accesses all
14869 constant addresses through a global offset table (GOT)@. The dynamic
14870 loader resolves the GOT entries when the program starts (the dynamic
14871 loader is not part of GCC; it is part of the operating system). If
14872 the GOT size for the linked executable exceeds a machine-specific
14873 maximum size, you get an error message from the linker indicating that
14874 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14875 instead. (These maximums are 8k on the SPARC and 32k
14876 on the m68k and RS/6000. The 386 has no such limit.)
14878 Position-independent code requires special support, and therefore works
14879 only on certain machines. For the 386, GCC supports PIC for System V
14880 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14881 position-independent.
14883 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14888 If supported for the target machine, emit position-independent code,
14889 suitable for dynamic linking and avoiding any limit on the size of the
14890 global offset table. This option makes a difference on the m68k,
14891 PowerPC and SPARC@.
14893 Position-independent code requires special support, and therefore works
14894 only on certain machines.
14896 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14903 These options are similar to @option{-fpic} and @option{-fPIC}, but
14904 generated position independent code can be only linked into executables.
14905 Usually these options are used when @option{-pie} GCC option will be
14906 used during linking.
14908 @option{-fpie} and @option{-fPIE} both define the macros
14909 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14910 for @option{-fpie} and 2 for @option{-fPIE}.
14912 @item -fno-jump-tables
14913 @opindex fno-jump-tables
14914 Do not use jump tables for switch statements even where it would be
14915 more efficient than other code generation strategies. This option is
14916 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14917 building code which forms part of a dynamic linker and cannot
14918 reference the address of a jump table. On some targets, jump tables
14919 do not require a GOT and this option is not needed.
14921 @item -ffixed-@var{reg}
14923 Treat the register named @var{reg} as a fixed register; generated code
14924 should never refer to it (except perhaps as a stack pointer, frame
14925 pointer or in some other fixed role).
14927 @var{reg} must be the name of a register. The register names accepted
14928 are machine-specific and are defined in the @code{REGISTER_NAMES}
14929 macro in the machine description macro file.
14931 This flag does not have a negative form, because it specifies a
14934 @item -fcall-used-@var{reg}
14935 @opindex fcall-used
14936 Treat the register named @var{reg} as an allocable register that is
14937 clobbered by function calls. It may be allocated for temporaries or
14938 variables that do not live across a call. Functions compiled this way
14939 will not save and restore the register @var{reg}.
14941 It is an error to used this flag with the frame pointer or stack pointer.
14942 Use of this flag for other registers that have fixed pervasive roles in
14943 the machine's execution model will produce disastrous results.
14945 This flag does not have a negative form, because it specifies a
14948 @item -fcall-saved-@var{reg}
14949 @opindex fcall-saved
14950 Treat the register named @var{reg} as an allocable register saved by
14951 functions. It may be allocated even for temporaries or variables that
14952 live across a call. Functions compiled this way will save and restore
14953 the register @var{reg} if they use it.
14955 It is an error to used this flag with the frame pointer or stack pointer.
14956 Use of this flag for other registers that have fixed pervasive roles in
14957 the machine's execution model will produce disastrous results.
14959 A different sort of disaster will result from the use of this flag for
14960 a register in which function values may be returned.
14962 This flag does not have a negative form, because it specifies a
14965 @item -fpack-struct[=@var{n}]
14966 @opindex fpack-struct
14967 Without a value specified, pack all structure members together without
14968 holes. When a value is specified (which must be a small power of two), pack
14969 structure members according to this value, representing the maximum
14970 alignment (that is, objects with default alignment requirements larger than
14971 this will be output potentially unaligned at the next fitting location.
14973 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14974 code that is not binary compatible with code generated without that switch.
14975 Additionally, it makes the code suboptimal.
14976 Use it to conform to a non-default application binary interface.
14978 @item -finstrument-functions
14979 @opindex finstrument-functions
14980 Generate instrumentation calls for entry and exit to functions. Just
14981 after function entry and just before function exit, the following
14982 profiling functions will be called with the address of the current
14983 function and its call site. (On some platforms,
14984 @code{__builtin_return_address} does not work beyond the current
14985 function, so the call site information may not be available to the
14986 profiling functions otherwise.)
14989 void __cyg_profile_func_enter (void *this_fn,
14991 void __cyg_profile_func_exit (void *this_fn,
14995 The first argument is the address of the start of the current function,
14996 which may be looked up exactly in the symbol table.
14998 This instrumentation is also done for functions expanded inline in other
14999 functions. The profiling calls will indicate where, conceptually, the
15000 inline function is entered and exited. This means that addressable
15001 versions of such functions must be available. If all your uses of a
15002 function are expanded inline, this may mean an additional expansion of
15003 code size. If you use @samp{extern inline} in your C code, an
15004 addressable version of such functions must be provided. (This is
15005 normally the case anyways, but if you get lucky and the optimizer always
15006 expands the functions inline, you might have gotten away without
15007 providing static copies.)
15009 A function may be given the attribute @code{no_instrument_function}, in
15010 which case this instrumentation will not be done. This can be used, for
15011 example, for the profiling functions listed above, high-priority
15012 interrupt routines, and any functions from which the profiling functions
15013 cannot safely be called (perhaps signal handlers, if the profiling
15014 routines generate output or allocate memory).
15016 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15017 @opindex finstrument-functions-exclude-file-list
15019 Set the list of functions that are excluded from instrumentation (see
15020 the description of @code{-finstrument-functions}). If the file that
15021 contains a function definition matches with one of @var{file}, then
15022 that function is not instrumented. The match is done on substrings:
15023 if the @var{file} parameter is a substring of the file name, it is
15024 considered to be a match.
15027 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15028 will exclude any inline function defined in files whose pathnames
15029 contain @code{/bits/stl} or @code{include/sys}.
15031 If, for some reason, you want to include letter @code{','} in one of
15032 @var{sym}, write @code{'\,'}. For example,
15033 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15034 (note the single quote surrounding the option).
15036 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15037 @opindex finstrument-functions-exclude-function-list
15039 This is similar to @code{-finstrument-functions-exclude-file-list},
15040 but this option sets the list of function names to be excluded from
15041 instrumentation. The function name to be matched is its user-visible
15042 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15043 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15044 match is done on substrings: if the @var{sym} parameter is a substring
15045 of the function name, it is considered to be a match.
15047 @item -fstack-check
15048 @opindex fstack-check
15049 Generate code to verify that you do not go beyond the boundary of the
15050 stack. You should specify this flag if you are running in an
15051 environment with multiple threads, but only rarely need to specify it in
15052 a single-threaded environment since stack overflow is automatically
15053 detected on nearly all systems if there is only one stack.
15055 Note that this switch does not actually cause checking to be done; the
15056 operating system must do that. The switch causes generation of code
15057 to ensure that the operating system sees the stack being extended.
15059 @item -fstack-limit-register=@var{reg}
15060 @itemx -fstack-limit-symbol=@var{sym}
15061 @itemx -fno-stack-limit
15062 @opindex fstack-limit-register
15063 @opindex fstack-limit-symbol
15064 @opindex fno-stack-limit
15065 Generate code to ensure that the stack does not grow beyond a certain value,
15066 either the value of a register or the address of a symbol. If the stack
15067 would grow beyond the value, a signal is raised. For most targets,
15068 the signal is raised before the stack overruns the boundary, so
15069 it is possible to catch the signal without taking special precautions.
15071 For instance, if the stack starts at absolute address @samp{0x80000000}
15072 and grows downwards, you can use the flags
15073 @option{-fstack-limit-symbol=__stack_limit} and
15074 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15075 of 128KB@. Note that this may only work with the GNU linker.
15077 @cindex aliasing of parameters
15078 @cindex parameters, aliased
15079 @item -fargument-alias
15080 @itemx -fargument-noalias
15081 @itemx -fargument-noalias-global
15082 @itemx -fargument-noalias-anything
15083 @opindex fargument-alias
15084 @opindex fargument-noalias
15085 @opindex fargument-noalias-global
15086 @opindex fargument-noalias-anything
15087 Specify the possible relationships among parameters and between
15088 parameters and global data.
15090 @option{-fargument-alias} specifies that arguments (parameters) may
15091 alias each other and may alias global storage.@*
15092 @option{-fargument-noalias} specifies that arguments do not alias
15093 each other, but may alias global storage.@*
15094 @option{-fargument-noalias-global} specifies that arguments do not
15095 alias each other and do not alias global storage.
15096 @option{-fargument-noalias-anything} specifies that arguments do not
15097 alias any other storage.
15099 Each language will automatically use whatever option is required by
15100 the language standard. You should not need to use these options yourself.
15102 @item -fleading-underscore
15103 @opindex fleading-underscore
15104 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15105 change the way C symbols are represented in the object file. One use
15106 is to help link with legacy assembly code.
15108 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15109 generate code that is not binary compatible with code generated without that
15110 switch. Use it to conform to a non-default application binary interface.
15111 Not all targets provide complete support for this switch.
15113 @item -ftls-model=@var{model}
15114 @opindex ftls-model
15115 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15116 The @var{model} argument should be one of @code{global-dynamic},
15117 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15119 The default without @option{-fpic} is @code{initial-exec}; with
15120 @option{-fpic} the default is @code{global-dynamic}.
15122 @item -fvisibility=@var{default|internal|hidden|protected}
15123 @opindex fvisibility
15124 Set the default ELF image symbol visibility to the specified option---all
15125 symbols will be marked with this unless overridden within the code.
15126 Using this feature can very substantially improve linking and
15127 load times of shared object libraries, produce more optimized
15128 code, provide near-perfect API export and prevent symbol clashes.
15129 It is @strong{strongly} recommended that you use this in any shared objects
15132 Despite the nomenclature, @code{default} always means public ie;
15133 available to be linked against from outside the shared object.
15134 @code{protected} and @code{internal} are pretty useless in real-world
15135 usage so the only other commonly used option will be @code{hidden}.
15136 The default if @option{-fvisibility} isn't specified is
15137 @code{default}, i.e., make every
15138 symbol public---this causes the same behavior as previous versions of
15141 A good explanation of the benefits offered by ensuring ELF
15142 symbols have the correct visibility is given by ``How To Write
15143 Shared Libraries'' by Ulrich Drepper (which can be found at
15144 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15145 solution made possible by this option to marking things hidden when
15146 the default is public is to make the default hidden and mark things
15147 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15148 and @code{__attribute__ ((visibility("default")))} instead of
15149 @code{__declspec(dllexport)} you get almost identical semantics with
15150 identical syntax. This is a great boon to those working with
15151 cross-platform projects.
15153 For those adding visibility support to existing code, you may find
15154 @samp{#pragma GCC visibility} of use. This works by you enclosing
15155 the declarations you wish to set visibility for with (for example)
15156 @samp{#pragma GCC visibility push(hidden)} and
15157 @samp{#pragma GCC visibility pop}.
15158 Bear in mind that symbol visibility should be viewed @strong{as
15159 part of the API interface contract} and thus all new code should
15160 always specify visibility when it is not the default ie; declarations
15161 only for use within the local DSO should @strong{always} be marked explicitly
15162 as hidden as so to avoid PLT indirection overheads---making this
15163 abundantly clear also aids readability and self-documentation of the code.
15164 Note that due to ISO C++ specification requirements, operator new and
15165 operator delete must always be of default visibility.
15167 Be aware that headers from outside your project, in particular system
15168 headers and headers from any other library you use, may not be
15169 expecting to be compiled with visibility other than the default. You
15170 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15171 before including any such headers.
15173 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15174 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15175 no modifications. However, this means that calls to @samp{extern}
15176 functions with no explicit visibility will use the PLT, so it is more
15177 effective to use @samp{__attribute ((visibility))} and/or
15178 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15179 declarations should be treated as hidden.
15181 Note that @samp{-fvisibility} does affect C++ vague linkage
15182 entities. This means that, for instance, an exception class that will
15183 be thrown between DSOs must be explicitly marked with default
15184 visibility so that the @samp{type_info} nodes will be unified between
15187 An overview of these techniques, their benefits and how to use them
15188 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15194 @node Environment Variables
15195 @section Environment Variables Affecting GCC
15196 @cindex environment variables
15198 @c man begin ENVIRONMENT
15199 This section describes several environment variables that affect how GCC
15200 operates. Some of them work by specifying directories or prefixes to use
15201 when searching for various kinds of files. Some are used to specify other
15202 aspects of the compilation environment.
15204 Note that you can also specify places to search using options such as
15205 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15206 take precedence over places specified using environment variables, which
15207 in turn take precedence over those specified by the configuration of GCC@.
15208 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15209 GNU Compiler Collection (GCC) Internals}.
15214 @c @itemx LC_COLLATE
15216 @c @itemx LC_MONETARY
15217 @c @itemx LC_NUMERIC
15222 @c @findex LC_COLLATE
15223 @findex LC_MESSAGES
15224 @c @findex LC_MONETARY
15225 @c @findex LC_NUMERIC
15229 These environment variables control the way that GCC uses
15230 localization information that allow GCC to work with different
15231 national conventions. GCC inspects the locale categories
15232 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15233 so. These locale categories can be set to any value supported by your
15234 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15235 Kingdom encoded in UTF-8.
15237 The @env{LC_CTYPE} environment variable specifies character
15238 classification. GCC uses it to determine the character boundaries in
15239 a string; this is needed for some multibyte encodings that contain quote
15240 and escape characters that would otherwise be interpreted as a string
15243 The @env{LC_MESSAGES} environment variable specifies the language to
15244 use in diagnostic messages.
15246 If the @env{LC_ALL} environment variable is set, it overrides the value
15247 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15248 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15249 environment variable. If none of these variables are set, GCC
15250 defaults to traditional C English behavior.
15254 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15255 files. GCC uses temporary files to hold the output of one stage of
15256 compilation which is to be used as input to the next stage: for example,
15257 the output of the preprocessor, which is the input to the compiler
15260 @item GCC_EXEC_PREFIX
15261 @findex GCC_EXEC_PREFIX
15262 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15263 names of the subprograms executed by the compiler. No slash is added
15264 when this prefix is combined with the name of a subprogram, but you can
15265 specify a prefix that ends with a slash if you wish.
15267 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15268 an appropriate prefix to use based on the pathname it was invoked with.
15270 If GCC cannot find the subprogram using the specified prefix, it
15271 tries looking in the usual places for the subprogram.
15273 The default value of @env{GCC_EXEC_PREFIX} is
15274 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15275 the installed compiler. In many cases @var{prefix} is the value
15276 of @code{prefix} when you ran the @file{configure} script.
15278 Other prefixes specified with @option{-B} take precedence over this prefix.
15280 This prefix is also used for finding files such as @file{crt0.o} that are
15283 In addition, the prefix is used in an unusual way in finding the
15284 directories to search for header files. For each of the standard
15285 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15286 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15287 replacing that beginning with the specified prefix to produce an
15288 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15289 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15290 These alternate directories are searched first; the standard directories
15291 come next. If a standard directory begins with the configured
15292 @var{prefix} then the value of @var{prefix} is replaced by
15293 @env{GCC_EXEC_PREFIX} when looking for header files.
15295 @item COMPILER_PATH
15296 @findex COMPILER_PATH
15297 The value of @env{COMPILER_PATH} is a colon-separated list of
15298 directories, much like @env{PATH}. GCC tries the directories thus
15299 specified when searching for subprograms, if it can't find the
15300 subprograms using @env{GCC_EXEC_PREFIX}.
15303 @findex LIBRARY_PATH
15304 The value of @env{LIBRARY_PATH} is a colon-separated list of
15305 directories, much like @env{PATH}. When configured as a native compiler,
15306 GCC tries the directories thus specified when searching for special
15307 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15308 using GCC also uses these directories when searching for ordinary
15309 libraries for the @option{-l} option (but directories specified with
15310 @option{-L} come first).
15314 @cindex locale definition
15315 This variable is used to pass locale information to the compiler. One way in
15316 which this information is used is to determine the character set to be used
15317 when character literals, string literals and comments are parsed in C and C++.
15318 When the compiler is configured to allow multibyte characters,
15319 the following values for @env{LANG} are recognized:
15323 Recognize JIS characters.
15325 Recognize SJIS characters.
15327 Recognize EUCJP characters.
15330 If @env{LANG} is not defined, or if it has some other value, then the
15331 compiler will use mblen and mbtowc as defined by the default locale to
15332 recognize and translate multibyte characters.
15336 Some additional environments variables affect the behavior of the
15339 @include cppenv.texi
15343 @node Precompiled Headers
15344 @section Using Precompiled Headers
15345 @cindex precompiled headers
15346 @cindex speed of compilation
15348 Often large projects have many header files that are included in every
15349 source file. The time the compiler takes to process these header files
15350 over and over again can account for nearly all of the time required to
15351 build the project. To make builds faster, GCC allows users to
15352 `precompile' a header file; then, if builds can use the precompiled
15353 header file they will be much faster.
15355 To create a precompiled header file, simply compile it as you would any
15356 other file, if necessary using the @option{-x} option to make the driver
15357 treat it as a C or C++ header file. You will probably want to use a
15358 tool like @command{make} to keep the precompiled header up-to-date when
15359 the headers it contains change.
15361 A precompiled header file will be searched for when @code{#include} is
15362 seen in the compilation. As it searches for the included file
15363 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15364 compiler looks for a precompiled header in each directory just before it
15365 looks for the include file in that directory. The name searched for is
15366 the name specified in the @code{#include} with @samp{.gch} appended. If
15367 the precompiled header file can't be used, it is ignored.
15369 For instance, if you have @code{#include "all.h"}, and you have
15370 @file{all.h.gch} in the same directory as @file{all.h}, then the
15371 precompiled header file will be used if possible, and the original
15372 header will be used otherwise.
15374 Alternatively, you might decide to put the precompiled header file in a
15375 directory and use @option{-I} to ensure that directory is searched
15376 before (or instead of) the directory containing the original header.
15377 Then, if you want to check that the precompiled header file is always
15378 used, you can put a file of the same name as the original header in this
15379 directory containing an @code{#error} command.
15381 This also works with @option{-include}. So yet another way to use
15382 precompiled headers, good for projects not designed with precompiled
15383 header files in mind, is to simply take most of the header files used by
15384 a project, include them from another header file, precompile that header
15385 file, and @option{-include} the precompiled header. If the header files
15386 have guards against multiple inclusion, they will be skipped because
15387 they've already been included (in the precompiled header).
15389 If you need to precompile the same header file for different
15390 languages, targets, or compiler options, you can instead make a
15391 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15392 header in the directory, perhaps using @option{-o}. It doesn't matter
15393 what you call the files in the directory, every precompiled header in
15394 the directory will be considered. The first precompiled header
15395 encountered in the directory that is valid for this compilation will
15396 be used; they're searched in no particular order.
15398 There are many other possibilities, limited only by your imagination,
15399 good sense, and the constraints of your build system.
15401 A precompiled header file can be used only when these conditions apply:
15405 Only one precompiled header can be used in a particular compilation.
15408 A precompiled header can't be used once the first C token is seen. You
15409 can have preprocessor directives before a precompiled header; you can
15410 even include a precompiled header from inside another header, so long as
15411 there are no C tokens before the @code{#include}.
15414 The precompiled header file must be produced for the same language as
15415 the current compilation. You can't use a C precompiled header for a C++
15419 The precompiled header file must have been produced by the same compiler
15420 binary as the current compilation is using.
15423 Any macros defined before the precompiled header is included must
15424 either be defined in the same way as when the precompiled header was
15425 generated, or must not affect the precompiled header, which usually
15426 means that they don't appear in the precompiled header at all.
15428 The @option{-D} option is one way to define a macro before a
15429 precompiled header is included; using a @code{#define} can also do it.
15430 There are also some options that define macros implicitly, like
15431 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15434 @item If debugging information is output when using the precompiled
15435 header, using @option{-g} or similar, the same kind of debugging information
15436 must have been output when building the precompiled header. However,
15437 a precompiled header built using @option{-g} can be used in a compilation
15438 when no debugging information is being output.
15440 @item The same @option{-m} options must generally be used when building
15441 and using the precompiled header. @xref{Submodel Options},
15442 for any cases where this rule is relaxed.
15444 @item Each of the following options must be the same when building and using
15445 the precompiled header:
15447 @gccoptlist{-fexceptions -funit-at-a-time}
15450 Some other command-line options starting with @option{-f},
15451 @option{-p}, or @option{-O} must be defined in the same way as when
15452 the precompiled header was generated. At present, it's not clear
15453 which options are safe to change and which are not; the safest choice
15454 is to use exactly the same options when generating and using the
15455 precompiled header. The following are known to be safe:
15457 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15458 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15459 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15464 For all of these except the last, the compiler will automatically
15465 ignore the precompiled header if the conditions aren't met. If you
15466 find an option combination that doesn't work and doesn't cause the
15467 precompiled header to be ignored, please consider filing a bug report,
15470 If you do use differing options when generating and using the
15471 precompiled header, the actual behavior will be a mixture of the
15472 behavior for the options. For instance, if you use @option{-g} to
15473 generate the precompiled header but not when using it, you may or may
15474 not get debugging information for routines in the precompiled header.
15476 @node Running Protoize
15477 @section Running Protoize
15479 The program @code{protoize} is an optional part of GCC@. You can use
15480 it to add prototypes to a program, thus converting the program to ISO
15481 C in one respect. The companion program @code{unprotoize} does the
15482 reverse: it removes argument types from any prototypes that are found.
15484 When you run these programs, you must specify a set of source files as
15485 command line arguments. The conversion programs start out by compiling
15486 these files to see what functions they define. The information gathered
15487 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15489 After scanning comes actual conversion. The specified files are all
15490 eligible to be converted; any files they include (whether sources or
15491 just headers) are eligible as well.
15493 But not all the eligible files are converted. By default,
15494 @code{protoize} and @code{unprotoize} convert only source and header
15495 files in the current directory. You can specify additional directories
15496 whose files should be converted with the @option{-d @var{directory}}
15497 option. You can also specify particular files to exclude with the
15498 @option{-x @var{file}} option. A file is converted if it is eligible, its
15499 directory name matches one of the specified directory names, and its
15500 name within the directory has not been excluded.
15502 Basic conversion with @code{protoize} consists of rewriting most
15503 function definitions and function declarations to specify the types of
15504 the arguments. The only ones not rewritten are those for varargs
15507 @code{protoize} optionally inserts prototype declarations at the
15508 beginning of the source file, to make them available for any calls that
15509 precede the function's definition. Or it can insert prototype
15510 declarations with block scope in the blocks where undeclared functions
15513 Basic conversion with @code{unprotoize} consists of rewriting most
15514 function declarations to remove any argument types, and rewriting
15515 function definitions to the old-style pre-ISO form.
15517 Both conversion programs print a warning for any function declaration or
15518 definition that they can't convert. You can suppress these warnings
15521 The output from @code{protoize} or @code{unprotoize} replaces the
15522 original source file. The original file is renamed to a name ending
15523 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15524 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15525 for DOS) file already exists, then the source file is simply discarded.
15527 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15528 scan the program and collect information about the functions it uses.
15529 So neither of these programs will work until GCC is installed.
15531 Here is a table of the options you can use with @code{protoize} and
15532 @code{unprotoize}. Each option works with both programs unless
15536 @item -B @var{directory}
15537 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15538 usual directory (normally @file{/usr/local/lib}). This file contains
15539 prototype information about standard system functions. This option
15540 applies only to @code{protoize}.
15542 @item -c @var{compilation-options}
15543 Use @var{compilation-options} as the options when running @command{gcc} to
15544 produce the @samp{.X} files. The special option @option{-aux-info} is
15545 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15547 Note that the compilation options must be given as a single argument to
15548 @code{protoize} or @code{unprotoize}. If you want to specify several
15549 @command{gcc} options, you must quote the entire set of compilation options
15550 to make them a single word in the shell.
15552 There are certain @command{gcc} arguments that you cannot use, because they
15553 would produce the wrong kind of output. These include @option{-g},
15554 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15555 the @var{compilation-options}, they are ignored.
15558 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15559 systems) instead of @samp{.c}. This is convenient if you are converting
15560 a C program to C++. This option applies only to @code{protoize}.
15563 Add explicit global declarations. This means inserting explicit
15564 declarations at the beginning of each source file for each function
15565 that is called in the file and was not declared. These declarations
15566 precede the first function definition that contains a call to an
15567 undeclared function. This option applies only to @code{protoize}.
15569 @item -i @var{string}
15570 Indent old-style parameter declarations with the string @var{string}.
15571 This option applies only to @code{protoize}.
15573 @code{unprotoize} converts prototyped function definitions to old-style
15574 function definitions, where the arguments are declared between the
15575 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15576 uses five spaces as the indentation. If you want to indent with just
15577 one space instead, use @option{-i " "}.
15580 Keep the @samp{.X} files. Normally, they are deleted after conversion
15584 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15585 a prototype declaration for each function in each block which calls the
15586 function without any declaration. This option applies only to
15590 Make no real changes. This mode just prints information about the conversions
15591 that would have been done without @option{-n}.
15594 Make no @samp{.save} files. The original files are simply deleted.
15595 Use this option with caution.
15597 @item -p @var{program}
15598 Use the program @var{program} as the compiler. Normally, the name
15599 @file{gcc} is used.
15602 Work quietly. Most warnings are suppressed.
15605 Print the version number, just like @option{-v} for @command{gcc}.
15608 If you need special compiler options to compile one of your program's
15609 source files, then you should generate that file's @samp{.X} file
15610 specially, by running @command{gcc} on that source file with the
15611 appropriate options and the option @option{-aux-info}. Then run
15612 @code{protoize} on the entire set of files. @code{protoize} will use
15613 the existing @samp{.X} file because it is newer than the source file.
15617 gcc -Dfoo=bar file1.c -aux-info file1.X
15622 You need to include the special files along with the rest in the
15623 @code{protoize} command, even though their @samp{.X} files already
15624 exist, because otherwise they won't get converted.
15626 @xref{Protoize Caveats}, for more information on how to use
15627 @code{protoize} successfully.