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}.
318 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
319 -falign-labels=@var{n} -falign-loops=@var{n} @gol
320 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
321 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
322 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
323 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
324 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
325 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
326 -fexpensive-optimizations -ffast-math -ffloat-store @gol
327 -fforce-addr -fforward-propagate -ffunction-sections @gol
328 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
329 -fcrossjumping -fif-conversion -fif-conversion2 @gol
330 -finline-functions -finline-functions-called-once @gol
331 -finline-small-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
332 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
333 -fmodulo-sched -fmodulo-sched-allow-regmoves -fno-branch-count-reg @gol
334 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
335 -fno-function-cse -fno-guess-branch-probability @gol
336 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
337 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
338 -ffinite-math-only -fno-signed-zeros @gol
339 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
340 -fomit-frame-pointer -foptimize-register-move @gol
341 -foptimize-sibling-calls -fpredictive-commoning -fprefetch-loop-arrays @gol
342 -fprofile-generate -fprofile-use @gol
343 -fregmove -frename-registers @gol
344 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
345 -frerun-cse-after-loop @gol
346 -frounding-math -frtl-abstract-sequences @gol
347 -fschedule-insns -fschedule-insns2 @gol
348 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
349 -fsched-spec-load-dangerous @gol
350 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
351 -fsched2-use-superblocks @gol
352 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
353 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
354 -fno-split-wide-types -fstack-protector -fstack-protector-all @gol
355 -fstrict-aliasing -fstrict-overflow -ftracer -fthread-jumps @gol
356 -funroll-all-loops -funroll-loops -fpeel-loops @gol
357 -fsplit-ivs-in-unroller -funswitch-loops @gol
358 -fvariable-expansion-in-unroller -ftree-reassoc @gol
359 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
360 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
361 -fcheck-data-deps @gol
362 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
363 -ftree-ch -ftree-sra -ftree-ter -ftree-fre -ftree-vectorize @gol
364 -ftree-vect-loop-version -fvect-cost-model -ftree-salias -fipa-pta -fweb @gol
365 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
366 --param @var{name}=@var{value}
367 -O -O0 -O1 -O2 -O3 -Os}
369 @item Preprocessor Options
370 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
371 @gccoptlist{-A@var{question}=@var{answer} @gol
372 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
373 -C -dD -dI -dM -dN @gol
374 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
375 -idirafter @var{dir} @gol
376 -include @var{file} -imacros @var{file} @gol
377 -iprefix @var{file} -iwithprefix @var{dir} @gol
378 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
379 -imultilib @var{dir} -isysroot @var{dir} @gol
380 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
381 -P -fworking-directory -remap @gol
382 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
383 -Xpreprocessor @var{option}}
385 @item Assembler Option
386 @xref{Assembler Options,,Passing Options to the Assembler}.
387 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
390 @xref{Link Options,,Options for Linking}.
391 @gccoptlist{@var{object-file-name} -l@var{library} @gol
392 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
393 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
394 -Wl,@var{option} -Xlinker @var{option} @gol
397 @item Directory Options
398 @xref{Directory Options,,Options for Directory Search}.
399 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
400 -specs=@var{file} -I- --sysroot=@var{dir}}
403 @c I wrote this xref this way to avoid overfull hbox. -- rms
404 @xref{Target Options}.
405 @gccoptlist{-V @var{version} -b @var{machine}}
407 @item Machine Dependent Options
408 @xref{Submodel Options,,Hardware Models and Configurations}.
409 @c This list is ordered alphanumerically by subsection name.
410 @c Try and put the significant identifier (CPU or system) first,
411 @c so users have a clue at guessing where the ones they want will be.
414 @gccoptlist{-EB -EL @gol
415 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
416 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
419 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
420 -mabi=@var{name} @gol
421 -mapcs-stack-check -mno-apcs-stack-check @gol
422 -mapcs-float -mno-apcs-float @gol
423 -mapcs-reentrant -mno-apcs-reentrant @gol
424 -msched-prolog -mno-sched-prolog @gol
425 -mlittle-endian -mbig-endian -mwords-little-endian @gol
426 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
427 -mthumb-interwork -mno-thumb-interwork @gol
428 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
429 -mstructure-size-boundary=@var{n} @gol
430 -mabort-on-noreturn @gol
431 -mlong-calls -mno-long-calls @gol
432 -msingle-pic-base -mno-single-pic-base @gol
433 -mpic-register=@var{reg} @gol
434 -mnop-fun-dllimport @gol
435 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
436 -mpoke-function-name @gol
438 -mtpcs-frame -mtpcs-leaf-frame @gol
439 -mcaller-super-interworking -mcallee-super-interworking @gol
443 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
444 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
446 @emph{Blackfin Options}
447 @gccoptlist{-mcpu=@var{cpu} -msim -momit-leaf-frame-pointer @gol
448 -mno-omit-leaf-frame-pointer -mspecld-anomaly -mno-specld-anomaly @gol
449 -mcsync-anomaly -mno-csync-anomaly -mlow-64k -mno-low64k @gol
450 -mstack-check-l1 -mid-shared-library -mno-id-shared-library @gol
451 -mshared-library-id=@var{n} -mleaf-id-shared-library @gol
452 -mno-leaf-id-shared-library -msep-data -mno-sep-data -mlong-calls @gol
453 -mno-long-calls -mfast-fp -minline-plt}
456 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
457 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
458 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
459 -mstack-align -mdata-align -mconst-align @gol
460 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
461 -melf -maout -melinux -mlinux -sim -sim2 @gol
462 -mmul-bug-workaround -mno-mul-bug-workaround}
465 @gccoptlist{-mmac -mpush-args}
467 @emph{Darwin Options}
468 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
469 -arch_only -bind_at_load -bundle -bundle_loader @gol
470 -client_name -compatibility_version -current_version @gol
472 -dependency-file -dylib_file -dylinker_install_name @gol
473 -dynamic -dynamiclib -exported_symbols_list @gol
474 -filelist -flat_namespace -force_cpusubtype_ALL @gol
475 -force_flat_namespace -headerpad_max_install_names @gol
477 -image_base -init -install_name -keep_private_externs @gol
478 -multi_module -multiply_defined -multiply_defined_unused @gol
479 -noall_load -no_dead_strip_inits_and_terms @gol
480 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
481 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
482 -private_bundle -read_only_relocs -sectalign @gol
483 -sectobjectsymbols -whyload -seg1addr @gol
484 -sectcreate -sectobjectsymbols -sectorder @gol
485 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
486 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
487 -segprot -segs_read_only_addr -segs_read_write_addr @gol
488 -single_module -static -sub_library -sub_umbrella @gol
489 -twolevel_namespace -umbrella -undefined @gol
490 -unexported_symbols_list -weak_reference_mismatches @gol
491 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
492 -mkernel -mone-byte-bool}
494 @emph{DEC Alpha Options}
495 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
496 -mieee -mieee-with-inexact -mieee-conformant @gol
497 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
498 -mtrap-precision=@var{mode} -mbuild-constants @gol
499 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
500 -mbwx -mmax -mfix -mcix @gol
501 -mfloat-vax -mfloat-ieee @gol
502 -mexplicit-relocs -msmall-data -mlarge-data @gol
503 -msmall-text -mlarge-text @gol
504 -mmemory-latency=@var{time}}
506 @emph{DEC Alpha/VMS Options}
507 @gccoptlist{-mvms-return-codes}
510 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
511 -mhard-float -msoft-float @gol
512 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
513 -mdouble -mno-double @gol
514 -mmedia -mno-media -mmuladd -mno-muladd @gol
515 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
516 -mlinked-fp -mlong-calls -malign-labels @gol
517 -mlibrary-pic -macc-4 -macc-8 @gol
518 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
519 -moptimize-membar -mno-optimize-membar @gol
520 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
521 -mvliw-branch -mno-vliw-branch @gol
522 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
523 -mno-nested-cond-exec -mtomcat-stats @gol
527 @emph{GNU/Linux Options}
528 @gccoptlist{-muclibc}
530 @emph{H8/300 Options}
531 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
534 @gccoptlist{-march=@var{architecture-type} @gol
535 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
536 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
537 -mfixed-range=@var{register-range} @gol
538 -mjump-in-delay -mlinker-opt -mlong-calls @gol
539 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
540 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
541 -mno-jump-in-delay -mno-long-load-store @gol
542 -mno-portable-runtime -mno-soft-float @gol
543 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
544 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
545 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
546 -munix=@var{unix-std} -nolibdld -static -threads}
548 @emph{i386 and x86-64 Options}
549 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
550 -mfpmath=@var{unit} @gol
551 -masm=@var{dialect} -mno-fancy-math-387 @gol
552 -mno-fp-ret-in-387 -msoft-float @gol
553 -mno-wide-multiply -mrtd -malign-double @gol
554 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
555 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
556 -msse4a -m3dnow -mpopcnt -mabm @gol
557 -mthreads -mno-align-stringops -minline-all-stringops @gol
558 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
559 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
560 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
561 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
562 -mcmodel=@var{code-model} @gol
563 -m32 -m64 -mlarge-data-threshold=@var{num}}
566 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
567 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
568 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
569 -minline-float-divide-max-throughput @gol
570 -minline-int-divide-min-latency @gol
571 -minline-int-divide-max-throughput @gol
572 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
573 -mno-dwarf2-asm -mearly-stop-bits @gol
574 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
575 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
576 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
577 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
578 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
579 -mno-sched-prefer-non-data-spec-insns @gol
580 -mno-sched-prefer-non-control-spec-insns @gol
581 -mno-sched-count-spec-in-critical-path}
583 @emph{M32R/D Options}
584 @gccoptlist{-m32r2 -m32rx -m32r @gol
586 -malign-loops -mno-align-loops @gol
587 -missue-rate=@var{number} @gol
588 -mbranch-cost=@var{number} @gol
589 -mmodel=@var{code-size-model-type} @gol
590 -msdata=@var{sdata-type} @gol
591 -mno-flush-func -mflush-func=@var{name} @gol
592 -mno-flush-trap -mflush-trap=@var{number} @gol
596 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
598 @emph{M680x0 Options}
599 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
600 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
601 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
602 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
603 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
604 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
605 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
606 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
608 @emph{M68hc1x Options}
609 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
610 -mauto-incdec -minmax -mlong-calls -mshort @gol
611 -msoft-reg-count=@var{count}}
614 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
615 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
616 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
617 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
618 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
621 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
622 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
623 -mips16 -mno-mips16 -mflip-mips16 @gol
624 -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 @gol
631 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
632 -G@var{num} -membedded-data -mno-embedded-data @gol
633 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
634 -mcode-readable=@var{setting} @gol
635 -msplit-addresses -mno-split-addresses @gol
636 -mexplicit-relocs -mno-explicit-relocs @gol
637 -mcheck-zero-division -mno-check-zero-division @gol
638 -mdivide-traps -mdivide-breaks @gol
639 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
640 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
641 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
642 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
643 -mfix-sb1 -mno-fix-sb1 @gol
644 -mflush-func=@var{func} -mno-flush-func @gol
645 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
646 -mfp-exceptions -mno-fp-exceptions @gol
647 -mvr4130-align -mno-vr4130-align}
650 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
651 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
652 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
653 -mno-base-addresses -msingle-exit -mno-single-exit}
655 @emph{MN10300 Options}
656 @gccoptlist{-mmult-bug -mno-mult-bug @gol
657 -mam33 -mno-am33 @gol
658 -mam33-2 -mno-am33-2 @gol
659 -mreturn-pointer-on-d0 @gol
663 @gccoptlist{-mno-crt0 -mbacc -msim @gol
664 -march=@var{cpu-type} }
666 @emph{PDP-11 Options}
667 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
668 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
669 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
670 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
671 -mbranch-expensive -mbranch-cheap @gol
672 -msplit -mno-split -munix-asm -mdec-asm}
674 @emph{PowerPC Options}
675 See RS/6000 and PowerPC Options.
677 @emph{RS/6000 and PowerPC Options}
678 @gccoptlist{-mcpu=@var{cpu-type} @gol
679 -mtune=@var{cpu-type} @gol
680 -mpower -mno-power -mpower2 -mno-power2 @gol
681 -mpowerpc -mpowerpc64 -mno-powerpc @gol
682 -maltivec -mno-altivec @gol
683 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
684 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
685 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
686 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mdfp -mno-dfp @gol
687 -mnew-mnemonics -mold-mnemonics @gol
688 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
689 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
690 -malign-power -malign-natural @gol
691 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
692 -mstring -mno-string -mupdate -mno-update @gol
693 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
694 -mstrict-align -mno-strict-align -mrelocatable @gol
695 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
696 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
697 -mdynamic-no-pic -maltivec -mswdiv @gol
698 -mprioritize-restricted-insns=@var{priority} @gol
699 -msched-costly-dep=@var{dependence_type} @gol
700 -minsert-sched-nops=@var{scheme} @gol
701 -mcall-sysv -mcall-netbsd @gol
702 -maix-struct-return -msvr4-struct-return @gol
703 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
704 -misel -mno-isel @gol
705 -misel=yes -misel=no @gol
707 -mspe=yes -mspe=no @gol
709 -mvrsave -mno-vrsave @gol
710 -mmulhw -mno-mulhw @gol
711 -mdlmzb -mno-dlmzb @gol
712 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
713 -mprototype -mno-prototype @gol
714 -msim -mmvme -mads -myellowknife -memb -msdata @gol
715 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
717 @emph{S/390 and zSeries Options}
718 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
719 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
720 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
721 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
722 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
723 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
724 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
727 @gccoptlist{-meb -mel @gol
731 -mscore5 -mscore5u -mscore7 -mscore7d}
734 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
735 -m4-nofpu -m4-single-only -m4-single -m4 @gol
736 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
737 -m5-64media -m5-64media-nofpu @gol
738 -m5-32media -m5-32media-nofpu @gol
739 -m5-compact -m5-compact-nofpu @gol
740 -mb -ml -mdalign -mrelax @gol
741 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
742 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
743 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
744 -mdivsi3_libfunc=@var{name} @gol
745 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
749 @gccoptlist{-mcpu=@var{cpu-type} @gol
750 -mtune=@var{cpu-type} @gol
751 -mcmodel=@var{code-model} @gol
752 -m32 -m64 -mapp-regs -mno-app-regs @gol
753 -mfaster-structs -mno-faster-structs @gol
754 -mfpu -mno-fpu -mhard-float -msoft-float @gol
755 -mhard-quad-float -msoft-quad-float @gol
756 -mimpure-text -mno-impure-text -mlittle-endian @gol
757 -mstack-bias -mno-stack-bias @gol
758 -munaligned-doubles -mno-unaligned-doubles @gol
759 -mv8plus -mno-v8plus -mvis -mno-vis
760 -threads -pthreads -pthread}
763 @gccoptlist{-mwarn-reloc -merror-reloc @gol
764 -msafe-dma -munsafe-dma @gol
766 -msmall-mem -mlarge-mem -mstdmain @gol
767 -mfixed-range=@var{register-range}}
769 @emph{System V Options}
770 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
772 @emph{TMS320C3x/C4x Options}
773 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
774 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
775 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
776 -mparallel-insns -mparallel-mpy -mpreserve-float}
779 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
780 -mprolog-function -mno-prolog-function -mspace @gol
781 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
782 -mapp-regs -mno-app-regs @gol
783 -mdisable-callt -mno-disable-callt @gol
789 @gccoptlist{-mg -mgnu -munix}
791 @emph{VxWorks Options}
792 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
793 -Xbind-lazy -Xbind-now}
795 @emph{x86-64 Options}
796 See i386 and x86-64 Options.
798 @emph{Xstormy16 Options}
801 @emph{Xtensa Options}
802 @gccoptlist{-mconst16 -mno-const16 @gol
803 -mfused-madd -mno-fused-madd @gol
804 -mtext-section-literals -mno-text-section-literals @gol
805 -mtarget-align -mno-target-align @gol
806 -mlongcalls -mno-longcalls}
808 @emph{zSeries Options}
809 See S/390 and zSeries Options.
811 @item Code Generation Options
812 @xref{Code Gen Options,,Options for Code Generation Conventions}.
813 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
814 -ffixed-@var{reg} -fexceptions @gol
815 -fnon-call-exceptions -funwind-tables @gol
816 -fasynchronous-unwind-tables @gol
817 -finhibit-size-directive -finstrument-functions @gol
818 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
819 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
820 -fno-common -fno-ident @gol
821 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
822 -fno-jump-tables @gol
823 -frecord-gcc-switches @gol
824 -freg-struct-return -fshort-enums @gol
825 -fshort-double -fshort-wchar @gol
826 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
827 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
828 -fargument-alias -fargument-noalias @gol
829 -fargument-noalias-global -fargument-noalias-anything
830 -fleading-underscore -ftls-model=@var{model} @gol
831 -ftrapv -fwrapv -fbounds-check @gol
836 * Overall Options:: Controlling the kind of output:
837 an executable, object files, assembler files,
838 or preprocessed source.
839 * C Dialect Options:: Controlling the variant of C language compiled.
840 * C++ Dialect Options:: Variations on C++.
841 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
843 * Language Independent Options:: Controlling how diagnostics should be
845 * Warning Options:: How picky should the compiler be?
846 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
847 * Optimize Options:: How much optimization?
848 * Preprocessor Options:: Controlling header files and macro definitions.
849 Also, getting dependency information for Make.
850 * Assembler Options:: Passing options to the assembler.
851 * Link Options:: Specifying libraries and so on.
852 * Directory Options:: Where to find header files and libraries.
853 Where to find the compiler executable files.
854 * Spec Files:: How to pass switches to sub-processes.
855 * Target Options:: Running a cross-compiler, or an old version of GCC.
858 @node Overall Options
859 @section Options Controlling the Kind of Output
861 Compilation can involve up to four stages: preprocessing, compilation
862 proper, assembly and linking, always in that order. GCC is capable of
863 preprocessing and compiling several files either into several
864 assembler input files, or into one assembler input file; then each
865 assembler input file produces an object file, and linking combines all
866 the object files (those newly compiled, and those specified as input)
867 into an executable file.
869 @cindex file name suffix
870 For any given input file, the file name suffix determines what kind of
875 C source code which must be preprocessed.
878 C source code which should not be preprocessed.
881 C++ source code which should not be preprocessed.
884 Objective-C source code. Note that you must link with the @file{libobjc}
885 library to make an Objective-C program work.
888 Objective-C source code which should not be preprocessed.
892 Objective-C++ source code. Note that you must link with the @file{libobjc}
893 library to make an Objective-C++ program work. Note that @samp{.M} refers
894 to a literal capital M@.
897 Objective-C++ source code which should not be preprocessed.
900 C, C++, Objective-C or Objective-C++ header file to be turned into a
905 @itemx @var{file}.cxx
906 @itemx @var{file}.cpp
907 @itemx @var{file}.CPP
908 @itemx @var{file}.c++
910 C++ source code which must be preprocessed. Note that in @samp{.cxx},
911 the last two letters must both be literally @samp{x}. Likewise,
912 @samp{.C} refers to a literal capital C@.
916 Objective-C++ source code which must be preprocessed.
919 Objective-C++ source code which should not be preprocessed.
924 @itemx @var{file}.hxx
925 @itemx @var{file}.hpp
926 @itemx @var{file}.HPP
927 @itemx @var{file}.h++
928 @itemx @var{file}.tcc
929 C++ header file to be turned into a precompiled header.
932 @itemx @var{file}.for
933 @itemx @var{file}.FOR
934 Fixed form Fortran source code which should not be preprocessed.
937 @itemx @var{file}.fpp
938 @itemx @var{file}.FPP
939 Fixed form Fortran source code which must be preprocessed (with the traditional
943 @itemx @var{file}.f95
944 Free form Fortran source code which should not be preprocessed.
947 @itemx @var{file}.F95
948 Free form Fortran source code which must be preprocessed (with the
949 traditional preprocessor).
951 @c FIXME: Descriptions of Java file types.
958 Ada source code file which contains a library unit declaration (a
959 declaration of a package, subprogram, or generic, or a generic
960 instantiation), or a library unit renaming declaration (a package,
961 generic, or subprogram renaming declaration). Such files are also
964 @itemx @var{file}.adb
965 Ada source code file containing a library unit body (a subprogram or
966 package body). Such files are also called @dfn{bodies}.
968 @c GCC also knows about some suffixes for languages not yet included:
979 Assembler code which must be preprocessed.
982 An object file to be fed straight into linking.
983 Any file name with no recognized suffix is treated this way.
987 You can specify the input language explicitly with the @option{-x} option:
990 @item -x @var{language}
991 Specify explicitly the @var{language} for the following input files
992 (rather than letting the compiler choose a default based on the file
993 name suffix). This option applies to all following input files until
994 the next @option{-x} option. Possible values for @var{language} are:
996 c c-header c-cpp-output
997 c++ c++-header c++-cpp-output
998 objective-c objective-c-header objective-c-cpp-output
999 objective-c++ objective-c++-header objective-c++-cpp-output
1000 assembler assembler-with-cpp
1008 Turn off any specification of a language, so that subsequent files are
1009 handled according to their file name suffixes (as they are if @option{-x}
1010 has not been used at all).
1012 @item -pass-exit-codes
1013 @opindex pass-exit-codes
1014 Normally the @command{gcc} program will exit with the code of 1 if any
1015 phase of the compiler returns a non-success return code. If you specify
1016 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1017 numerically highest error produced by any phase that returned an error
1018 indication. The C, C++, and Fortran frontends return 4, if an internal
1019 compiler error is encountered.
1022 If you only want some of the stages of compilation, you can use
1023 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1024 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1025 @command{gcc} is to stop. Note that some combinations (for example,
1026 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1031 Compile or assemble the source files, but do not link. The linking
1032 stage simply is not done. The ultimate output is in the form of an
1033 object file for each source file.
1035 By default, the object file name for a source file is made by replacing
1036 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1038 Unrecognized input files, not requiring compilation or assembly, are
1043 Stop after the stage of compilation proper; do not assemble. The output
1044 is in the form of an assembler code file for each non-assembler input
1047 By default, the assembler file name for a source file is made by
1048 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1050 Input files that don't require compilation are ignored.
1054 Stop after the preprocessing stage; do not run the compiler proper. The
1055 output is in the form of preprocessed source code, which is sent to the
1058 Input files which don't require preprocessing are ignored.
1060 @cindex output file option
1063 Place output in file @var{file}. This applies regardless to whatever
1064 sort of output is being produced, whether it be an executable file,
1065 an object file, an assembler file or preprocessed C code.
1067 If @option{-o} is not specified, the default is to put an executable
1068 file in @file{a.out}, the object file for
1069 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1070 assembler file in @file{@var{source}.s}, a precompiled header file in
1071 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1076 Print (on standard error output) the commands executed to run the stages
1077 of compilation. Also print the version number of the compiler driver
1078 program and of the preprocessor and the compiler proper.
1082 Like @option{-v} except the commands are not executed and all command
1083 arguments are quoted. This is useful for shell scripts to capture the
1084 driver-generated command lines.
1088 Use pipes rather than temporary files for communication between the
1089 various stages of compilation. This fails to work on some systems where
1090 the assembler is unable to read from a pipe; but the GNU assembler has
1095 If you are compiling multiple source files, this option tells the driver
1096 to pass all the source files to the compiler at once (for those
1097 languages for which the compiler can handle this). This will allow
1098 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1099 language for which this is supported is C@. If you pass source files for
1100 multiple languages to the driver, using this option, the driver will invoke
1101 the compiler(s) that support IMA once each, passing each compiler all the
1102 source files appropriate for it. For those languages that do not support
1103 IMA this option will be ignored, and the compiler will be invoked once for
1104 each source file in that language. If you use this option in conjunction
1105 with @option{-save-temps}, the compiler will generate multiple
1107 (one for each source file), but only one (combined) @file{.o} or
1112 Print (on the standard output) a description of the command line options
1113 understood by @command{gcc}. If the @option{-v} option is also specified
1114 then @option{--help} will also be passed on to the various processes
1115 invoked by @command{gcc}, so that they can display the command line options
1116 they accept. If the @option{-Wextra} option has also been specified
1117 (prior to the @option{--help} option), then command line options which
1118 have no documentation associated with them will also be displayed.
1121 @opindex target-help
1122 Print (on the standard output) a description of target-specific command
1123 line options for each tool. For some targets extra target-specific
1124 information may also be printed.
1126 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1127 Print (on the standard output) a description of the command line
1128 options understood by the compiler that fit into a specific class.
1129 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1130 @samp{params}, or @var{language}:
1133 @item @samp{optimizers}
1134 This will display all of the optimization options supported by the
1137 @item @samp{warnings}
1138 This will display all of the options controlling warning messages
1139 produced by the compiler.
1142 This will display target-specific options. Unlike the
1143 @option{--target-help} option however, target-specific options of the
1144 linker and assembler will not be displayed. This is because those
1145 tools do not currently support the extended @option{--help=} syntax.
1148 This will display the values recognized by the @option{--param}
1151 @item @var{language}
1152 This will display the options supported for @var{language}, where
1153 @var{language} is the name of one of the languages supported in this
1157 This will display the options that are common to all languages.
1160 It is possible to further refine the output of the @option{--help=}
1161 option by adding a comma separated list of qualifiers after the
1162 class. These can be any from the following list:
1165 @item @samp{undocumented}
1166 Display only those options which are undocumented.
1169 Display options which take an argument that appears after an equal
1170 sign in the same continuous piece of text, such as:
1171 @samp{--help=target}.
1173 @item @samp{separate}
1174 Display options which take an argument that appears as a separate word
1175 following the original option, such as: @samp{-o output-file}.
1178 Thus for example to display all the undocumented target-specific
1179 switches supported by the compiler the following can be used:
1182 --help=target,undocumented
1185 The sense of a qualifier can be inverted by prefixing it with the
1186 @var{^} character, so for example to display all binary warning
1187 options (i.e. ones that are either on or off and that do not take an
1188 argument), which have a description the following can be used:
1191 --help=warnings,^joined,^undocumented
1194 A class can also be used as a qualifier, although this usually
1195 restricts the output by so much that there is nothing to display. One
1196 case where it does work however is when one of the classes is
1197 @var{target}. So for example to display all the target-specific
1198 optimization options the following can be used:
1201 --help=target,optimizers
1204 The @option{--help=} option can be repeated on the command line. Each
1205 successive use will display its requested class of options, skipping
1206 those that have already been displayed.
1208 If the @option{-Q} option appears on the command line before the
1209 @option{--help=} option, then the descriptive text displayed by
1210 @option{--help=} is changed. Instead of describing the displayed
1211 options, an indication is given as to whether the option is enabled,
1212 disabled or set to a specific value (assuming that the compiler
1213 knows this at the point where the @option{--help=} option is used).
1215 Here is a truncated example from the ARM port of @command{gcc}:
1218 % gcc -Q -mabi=2 --help=target -c
1219 The following options are target specific:
1221 -mabort-on-noreturn [disabled]
1225 The output is sensitive to the effects of previous command line
1226 options, so for example it is possible to find out which optimizations
1227 are enabled at @option{-O2} by using:
1230 -O2 --help=optimizers
1233 Alternatively you can discover which binary optimizations are enabled
1234 by @option{-O3} by using:
1237 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1238 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1239 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1244 Display the version number and copyrights of the invoked GCC@.
1246 @include @value{srcdir}/../libiberty/at-file.texi
1250 @section Compiling C++ Programs
1252 @cindex suffixes for C++ source
1253 @cindex C++ source file suffixes
1254 C++ source files conventionally use one of the suffixes @samp{.C},
1255 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1256 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1257 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1258 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1259 files with these names and compiles them as C++ programs even if you
1260 call the compiler the same way as for compiling C programs (usually
1261 with the name @command{gcc}).
1265 However, the use of @command{gcc} does not add the C++ library.
1266 @command{g++} is a program that calls GCC and treats @samp{.c},
1267 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1268 files unless @option{-x} is used, and automatically specifies linking
1269 against the C++ library. This program is also useful when
1270 precompiling a C header file with a @samp{.h} extension for use in C++
1271 compilations. On many systems, @command{g++} is also installed with
1272 the name @command{c++}.
1274 @cindex invoking @command{g++}
1275 When you compile C++ programs, you may specify many of the same
1276 command-line options that you use for compiling programs in any
1277 language; or command-line options meaningful for C and related
1278 languages; or options that are meaningful only for C++ programs.
1279 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1280 explanations of options for languages related to C@.
1281 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1282 explanations of options that are meaningful only for C++ programs.
1284 @node C Dialect Options
1285 @section Options Controlling C Dialect
1286 @cindex dialect options
1287 @cindex language dialect options
1288 @cindex options, dialect
1290 The following options control the dialect of C (or languages derived
1291 from C, such as C++, Objective-C and Objective-C++) that the compiler
1295 @cindex ANSI support
1299 In C mode, support all ISO C90 programs. In C++ mode,
1300 remove GNU extensions that conflict with ISO C++.
1302 This turns off certain features of GCC that are incompatible with ISO
1303 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1304 such as the @code{asm} and @code{typeof} keywords, and
1305 predefined macros such as @code{unix} and @code{vax} that identify the
1306 type of system you are using. It also enables the undesirable and
1307 rarely used ISO trigraph feature. For the C compiler,
1308 it disables recognition of C++ style @samp{//} comments as well as
1309 the @code{inline} keyword.
1311 The alternate keywords @code{__asm__}, @code{__extension__},
1312 @code{__inline__} and @code{__typeof__} continue to work despite
1313 @option{-ansi}. You would not want to use them in an ISO C program, of
1314 course, but it is useful to put them in header files that might be included
1315 in compilations done with @option{-ansi}. Alternate predefined macros
1316 such as @code{__unix__} and @code{__vax__} are also available, with or
1317 without @option{-ansi}.
1319 The @option{-ansi} option does not cause non-ISO programs to be
1320 rejected gratuitously. For that, @option{-pedantic} is required in
1321 addition to @option{-ansi}. @xref{Warning Options}.
1323 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1324 option is used. Some header files may notice this macro and refrain
1325 from declaring certain functions or defining certain macros that the
1326 ISO standard doesn't call for; this is to avoid interfering with any
1327 programs that might use these names for other things.
1329 Functions which would normally be built in but do not have semantics
1330 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1331 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1332 built-in functions provided by GCC}, for details of the functions
1337 Determine the language standard. This option is currently only
1338 supported when compiling C or C++. A value for this option must be
1339 provided; possible values are
1344 ISO C90 (same as @option{-ansi}).
1346 @item iso9899:199409
1347 ISO C90 as modified in amendment 1.
1353 ISO C99. Note that this standard is not yet fully supported; see
1354 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1355 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1358 Default, ISO C90 plus GNU extensions (including some C99 features).
1362 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1363 this will become the default. The name @samp{gnu9x} is deprecated.
1366 The 1998 ISO C++ standard plus amendments.
1369 The same as @option{-std=c++98} plus GNU extensions. This is the
1370 default for C++ code.
1373 The working draft of the upcoming ISO C++0x standard. This option
1374 enables experimental features that are likely to be included in
1375 C++0x. The working draft is constantly changing, and any feature that is
1376 enabled by this flag may be removed from future versions of GCC if it is
1377 not part of the C++0x standard.
1380 The same as @option{-std=c++0x} plus GNU extensions. As with
1381 @option{-std=c++0x}, this option enables experimental features that may
1382 be removed in future versions of GCC.
1385 Even when this option is not specified, you can still use some of the
1386 features of newer standards in so far as they do not conflict with
1387 previous C standards. For example, you may use @code{__restrict__} even
1388 when @option{-std=c99} is not specified.
1390 The @option{-std} options specifying some version of ISO C have the same
1391 effects as @option{-ansi}, except that features that were not in ISO C90
1392 but are in the specified version (for example, @samp{//} comments and
1393 the @code{inline} keyword in ISO C99) are not disabled.
1395 @xref{Standards,,Language Standards Supported by GCC}, for details of
1396 these standard versions.
1398 @item -fgnu89-inline
1399 @opindex fgnu89-inline
1400 The option @option{-fgnu89-inline} tells GCC to use the traditional
1401 GNU semantics for @code{inline} functions when in C99 mode.
1402 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1403 is accepted and ignored by GCC versions 4.1.3 up to but not including
1404 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1405 C99 mode. Using this option is roughly equivalent to adding the
1406 @code{gnu_inline} function attribute to all inline functions
1407 (@pxref{Function Attributes}).
1409 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1410 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1411 specifies the default behavior). This option was first supported in
1412 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1414 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1415 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1416 in effect for @code{inline} functions. @xref{Common Predefined
1417 Macros,,,cpp,The C Preprocessor}.
1419 @item -aux-info @var{filename}
1421 Output to the given filename prototyped declarations for all functions
1422 declared and/or defined in a translation unit, including those in header
1423 files. This option is silently ignored in any language other than C@.
1425 Besides declarations, the file indicates, in comments, the origin of
1426 each declaration (source file and line), whether the declaration was
1427 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1428 @samp{O} for old, respectively, in the first character after the line
1429 number and the colon), and whether it came from a declaration or a
1430 definition (@samp{C} or @samp{F}, respectively, in the following
1431 character). In the case of function definitions, a K&R-style list of
1432 arguments followed by their declarations is also provided, inside
1433 comments, after the declaration.
1437 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1438 keyword, so that code can use these words as identifiers. You can use
1439 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1440 instead. @option{-ansi} implies @option{-fno-asm}.
1442 In C++, this switch only affects the @code{typeof} keyword, since
1443 @code{asm} and @code{inline} are standard keywords. You may want to
1444 use the @option{-fno-gnu-keywords} flag instead, which has the same
1445 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1446 switch only affects the @code{asm} and @code{typeof} keywords, since
1447 @code{inline} is a standard keyword in ISO C99.
1450 @itemx -fno-builtin-@var{function}
1451 @opindex fno-builtin
1452 @cindex built-in functions
1453 Don't recognize built-in functions that do not begin with
1454 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1455 functions provided by GCC}, for details of the functions affected,
1456 including those which are not built-in functions when @option{-ansi} or
1457 @option{-std} options for strict ISO C conformance are used because they
1458 do not have an ISO standard meaning.
1460 GCC normally generates special code to handle certain built-in functions
1461 more efficiently; for instance, calls to @code{alloca} may become single
1462 instructions that adjust the stack directly, and calls to @code{memcpy}
1463 may become inline copy loops. The resulting code is often both smaller
1464 and faster, but since the function calls no longer appear as such, you
1465 cannot set a breakpoint on those calls, nor can you change the behavior
1466 of the functions by linking with a different library. In addition,
1467 when a function is recognized as a built-in function, GCC may use
1468 information about that function to warn about problems with calls to
1469 that function, or to generate more efficient code, even if the
1470 resulting code still contains calls to that function. For example,
1471 warnings are given with @option{-Wformat} for bad calls to
1472 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1473 known not to modify global memory.
1475 With the @option{-fno-builtin-@var{function}} option
1476 only the built-in function @var{function} is
1477 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1478 function is named this is not built-in in this version of GCC, this
1479 option is ignored. There is no corresponding
1480 @option{-fbuiltin-@var{function}} option; if you wish to enable
1481 built-in functions selectively when using @option{-fno-builtin} or
1482 @option{-ffreestanding}, you may define macros such as:
1485 #define abs(n) __builtin_abs ((n))
1486 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1491 @cindex hosted environment
1493 Assert that compilation takes place in a hosted environment. This implies
1494 @option{-fbuiltin}. A hosted environment is one in which the
1495 entire standard library is available, and in which @code{main} has a return
1496 type of @code{int}. Examples are nearly everything except a kernel.
1497 This is equivalent to @option{-fno-freestanding}.
1499 @item -ffreestanding
1500 @opindex ffreestanding
1501 @cindex hosted environment
1503 Assert that compilation takes place in a freestanding environment. This
1504 implies @option{-fno-builtin}. A freestanding environment
1505 is one in which the standard library may not exist, and program startup may
1506 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1507 This is equivalent to @option{-fno-hosted}.
1509 @xref{Standards,,Language Standards Supported by GCC}, for details of
1510 freestanding and hosted environments.
1514 @cindex openmp parallel
1515 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1516 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1517 compiler generates parallel code according to the OpenMP Application
1518 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1520 @item -fms-extensions
1521 @opindex fms-extensions
1522 Accept some non-standard constructs used in Microsoft header files.
1524 Some cases of unnamed fields in structures and unions are only
1525 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1526 fields within structs/unions}, for details.
1530 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1531 options for strict ISO C conformance) implies @option{-trigraphs}.
1533 @item -no-integrated-cpp
1534 @opindex no-integrated-cpp
1535 Performs a compilation in two passes: preprocessing and compiling. This
1536 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1537 @option{-B} option. The user supplied compilation step can then add in
1538 an additional preprocessing step after normal preprocessing but before
1539 compiling. The default is to use the integrated cpp (internal cpp)
1541 The semantics of this option will change if "cc1", "cc1plus", and
1542 "cc1obj" are merged.
1544 @cindex traditional C language
1545 @cindex C language, traditional
1547 @itemx -traditional-cpp
1548 @opindex traditional-cpp
1549 @opindex traditional
1550 Formerly, these options caused GCC to attempt to emulate a pre-standard
1551 C compiler. They are now only supported with the @option{-E} switch.
1552 The preprocessor continues to support a pre-standard mode. See the GNU
1553 CPP manual for details.
1555 @item -fcond-mismatch
1556 @opindex fcond-mismatch
1557 Allow conditional expressions with mismatched types in the second and
1558 third arguments. The value of such an expression is void. This option
1559 is not supported for C++.
1561 @item -flax-vector-conversions
1562 @opindex flax-vector-conversions
1563 Allow implicit conversions between vectors with differing numbers of
1564 elements and/or incompatible element types. This option should not be
1567 @item -funsigned-char
1568 @opindex funsigned-char
1569 Let the type @code{char} be unsigned, like @code{unsigned char}.
1571 Each kind of machine has a default for what @code{char} should
1572 be. It is either like @code{unsigned char} by default or like
1573 @code{signed char} by default.
1575 Ideally, a portable program should always use @code{signed char} or
1576 @code{unsigned char} when it depends on the signedness of an object.
1577 But many programs have been written to use plain @code{char} and
1578 expect it to be signed, or expect it to be unsigned, depending on the
1579 machines they were written for. This option, and its inverse, let you
1580 make such a program work with the opposite default.
1582 The type @code{char} is always a distinct type from each of
1583 @code{signed char} or @code{unsigned char}, even though its behavior
1584 is always just like one of those two.
1587 @opindex fsigned-char
1588 Let the type @code{char} be signed, like @code{signed char}.
1590 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1591 the negative form of @option{-funsigned-char}. Likewise, the option
1592 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1594 @item -fsigned-bitfields
1595 @itemx -funsigned-bitfields
1596 @itemx -fno-signed-bitfields
1597 @itemx -fno-unsigned-bitfields
1598 @opindex fsigned-bitfields
1599 @opindex funsigned-bitfields
1600 @opindex fno-signed-bitfields
1601 @opindex fno-unsigned-bitfields
1602 These options control whether a bit-field is signed or unsigned, when the
1603 declaration does not use either @code{signed} or @code{unsigned}. By
1604 default, such a bit-field is signed, because this is consistent: the
1605 basic integer types such as @code{int} are signed types.
1608 @node C++ Dialect Options
1609 @section Options Controlling C++ Dialect
1611 @cindex compiler options, C++
1612 @cindex C++ options, command line
1613 @cindex options, C++
1614 This section describes the command-line options that are only meaningful
1615 for C++ programs; but you can also use most of the GNU compiler options
1616 regardless of what language your program is in. For example, you
1617 might compile a file @code{firstClass.C} like this:
1620 g++ -g -frepo -O -c firstClass.C
1624 In this example, only @option{-frepo} is an option meant
1625 only for C++ programs; you can use the other options with any
1626 language supported by GCC@.
1628 Here is a list of options that are @emph{only} for compiling C++ programs:
1632 @item -fabi-version=@var{n}
1633 @opindex fabi-version
1634 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1635 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1636 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1637 the version that conforms most closely to the C++ ABI specification.
1638 Therefore, the ABI obtained using version 0 will change as ABI bugs
1641 The default is version 2.
1643 @item -fno-access-control
1644 @opindex fno-access-control
1645 Turn off all access checking. This switch is mainly useful for working
1646 around bugs in the access control code.
1650 Check that the pointer returned by @code{operator new} is non-null
1651 before attempting to modify the storage allocated. This check is
1652 normally unnecessary because the C++ standard specifies that
1653 @code{operator new} will only return @code{0} if it is declared
1654 @samp{throw()}, in which case the compiler will always check the
1655 return value even without this option. In all other cases, when
1656 @code{operator new} has a non-empty exception specification, memory
1657 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1658 @samp{new (nothrow)}.
1660 @item -fconserve-space
1661 @opindex fconserve-space
1662 Put uninitialized or runtime-initialized global variables into the
1663 common segment, as C does. This saves space in the executable at the
1664 cost of not diagnosing duplicate definitions. If you compile with this
1665 flag and your program mysteriously crashes after @code{main()} has
1666 completed, you may have an object that is being destroyed twice because
1667 two definitions were merged.
1669 This option is no longer useful on most targets, now that support has
1670 been added for putting variables into BSS without making them common.
1672 @item -ffriend-injection
1673 @opindex ffriend-injection
1674 Inject friend functions into the enclosing namespace, so that they are
1675 visible outside the scope of the class in which they are declared.
1676 Friend functions were documented to work this way in the old Annotated
1677 C++ Reference Manual, and versions of G++ before 4.1 always worked
1678 that way. However, in ISO C++ a friend function which is not declared
1679 in an enclosing scope can only be found using argument dependent
1680 lookup. This option causes friends to be injected as they were in
1683 This option is for compatibility, and may be removed in a future
1686 @item -fno-elide-constructors
1687 @opindex fno-elide-constructors
1688 The C++ standard allows an implementation to omit creating a temporary
1689 which is only used to initialize another object of the same type.
1690 Specifying this option disables that optimization, and forces G++ to
1691 call the copy constructor in all cases.
1693 @item -fno-enforce-eh-specs
1694 @opindex fno-enforce-eh-specs
1695 Don't generate code to check for violation of exception specifications
1696 at runtime. This option violates the C++ standard, but may be useful
1697 for reducing code size in production builds, much like defining
1698 @samp{NDEBUG}. This does not give user code permission to throw
1699 exceptions in violation of the exception specifications; the compiler
1700 will still optimize based on the specifications, so throwing an
1701 unexpected exception will result in undefined behavior.
1704 @itemx -fno-for-scope
1706 @opindex fno-for-scope
1707 If @option{-ffor-scope} is specified, the scope of variables declared in
1708 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1709 as specified by the C++ standard.
1710 If @option{-fno-for-scope} is specified, the scope of variables declared in
1711 a @i{for-init-statement} extends to the end of the enclosing scope,
1712 as was the case in old versions of G++, and other (traditional)
1713 implementations of C++.
1715 The default if neither flag is given to follow the standard,
1716 but to allow and give a warning for old-style code that would
1717 otherwise be invalid, or have different behavior.
1719 @item -fno-gnu-keywords
1720 @opindex fno-gnu-keywords
1721 Do not recognize @code{typeof} as a keyword, so that code can use this
1722 word as an identifier. You can use the keyword @code{__typeof__} instead.
1723 @option{-ansi} implies @option{-fno-gnu-keywords}.
1725 @item -fno-implicit-templates
1726 @opindex fno-implicit-templates
1727 Never emit code for non-inline templates which are instantiated
1728 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1729 @xref{Template Instantiation}, for more information.
1731 @item -fno-implicit-inline-templates
1732 @opindex fno-implicit-inline-templates
1733 Don't emit code for implicit instantiations of inline templates, either.
1734 The default is to handle inlines differently so that compiles with and
1735 without optimization will need the same set of explicit instantiations.
1737 @item -fno-implement-inlines
1738 @opindex fno-implement-inlines
1739 To save space, do not emit out-of-line copies of inline functions
1740 controlled by @samp{#pragma implementation}. This will cause linker
1741 errors if these functions are not inlined everywhere they are called.
1743 @item -fms-extensions
1744 @opindex fms-extensions
1745 Disable pedantic warnings about constructs used in MFC, such as implicit
1746 int and getting a pointer to member function via non-standard syntax.
1748 @item -fno-nonansi-builtins
1749 @opindex fno-nonansi-builtins
1750 Disable built-in declarations of functions that are not mandated by
1751 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1752 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1754 @item -fno-operator-names
1755 @opindex fno-operator-names
1756 Do not treat the operator name keywords @code{and}, @code{bitand},
1757 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1758 synonyms as keywords.
1760 @item -fno-optional-diags
1761 @opindex fno-optional-diags
1762 Disable diagnostics that the standard says a compiler does not need to
1763 issue. Currently, the only such diagnostic issued by G++ is the one for
1764 a name having multiple meanings within a class.
1767 @opindex fpermissive
1768 Downgrade some diagnostics about nonconformant code from errors to
1769 warnings. Thus, using @option{-fpermissive} will allow some
1770 nonconforming code to compile.
1774 Enable automatic template instantiation at link time. This option also
1775 implies @option{-fno-implicit-templates}. @xref{Template
1776 Instantiation}, for more information.
1780 Disable generation of information about every class with virtual
1781 functions for use by the C++ runtime type identification features
1782 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1783 of the language, you can save some space by using this flag. Note that
1784 exception handling uses the same information, but it will generate it as
1785 needed. The @samp{dynamic_cast} operator can still be used for casts that
1786 do not require runtime type information, i.e. casts to @code{void *} or to
1787 unambiguous base classes.
1791 Emit statistics about front-end processing at the end of the compilation.
1792 This information is generally only useful to the G++ development team.
1794 @item -ftemplate-depth-@var{n}
1795 @opindex ftemplate-depth
1796 Set the maximum instantiation depth for template classes to @var{n}.
1797 A limit on the template instantiation depth is needed to detect
1798 endless recursions during template class instantiation. ANSI/ISO C++
1799 conforming programs must not rely on a maximum depth greater than 17.
1801 @item -fno-threadsafe-statics
1802 @opindex fno-threadsafe-statics
1803 Do not emit the extra code to use the routines specified in the C++
1804 ABI for thread-safe initialization of local statics. You can use this
1805 option to reduce code size slightly in code that doesn't need to be
1808 @item -fuse-cxa-atexit
1809 @opindex fuse-cxa-atexit
1810 Register destructors for objects with static storage duration with the
1811 @code{__cxa_atexit} function rather than the @code{atexit} function.
1812 This option is required for fully standards-compliant handling of static
1813 destructors, but will only work if your C library supports
1814 @code{__cxa_atexit}.
1816 @item -fno-use-cxa-get-exception-ptr
1817 @opindex fno-use-cxa-get-exception-ptr
1818 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1819 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1820 if the runtime routine is not available.
1822 @item -fvisibility-inlines-hidden
1823 @opindex fvisibility-inlines-hidden
1824 This switch declares that the user does not attempt to compare
1825 pointers to inline methods where the addresses of the two functions
1826 were taken in different shared objects.
1828 The effect of this is that GCC may, effectively, mark inline methods with
1829 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1830 appear in the export table of a DSO and do not require a PLT indirection
1831 when used within the DSO@. Enabling this option can have a dramatic effect
1832 on load and link times of a DSO as it massively reduces the size of the
1833 dynamic export table when the library makes heavy use of templates.
1835 The behavior of this switch is not quite the same as marking the
1836 methods as hidden directly, because it does not affect static variables
1837 local to the function or cause the compiler to deduce that
1838 the function is defined in only one shared object.
1840 You may mark a method as having a visibility explicitly to negate the
1841 effect of the switch for that method. For example, if you do want to
1842 compare pointers to a particular inline method, you might mark it as
1843 having default visibility. Marking the enclosing class with explicit
1844 visibility will have no effect.
1846 Explicitly instantiated inline methods are unaffected by this option
1847 as their linkage might otherwise cross a shared library boundary.
1848 @xref{Template Instantiation}.
1850 @item -fvisibility-ms-compat
1851 @opindex fvisibility-ms-compat
1852 This flag attempts to use visibility settings to make GCC's C++
1853 linkage model compatible with that of Microsoft Visual Studio.
1855 The flag makes these changes to GCC's linkage model:
1859 It sets the default visibility to @code{hidden}, like
1860 @option{-fvisibility=hidden}.
1863 Types, but not their members, are not hidden by default.
1866 The One Definition Rule is relaxed for types without explicit
1867 visibility specifications which are defined in more than one different
1868 shared object: those declarations are permitted if they would have
1869 been permitted when this option was not used.
1872 In new code it is better to use @option{-fvisibility=hidden} and
1873 export those classes which are intended to be externally visible.
1874 Unfortunately it is possible for code to rely, perhaps accidentally,
1875 on the Visual Studio behavior.
1877 Among the consequences of these changes are that static data members
1878 of the same type with the same name but defined in different shared
1879 objects will be different, so changing one will not change the other;
1880 and that pointers to function members defined in different shared
1881 objects may not compare equal. When this flag is given, it is a
1882 violation of the ODR to define types with the same name differently.
1886 Do not use weak symbol support, even if it is provided by the linker.
1887 By default, G++ will use weak symbols if they are available. This
1888 option exists only for testing, and should not be used by end-users;
1889 it will result in inferior code and has no benefits. This option may
1890 be removed in a future release of G++.
1894 Do not search for header files in the standard directories specific to
1895 C++, but do still search the other standard directories. (This option
1896 is used when building the C++ library.)
1899 In addition, these optimization, warning, and code generation options
1900 have meanings only for C++ programs:
1903 @item -fno-default-inline
1904 @opindex fno-default-inline
1905 Do not assume @samp{inline} for functions defined inside a class scope.
1906 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1907 functions will have linkage like inline functions; they just won't be
1910 @item -Wabi @r{(C++ only)}
1912 Warn when G++ generates code that is probably not compatible with the
1913 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1914 all such cases, there are probably some cases that are not warned about,
1915 even though G++ is generating incompatible code. There may also be
1916 cases where warnings are emitted even though the code that is generated
1919 You should rewrite your code to avoid these warnings if you are
1920 concerned about the fact that code generated by G++ may not be binary
1921 compatible with code generated by other compilers.
1923 The known incompatibilities at this point include:
1928 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1929 pack data into the same byte as a base class. For example:
1932 struct A @{ virtual void f(); int f1 : 1; @};
1933 struct B : public A @{ int f2 : 1; @};
1937 In this case, G++ will place @code{B::f2} into the same byte
1938 as@code{A::f1}; other compilers will not. You can avoid this problem
1939 by explicitly padding @code{A} so that its size is a multiple of the
1940 byte size on your platform; that will cause G++ and other compilers to
1941 layout @code{B} identically.
1944 Incorrect handling of tail-padding for virtual bases. G++ does not use
1945 tail padding when laying out virtual bases. For example:
1948 struct A @{ virtual void f(); char c1; @};
1949 struct B @{ B(); char c2; @};
1950 struct C : public A, public virtual B @{@};
1954 In this case, G++ will not place @code{B} into the tail-padding for
1955 @code{A}; other compilers will. You can avoid this problem by
1956 explicitly padding @code{A} so that its size is a multiple of its
1957 alignment (ignoring virtual base classes); that will cause G++ and other
1958 compilers to layout @code{C} identically.
1961 Incorrect handling of bit-fields with declared widths greater than that
1962 of their underlying types, when the bit-fields appear in a union. For
1966 union U @{ int i : 4096; @};
1970 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1971 union too small by the number of bits in an @code{int}.
1974 Empty classes can be placed at incorrect offsets. For example:
1984 struct C : public B, public A @{@};
1988 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1989 it should be placed at offset zero. G++ mistakenly believes that the
1990 @code{A} data member of @code{B} is already at offset zero.
1993 Names of template functions whose types involve @code{typename} or
1994 template template parameters can be mangled incorrectly.
1997 template <typename Q>
1998 void f(typename Q::X) @{@}
2000 template <template <typename> class Q>
2001 void f(typename Q<int>::X) @{@}
2005 Instantiations of these templates may be mangled incorrectly.
2009 @item -Wctor-dtor-privacy @r{(C++ only)}
2010 @opindex Wctor-dtor-privacy
2011 Warn when a class seems unusable because all the constructors or
2012 destructors in that class are private, and it has neither friends nor
2013 public static member functions.
2015 @item -Wnon-virtual-dtor @r{(C++ only)}
2016 @opindex Wnon-virtual-dtor
2017 Warn when a class has virtual functions and accessible non-virtual
2018 destructor, in which case it would be possible but unsafe to delete
2019 an instance of a derived class through a pointer to the base class.
2020 This warning is also enabled if -Weffc++ is specified.
2022 @item -Wreorder @r{(C++ only)}
2024 @cindex reordering, warning
2025 @cindex warning for reordering of member initializers
2026 Warn when the order of member initializers given in the code does not
2027 match the order in which they must be executed. For instance:
2033 A(): j (0), i (1) @{ @}
2037 The compiler will rearrange the member initializers for @samp{i}
2038 and @samp{j} to match the declaration order of the members, emitting
2039 a warning to that effect. This warning is enabled by @option{-Wall}.
2042 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2045 @item -Weffc++ @r{(C++ only)}
2047 Warn about violations of the following style guidelines from Scott Meyers'
2048 @cite{Effective C++} book:
2052 Item 11: Define a copy constructor and an assignment operator for classes
2053 with dynamically allocated memory.
2056 Item 12: Prefer initialization to assignment in constructors.
2059 Item 14: Make destructors virtual in base classes.
2062 Item 15: Have @code{operator=} return a reference to @code{*this}.
2065 Item 23: Don't try to return a reference when you must return an object.
2069 Also warn about violations of the following style guidelines from
2070 Scott Meyers' @cite{More Effective C++} book:
2074 Item 6: Distinguish between prefix and postfix forms of increment and
2075 decrement operators.
2078 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2082 When selecting this option, be aware that the standard library
2083 headers do not obey all of these guidelines; use @samp{grep -v}
2084 to filter out those warnings.
2086 @item -Wno-deprecated @r{(C++ only)}
2087 @opindex Wno-deprecated
2088 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2090 @item -Wstrict-null-sentinel @r{(C++ only)}
2091 @opindex Wstrict-null-sentinel
2092 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2093 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2094 to @code{__null}. Although it is a null pointer constant not a null pointer,
2095 it is guaranteed to of the same size as a pointer. But this use is
2096 not portable across different compilers.
2098 @item -Wno-non-template-friend @r{(C++ only)}
2099 @opindex Wno-non-template-friend
2100 Disable warnings when non-templatized friend functions are declared
2101 within a template. Since the advent of explicit template specification
2102 support in G++, if the name of the friend is an unqualified-id (i.e.,
2103 @samp{friend foo(int)}), the C++ language specification demands that the
2104 friend declare or define an ordinary, nontemplate function. (Section
2105 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2106 could be interpreted as a particular specialization of a templatized
2107 function. Because this non-conforming behavior is no longer the default
2108 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2109 check existing code for potential trouble spots and is on by default.
2110 This new compiler behavior can be turned off with
2111 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2112 but disables the helpful warning.
2114 @item -Wold-style-cast @r{(C++ only)}
2115 @opindex Wold-style-cast
2116 Warn if an old-style (C-style) cast to a non-void type is used within
2117 a C++ program. The new-style casts (@samp{dynamic_cast},
2118 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2119 less vulnerable to unintended effects and much easier to search for.
2121 @item -Woverloaded-virtual @r{(C++ only)}
2122 @opindex Woverloaded-virtual
2123 @cindex overloaded virtual fn, warning
2124 @cindex warning for overloaded virtual fn
2125 Warn when a function declaration hides virtual functions from a
2126 base class. For example, in:
2133 struct B: public A @{
2138 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2146 will fail to compile.
2148 @item -Wno-pmf-conversions @r{(C++ only)}
2149 @opindex Wno-pmf-conversions
2150 Disable the diagnostic for converting a bound pointer to member function
2153 @item -Wsign-promo @r{(C++ only)}
2154 @opindex Wsign-promo
2155 Warn when overload resolution chooses a promotion from unsigned or
2156 enumerated type to a signed type, over a conversion to an unsigned type of
2157 the same size. Previous versions of G++ would try to preserve
2158 unsignedness, but the standard mandates the current behavior.
2163 A& operator = (int);
2173 In this example, G++ will synthesize a default @samp{A& operator =
2174 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2177 @node Objective-C and Objective-C++ Dialect Options
2178 @section Options Controlling Objective-C and Objective-C++ Dialects
2180 @cindex compiler options, Objective-C and Objective-C++
2181 @cindex Objective-C and Objective-C++ options, command line
2182 @cindex options, Objective-C and Objective-C++
2183 (NOTE: This manual does not describe the Objective-C and Objective-C++
2184 languages themselves. See @xref{Standards,,Language Standards
2185 Supported by GCC}, for references.)
2187 This section describes the command-line options that are only meaningful
2188 for Objective-C and Objective-C++ programs, but you can also use most of
2189 the language-independent GNU compiler options.
2190 For example, you might compile a file @code{some_class.m} like this:
2193 gcc -g -fgnu-runtime -O -c some_class.m
2197 In this example, @option{-fgnu-runtime} is an option meant only for
2198 Objective-C and Objective-C++ programs; you can use the other options with
2199 any language supported by GCC@.
2201 Note that since Objective-C is an extension of the C language, Objective-C
2202 compilations may also use options specific to the C front-end (e.g.,
2203 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2204 C++-specific options (e.g., @option{-Wabi}).
2206 Here is a list of options that are @emph{only} for compiling Objective-C
2207 and Objective-C++ programs:
2210 @item -fconstant-string-class=@var{class-name}
2211 @opindex fconstant-string-class
2212 Use @var{class-name} as the name of the class to instantiate for each
2213 literal string specified with the syntax @code{@@"@dots{}"}. The default
2214 class name is @code{NXConstantString} if the GNU runtime is being used, and
2215 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2216 @option{-fconstant-cfstrings} option, if also present, will override the
2217 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2218 to be laid out as constant CoreFoundation strings.
2221 @opindex fgnu-runtime
2222 Generate object code compatible with the standard GNU Objective-C
2223 runtime. This is the default for most types of systems.
2225 @item -fnext-runtime
2226 @opindex fnext-runtime
2227 Generate output compatible with the NeXT runtime. This is the default
2228 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2229 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2232 @item -fno-nil-receivers
2233 @opindex fno-nil-receivers
2234 Assume that all Objective-C message dispatches (e.g.,
2235 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2236 is not @code{nil}. This allows for more efficient entry points in the runtime
2237 to be used. Currently, this option is only available in conjunction with
2238 the NeXT runtime on Mac OS X 10.3 and later.
2240 @item -fobjc-call-cxx-cdtors
2241 @opindex fobjc-call-cxx-cdtors
2242 For each Objective-C class, check if any of its instance variables is a
2243 C++ object with a non-trivial default constructor. If so, synthesize a
2244 special @code{- (id) .cxx_construct} instance method that will run
2245 non-trivial default constructors on any such instance variables, in order,
2246 and then return @code{self}. Similarly, check if any instance variable
2247 is a C++ object with a non-trivial destructor, and if so, synthesize a
2248 special @code{- (void) .cxx_destruct} method that will run
2249 all such default destructors, in reverse order.
2251 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2252 thusly generated will only operate on instance variables declared in the
2253 current Objective-C class, and not those inherited from superclasses. It
2254 is the responsibility of the Objective-C runtime to invoke all such methods
2255 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2256 will be invoked by the runtime immediately after a new object
2257 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2258 be invoked immediately before the runtime deallocates an object instance.
2260 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2261 support for invoking the @code{- (id) .cxx_construct} and
2262 @code{- (void) .cxx_destruct} methods.
2264 @item -fobjc-direct-dispatch
2265 @opindex fobjc-direct-dispatch
2266 Allow fast jumps to the message dispatcher. On Darwin this is
2267 accomplished via the comm page.
2269 @item -fobjc-exceptions
2270 @opindex fobjc-exceptions
2271 Enable syntactic support for structured exception handling in Objective-C,
2272 similar to what is offered by C++ and Java. This option is
2273 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2282 @@catch (AnObjCClass *exc) @{
2289 @@catch (AnotherClass *exc) @{
2292 @@catch (id allOthers) @{
2302 The @code{@@throw} statement may appear anywhere in an Objective-C or
2303 Objective-C++ program; when used inside of a @code{@@catch} block, the
2304 @code{@@throw} may appear without an argument (as shown above), in which case
2305 the object caught by the @code{@@catch} will be rethrown.
2307 Note that only (pointers to) Objective-C objects may be thrown and
2308 caught using this scheme. When an object is thrown, it will be caught
2309 by the nearest @code{@@catch} clause capable of handling objects of that type,
2310 analogously to how @code{catch} blocks work in C++ and Java. A
2311 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2312 any and all Objective-C exceptions not caught by previous @code{@@catch}
2315 The @code{@@finally} clause, if present, will be executed upon exit from the
2316 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2317 regardless of whether any exceptions are thrown, caught or rethrown
2318 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2319 of the @code{finally} clause in Java.
2321 There are several caveats to using the new exception mechanism:
2325 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2326 idioms provided by the @code{NSException} class, the new
2327 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2328 systems, due to additional functionality needed in the (NeXT) Objective-C
2332 As mentioned above, the new exceptions do not support handling
2333 types other than Objective-C objects. Furthermore, when used from
2334 Objective-C++, the Objective-C exception model does not interoperate with C++
2335 exceptions at this time. This means you cannot @code{@@throw} an exception
2336 from Objective-C and @code{catch} it in C++, or vice versa
2337 (i.e., @code{throw @dots{} @@catch}).
2340 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2341 blocks for thread-safe execution:
2344 @@synchronized (ObjCClass *guard) @{
2349 Upon entering the @code{@@synchronized} block, a thread of execution shall
2350 first check whether a lock has been placed on the corresponding @code{guard}
2351 object by another thread. If it has, the current thread shall wait until
2352 the other thread relinquishes its lock. Once @code{guard} becomes available,
2353 the current thread will place its own lock on it, execute the code contained in
2354 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2355 making @code{guard} available to other threads).
2357 Unlike Java, Objective-C does not allow for entire methods to be marked
2358 @code{@@synchronized}. Note that throwing exceptions out of
2359 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2360 to be unlocked properly.
2364 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2366 @item -freplace-objc-classes
2367 @opindex freplace-objc-classes
2368 Emit a special marker instructing @command{ld(1)} not to statically link in
2369 the resulting object file, and allow @command{dyld(1)} to load it in at
2370 run time instead. This is used in conjunction with the Fix-and-Continue
2371 debugging mode, where the object file in question may be recompiled and
2372 dynamically reloaded in the course of program execution, without the need
2373 to restart the program itself. Currently, Fix-and-Continue functionality
2374 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2379 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2380 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2381 compile time) with static class references that get initialized at load time,
2382 which improves run-time performance. Specifying the @option{-fzero-link} flag
2383 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2384 to be retained. This is useful in Zero-Link debugging mode, since it allows
2385 for individual class implementations to be modified during program execution.
2389 Dump interface declarations for all classes seen in the source file to a
2390 file named @file{@var{sourcename}.decl}.
2392 @item -Wassign-intercept
2393 @opindex Wassign-intercept
2394 Warn whenever an Objective-C assignment is being intercepted by the
2398 @opindex Wno-protocol
2399 If a class is declared to implement a protocol, a warning is issued for
2400 every method in the protocol that is not implemented by the class. The
2401 default behavior is to issue a warning for every method not explicitly
2402 implemented in the class, even if a method implementation is inherited
2403 from the superclass. If you use the @option{-Wno-protocol} option, then
2404 methods inherited from the superclass are considered to be implemented,
2405 and no warning is issued for them.
2409 Warn if multiple methods of different types for the same selector are
2410 found during compilation. The check is performed on the list of methods
2411 in the final stage of compilation. Additionally, a check is performed
2412 for each selector appearing in a @code{@@selector(@dots{})}
2413 expression, and a corresponding method for that selector has been found
2414 during compilation. Because these checks scan the method table only at
2415 the end of compilation, these warnings are not produced if the final
2416 stage of compilation is not reached, for example because an error is
2417 found during compilation, or because the @option{-fsyntax-only} option is
2420 @item -Wstrict-selector-match
2421 @opindex Wstrict-selector-match
2422 Warn if multiple methods with differing argument and/or return types are
2423 found for a given selector when attempting to send a message using this
2424 selector to a receiver of type @code{id} or @code{Class}. When this flag
2425 is off (which is the default behavior), the compiler will omit such warnings
2426 if any differences found are confined to types which share the same size
2429 @item -Wundeclared-selector
2430 @opindex Wundeclared-selector
2431 Warn if a @code{@@selector(@dots{})} expression referring to an
2432 undeclared selector is found. A selector is considered undeclared if no
2433 method with that name has been declared before the
2434 @code{@@selector(@dots{})} expression, either explicitly in an
2435 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2436 an @code{@@implementation} section. This option always performs its
2437 checks as soon as a @code{@@selector(@dots{})} expression is found,
2438 while @option{-Wselector} only performs its checks in the final stage of
2439 compilation. This also enforces the coding style convention
2440 that methods and selectors must be declared before being used.
2442 @item -print-objc-runtime-info
2443 @opindex print-objc-runtime-info
2444 Generate C header describing the largest structure that is passed by
2449 @node Language Independent Options
2450 @section Options to Control Diagnostic Messages Formatting
2451 @cindex options to control diagnostics formatting
2452 @cindex diagnostic messages
2453 @cindex message formatting
2455 Traditionally, diagnostic messages have been formatted irrespective of
2456 the output device's aspect (e.g.@: its width, @dots{}). The options described
2457 below can be used to control the diagnostic messages formatting
2458 algorithm, e.g.@: how many characters per line, how often source location
2459 information should be reported. Right now, only the C++ front end can
2460 honor these options. However it is expected, in the near future, that
2461 the remaining front ends would be able to digest them correctly.
2464 @item -fmessage-length=@var{n}
2465 @opindex fmessage-length
2466 Try to format error messages so that they fit on lines of about @var{n}
2467 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2468 the front ends supported by GCC@. If @var{n} is zero, then no
2469 line-wrapping will be done; each error message will appear on a single
2472 @opindex fdiagnostics-show-location
2473 @item -fdiagnostics-show-location=once
2474 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2475 reporter to emit @emph{once} source location information; that is, in
2476 case the message is too long to fit on a single physical line and has to
2477 be wrapped, the source location won't be emitted (as prefix) again,
2478 over and over, in subsequent continuation lines. This is the default
2481 @item -fdiagnostics-show-location=every-line
2482 Only meaningful in line-wrapping mode. Instructs the diagnostic
2483 messages reporter to emit the same source location information (as
2484 prefix) for physical lines that result from the process of breaking
2485 a message which is too long to fit on a single line.
2487 @item -fdiagnostics-show-option
2488 @opindex fdiagnostics-show-option
2489 This option instructs the diagnostic machinery to add text to each
2490 diagnostic emitted, which indicates which command line option directly
2491 controls that diagnostic, when such an option is known to the
2492 diagnostic machinery.
2494 @item -Wcoverage-mismatch
2495 @opindex Wcoverage-mismatch
2496 Warn if feedback profiles do not match when using the
2497 @option{-fprofile-use} option.
2498 If a source file was changed between @option{-fprofile-gen} and
2499 @option{-fprofile-use}, the files with the profile feedback can fail
2500 to match the source file and GCC can not use the profile feedback
2501 information. By default, GCC emits an error message in this case.
2502 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2503 error. GCC does not use appropriate feedback profiles, so using this
2504 option can result in poorly optimized code. This option is useful
2505 only in the case of very minor changes such as bug fixes to an
2510 @node Warning Options
2511 @section Options to Request or Suppress Warnings
2512 @cindex options to control warnings
2513 @cindex warning messages
2514 @cindex messages, warning
2515 @cindex suppressing warnings
2517 Warnings are diagnostic messages that report constructions which
2518 are not inherently erroneous but which are risky or suggest there
2519 may have been an error.
2521 You can request many specific warnings with options beginning @samp{-W},
2522 for example @option{-Wimplicit} to request warnings on implicit
2523 declarations. Each of these specific warning options also has a
2524 negative form beginning @samp{-Wno-} to turn off warnings;
2525 for example, @option{-Wno-implicit}. This manual lists only one of the
2526 two forms, whichever is not the default.
2528 The following options control the amount and kinds of warnings produced
2529 by GCC; for further, language-specific options also refer to
2530 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2534 @cindex syntax checking
2536 @opindex fsyntax-only
2537 Check the code for syntax errors, but don't do anything beyond that.
2541 Issue all the warnings demanded by strict ISO C and ISO C++;
2542 reject all programs that use forbidden extensions, and some other
2543 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2544 version of the ISO C standard specified by any @option{-std} option used.
2546 Valid ISO C and ISO C++ programs should compile properly with or without
2547 this option (though a rare few will require @option{-ansi} or a
2548 @option{-std} option specifying the required version of ISO C)@. However,
2549 without this option, certain GNU extensions and traditional C and C++
2550 features are supported as well. With this option, they are rejected.
2552 @option{-pedantic} does not cause warning messages for use of the
2553 alternate keywords whose names begin and end with @samp{__}. Pedantic
2554 warnings are also disabled in the expression that follows
2555 @code{__extension__}. However, only system header files should use
2556 these escape routes; application programs should avoid them.
2557 @xref{Alternate Keywords}.
2559 Some users try to use @option{-pedantic} to check programs for strict ISO
2560 C conformance. They soon find that it does not do quite what they want:
2561 it finds some non-ISO practices, but not all---only those for which
2562 ISO C @emph{requires} a diagnostic, and some others for which
2563 diagnostics have been added.
2565 A feature to report any failure to conform to ISO C might be useful in
2566 some instances, but would require considerable additional work and would
2567 be quite different from @option{-pedantic}. We don't have plans to
2568 support such a feature in the near future.
2570 Where the standard specified with @option{-std} represents a GNU
2571 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2572 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2573 extended dialect is based. Warnings from @option{-pedantic} are given
2574 where they are required by the base standard. (It would not make sense
2575 for such warnings to be given only for features not in the specified GNU
2576 C dialect, since by definition the GNU dialects of C include all
2577 features the compiler supports with the given option, and there would be
2578 nothing to warn about.)
2580 @item -pedantic-errors
2581 @opindex pedantic-errors
2582 Like @option{-pedantic}, except that errors are produced rather than
2587 Inhibit all warning messages.
2591 Inhibit warning messages about the use of @samp{#import}.
2593 @item -Wchar-subscripts
2594 @opindex Wchar-subscripts
2595 Warn if an array subscript has type @code{char}. This is a common cause
2596 of error, as programmers often forget that this type is signed on some
2598 This warning is enabled by @option{-Wall}.
2602 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2603 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2604 This warning is enabled by @option{-Wall}.
2606 @item -Wfatal-errors
2607 @opindex Wfatal-errors
2608 This option causes the compiler to abort compilation on the first error
2609 occurred rather than trying to keep going and printing further error
2614 @opindex ffreestanding
2615 @opindex fno-builtin
2616 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2617 the arguments supplied have types appropriate to the format string
2618 specified, and that the conversions specified in the format string make
2619 sense. This includes standard functions, and others specified by format
2620 attributes (@pxref{Function Attributes}), in the @code{printf},
2621 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2622 not in the C standard) families (or other target-specific families).
2623 Which functions are checked without format attributes having been
2624 specified depends on the standard version selected, and such checks of
2625 functions without the attribute specified are disabled by
2626 @option{-ffreestanding} or @option{-fno-builtin}.
2628 The formats are checked against the format features supported by GNU
2629 libc version 2.2. These include all ISO C90 and C99 features, as well
2630 as features from the Single Unix Specification and some BSD and GNU
2631 extensions. Other library implementations may not support all these
2632 features; GCC does not support warning about features that go beyond a
2633 particular library's limitations. However, if @option{-pedantic} is used
2634 with @option{-Wformat}, warnings will be given about format features not
2635 in the selected standard version (but not for @code{strfmon} formats,
2636 since those are not in any version of the C standard). @xref{C Dialect
2637 Options,,Options Controlling C Dialect}.
2639 Since @option{-Wformat} also checks for null format arguments for
2640 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2642 @option{-Wformat} is included in @option{-Wall}. For more control over some
2643 aspects of format checking, the options @option{-Wformat-y2k},
2644 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2645 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2646 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2649 @opindex Wformat-y2k
2650 If @option{-Wformat} is specified, also warn about @code{strftime}
2651 formats which may yield only a two-digit year.
2653 @item -Wno-format-extra-args
2654 @opindex Wno-format-extra-args
2655 If @option{-Wformat} is specified, do not warn about excess arguments to a
2656 @code{printf} or @code{scanf} format function. The C standard specifies
2657 that such arguments are ignored.
2659 Where the unused arguments lie between used arguments that are
2660 specified with @samp{$} operand number specifications, normally
2661 warnings are still given, since the implementation could not know what
2662 type to pass to @code{va_arg} to skip the unused arguments. However,
2663 in the case of @code{scanf} formats, this option will suppress the
2664 warning if the unused arguments are all pointers, since the Single
2665 Unix Specification says that such unused arguments are allowed.
2667 @item -Wno-format-zero-length
2668 @opindex Wno-format-zero-length
2669 If @option{-Wformat} is specified, do not warn about zero-length formats.
2670 The C standard specifies that zero-length formats are allowed.
2672 @item -Wformat-nonliteral
2673 @opindex Wformat-nonliteral
2674 If @option{-Wformat} is specified, also warn if the format string is not a
2675 string literal and so cannot be checked, unless the format function
2676 takes its format arguments as a @code{va_list}.
2678 @item -Wformat-security
2679 @opindex Wformat-security
2680 If @option{-Wformat} is specified, also warn about uses of format
2681 functions that represent possible security problems. At present, this
2682 warns about calls to @code{printf} and @code{scanf} functions where the
2683 format string is not a string literal and there are no format arguments,
2684 as in @code{printf (foo);}. This may be a security hole if the format
2685 string came from untrusted input and contains @samp{%n}. (This is
2686 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2687 in future warnings may be added to @option{-Wformat-security} that are not
2688 included in @option{-Wformat-nonliteral}.)
2692 Enable @option{-Wformat} plus format checks not included in
2693 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2694 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2698 Warn about passing a null pointer for arguments marked as
2699 requiring a non-null value by the @code{nonnull} function attribute.
2701 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2702 can be disabled with the @option{-Wno-nonnull} option.
2704 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2706 Warn about uninitialized variables which are initialized with themselves.
2707 Note this option can only be used with the @option{-Wuninitialized} option,
2708 which in turn only works with @option{-O1} and above.
2710 For example, GCC will warn about @code{i} being uninitialized in the
2711 following snippet only when @option{-Winit-self} has been specified:
2722 @item -Wimplicit-int
2723 @opindex Wimplicit-int
2724 Warn when a declaration does not specify a type.
2725 This warning is enabled by @option{-Wall}.
2727 @item -Wimplicit-function-declaration
2728 @opindex Wimplicit-function-declaration
2729 @opindex Wno-implicit-function-declaration
2730 Give a warning whenever a function is used before being declared. In
2731 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2732 enabled by default and it is made into an error by
2733 @option{-pedantic-errors}. This warning is also enabled by
2738 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2739 This warning is enabled by @option{-Wall}.
2743 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2744 function with external linkage, returning int, taking either zero
2745 arguments, two, or three arguments of appropriate types.
2746 This warning is enabled by @option{-Wall}.
2748 @item -Wmissing-braces
2749 @opindex Wmissing-braces
2750 Warn if an aggregate or union initializer is not fully bracketed. In
2751 the following example, the initializer for @samp{a} is not fully
2752 bracketed, but that for @samp{b} is fully bracketed.
2755 int a[2][2] = @{ 0, 1, 2, 3 @};
2756 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2759 This warning is enabled by @option{-Wall}.
2761 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2762 @opindex Wmissing-include-dirs
2763 Warn if a user-supplied include directory does not exist.
2766 @opindex Wparentheses
2767 Warn if parentheses are omitted in certain contexts, such
2768 as when there is an assignment in a context where a truth value
2769 is expected, or when operators are nested whose precedence people
2770 often get confused about.
2772 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2773 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2774 interpretation from that of ordinary mathematical notation.
2776 Also warn about constructions where there may be confusion to which
2777 @code{if} statement an @code{else} branch belongs. Here is an example of
2792 In C/C++, every @code{else} branch belongs to the innermost possible
2793 @code{if} statement, which in this example is @code{if (b)}. This is
2794 often not what the programmer expected, as illustrated in the above
2795 example by indentation the programmer chose. When there is the
2796 potential for this confusion, GCC will issue a warning when this flag
2797 is specified. To eliminate the warning, add explicit braces around
2798 the innermost @code{if} statement so there is no way the @code{else}
2799 could belong to the enclosing @code{if}. The resulting code would
2816 This warning is enabled by @option{-Wall}.
2818 @item -Wsequence-point
2819 @opindex Wsequence-point
2820 Warn about code that may have undefined semantics because of violations
2821 of sequence point rules in the C and C++ standards.
2823 The C and C++ standards defines the order in which expressions in a C/C++
2824 program are evaluated in terms of @dfn{sequence points}, which represent
2825 a partial ordering between the execution of parts of the program: those
2826 executed before the sequence point, and those executed after it. These
2827 occur after the evaluation of a full expression (one which is not part
2828 of a larger expression), after the evaluation of the first operand of a
2829 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2830 function is called (but after the evaluation of its arguments and the
2831 expression denoting the called function), and in certain other places.
2832 Other than as expressed by the sequence point rules, the order of
2833 evaluation of subexpressions of an expression is not specified. All
2834 these rules describe only a partial order rather than a total order,
2835 since, for example, if two functions are called within one expression
2836 with no sequence point between them, the order in which the functions
2837 are called is not specified. However, the standards committee have
2838 ruled that function calls do not overlap.
2840 It is not specified when between sequence points modifications to the
2841 values of objects take effect. Programs whose behavior depends on this
2842 have undefined behavior; the C and C++ standards specify that ``Between
2843 the previous and next sequence point an object shall have its stored
2844 value modified at most once by the evaluation of an expression.
2845 Furthermore, the prior value shall be read only to determine the value
2846 to be stored.''. If a program breaks these rules, the results on any
2847 particular implementation are entirely unpredictable.
2849 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2850 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2851 diagnosed by this option, and it may give an occasional false positive
2852 result, but in general it has been found fairly effective at detecting
2853 this sort of problem in programs.
2855 The standard is worded confusingly, therefore there is some debate
2856 over the precise meaning of the sequence point rules in subtle cases.
2857 Links to discussions of the problem, including proposed formal
2858 definitions, may be found on the GCC readings page, at
2859 @w{@uref{http://gcc.gnu.org/readings.html}}.
2861 This warning is enabled by @option{-Wall} for C and C++.
2864 @opindex Wreturn-type
2865 @opindex Wno-return-type
2866 Warn whenever a function is defined with a return-type that defaults
2867 to @code{int}. Also warn about any @code{return} statement with no
2868 return-value in a function whose return-type is not @code{void}
2869 (falling off the end of the function body is considered returning
2870 without a value), and about a @code{return} statement with a
2871 expression in a function whose return-type is @code{void}.
2873 Also warn if the return type of a function has a type qualifier
2874 such as @code{const}. For ISO C such a type qualifier has no effect,
2875 since the value returned by a function is not an lvalue.
2876 For C++, the warning is only emitted for scalar types or @code{void}.
2877 ISO C prohibits qualified @code{void} return types on function
2878 definitions, so such return types always receive a warning
2879 even without this option.
2881 For C++, a function without return type always produces a diagnostic
2882 message, even when @option{-Wno-return-type} is specified. The only
2883 exceptions are @samp{main} and functions defined in system headers.
2885 This warning is enabled by @option{-Wall}.
2889 Warn whenever a @code{switch} statement has an index of enumerated type
2890 and lacks a @code{case} for one or more of the named codes of that
2891 enumeration. (The presence of a @code{default} label prevents this
2892 warning.) @code{case} labels outside the enumeration range also
2893 provoke warnings when this option is used.
2894 This warning is enabled by @option{-Wall}.
2896 @item -Wswitch-default
2897 @opindex Wswitch-switch
2898 Warn whenever a @code{switch} statement does not have a @code{default}
2902 @opindex Wswitch-enum
2903 Warn whenever a @code{switch} statement has an index of enumerated type
2904 and lacks a @code{case} for one or more of the named codes of that
2905 enumeration. @code{case} labels outside the enumeration range also
2906 provoke warnings when this option is used.
2910 Warn if any trigraphs are encountered that might change the meaning of
2911 the program (trigraphs within comments are not warned about).
2912 This warning is enabled by @option{-Wall}.
2914 @item -Wunused-function
2915 @opindex Wunused-function
2916 Warn whenever a static function is declared but not defined or a
2917 non-inline static function is unused.
2918 This warning is enabled by @option{-Wall}.
2920 @item -Wunused-label
2921 @opindex Wunused-label
2922 Warn whenever a label is declared but not used.
2923 This warning is enabled by @option{-Wall}.
2925 To suppress this warning use the @samp{unused} attribute
2926 (@pxref{Variable Attributes}).
2928 @item -Wunused-parameter
2929 @opindex Wunused-parameter
2930 Warn whenever a function parameter is unused aside from its declaration.
2932 To suppress this warning use the @samp{unused} attribute
2933 (@pxref{Variable Attributes}).
2935 @item -Wunused-variable
2936 @opindex Wunused-variable
2937 Warn whenever a local variable or non-constant static variable is unused
2938 aside from its declaration.
2939 This warning is enabled by @option{-Wall}.
2941 To suppress this warning use the @samp{unused} attribute
2942 (@pxref{Variable Attributes}).
2944 @item -Wunused-value
2945 @opindex Wunused-value
2946 Warn whenever a statement computes a result that is explicitly not
2947 used. To suppress this warning cast the unused expression to
2948 @samp{void}. This includes an expression-statement or the left-hand
2949 side of a comma expression that contains no side effects. For example,
2950 an expression such as @samp{x[i,j]} will cause a warning, while
2951 @samp{x[(void)i,j]} will not.
2953 This warning is enabled by @option{-Wall}.
2957 All the above @option{-Wunused} options combined.
2959 In order to get a warning about an unused function parameter, you must
2960 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2961 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2963 @item -Wuninitialized
2964 @opindex Wuninitialized
2965 Warn if an automatic variable is used without first being initialized or
2966 if a variable may be clobbered by a @code{setjmp} call.
2968 These warnings are possible only in optimizing compilation,
2969 because they require data flow information that is computed only
2970 when optimizing. If you do not specify @option{-O}, you will not get
2971 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2972 requiring @option{-O}.
2974 If you want to warn about code which uses the uninitialized value of the
2975 variable in its own initializer, use the @option{-Winit-self} option.
2977 These warnings occur for individual uninitialized or clobbered
2978 elements of structure, union or array variables as well as for
2979 variables which are uninitialized or clobbered as a whole. They do
2980 not occur for variables or elements declared @code{volatile}. Because
2981 these warnings depend on optimization, the exact variables or elements
2982 for which there are warnings will depend on the precise optimization
2983 options and version of GCC used.
2985 Note that there may be no warning about a variable that is used only
2986 to compute a value that itself is never used, because such
2987 computations may be deleted by data flow analysis before the warnings
2990 These warnings are made optional because GCC is not smart
2991 enough to see all the reasons why the code might be correct
2992 despite appearing to have an error. Here is one example of how
3013 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3014 always initialized, but GCC doesn't know this. Here is
3015 another common case:
3020 if (change_y) save_y = y, y = new_y;
3022 if (change_y) y = save_y;
3027 This has no bug because @code{save_y} is used only if it is set.
3029 @cindex @code{longjmp} warnings
3030 This option also warns when a non-volatile automatic variable might be
3031 changed by a call to @code{longjmp}. These warnings as well are possible
3032 only in optimizing compilation.
3034 The compiler sees only the calls to @code{setjmp}. It cannot know
3035 where @code{longjmp} will be called; in fact, a signal handler could
3036 call it at any point in the code. As a result, you may get a warning
3037 even when there is in fact no problem because @code{longjmp} cannot
3038 in fact be called at the place which would cause a problem.
3040 Some spurious warnings can be avoided if you declare all the functions
3041 you use that never return as @code{noreturn}. @xref{Function
3044 This warning is enabled by @option{-Wall}.
3046 @item -Wunknown-pragmas
3047 @opindex Wunknown-pragmas
3048 @cindex warning for unknown pragmas
3049 @cindex unknown pragmas, warning
3050 @cindex pragmas, warning of unknown
3051 Warn when a #pragma directive is encountered which is not understood by
3052 GCC@. If this command line option is used, warnings will even be issued
3053 for unknown pragmas in system header files. This is not the case if
3054 the warnings were only enabled by the @option{-Wall} command line option.
3057 @opindex Wno-pragmas
3059 Do not warn about misuses of pragmas, such as incorrect parameters,
3060 invalid syntax, or conflicts between pragmas. See also
3061 @samp{-Wunknown-pragmas}.
3063 @item -Wstrict-aliasing
3064 @opindex Wstrict-aliasing
3065 This option is only active when @option{-fstrict-aliasing} is active.
3066 It warns about code which might break the strict aliasing rules that the
3067 compiler is using for optimization. The warning does not catch all
3068 cases, but does attempt to catch the more common pitfalls. It is
3069 included in @option{-Wall}.
3070 It is equivalent to -Wstrict-aliasing=3
3072 @item -Wstrict-aliasing=n
3073 @opindex Wstrict-aliasing=n
3074 This option is only active when @option{-fstrict-aliasing} is active.
3075 It warns about code which might break the strict aliasing rules that the
3076 compiler is using for optimization.
3077 Higher levels correspond to higher accuracy (fewer false positives).
3078 Higher levels also correspond to more effort, similar to the way -O works.
3079 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3082 Level 1: Most aggressive, quick, least accurate.
3083 Possibly useful when higher levels
3084 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3085 false negatives. However, it has many false positives.
3086 Warns for all pointer conversions between possibly incompatible types,
3087 even if never dereferenced. Runs in the frontend only.
3089 Level 2: Aggressive, quick, not too precise.
3090 May still have many false positives (not as many as level 1 though),
3091 and few false negatives (but possibly more than level 1).
3092 Unlike level 1, it only warns when an address is taken. Warns about
3093 incomplete types. Runs in the frontend only.
3095 Level 3 (default for @option{-Wstrict-aliasing}):
3096 Should have very few false positives and few false
3097 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3098 Takes care of the common punn+dereference pattern in the frontend:
3099 @code{*(int*)&some_float}.
3100 If optimization is enabled, it also runs in the backend, where it deals
3101 with multiple statement cases using flow-sensitive points-to information.
3102 Only warns when the converted pointer is dereferenced.
3103 Does not warn about incomplete types.
3105 @item -Wstrict-overflow
3106 @item -Wstrict-overflow=@var{n}
3107 @opindex Wstrict-overflow
3108 This option is only active when @option{-fstrict-overflow} is active.
3109 It warns about cases where the compiler optimizes based on the
3110 assumption that signed overflow does not occur. Note that it does not
3111 warn about all cases where the code might overflow: it only warns
3112 about cases where the compiler implements some optimization. Thus
3113 this warning depends on the optimization level.
3115 An optimization which assumes that signed overflow does not occur is
3116 perfectly safe if the values of the variables involved are such that
3117 overflow never does, in fact, occur. Therefore this warning can
3118 easily give a false positive: a warning about code which is not
3119 actually a problem. To help focus on important issues, several
3120 warning levels are defined. No warnings are issued for the use of
3121 undefined signed overflow when estimating how many iterations a loop
3122 will require, in particular when determining whether a loop will be
3126 @item -Wstrict-overflow=1
3127 Warn about cases which are both questionable and easy to avoid. For
3128 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3129 compiler will simplify this to @code{1}. This level of
3130 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3131 are not, and must be explicitly requested.
3133 @item -Wstrict-overflow=2
3134 Also warn about other cases where a comparison is simplified to a
3135 constant. For example: @code{abs (x) >= 0}. This can only be
3136 simplified when @option{-fstrict-overflow} is in effect, because
3137 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3138 zero. @option{-Wstrict-overflow} (with no level) is the same as
3139 @option{-Wstrict-overflow=2}.
3141 @item -Wstrict-overflow=3
3142 Also warn about other cases where a comparison is simplified. For
3143 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3145 @item -Wstrict-overflow=4
3146 Also warn about other simplifications not covered by the above cases.
3147 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3149 @item -Wstrict-overflow=5
3150 Also warn about cases where the compiler reduces the magnitude of a
3151 constant involved in a comparison. For example: @code{x + 2 > y} will
3152 be simplified to @code{x + 1 >= y}. This is reported only at the
3153 highest warning level because this simplification applies to many
3154 comparisons, so this warning level will give a very large number of
3158 @item -Warray-bounds
3159 @opindex Wno-array-bounds
3160 @opindex Warray-bounds
3161 This option is only active when @option{-ftree-vrp} is active
3162 (default for -O2 and above). It warns about subscripts to arrays
3163 that are always out of bounds. This warning is enabled by @option{-Wall}.
3167 All of the above @samp{-W} options combined. This enables all the
3168 warnings about constructions that some users consider questionable, and
3169 that are easy to avoid (or modify to prevent the warning), even in
3170 conjunction with macros. This also enables some language-specific
3171 warnings described in @ref{C++ Dialect Options} and
3172 @ref{Objective-C and Objective-C++ Dialect Options}.
3175 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
3176 Some of them warn about constructions that users generally do not
3177 consider questionable, but which occasionally you might wish to check
3178 for; others warn about constructions that are necessary or hard to avoid
3179 in some cases, and there is no simple way to modify the code to suppress
3186 (This option used to be called @option{-W}. The older name is still
3187 supported, but the newer name is more descriptive.) Print extra warning
3188 messages for these events:
3192 Warn if a comparison is always true or always false due to the limited
3193 range of the data type, but do not warn for constant expressions. For
3194 example, warn if an unsigned variable is compared against zero with
3195 @samp{<} or @samp{>=}. This warning can be independently controlled
3196 by @option{-Wtype-limits}.
3199 Storage-class specifiers like @code{static} are not the first things
3200 in a declaration. According to the C Standard, this usage is
3201 obsolescent. This warning can be independently controlled by
3202 @option{-Wold-style-declaration}.
3205 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
3209 A comparison between signed and unsigned values could produce an
3210 incorrect result when the signed value is converted to unsigned.
3211 (But don't warn if @option{-Wno-sign-compare} is also specified.)
3214 An aggregate has an initializer which does not initialize all members.
3215 This warning can be independently controlled by
3216 @option{-Wmissing-field-initializers}.
3219 An initialized field without side effects is overridden when using
3220 designated initializers (@pxref{Designated Inits, , Designated
3221 Initializers}). This warning can be independently controlled by
3222 @option{-Woverride-init}.
3225 A function parameter is declared without a type specifier in K&R-style
3226 functions. This warning can be independently controlled by
3227 @option{-Wmissing-parameter-type}.
3230 An empty body occurs in an @samp{if}, @samp{else} or
3231 @samp{do while} statement. This warning can be independently
3232 controlled by @option{-Wempty-body}.
3234 @item @r{(C++ only)}
3235 An empty body occurs in a @samp{while} or @samp{for} statement with no
3236 whitespacing before the semicolon. This warning can be independently
3237 controlled by @option{-Wempty-body}.
3240 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3241 @samp{>}, or @samp{>=}.
3244 A variable might be changed by @samp{longjmp} or @samp{vfork}.
3245 This warning can be independently controlled by @option{-Wclobbered}.
3247 @item @r{(C++ only)}
3248 An enumerator and a non-enumerator both appear in a conditional expression.
3250 @item @r{(C++ only)}
3251 A non-static reference or non-static @samp{const} member appears in a
3252 class without constructors.
3254 @item @r{(C++ only)}
3255 Ambiguous virtual bases.
3257 @item @r{(C++ only)}
3258 Subscripting an array which has been declared @samp{register}.
3260 @item @r{(C++ only)}
3261 Taking the address of a variable which has been declared @samp{register}.
3263 @item @r{(C++ only)}
3264 A base class is not initialized in a derived class' copy constructor.
3267 @item -Wno-div-by-zero
3268 @opindex Wno-div-by-zero
3269 @opindex Wdiv-by-zero
3270 Do not warn about compile-time integer division by zero. Floating point
3271 division by zero is not warned about, as it can be a legitimate way of
3272 obtaining infinities and NaNs.
3274 @item -Wsystem-headers
3275 @opindex Wsystem-headers
3276 @cindex warnings from system headers
3277 @cindex system headers, warnings from
3278 Print warning messages for constructs found in system header files.
3279 Warnings from system headers are normally suppressed, on the assumption
3280 that they usually do not indicate real problems and would only make the
3281 compiler output harder to read. Using this command line option tells
3282 GCC to emit warnings from system headers as if they occurred in user
3283 code. However, note that using @option{-Wall} in conjunction with this
3284 option will @emph{not} warn about unknown pragmas in system
3285 headers---for that, @option{-Wunknown-pragmas} must also be used.
3288 @opindex Wfloat-equal
3289 Warn if floating point values are used in equality comparisons.
3291 The idea behind this is that sometimes it is convenient (for the
3292 programmer) to consider floating-point values as approximations to
3293 infinitely precise real numbers. If you are doing this, then you need
3294 to compute (by analyzing the code, or in some other way) the maximum or
3295 likely maximum error that the computation introduces, and allow for it
3296 when performing comparisons (and when producing output, but that's a
3297 different problem). In particular, instead of testing for equality, you
3298 would check to see whether the two values have ranges that overlap; and
3299 this is done with the relational operators, so equality comparisons are
3302 @item -Wtraditional @r{(C only)}
3303 @opindex Wtraditional
3304 Warn about certain constructs that behave differently in traditional and
3305 ISO C@. Also warn about ISO C constructs that have no traditional C
3306 equivalent, and/or problematic constructs which should be avoided.
3310 Macro parameters that appear within string literals in the macro body.
3311 In traditional C macro replacement takes place within string literals,
3312 but does not in ISO C@.
3315 In traditional C, some preprocessor directives did not exist.
3316 Traditional preprocessors would only consider a line to be a directive
3317 if the @samp{#} appeared in column 1 on the line. Therefore
3318 @option{-Wtraditional} warns about directives that traditional C
3319 understands but would ignore because the @samp{#} does not appear as the
3320 first character on the line. It also suggests you hide directives like
3321 @samp{#pragma} not understood by traditional C by indenting them. Some
3322 traditional implementations would not recognize @samp{#elif}, so it
3323 suggests avoiding it altogether.
3326 A function-like macro that appears without arguments.
3329 The unary plus operator.
3332 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3333 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3334 constants.) Note, these suffixes appear in macros defined in the system
3335 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3336 Use of these macros in user code might normally lead to spurious
3337 warnings, however GCC's integrated preprocessor has enough context to
3338 avoid warning in these cases.
3341 A function declared external in one block and then used after the end of
3345 A @code{switch} statement has an operand of type @code{long}.
3348 A non-@code{static} function declaration follows a @code{static} one.
3349 This construct is not accepted by some traditional C compilers.
3352 The ISO type of an integer constant has a different width or
3353 signedness from its traditional type. This warning is only issued if
3354 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3355 typically represent bit patterns, are not warned about.
3358 Usage of ISO string concatenation is detected.
3361 Initialization of automatic aggregates.
3364 Identifier conflicts with labels. Traditional C lacks a separate
3365 namespace for labels.
3368 Initialization of unions. If the initializer is zero, the warning is
3369 omitted. This is done under the assumption that the zero initializer in
3370 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3371 initializer warnings and relies on default initialization to zero in the
3375 Conversions by prototypes between fixed/floating point values and vice
3376 versa. The absence of these prototypes when compiling with traditional
3377 C would cause serious problems. This is a subset of the possible
3378 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3381 Use of ISO C style function definitions. This warning intentionally is
3382 @emph{not} issued for prototype declarations or variadic functions
3383 because these ISO C features will appear in your code when using
3384 libiberty's traditional C compatibility macros, @code{PARAMS} and
3385 @code{VPARAMS}. This warning is also bypassed for nested functions
3386 because that feature is already a GCC extension and thus not relevant to
3387 traditional C compatibility.
3390 @item -Wtraditional-conversion @r{(C only)}
3391 @opindex Wtraditional-conversion
3392 Warn if a prototype causes a type conversion that is different from what
3393 would happen to the same argument in the absence of a prototype. This
3394 includes conversions of fixed point to floating and vice versa, and
3395 conversions changing the width or signedness of a fixed point argument
3396 except when the same as the default promotion.
3398 @item -Wdeclaration-after-statement @r{(C only)}
3399 @opindex Wdeclaration-after-statement
3400 Warn when a declaration is found after a statement in a block. This
3401 construct, known from C++, was introduced with ISO C99 and is by default
3402 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3403 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3407 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3409 @item -Wno-endif-labels
3410 @opindex Wno-endif-labels
3411 @opindex Wendif-labels
3412 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3416 Warn whenever a local variable shadows another local variable, parameter or
3417 global variable or whenever a built-in function is shadowed.
3419 @item -Wlarger-than-@var{len}
3420 @opindex Wlarger-than
3421 Warn whenever an object of larger than @var{len} bytes is defined.
3423 @item -Wunsafe-loop-optimizations
3424 @opindex Wunsafe-loop-optimizations
3425 Warn if the loop cannot be optimized because the compiler could not
3426 assume anything on the bounds of the loop indices. With
3427 @option{-funsafe-loop-optimizations} warn if the compiler made
3430 @item -Wpointer-arith
3431 @opindex Wpointer-arith
3432 Warn about anything that depends on the ``size of'' a function type or
3433 of @code{void}. GNU C assigns these types a size of 1, for
3434 convenience in calculations with @code{void *} pointers and pointers
3435 to functions. In C++, warn also when an arithmetic operation involves
3436 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3439 @opindex Wtype-limits
3440 @opindex Wno-type-limits
3441 Warn if a comparison is always true or always false due to the limited
3442 range of the data type, but do not warn for constant expressions. For
3443 example, warn if an unsigned variable is compared against zero with
3444 @samp{<} or @samp{>=}. This warning is also enabled by
3447 @item -Wbad-function-cast @r{(C only)}
3448 @opindex Wbad-function-cast
3449 Warn whenever a function call is cast to a non-matching type.
3450 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3453 Warn about ISO C constructs that are outside of the common subset of
3454 ISO C and ISO C++, e.g.@: request for implicit conversion from
3455 @code{void *} to a pointer to non-@code{void} type.
3457 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3458 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3459 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3460 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3464 Warn whenever a pointer is cast so as to remove a type qualifier from
3465 the target type. For example, warn if a @code{const char *} is cast
3466 to an ordinary @code{char *}.
3469 @opindex Wcast-align
3470 Warn whenever a pointer is cast such that the required alignment of the
3471 target is increased. For example, warn if a @code{char *} is cast to
3472 an @code{int *} on machines where integers can only be accessed at
3473 two- or four-byte boundaries.
3475 @item -Wwrite-strings
3476 @opindex Wwrite-strings
3477 When compiling C, give string constants the type @code{const
3478 char[@var{length}]} so that
3479 copying the address of one into a non-@code{const} @code{char *}
3480 pointer will get a warning; when compiling C++, warn about the
3481 deprecated conversion from string literals to @code{char *}. This
3482 warning, by default, is enabled for C++ programs.
3483 These warnings will help you find at
3484 compile time code that can try to write into a string constant, but
3485 only if you have been very careful about using @code{const} in
3486 declarations and prototypes. Otherwise, it will just be a nuisance;
3487 this is why we did not make @option{-Wall} request these warnings.
3491 Warn for variables that might be changed by @samp{longjmp} or
3492 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3495 @opindex Wconversion
3496 @opindex Wno-conversion
3497 Warn for implicit conversions that may alter a value. This includes
3498 conversions between real and integer, like @code{abs (x)} when
3499 @code{x} is @code{double}; conversions between signed and unsigned,
3500 like @code{unsigned ui = -1}; and conversions to smaller types, like
3501 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3502 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3503 changed by the conversion like in @code{abs (2.0)}. Warnings about
3504 conversions between signed and unsigned integers can be disabled by
3505 using @option{-Wno-sign-conversion}.
3507 For C++, also warn for conversions between @code{NULL} and non-pointer
3508 types; confusing overload resolution for user-defined conversions; and
3509 conversions that will never use a type conversion operator:
3510 conversions to @code{void}, the same type, a base class or a reference
3511 to them. Warnings about conversions between signed and unsigned
3512 integers are disabled by default in C++ unless
3513 @option{-Wsign-conversion} is explicitly enabled.
3516 @opindex Wempty-body
3517 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3518 while} statement. Additionally, in C++, warn when an empty body occurs
3519 in a @samp{while} or @samp{for} statement with no whitespacing before
3520 the semicolon. This warning is also enabled by @option{-Wextra}.
3522 @item -Wsign-compare
3523 @opindex Wsign-compare
3524 @cindex warning for comparison of signed and unsigned values
3525 @cindex comparison of signed and unsigned values, warning
3526 @cindex signed and unsigned values, comparison warning
3527 Warn when a comparison between signed and unsigned values could produce
3528 an incorrect result when the signed value is converted to unsigned.
3529 This warning is also enabled by @option{-Wextra}; to get the other warnings
3530 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3532 @item -Wsign-conversion
3533 @opindex Wsign-conversion
3534 @opindex Wno-sign-conversion
3535 Warn for implicit conversions that may change the sign of an integer
3536 value, like assigning a signed integer expression to an unsigned
3537 integer variable. An explicit cast silences the warning. In C, this
3538 option is enabled also by @option{-Wconversion}.
3542 @opindex Wno-address
3543 Warn about suspicious uses of memory addresses. These include using
3544 the address of a function in a conditional expression, such as
3545 @code{void func(void); if (func)}, and comparisons against the memory
3546 address of a string literal, such as @code{if (x == "abc")}. Such
3547 uses typically indicate a programmer error: the address of a function
3548 always evaluates to true, so their use in a conditional usually
3549 indicate that the programmer forgot the parentheses in a function
3550 call; and comparisons against string literals result in unspecified
3551 behavior and are not portable in C, so they usually indicate that the
3552 programmer intended to use @code{strcmp}. This warning is enabled by
3556 @opindex Wlogical-op
3557 @opindex Wno-logical-op
3558 Warn about suspicious uses of logical operators in expressions.
3559 This includes using logical operators in contexts where a
3560 bit-wise operator is likely to be expected.
3562 @item -Waggregate-return
3563 @opindex Waggregate-return
3564 Warn if any functions that return structures or unions are defined or
3565 called. (In languages where you can return an array, this also elicits
3568 @item -Wno-attributes
3569 @opindex Wno-attributes
3570 @opindex Wattributes
3571 Do not warn if an unexpected @code{__attribute__} is used, such as
3572 unrecognized attributes, function attributes applied to variables,
3573 etc. This will not stop errors for incorrect use of supported
3576 @item -Wstrict-prototypes @r{(C only)}
3577 @opindex Wstrict-prototypes
3578 Warn if a function is declared or defined without specifying the
3579 argument types. (An old-style function definition is permitted without
3580 a warning if preceded by a declaration which specifies the argument
3583 @item -Wold-style-declaration @r{(C only)}
3584 @opindex Wold-style-declaration
3585 Warn for obsolescent usages, according to the C Standard, in a
3586 declaration. For example, warn if storage-class specifiers like
3587 @code{static} are not the first things in a declaration. This warning
3588 is also enabled by @option{-Wextra}.
3590 @item -Wold-style-definition @r{(C only)}
3591 @opindex Wold-style-definition
3592 Warn if an old-style function definition is used. A warning is given
3593 even if there is a previous prototype.
3595 @item -Wmissing-parameter-type @r{(C only)}
3596 @opindex Wmissing-parameter-type
3597 A function parameter is declared without a type specifier in K&R-style
3604 This warning is also enabled by @option{-Wextra}.
3606 @item -Wmissing-prototypes @r{(C only)}
3607 @opindex Wmissing-prototypes
3608 Warn if a global function is defined without a previous prototype
3609 declaration. This warning is issued even if the definition itself
3610 provides a prototype. The aim is to detect global functions that fail
3611 to be declared in header files.
3613 @item -Wmissing-declarations @r{(C and C++ only)}
3614 @opindex Wmissing-declarations
3615 Warn if a global function is defined without a previous declaration.
3616 Do so even if the definition itself provides a prototype.
3617 Use this option to detect global functions that are not declared in
3618 header files. In C++, no warnings are issued for function templates,
3619 or for inline functions, or for functions in anonymous namespaces.
3621 @item -Wmissing-field-initializers
3622 @opindex Wmissing-field-initializers
3625 Warn if a structure's initializer has some fields missing. For
3626 example, the following code would cause such a warning, because
3627 @code{x.h} is implicitly zero:
3630 struct s @{ int f, g, h; @};
3631 struct s x = @{ 3, 4 @};
3634 This option does not warn about designated initializers, so the following
3635 modification would not trigger a warning:
3638 struct s @{ int f, g, h; @};
3639 struct s x = @{ .f = 3, .g = 4 @};
3642 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3643 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3645 @item -Wmissing-noreturn
3646 @opindex Wmissing-noreturn
3647 Warn about functions which might be candidates for attribute @code{noreturn}.
3648 Note these are only possible candidates, not absolute ones. Care should
3649 be taken to manually verify functions actually do not ever return before
3650 adding the @code{noreturn} attribute, otherwise subtle code generation
3651 bugs could be introduced. You will not get a warning for @code{main} in
3652 hosted C environments.
3654 @item -Wmissing-format-attribute
3655 @opindex Wmissing-format-attribute
3657 Warn about function pointers which might be candidates for @code{format}
3658 attributes. Note these are only possible candidates, not absolute ones.
3659 GCC will guess that function pointers with @code{format} attributes that
3660 are used in assignment, initialization, parameter passing or return
3661 statements should have a corresponding @code{format} attribute in the
3662 resulting type. I.e.@: the left-hand side of the assignment or
3663 initialization, the type of the parameter variable, or the return type
3664 of the containing function respectively should also have a @code{format}
3665 attribute to avoid the warning.
3667 GCC will also warn about function definitions which might be
3668 candidates for @code{format} attributes. Again, these are only
3669 possible candidates. GCC will guess that @code{format} attributes
3670 might be appropriate for any function that calls a function like
3671 @code{vprintf} or @code{vscanf}, but this might not always be the
3672 case, and some functions for which @code{format} attributes are
3673 appropriate may not be detected.
3675 @item -Wno-multichar
3676 @opindex Wno-multichar
3678 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3679 Usually they indicate a typo in the user's code, as they have
3680 implementation-defined values, and should not be used in portable code.
3682 @item -Wnormalized=<none|id|nfc|nfkc>
3683 @opindex Wnormalized
3686 @cindex character set, input normalization
3687 In ISO C and ISO C++, two identifiers are different if they are
3688 different sequences of characters. However, sometimes when characters
3689 outside the basic ASCII character set are used, you can have two
3690 different character sequences that look the same. To avoid confusion,
3691 the ISO 10646 standard sets out some @dfn{normalization rules} which
3692 when applied ensure that two sequences that look the same are turned into
3693 the same sequence. GCC can warn you if you are using identifiers which
3694 have not been normalized; this option controls that warning.
3696 There are four levels of warning that GCC supports. The default is
3697 @option{-Wnormalized=nfc}, which warns about any identifier which is
3698 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3699 recommended form for most uses.
3701 Unfortunately, there are some characters which ISO C and ISO C++ allow
3702 in identifiers that when turned into NFC aren't allowable as
3703 identifiers. That is, there's no way to use these symbols in portable
3704 ISO C or C++ and have all your identifiers in NFC.
3705 @option{-Wnormalized=id} suppresses the warning for these characters.
3706 It is hoped that future versions of the standards involved will correct
3707 this, which is why this option is not the default.
3709 You can switch the warning off for all characters by writing
3710 @option{-Wnormalized=none}. You would only want to do this if you
3711 were using some other normalization scheme (like ``D''), because
3712 otherwise you can easily create bugs that are literally impossible to see.
3714 Some characters in ISO 10646 have distinct meanings but look identical
3715 in some fonts or display methodologies, especially once formatting has
3716 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3717 LETTER N'', will display just like a regular @code{n} which has been
3718 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3719 normalization scheme to convert all these into a standard form as
3720 well, and GCC will warn if your code is not in NFKC if you use
3721 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3722 about every identifier that contains the letter O because it might be
3723 confused with the digit 0, and so is not the default, but may be
3724 useful as a local coding convention if the programming environment is
3725 unable to be fixed to display these characters distinctly.
3727 @item -Wno-deprecated-declarations
3728 @opindex Wno-deprecated-declarations
3729 Do not warn about uses of functions (@pxref{Function Attributes}),
3730 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3731 Attributes}) marked as deprecated by using the @code{deprecated}
3735 @opindex Wno-overflow
3736 Do not warn about compile-time overflow in constant expressions.
3738 @item -Woverride-init
3739 @opindex Woverride-init
3742 Warn if an initialized field without side effects is overridden when
3743 using designated initializers (@pxref{Designated Inits, , Designated
3746 This warning is included in @option{-Wextra}. To get other
3747 @option{-Wextra} warnings without this one, use @samp{-Wextra
3748 -Wno-override-init}.
3752 Warn if a structure is given the packed attribute, but the packed
3753 attribute has no effect on the layout or size of the structure.
3754 Such structures may be mis-aligned for little benefit. For
3755 instance, in this code, the variable @code{f.x} in @code{struct bar}
3756 will be misaligned even though @code{struct bar} does not itself
3757 have the packed attribute:
3764 @} __attribute__((packed));
3774 Warn if padding is included in a structure, either to align an element
3775 of the structure or to align the whole structure. Sometimes when this
3776 happens it is possible to rearrange the fields of the structure to
3777 reduce the padding and so make the structure smaller.
3779 @item -Wredundant-decls
3780 @opindex Wredundant-decls
3781 Warn if anything is declared more than once in the same scope, even in
3782 cases where multiple declaration is valid and changes nothing.
3784 @item -Wnested-externs @r{(C only)}
3785 @opindex Wnested-externs
3786 Warn if an @code{extern} declaration is encountered within a function.
3788 @item -Wunreachable-code
3789 @opindex Wunreachable-code
3790 Warn if the compiler detects that code will never be executed.
3792 This option is intended to warn when the compiler detects that at
3793 least a whole line of source code will never be executed, because
3794 some condition is never satisfied or because it is after a
3795 procedure that never returns.
3797 It is possible for this option to produce a warning even though there
3798 are circumstances under which part of the affected line can be executed,
3799 so care should be taken when removing apparently-unreachable code.
3801 For instance, when a function is inlined, a warning may mean that the
3802 line is unreachable in only one inlined copy of the function.
3804 This option is not made part of @option{-Wall} because in a debugging
3805 version of a program there is often substantial code which checks
3806 correct functioning of the program and is, hopefully, unreachable
3807 because the program does work. Another common use of unreachable
3808 code is to provide behavior which is selectable at compile-time.
3812 Warn if a function can not be inlined and it was declared as inline.
3813 Even with this option, the compiler will not warn about failures to
3814 inline functions declared in system headers.
3816 The compiler uses a variety of heuristics to determine whether or not
3817 to inline a function. For example, the compiler takes into account
3818 the size of the function being inlined and the amount of inlining
3819 that has already been done in the current function. Therefore,
3820 seemingly insignificant changes in the source program can cause the
3821 warnings produced by @option{-Winline} to appear or disappear.
3823 @item -Wno-invalid-offsetof @r{(C++ only)}
3824 @opindex Wno-invalid-offsetof
3825 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3826 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3827 to a non-POD type is undefined. In existing C++ implementations,
3828 however, @samp{offsetof} typically gives meaningful results even when
3829 applied to certain kinds of non-POD types. (Such as a simple
3830 @samp{struct} that fails to be a POD type only by virtue of having a
3831 constructor.) This flag is for users who are aware that they are
3832 writing nonportable code and who have deliberately chosen to ignore the
3835 The restrictions on @samp{offsetof} may be relaxed in a future version
3836 of the C++ standard.
3838 @item -Wno-int-to-pointer-cast @r{(C only)}
3839 @opindex Wno-int-to-pointer-cast
3840 Suppress warnings from casts to pointer type of an integer of a
3843 @item -Wno-pointer-to-int-cast @r{(C only)}
3844 @opindex Wno-pointer-to-int-cast
3845 Suppress warnings from casts from a pointer to an integer type of a
3849 @opindex Winvalid-pch
3850 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3851 the search path but can't be used.
3855 @opindex Wno-long-long
3856 Warn if @samp{long long} type is used. This is default. To inhibit
3857 the warning messages, use @option{-Wno-long-long}. Flags
3858 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3859 only when @option{-pedantic} flag is used.
3861 @item -Wvariadic-macros
3862 @opindex Wvariadic-macros
3863 @opindex Wno-variadic-macros
3864 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3865 alternate syntax when in pedantic ISO C99 mode. This is default.
3866 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3871 Warn if variable length array is used in the code.
3872 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3873 the variable length array.
3875 @item -Wvolatile-register-var
3876 @opindex Wvolatile-register-var
3877 @opindex Wno-volatile-register-var
3878 Warn if a register variable is declared volatile. The volatile
3879 modifier does not inhibit all optimizations that may eliminate reads
3880 and/or writes to register variables.
3882 @item -Wdisabled-optimization
3883 @opindex Wdisabled-optimization
3884 Warn if a requested optimization pass is disabled. This warning does
3885 not generally indicate that there is anything wrong with your code; it
3886 merely indicates that GCC's optimizers were unable to handle the code
3887 effectively. Often, the problem is that your code is too big or too
3888 complex; GCC will refuse to optimize programs when the optimization
3889 itself is likely to take inordinate amounts of time.
3891 @item -Wpointer-sign
3892 @opindex Wpointer-sign
3893 @opindex Wno-pointer-sign
3894 Warn for pointer argument passing or assignment with different signedness.
3895 This option is only supported for C and Objective-C@. It is implied by
3896 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3897 @option{-Wno-pointer-sign}.
3901 Make all warnings into errors.
3905 Make the specified warning into an errors. The specifier for a
3906 warning is appended, for example @option{-Werror=switch} turns the
3907 warnings controlled by @option{-Wswitch} into errors. This switch
3908 takes a negative form, to be used to negate @option{-Werror} for
3909 specific warnings, for example @option{-Wno-error=switch} makes
3910 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3911 is in effect. You can use the @option{-fdiagnostics-show-option}
3912 option to have each controllable warning amended with the option which
3913 controls it, to determine what to use with this option.
3915 Note that specifying @option{-Werror=}@var{foo} automatically implies
3916 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3919 @item -Wstack-protector
3920 @opindex Wstack-protector
3921 This option is only active when @option{-fstack-protector} is active. It
3922 warns about functions that will not be protected against stack smashing.
3924 @item -Woverlength-strings
3925 @opindex Woverlength-strings
3926 Warn about string constants which are longer than the ``minimum
3927 maximum'' length specified in the C standard. Modern compilers
3928 generally allow string constants which are much longer than the
3929 standard's minimum limit, but very portable programs should avoid
3930 using longer strings.
3932 The limit applies @emph{after} string constant concatenation, and does
3933 not count the trailing NUL@. In C89, the limit was 509 characters; in
3934 C99, it was raised to 4095. C++98 does not specify a normative
3935 minimum maximum, so we do not diagnose overlength strings in C++@.
3937 This option is implied by @option{-pedantic}, and can be disabled with
3938 @option{-Wno-overlength-strings}.
3941 @node Debugging Options
3942 @section Options for Debugging Your Program or GCC
3943 @cindex options, debugging
3944 @cindex debugging information options
3946 GCC has various special options that are used for debugging
3947 either your program or GCC:
3952 Produce debugging information in the operating system's native format
3953 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3956 On most systems that use stabs format, @option{-g} enables use of extra
3957 debugging information that only GDB can use; this extra information
3958 makes debugging work better in GDB but will probably make other debuggers
3960 refuse to read the program. If you want to control for certain whether
3961 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3962 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3964 GCC allows you to use @option{-g} with
3965 @option{-O}. The shortcuts taken by optimized code may occasionally
3966 produce surprising results: some variables you declared may not exist
3967 at all; flow of control may briefly move where you did not expect it;
3968 some statements may not be executed because they compute constant
3969 results or their values were already at hand; some statements may
3970 execute in different places because they were moved out of loops.
3972 Nevertheless it proves possible to debug optimized output. This makes
3973 it reasonable to use the optimizer for programs that might have bugs.
3975 The following options are useful when GCC is generated with the
3976 capability for more than one debugging format.
3980 Produce debugging information for use by GDB@. This means to use the
3981 most expressive format available (DWARF 2, stabs, or the native format
3982 if neither of those are supported), including GDB extensions if at all
3987 Produce debugging information in stabs format (if that is supported),
3988 without GDB extensions. This is the format used by DBX on most BSD
3989 systems. On MIPS, Alpha and System V Release 4 systems this option
3990 produces stabs debugging output which is not understood by DBX or SDB@.
3991 On System V Release 4 systems this option requires the GNU assembler.
3993 @item -feliminate-unused-debug-symbols
3994 @opindex feliminate-unused-debug-symbols
3995 Produce debugging information in stabs format (if that is supported),
3996 for only symbols that are actually used.
3998 @item -femit-class-debug-always
3999 Instead of emitting debugging information for a C++ class in only one
4000 object file, emit it in all object files using the class. This option
4001 should be used only with debuggers that are unable to handle the way GCC
4002 normally emits debugging information for classes because using this
4003 option will increase the size of debugging information by as much as a
4008 Produce debugging information in stabs format (if that is supported),
4009 using GNU extensions understood only by the GNU debugger (GDB)@. The
4010 use of these extensions is likely to make other debuggers crash or
4011 refuse to read the program.
4015 Produce debugging information in COFF format (if that is supported).
4016 This is the format used by SDB on most System V systems prior to
4021 Produce debugging information in XCOFF format (if that is supported).
4022 This is the format used by the DBX debugger on IBM RS/6000 systems.
4026 Produce debugging information in XCOFF format (if that is supported),
4027 using GNU extensions understood only by the GNU debugger (GDB)@. The
4028 use of these extensions is likely to make other debuggers crash or
4029 refuse to read the program, and may cause assemblers other than the GNU
4030 assembler (GAS) to fail with an error.
4034 Produce debugging information in DWARF version 2 format (if that is
4035 supported). This is the format used by DBX on IRIX 6. With this
4036 option, GCC uses features of DWARF version 3 when they are useful;
4037 version 3 is upward compatible with version 2, but may still cause
4038 problems for older debuggers.
4042 Produce debugging information in VMS debug format (if that is
4043 supported). This is the format used by DEBUG on VMS systems.
4046 @itemx -ggdb@var{level}
4047 @itemx -gstabs@var{level}
4048 @itemx -gcoff@var{level}
4049 @itemx -gxcoff@var{level}
4050 @itemx -gvms@var{level}
4051 Request debugging information and also use @var{level} to specify how
4052 much information. The default level is 2.
4054 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4057 Level 1 produces minimal information, enough for making backtraces in
4058 parts of the program that you don't plan to debug. This includes
4059 descriptions of functions and external variables, but no information
4060 about local variables and no line numbers.
4062 Level 3 includes extra information, such as all the macro definitions
4063 present in the program. Some debuggers support macro expansion when
4064 you use @option{-g3}.
4066 @option{-gdwarf-2} does not accept a concatenated debug level, because
4067 GCC used to support an option @option{-gdwarf} that meant to generate
4068 debug information in version 1 of the DWARF format (which is very
4069 different from version 2), and it would have been too confusing. That
4070 debug format is long obsolete, but the option cannot be changed now.
4071 Instead use an additional @option{-g@var{level}} option to change the
4072 debug level for DWARF2.
4074 @item -feliminate-dwarf2-dups
4075 @opindex feliminate-dwarf2-dups
4076 Compress DWARF2 debugging information by eliminating duplicated
4077 information about each symbol. This option only makes sense when
4078 generating DWARF2 debugging information with @option{-gdwarf-2}.
4080 @item -femit-struct-debug-baseonly
4081 Emit debug information for struct-like types
4082 only when the base name of the compilation source file
4083 matches the base name of file in which the struct was defined.
4085 This option substantially reduces the size of debugging information,
4086 but at significant potential loss in type information to the debugger.
4087 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4088 See @option{-femit-struct-debug-detailed} for more detailed control.
4090 This option works only with DWARF 2.
4092 @item -femit-struct-debug-reduced
4093 Emit debug information for struct-like types
4094 only when the base name of the compilation source file
4095 matches the base name of file in which the type was defined,
4096 unless the struct is a template or defined in a system header.
4098 This option significantly reduces the size of debugging information,
4099 with some potential loss in type information to the debugger.
4100 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4101 See @option{-femit-struct-debug-detailed} for more detailed control.
4103 This option works only with DWARF 2.
4105 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4106 Specify the struct-like types
4107 for which the compiler will generate debug information.
4108 The intent is to reduce duplicate struct debug information
4109 between different object files within the same program.
4111 This option is a detailed version of
4112 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4113 which will serve for most needs.
4115 A specification has the syntax
4116 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4118 The optional first word limits the specification to
4119 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4120 A struct type is used directly when it is the type of a variable, member.
4121 Indirect uses arise through pointers to structs.
4122 That is, when use of an incomplete struct would be legal, the use is indirect.
4124 @samp{struct one direct; struct two * indirect;}.
4126 The optional second word limits the specification to
4127 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4128 Generic structs are a bit complicated to explain.
4129 For C++, these are non-explicit specializations of template classes,
4130 or non-template classes within the above.
4131 Other programming languages have generics,
4132 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4134 The third word specifies the source files for those
4135 structs for which the compiler will emit debug information.
4136 The values @samp{none} and @samp{any} have the normal meaning.
4137 The value @samp{base} means that
4138 the base of name of the file in which the type declaration appears
4139 must match the base of the name of the main compilation file.
4140 In practice, this means that
4141 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4142 but types declared in other header will not.
4143 The value @samp{sys} means those types satisfying @samp{base}
4144 or declared in system or compiler headers.
4146 You may need to experiment to determine the best settings for your application.
4148 The default is @samp{-femit-struct-debug-detailed=all}.
4150 This option works only with DWARF 2.
4152 @item -fdebug-prefix-map=@var{old}=@var{new}
4153 @opindex fdebug-prefix-map
4154 When compiling files in directory @file{@var{old}}, record debugging
4155 information describing them as in @file{@var{new}} instead.
4157 @cindex @command{prof}
4160 Generate extra code to write profile information suitable for the
4161 analysis program @command{prof}. You must use this option when compiling
4162 the source files you want data about, and you must also use it when
4165 @cindex @command{gprof}
4168 Generate extra code to write profile information suitable for the
4169 analysis program @command{gprof}. You must use this option when compiling
4170 the source files you want data about, and you must also use it when
4175 Makes the compiler print out each function name as it is compiled, and
4176 print some statistics about each pass when it finishes.
4179 @opindex ftime-report
4180 Makes the compiler print some statistics about the time consumed by each
4181 pass when it finishes.
4184 @opindex fmem-report
4185 Makes the compiler print some statistics about permanent memory
4186 allocation when it finishes.
4188 @item -fpre-ipa-mem-report
4189 @opindex fpre-ipa-mem-report
4190 @item -fpost-ipa-mem-report
4191 @opindex fpost-ipa-mem-report
4192 Makes the compiler print some statistics about permanent memory
4193 allocation before or after interprocedural optimization.
4195 @item -fprofile-arcs
4196 @opindex fprofile-arcs
4197 Add code so that program flow @dfn{arcs} are instrumented. During
4198 execution the program records how many times each branch and call is
4199 executed and how many times it is taken or returns. When the compiled
4200 program exits it saves this data to a file called
4201 @file{@var{auxname}.gcda} for each source file. The data may be used for
4202 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4203 test coverage analysis (@option{-ftest-coverage}). Each object file's
4204 @var{auxname} is generated from the name of the output file, if
4205 explicitly specified and it is not the final executable, otherwise it is
4206 the basename of the source file. In both cases any suffix is removed
4207 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4208 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4209 @xref{Cross-profiling}.
4211 @cindex @command{gcov}
4215 This option is used to compile and link code instrumented for coverage
4216 analysis. The option is a synonym for @option{-fprofile-arcs}
4217 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4218 linking). See the documentation for those options for more details.
4223 Compile the source files with @option{-fprofile-arcs} plus optimization
4224 and code generation options. For test coverage analysis, use the
4225 additional @option{-ftest-coverage} option. You do not need to profile
4226 every source file in a program.
4229 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4230 (the latter implies the former).
4233 Run the program on a representative workload to generate the arc profile
4234 information. This may be repeated any number of times. You can run
4235 concurrent instances of your program, and provided that the file system
4236 supports locking, the data files will be correctly updated. Also
4237 @code{fork} calls are detected and correctly handled (double counting
4241 For profile-directed optimizations, compile the source files again with
4242 the same optimization and code generation options plus
4243 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4244 Control Optimization}).
4247 For test coverage analysis, use @command{gcov} to produce human readable
4248 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4249 @command{gcov} documentation for further information.
4253 With @option{-fprofile-arcs}, for each function of your program GCC
4254 creates a program flow graph, then finds a spanning tree for the graph.
4255 Only arcs that are not on the spanning tree have to be instrumented: the
4256 compiler adds code to count the number of times that these arcs are
4257 executed. When an arc is the only exit or only entrance to a block, the
4258 instrumentation code can be added to the block; otherwise, a new basic
4259 block must be created to hold the instrumentation code.
4262 @item -ftest-coverage
4263 @opindex ftest-coverage
4264 Produce a notes file that the @command{gcov} code-coverage utility
4265 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4266 show program coverage. Each source file's note file is called
4267 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4268 above for a description of @var{auxname} and instructions on how to
4269 generate test coverage data. Coverage data will match the source files
4270 more closely, if you do not optimize.
4272 @item -fdbg-cnt-list
4273 @opindex fdbg-cnt-list
4274 Print the name and the counter upperbound for all debug counters.
4276 @item -fdbg-cnt=@var{counter-value-list}
4278 Set the internal debug counter upperbound. @var{counter-value-list}
4279 is a comma-separated list of @var{name}:@var{value} pairs
4280 which sets the upperbound of each debug counter @var{name} to @var{value}.
4281 All debug counters have the initial upperbound of @var{UINT_MAX},
4282 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4283 e.g. With -fdbg-cnt=dce:10,tail_call:0
4284 dbg_cnt(dce) will return true only for first 10 invocations
4285 and dbg_cnt(tail_call) will return false always.
4287 @item -d@var{letters}
4288 @item -fdump-rtl-@var{pass}
4290 Says to make debugging dumps during compilation at times specified by
4291 @var{letters}. This is used for debugging the RTL-based passes of the
4292 compiler. The file names for most of the dumps are made by appending a
4293 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4294 from the name of the output file, if explicitly specified and it is not
4295 an executable, otherwise it is the basename of the source file. These
4296 switches may have different effects when @option{-E} is used for
4299 Most debug dumps can be enabled either passing a letter to the @option{-d}
4300 option, or with a long @option{-fdump-rtl} switch; here are the possible
4301 letters for use in @var{letters} and @var{pass}, and their meanings:
4306 Annotate the assembler output with miscellaneous debugging information.
4309 @itemx -fdump-rtl-bbro
4311 @opindex fdump-rtl-bbro
4312 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4315 @itemx -fdump-rtl-combine
4317 @opindex fdump-rtl-combine
4318 Dump after the RTL instruction combination pass, to the file
4319 @file{@var{file}.129r.combine}.
4322 @itemx -fdump-rtl-ce1
4323 @itemx -fdump-rtl-ce2
4325 @opindex fdump-rtl-ce1
4326 @opindex fdump-rtl-ce2
4327 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4328 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4329 and @option{-fdump-rtl-ce2} enable dumping after the second if
4330 conversion, to the file @file{@var{file}.130r.ce2}.
4333 @itemx -fdump-rtl-btl
4334 @itemx -fdump-rtl-dbr
4336 @opindex fdump-rtl-btl
4337 @opindex fdump-rtl-dbr
4338 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4339 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4340 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4341 scheduling, to @file{@var{file}.36.dbr}.
4345 Dump all macro definitions, at the end of preprocessing, in addition to
4349 @itemx -fdump-rtl-ce3
4351 @opindex fdump-rtl-ce3
4352 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4355 @itemx -fdump-rtl-cfg
4356 @itemx -fdump-rtl-life
4358 @opindex fdump-rtl-cfg
4359 @opindex fdump-rtl-life
4360 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4361 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4362 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4363 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4366 @itemx -fdump-rtl-greg
4368 @opindex fdump-rtl-greg
4369 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4372 @itemx -fdump-rtl-gcse
4373 @itemx -fdump-rtl-bypass
4375 @opindex fdump-rtl-gcse
4376 @opindex fdump-rtl-bypass
4377 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4378 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4379 enable dumping after jump bypassing and control flow optimizations, to
4380 @file{@var{file}.115r.bypass}.
4383 @itemx -fdump-rtl-eh
4385 @opindex fdump-rtl-eh
4386 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4389 @itemx -fdump-rtl-sibling
4391 @opindex fdump-rtl-sibling
4392 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4395 @itemx -fdump-rtl-jump
4397 @opindex fdump-rtl-jump
4398 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4401 @itemx -fdump-rtl-stack
4403 @opindex fdump-rtl-stack
4404 Dump after conversion from GCC's "flat register file" registers to the
4405 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4408 @itemx -fdump-rtl-lreg
4410 @opindex fdump-rtl-lreg
4411 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4414 @itemx -fdump-rtl-loop2
4416 @opindex fdump-rtl-loop2
4417 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4418 loop optimization pass, to @file{@var{file}.119r.loop2},
4419 @file{@var{file}.120r.loop2_init},
4420 @file{@var{file}.121r.loop2_invariant}, and
4421 @file{@var{file}.125r.loop2_done}.
4424 @itemx -fdump-rtl-sms
4426 @opindex fdump-rtl-sms
4427 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4430 @itemx -fdump-rtl-mach
4432 @opindex fdump-rtl-mach
4433 Dump after performing the machine dependent reorganization pass, to
4434 @file{@var{file}.155r.mach} if that pass exists.
4437 @itemx -fdump-rtl-rnreg
4439 @opindex fdump-rtl-rnreg
4440 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4443 @itemx -fdump-rtl-regmove
4445 @opindex fdump-rtl-regmove
4446 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4449 @itemx -fdump-rtl-postreload
4451 @opindex fdump-rtl-postreload
4452 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4455 @itemx -fdump-rtl-expand
4457 @opindex fdump-rtl-expand
4458 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4461 @itemx -fdump-rtl-sched2
4463 @opindex fdump-rtl-sched2
4464 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4467 @itemx -fdump-rtl-cse
4469 @opindex fdump-rtl-cse
4470 Dump after CSE (including the jump optimization that sometimes follows
4471 CSE), to @file{@var{file}.113r.cse}.
4474 @itemx -fdump-rtl-sched1
4476 @opindex fdump-rtl-sched1
4477 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4480 @itemx -fdump-rtl-cse2
4482 @opindex fdump-rtl-cse2
4483 Dump after the second CSE pass (including the jump optimization that
4484 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4487 @itemx -fdump-rtl-tracer
4489 @opindex fdump-rtl-tracer
4490 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4493 @itemx -fdump-rtl-vpt
4494 @itemx -fdump-rtl-vartrack
4496 @opindex fdump-rtl-vpt
4497 @opindex fdump-rtl-vartrack
4498 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4499 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4500 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4501 to @file{@var{file}.154r.vartrack}.
4504 @itemx -fdump-rtl-flow2
4506 @opindex fdump-rtl-flow2
4507 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4510 @itemx -fdump-rtl-peephole2
4512 @opindex fdump-rtl-peephole2
4513 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4516 @itemx -fdump-rtl-web
4518 @opindex fdump-rtl-web
4519 Dump after live range splitting, to @file{@var{file}.126r.web}.
4522 @itemx -fdump-rtl-all
4524 @opindex fdump-rtl-all
4525 Produce all the dumps listed above.
4529 Produce a core dump whenever an error occurs.
4533 Print statistics on memory usage, at the end of the run, to
4538 Annotate the assembler output with a comment indicating which
4539 pattern and alternative was used. The length of each instruction is
4544 Dump the RTL in the assembler output as a comment before each instruction.
4545 Also turns on @option{-dp} annotation.
4549 For each of the other indicated dump files (either with @option{-d} or
4550 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4551 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4555 Just generate RTL for a function instead of compiling it. Usually used
4556 with @samp{r} (@option{-fdump-rtl-expand}).
4560 Dump debugging information during parsing, to standard error.
4564 @opindex fdump-noaddr
4565 When doing debugging dumps (see @option{-d} option above), suppress
4566 address output. This makes it more feasible to use diff on debugging
4567 dumps for compiler invocations with different compiler binaries and/or
4568 different text / bss / data / heap / stack / dso start locations.
4570 @item -fdump-unnumbered
4571 @opindex fdump-unnumbered
4572 When doing debugging dumps (see @option{-d} option above), suppress instruction
4573 numbers and address output. This makes it more feasible to
4574 use diff on debugging dumps for compiler invocations with different
4575 options, in particular with and without @option{-g}.
4577 @item -fdump-translation-unit @r{(C++ only)}
4578 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4579 @opindex fdump-translation-unit
4580 Dump a representation of the tree structure for the entire translation
4581 unit to a file. The file name is made by appending @file{.tu} to the
4582 source file name. If the @samp{-@var{options}} form is used, @var{options}
4583 controls the details of the dump as described for the
4584 @option{-fdump-tree} options.
4586 @item -fdump-class-hierarchy @r{(C++ only)}
4587 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4588 @opindex fdump-class-hierarchy
4589 Dump a representation of each class's hierarchy and virtual function
4590 table layout to a file. The file name is made by appending @file{.class}
4591 to the source file name. If the @samp{-@var{options}} form is used,
4592 @var{options} controls the details of the dump as described for the
4593 @option{-fdump-tree} options.
4595 @item -fdump-ipa-@var{switch}
4597 Control the dumping at various stages of inter-procedural analysis
4598 language tree to a file. The file name is generated by appending a switch
4599 specific suffix to the source file name. The following dumps are possible:
4603 Enables all inter-procedural analysis dumps; currently the only produced
4604 dump is the @samp{cgraph} dump.
4607 Dumps information about call-graph optimization, unused function removal,
4608 and inlining decisions.
4611 @item -fdump-tree-@var{switch}
4612 @itemx -fdump-tree-@var{switch}-@var{options}
4614 Control the dumping at various stages of processing the intermediate
4615 language tree to a file. The file name is generated by appending a switch
4616 specific suffix to the source file name. If the @samp{-@var{options}}
4617 form is used, @var{options} is a list of @samp{-} separated options that
4618 control the details of the dump. Not all options are applicable to all
4619 dumps, those which are not meaningful will be ignored. The following
4620 options are available
4624 Print the address of each node. Usually this is not meaningful as it
4625 changes according to the environment and source file. Its primary use
4626 is for tying up a dump file with a debug environment.
4628 Inhibit dumping of members of a scope or body of a function merely
4629 because that scope has been reached. Only dump such items when they
4630 are directly reachable by some other path. When dumping pretty-printed
4631 trees, this option inhibits dumping the bodies of control structures.
4633 Print a raw representation of the tree. By default, trees are
4634 pretty-printed into a C-like representation.
4636 Enable more detailed dumps (not honored by every dump option).
4638 Enable dumping various statistics about the pass (not honored by every dump
4641 Enable showing basic block boundaries (disabled in raw dumps).
4643 Enable showing virtual operands for every statement.
4645 Enable showing line numbers for statements.
4647 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4649 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4652 The following tree dumps are possible:
4656 Dump before any tree based optimization, to @file{@var{file}.original}.
4659 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4662 Dump after function inlining, to @file{@var{file}.inlined}.
4665 @opindex fdump-tree-gimple
4666 Dump each function before and after the gimplification pass to a file. The
4667 file name is made by appending @file{.gimple} to the source file name.
4670 @opindex fdump-tree-cfg
4671 Dump the control flow graph of each function to a file. The file name is
4672 made by appending @file{.cfg} to the source file name.
4675 @opindex fdump-tree-vcg
4676 Dump the control flow graph of each function to a file in VCG format. The
4677 file name is made by appending @file{.vcg} to the source file name. Note
4678 that if the file contains more than one function, the generated file cannot
4679 be used directly by VCG@. You will need to cut and paste each function's
4680 graph into its own separate file first.
4683 @opindex fdump-tree-ch
4684 Dump each function after copying loop headers. The file name is made by
4685 appending @file{.ch} to the source file name.
4688 @opindex fdump-tree-ssa
4689 Dump SSA related information to a file. The file name is made by appending
4690 @file{.ssa} to the source file name.
4693 @opindex fdump-tree-salias
4694 Dump structure aliasing variable information to a file. This file name
4695 is made by appending @file{.salias} to the source file name.
4698 @opindex fdump-tree-alias
4699 Dump aliasing information for each function. The file name is made by
4700 appending @file{.alias} to the source file name.
4703 @opindex fdump-tree-ccp
4704 Dump each function after CCP@. The file name is made by appending
4705 @file{.ccp} to the source file name.
4708 @opindex fdump-tree-storeccp
4709 Dump each function after STORE-CCP. The file name is made by appending
4710 @file{.storeccp} to the source file name.
4713 @opindex fdump-tree-pre
4714 Dump trees after partial redundancy elimination. The file name is made
4715 by appending @file{.pre} to the source file name.
4718 @opindex fdump-tree-fre
4719 Dump trees after full redundancy elimination. The file name is made
4720 by appending @file{.fre} to the source file name.
4723 @opindex fdump-tree-copyprop
4724 Dump trees after copy propagation. The file name is made
4725 by appending @file{.copyprop} to the source file name.
4727 @item store_copyprop
4728 @opindex fdump-tree-store_copyprop
4729 Dump trees after store copy-propagation. The file name is made
4730 by appending @file{.store_copyprop} to the source file name.
4733 @opindex fdump-tree-dce
4734 Dump each function after dead code elimination. The file name is made by
4735 appending @file{.dce} to the source file name.
4738 @opindex fdump-tree-mudflap
4739 Dump each function after adding mudflap instrumentation. The file name is
4740 made by appending @file{.mudflap} to the source file name.
4743 @opindex fdump-tree-sra
4744 Dump each function after performing scalar replacement of aggregates. The
4745 file name is made by appending @file{.sra} to the source file name.
4748 @opindex fdump-tree-sink
4749 Dump each function after performing code sinking. The file name is made
4750 by appending @file{.sink} to the source file name.
4753 @opindex fdump-tree-dom
4754 Dump each function after applying dominator tree optimizations. The file
4755 name is made by appending @file{.dom} to the source file name.
4758 @opindex fdump-tree-dse
4759 Dump each function after applying dead store elimination. The file
4760 name is made by appending @file{.dse} to the source file name.
4763 @opindex fdump-tree-phiopt
4764 Dump each function after optimizing PHI nodes into straightline code. The file
4765 name is made by appending @file{.phiopt} to the source file name.
4768 @opindex fdump-tree-forwprop
4769 Dump each function after forward propagating single use variables. The file
4770 name is made by appending @file{.forwprop} to the source file name.
4773 @opindex fdump-tree-copyrename
4774 Dump each function after applying the copy rename optimization. The file
4775 name is made by appending @file{.copyrename} to the source file name.
4778 @opindex fdump-tree-nrv
4779 Dump each function after applying the named return value optimization on
4780 generic trees. The file name is made by appending @file{.nrv} to the source
4784 @opindex fdump-tree-vect
4785 Dump each function after applying vectorization of loops. The file name is
4786 made by appending @file{.vect} to the source file name.
4789 @opindex fdump-tree-vrp
4790 Dump each function after Value Range Propagation (VRP). The file name
4791 is made by appending @file{.vrp} to the source file name.
4794 @opindex fdump-tree-all
4795 Enable all the available tree dumps with the flags provided in this option.
4798 @item -ftree-vectorizer-verbose=@var{n}
4799 @opindex ftree-vectorizer-verbose
4800 This option controls the amount of debugging output the vectorizer prints.
4801 This information is written to standard error, unless
4802 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4803 in which case it is output to the usual dump listing file, @file{.vect}.
4804 For @var{n}=0 no diagnostic information is reported.
4805 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4806 and the total number of loops that got vectorized.
4807 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4808 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4809 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4810 level that @option{-fdump-tree-vect-stats} uses.
4811 Higher verbosity levels mean either more information dumped for each
4812 reported loop, or same amount of information reported for more loops:
4813 If @var{n}=3, alignment related information is added to the reports.
4814 If @var{n}=4, data-references related information (e.g. memory dependences,
4815 memory access-patterns) is added to the reports.
4816 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4817 that did not pass the first analysis phase (i.e. may not be countable, or
4818 may have complicated control-flow).
4819 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4820 For @var{n}=7, all the information the vectorizer generates during its
4821 analysis and transformation is reported. This is the same verbosity level
4822 that @option{-fdump-tree-vect-details} uses.
4824 @item -frandom-seed=@var{string}
4825 @opindex frandom-string
4826 This option provides a seed that GCC uses when it would otherwise use
4827 random numbers. It is used to generate certain symbol names
4828 that have to be different in every compiled file. It is also used to
4829 place unique stamps in coverage data files and the object files that
4830 produce them. You can use the @option{-frandom-seed} option to produce
4831 reproducibly identical object files.
4833 The @var{string} should be different for every file you compile.
4835 @item -fsched-verbose=@var{n}
4836 @opindex fsched-verbose
4837 On targets that use instruction scheduling, this option controls the
4838 amount of debugging output the scheduler prints. This information is
4839 written to standard error, unless @option{-dS} or @option{-dR} is
4840 specified, in which case it is output to the usual dump
4841 listing file, @file{.sched} or @file{.sched2} respectively. However
4842 for @var{n} greater than nine, the output is always printed to standard
4845 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4846 same information as @option{-dRS}. For @var{n} greater than one, it
4847 also output basic block probabilities, detailed ready list information
4848 and unit/insn info. For @var{n} greater than two, it includes RTL
4849 at abort point, control-flow and regions info. And for @var{n} over
4850 four, @option{-fsched-verbose} also includes dependence info.
4854 Store the usual ``temporary'' intermediate files permanently; place them
4855 in the current directory and name them based on the source file. Thus,
4856 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4857 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4858 preprocessed @file{foo.i} output file even though the compiler now
4859 normally uses an integrated preprocessor.
4861 When used in combination with the @option{-x} command line option,
4862 @option{-save-temps} is sensible enough to avoid over writing an
4863 input source file with the same extension as an intermediate file.
4864 The corresponding intermediate file may be obtained by renaming the
4865 source file before using @option{-save-temps}.
4869 Report the CPU time taken by each subprocess in the compilation
4870 sequence. For C source files, this is the compiler proper and assembler
4871 (plus the linker if linking is done). The output looks like this:
4878 The first number on each line is the ``user time'', that is time spent
4879 executing the program itself. The second number is ``system time'',
4880 time spent executing operating system routines on behalf of the program.
4881 Both numbers are in seconds.
4883 @item -fvar-tracking
4884 @opindex fvar-tracking
4885 Run variable tracking pass. It computes where variables are stored at each
4886 position in code. Better debugging information is then generated
4887 (if the debugging information format supports this information).
4889 It is enabled by default when compiling with optimization (@option{-Os},
4890 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4891 the debug info format supports it.
4893 @item -print-file-name=@var{library}
4894 @opindex print-file-name
4895 Print the full absolute name of the library file @var{library} that
4896 would be used when linking---and don't do anything else. With this
4897 option, GCC does not compile or link anything; it just prints the
4900 @item -print-multi-directory
4901 @opindex print-multi-directory
4902 Print the directory name corresponding to the multilib selected by any
4903 other switches present in the command line. This directory is supposed
4904 to exist in @env{GCC_EXEC_PREFIX}.
4906 @item -print-multi-lib
4907 @opindex print-multi-lib
4908 Print the mapping from multilib directory names to compiler switches
4909 that enable them. The directory name is separated from the switches by
4910 @samp{;}, and each switch starts with an @samp{@@} instead of the
4911 @samp{-}, without spaces between multiple switches. This is supposed to
4912 ease shell-processing.
4914 @item -print-prog-name=@var{program}
4915 @opindex print-prog-name
4916 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4918 @item -print-libgcc-file-name
4919 @opindex print-libgcc-file-name
4920 Same as @option{-print-file-name=libgcc.a}.
4922 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4923 but you do want to link with @file{libgcc.a}. You can do
4926 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4929 @item -print-search-dirs
4930 @opindex print-search-dirs
4931 Print the name of the configured installation directory and a list of
4932 program and library directories @command{gcc} will search---and don't do anything else.
4934 This is useful when @command{gcc} prints the error message
4935 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4936 To resolve this you either need to put @file{cpp0} and the other compiler
4937 components where @command{gcc} expects to find them, or you can set the environment
4938 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4939 Don't forget the trailing @samp{/}.
4940 @xref{Environment Variables}.
4942 @item -print-sysroot-headers-suffix
4943 @opindex print-sysroot-headers-suffix
4944 Print the suffix added to the target sysroot when searching for
4945 headers, or give an error if the compiler is not configured with such
4946 a suffix---and don't do anything else.
4949 @opindex dumpmachine
4950 Print the compiler's target machine (for example,
4951 @samp{i686-pc-linux-gnu})---and don't do anything else.
4954 @opindex dumpversion
4955 Print the compiler version (for example, @samp{3.0})---and don't do
4960 Print the compiler's built-in specs---and don't do anything else. (This
4961 is used when GCC itself is being built.) @xref{Spec Files}.
4963 @item -feliminate-unused-debug-types
4964 @opindex feliminate-unused-debug-types
4965 Normally, when producing DWARF2 output, GCC will emit debugging
4966 information for all types declared in a compilation
4967 unit, regardless of whether or not they are actually used
4968 in that compilation unit. Sometimes this is useful, such as
4969 if, in the debugger, you want to cast a value to a type that is
4970 not actually used in your program (but is declared). More often,
4971 however, this results in a significant amount of wasted space.
4972 With this option, GCC will avoid producing debug symbol output
4973 for types that are nowhere used in the source file being compiled.
4976 @node Optimize Options
4977 @section Options That Control Optimization
4978 @cindex optimize options
4979 @cindex options, optimization
4981 These options control various sorts of optimizations.
4983 Without any optimization option, the compiler's goal is to reduce the
4984 cost of compilation and to make debugging produce the expected
4985 results. Statements are independent: if you stop the program with a
4986 breakpoint between statements, you can then assign a new value to any
4987 variable or change the program counter to any other statement in the
4988 function and get exactly the results you would expect from the source
4991 Turning on optimization flags makes the compiler attempt to improve
4992 the performance and/or code size at the expense of compilation time
4993 and possibly the ability to debug the program.
4995 The compiler performs optimization based on the knowledge it has of
4996 the program. Optimization levels @option{-O} and above, in
4997 particular, enable @emph{unit-at-a-time} mode, which allows the
4998 compiler to consider information gained from later functions in
4999 the file when compiling a function. Compiling multiple files at
5000 once to a single output file in @emph{unit-at-a-time} mode allows
5001 the compiler to use information gained from all of the files when
5002 compiling each of them.
5004 Not all optimizations are controlled directly by a flag. Only
5005 optimizations that have a flag are listed.
5012 Optimize. Optimizing compilation takes somewhat more time, and a lot
5013 more memory for a large function.
5015 With @option{-O}, the compiler tries to reduce code size and execution
5016 time, without performing any optimizations that take a great deal of
5019 @option{-O} turns on the following optimization flags:
5020 @gccoptlist{-fdefer-pop @gol
5021 -fdelayed-branch @gol
5022 -fguess-branch-probability @gol
5023 -fcprop-registers @gol
5024 -fif-conversion @gol
5025 -fif-conversion2 @gol
5026 -fsplit-wide-types @gol
5029 -ftree-dominator-opts @gol
5033 -ftree-copyrename @gol
5036 -funit-at-a-time @gol
5037 -finline-small-functions @gol
5040 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5041 where doing so does not interfere with debugging.
5045 Optimize even more. GCC performs nearly all supported optimizations
5046 that do not involve a space-speed tradeoff. The compiler does not
5047 perform loop unrolling or function inlining when you specify @option{-O2}.
5048 As compared to @option{-O}, this option increases both compilation time
5049 and the performance of the generated code.
5051 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5052 also turns on the following optimization flags:
5053 @gccoptlist{-fthread-jumps @gol
5055 -foptimize-sibling-calls @gol
5056 -fcse-follow-jumps -fcse-skip-blocks @gol
5057 -fgcse -fgcse-lm @gol
5058 -fexpensive-optimizations @gol
5059 -frerun-cse-after-loop @gol
5062 -fschedule-insns -fschedule-insns2 @gol
5063 -fsched-interblock -fsched-spec @gol
5065 -fstrict-aliasing -fstrict-overflow @gol
5066 -fdelete-null-pointer-checks @gol
5067 -freorder-blocks -freorder-functions @gol
5068 -falign-functions -falign-jumps @gol
5069 -falign-loops -falign-labels @gol
5073 Please note the warning under @option{-fgcse} about
5074 invoking @option{-O2} on programs that use computed gotos.
5078 Optimize yet more. @option{-O3} turns on all optimizations specified by
5079 @option{-O2} and also turns on the @option{-finline-functions},
5080 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5081 @option{-fgcse-after-reload} options.
5085 Reduce compilation time and make debugging produce the expected
5086 results. This is the default.
5090 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5091 do not typically increase code size. It also performs further
5092 optimizations designed to reduce code size.
5094 @option{-Os} disables the following optimization flags:
5095 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5096 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5097 -fprefetch-loop-arrays -ftree-vect-loop-version}
5099 If you use multiple @option{-O} options, with or without level numbers,
5100 the last such option is the one that is effective.
5103 Options of the form @option{-f@var{flag}} specify machine-independent
5104 flags. Most flags have both positive and negative forms; the negative
5105 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5106 below, only one of the forms is listed---the one you typically will
5107 use. You can figure out the other form by either removing @samp{no-}
5110 The following options control specific optimizations. They are either
5111 activated by @option{-O} options or are related to ones that are. You
5112 can use the following flags in the rare cases when ``fine-tuning'' of
5113 optimizations to be performed is desired.
5116 @item -fno-default-inline
5117 @opindex fno-default-inline
5118 Do not make member functions inline by default merely because they are
5119 defined inside the class scope (C++ only). Otherwise, when you specify
5120 @w{@option{-O}}, member functions defined inside class scope are compiled
5121 inline by default; i.e., you don't need to add @samp{inline} in front of
5122 the member function name.
5124 @item -fno-defer-pop
5125 @opindex fno-defer-pop
5126 Always pop the arguments to each function call as soon as that function
5127 returns. For machines which must pop arguments after a function call,
5128 the compiler normally lets arguments accumulate on the stack for several
5129 function calls and pops them all at once.
5131 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5134 @opindex fforce-addr
5135 Force memory address constants to be copied into registers before
5136 doing arithmetic on them.
5138 @item -fforward-propagate
5139 @opindex fforward-propagate
5140 Perform a forward propagation pass on RTL. The pass tries to combine two
5141 instructions and checks if the result can be simplified. If loop unrolling
5142 is active, two passes are performed and the second is scheduled after
5145 This option is enabled by default at optimization levels @option{-O2},
5146 @option{-O3}, @option{-Os}.
5148 @item -fomit-frame-pointer
5149 @opindex fomit-frame-pointer
5150 Don't keep the frame pointer in a register for functions that
5151 don't need one. This avoids the instructions to save, set up and
5152 restore frame pointers; it also makes an extra register available
5153 in many functions. @strong{It also makes debugging impossible on
5156 On some machines, such as the VAX, this flag has no effect, because
5157 the standard calling sequence automatically handles the frame pointer
5158 and nothing is saved by pretending it doesn't exist. The
5159 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5160 whether a target machine supports this flag. @xref{Registers,,Register
5161 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5163 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5165 @item -foptimize-sibling-calls
5166 @opindex foptimize-sibling-calls
5167 Optimize sibling and tail recursive calls.
5169 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5173 Don't pay attention to the @code{inline} keyword. Normally this option
5174 is used to keep the compiler from expanding any functions inline.
5175 Note that if you are not optimizing, no functions can be expanded inline.
5177 @item -finline-small-functions
5178 @opindex finline-small-functions
5179 Integrate functions into their callers when their body is smaller than expected
5180 function call code (so overall size of program gets smaller). The compiler
5181 heuristically decides which functions are simple enough to be worth integrating
5184 Enabled at level @option{-O2}.
5186 @item -finline-functions
5187 @opindex finline-functions
5188 Integrate all simple functions into their callers. The compiler
5189 heuristically decides which functions are simple enough to be worth
5190 integrating in this way.
5192 If all calls to a given function are integrated, and the function is
5193 declared @code{static}, then the function is normally not output as
5194 assembler code in its own right.
5196 Enabled at level @option{-O3}.
5198 @item -finline-functions-called-once
5199 @opindex finline-functions-called-once
5200 Consider all @code{static} functions called once for inlining into their
5201 caller even if they are not marked @code{inline}. If a call to a given
5202 function is integrated, then the function is not output as assembler code
5205 Enabled if @option{-funit-at-a-time} is enabled.
5207 @item -fearly-inlining
5208 @opindex fearly-inlining
5209 Inline functions marked by @code{always_inline} and functions whose body seems
5210 smaller than the function call overhead early before doing
5211 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5212 makes profiling significantly cheaper and usually inlining faster on programs
5213 having large chains of nested wrapper functions.
5217 @item -finline-limit=@var{n}
5218 @opindex finline-limit
5219 By default, GCC limits the size of functions that can be inlined. This flag
5220 allows the control of this limit for functions that are explicitly marked as
5221 inline (i.e., marked with the inline keyword or defined within the class
5222 definition in c++). @var{n} is the size of functions that can be inlined in
5223 number of pseudo instructions (not counting parameter handling). The default
5224 value of @var{n} is 600.
5225 Increasing this value can result in more inlined code at
5226 the cost of compilation time and memory consumption. Decreasing usually makes
5227 the compilation faster and less code will be inlined (which presumably
5228 means slower programs). This option is particularly useful for programs that
5229 use inlining heavily such as those based on recursive templates with C++.
5231 Inlining is actually controlled by a number of parameters, which may be
5232 specified individually by using @option{--param @var{name}=@var{value}}.
5233 The @option{-finline-limit=@var{n}} option sets some of these parameters
5237 @item max-inline-insns-single
5238 is set to @var{n}/2.
5239 @item max-inline-insns-auto
5240 is set to @var{n}/2.
5241 @item min-inline-insns
5242 is set to 130 or @var{n}/4, whichever is smaller.
5243 @item max-inline-insns-rtl
5247 See below for a documentation of the individual
5248 parameters controlling inlining.
5250 @emph{Note:} pseudo instruction represents, in this particular context, an
5251 abstract measurement of function's size. In no way does it represent a count
5252 of assembly instructions and as such its exact meaning might change from one
5253 release to an another.
5255 @item -fkeep-inline-functions
5256 @opindex fkeep-inline-functions
5257 In C, emit @code{static} functions that are declared @code{inline}
5258 into the object file, even if the function has been inlined into all
5259 of its callers. This switch does not affect functions using the
5260 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5261 inline functions into the object file.
5263 @item -fkeep-static-consts
5264 @opindex fkeep-static-consts
5265 Emit variables declared @code{static const} when optimization isn't turned
5266 on, even if the variables aren't referenced.
5268 GCC enables this option by default. If you want to force the compiler to
5269 check if the variable was referenced, regardless of whether or not
5270 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5272 @item -fmerge-constants
5273 Attempt to merge identical constants (string constants and floating point
5274 constants) across compilation units.
5276 This option is the default for optimized compilation if the assembler and
5277 linker support it. Use @option{-fno-merge-constants} to inhibit this
5280 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5282 @item -fmerge-all-constants
5283 Attempt to merge identical constants and identical variables.
5285 This option implies @option{-fmerge-constants}. In addition to
5286 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5287 arrays or initialized constant variables with integral or floating point
5288 types. Languages like C or C++ require each non-automatic variable to
5289 have distinct location, so using this option will result in non-conforming
5292 @item -fmodulo-sched
5293 @opindex fmodulo-sched
5294 Perform swing modulo scheduling immediately before the first scheduling
5295 pass. This pass looks at innermost loops and reorders their
5296 instructions by overlapping different iterations.
5298 @item -fmodulo-sched-allow-regmoves
5299 @opindex fmodulo-sched-allow-regmoves
5300 Perform more aggressive SMS based modulo scheduling with register moves
5301 allowed. By setting this flag certain anti-dependences edges will be
5302 deleted which will trigger the generation of reg-moves based on the
5303 life-range analysis. This option is effective only with
5304 @option{-fmodulo-sched} enabled.
5306 @item -fno-branch-count-reg
5307 @opindex fno-branch-count-reg
5308 Do not use ``decrement and branch'' instructions on a count register,
5309 but instead generate a sequence of instructions that decrement a
5310 register, compare it against zero, then branch based upon the result.
5311 This option is only meaningful on architectures that support such
5312 instructions, which include x86, PowerPC, IA-64 and S/390.
5314 The default is @option{-fbranch-count-reg}.
5316 @item -fno-function-cse
5317 @opindex fno-function-cse
5318 Do not put function addresses in registers; make each instruction that
5319 calls a constant function contain the function's address explicitly.
5321 This option results in less efficient code, but some strange hacks
5322 that alter the assembler output may be confused by the optimizations
5323 performed when this option is not used.
5325 The default is @option{-ffunction-cse}
5327 @item -fno-zero-initialized-in-bss
5328 @opindex fno-zero-initialized-in-bss
5329 If the target supports a BSS section, GCC by default puts variables that
5330 are initialized to zero into BSS@. This can save space in the resulting
5333 This option turns off this behavior because some programs explicitly
5334 rely on variables going to the data section. E.g., so that the
5335 resulting executable can find the beginning of that section and/or make
5336 assumptions based on that.
5338 The default is @option{-fzero-initialized-in-bss}.
5340 @item -fbounds-check
5341 @opindex fbounds-check
5342 For front-ends that support it, generate additional code to check that
5343 indices used to access arrays are within the declared range. This is
5344 currently only supported by the Java and Fortran front-ends, where
5345 this option defaults to true and false respectively.
5347 @item -fmudflap -fmudflapth -fmudflapir
5351 @cindex bounds checking
5353 For front-ends that support it (C and C++), instrument all risky
5354 pointer/array dereferencing operations, some standard library
5355 string/heap functions, and some other associated constructs with
5356 range/validity tests. Modules so instrumented should be immune to
5357 buffer overflows, invalid heap use, and some other classes of C/C++
5358 programming errors. The instrumentation relies on a separate runtime
5359 library (@file{libmudflap}), which will be linked into a program if
5360 @option{-fmudflap} is given at link time. Run-time behavior of the
5361 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5362 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5365 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5366 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5367 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5368 instrumentation should ignore pointer reads. This produces less
5369 instrumentation (and therefore faster execution) and still provides
5370 some protection against outright memory corrupting writes, but allows
5371 erroneously read data to propagate within a program.
5373 @item -fthread-jumps
5374 @opindex fthread-jumps
5375 Perform optimizations where we check to see if a jump branches to a
5376 location where another comparison subsumed by the first is found. If
5377 so, the first branch is redirected to either the destination of the
5378 second branch or a point immediately following it, depending on whether
5379 the condition is known to be true or false.
5381 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5383 @item -fsplit-wide-types
5384 @opindex fsplit-wide-types
5385 When using a type that occupies multiple registers, such as @code{long
5386 long} on a 32-bit system, split the registers apart and allocate them
5387 independently. This normally generates better code for those types,
5388 but may make debugging more difficult.
5390 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5393 @item -fcse-follow-jumps
5394 @opindex fcse-follow-jumps
5395 In common subexpression elimination, scan through jump instructions
5396 when the target of the jump is not reached by any other path. For
5397 example, when CSE encounters an @code{if} statement with an
5398 @code{else} clause, CSE will follow the jump when the condition
5401 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5403 @item -fcse-skip-blocks
5404 @opindex fcse-skip-blocks
5405 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5406 follow jumps which conditionally skip over blocks. When CSE
5407 encounters a simple @code{if} statement with no else clause,
5408 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5409 body of the @code{if}.
5411 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5413 @item -frerun-cse-after-loop
5414 @opindex frerun-cse-after-loop
5415 Re-run common subexpression elimination after loop optimizations has been
5418 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5422 Perform a global common subexpression elimination pass.
5423 This pass also performs global constant and copy propagation.
5425 @emph{Note:} When compiling a program using computed gotos, a GCC
5426 extension, you may get better runtime performance if you disable
5427 the global common subexpression elimination pass by adding
5428 @option{-fno-gcse} to the command line.
5430 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5434 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5435 attempt to move loads which are only killed by stores into themselves. This
5436 allows a loop containing a load/store sequence to be changed to a load outside
5437 the loop, and a copy/store within the loop.
5439 Enabled by default when gcse is enabled.
5443 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5444 global common subexpression elimination. This pass will attempt to move
5445 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5446 loops containing a load/store sequence can be changed to a load before
5447 the loop and a store after the loop.
5449 Not enabled at any optimization level.
5453 When @option{-fgcse-las} is enabled, the global common subexpression
5454 elimination pass eliminates redundant loads that come after stores to the
5455 same memory location (both partial and full redundancies).
5457 Not enabled at any optimization level.
5459 @item -fgcse-after-reload
5460 @opindex fgcse-after-reload
5461 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5462 pass is performed after reload. The purpose of this pass is to cleanup
5465 @item -funsafe-loop-optimizations
5466 @opindex funsafe-loop-optimizations
5467 If given, the loop optimizer will assume that loop indices do not
5468 overflow, and that the loops with nontrivial exit condition are not
5469 infinite. This enables a wider range of loop optimizations even if
5470 the loop optimizer itself cannot prove that these assumptions are valid.
5471 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5472 if it finds this kind of loop.
5474 @item -fcrossjumping
5475 @opindex crossjumping
5476 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5477 resulting code may or may not perform better than without cross-jumping.
5479 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5481 @item -fif-conversion
5482 @opindex if-conversion
5483 Attempt to transform conditional jumps into branch-less equivalents. This
5484 include use of conditional moves, min, max, set flags and abs instructions, and
5485 some tricks doable by standard arithmetics. The use of conditional execution
5486 on chips where it is available is controlled by @code{if-conversion2}.
5488 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5490 @item -fif-conversion2
5491 @opindex if-conversion2
5492 Use conditional execution (where available) to transform conditional jumps into
5493 branch-less equivalents.
5495 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5497 @item -fdelete-null-pointer-checks
5498 @opindex fdelete-null-pointer-checks
5499 Use global dataflow analysis to identify and eliminate useless checks
5500 for null pointers. The compiler assumes that dereferencing a null
5501 pointer would have halted the program. If a pointer is checked after
5502 it has already been dereferenced, it cannot be null.
5504 In some environments, this assumption is not true, and programs can
5505 safely dereference null pointers. Use
5506 @option{-fno-delete-null-pointer-checks} to disable this optimization
5507 for programs which depend on that behavior.
5509 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5511 @item -fexpensive-optimizations
5512 @opindex fexpensive-optimizations
5513 Perform a number of minor optimizations that are relatively expensive.
5515 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5517 @item -foptimize-register-move
5519 @opindex foptimize-register-move
5521 Attempt to reassign register numbers in move instructions and as
5522 operands of other simple instructions in order to maximize the amount of
5523 register tying. This is especially helpful on machines with two-operand
5526 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5529 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5531 @item -fdelayed-branch
5532 @opindex fdelayed-branch
5533 If supported for the target machine, attempt to reorder instructions
5534 to exploit instruction slots available after delayed branch
5537 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5539 @item -fschedule-insns
5540 @opindex fschedule-insns
5541 If supported for the target machine, attempt to reorder instructions to
5542 eliminate execution stalls due to required data being unavailable. This
5543 helps machines that have slow floating point or memory load instructions
5544 by allowing other instructions to be issued until the result of the load
5545 or floating point instruction is required.
5547 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5549 @item -fschedule-insns2
5550 @opindex fschedule-insns2
5551 Similar to @option{-fschedule-insns}, but requests an additional pass of
5552 instruction scheduling after register allocation has been done. This is
5553 especially useful on machines with a relatively small number of
5554 registers and where memory load instructions take more than one cycle.
5556 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5558 @item -fno-sched-interblock
5559 @opindex fno-sched-interblock
5560 Don't schedule instructions across basic blocks. This is normally
5561 enabled by default when scheduling before register allocation, i.e.@:
5562 with @option{-fschedule-insns} or at @option{-O2} or higher.
5564 @item -fno-sched-spec
5565 @opindex fno-sched-spec
5566 Don't allow speculative motion of non-load instructions. This is normally
5567 enabled by default when scheduling before register allocation, i.e.@:
5568 with @option{-fschedule-insns} or at @option{-O2} or higher.
5570 @item -fsched-spec-load
5571 @opindex fsched-spec-load
5572 Allow speculative motion of some load instructions. This only makes
5573 sense when scheduling before register allocation, i.e.@: with
5574 @option{-fschedule-insns} or at @option{-O2} or higher.
5576 @item -fsched-spec-load-dangerous
5577 @opindex fsched-spec-load-dangerous
5578 Allow speculative motion of more load instructions. This only makes
5579 sense when scheduling before register allocation, i.e.@: with
5580 @option{-fschedule-insns} or at @option{-O2} or higher.
5582 @item -fsched-stalled-insns=@var{n}
5583 @opindex fsched-stalled-insns
5584 Define how many insns (if any) can be moved prematurely from the queue
5585 of stalled insns into the ready list, during the second scheduling pass.
5587 @item -fsched-stalled-insns-dep=@var{n}
5588 @opindex fsched-stalled-insns-dep
5589 Define how many insn groups (cycles) will be examined for a dependency
5590 on a stalled insn that is candidate for premature removal from the queue
5591 of stalled insns. Has an effect only during the second scheduling pass,
5592 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5594 @item -fsched2-use-superblocks
5595 @opindex fsched2-use-superblocks
5596 When scheduling after register allocation, do use superblock scheduling
5597 algorithm. Superblock scheduling allows motion across basic block boundaries
5598 resulting on faster schedules. This option is experimental, as not all machine
5599 descriptions used by GCC model the CPU closely enough to avoid unreliable
5600 results from the algorithm.
5602 This only makes sense when scheduling after register allocation, i.e.@: with
5603 @option{-fschedule-insns2} or at @option{-O2} or higher.
5605 @item -fsched2-use-traces
5606 @opindex fsched2-use-traces
5607 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5608 allocation and additionally perform code duplication in order to increase the
5609 size of superblocks using tracer pass. See @option{-ftracer} for details on
5612 This mode should produce faster but significantly longer programs. Also
5613 without @option{-fbranch-probabilities} the traces constructed may not
5614 match the reality and hurt the performance. This only makes
5615 sense when scheduling after register allocation, i.e.@: with
5616 @option{-fschedule-insns2} or at @option{-O2} or higher.
5620 Eliminates redundant extension instructions and move the non redundant
5621 ones to optimal placement using LCM.
5623 @item -freschedule-modulo-scheduled-loops
5624 @opindex fscheduling-in-modulo-scheduled-loops
5625 The modulo scheduling comes before the traditional scheduling, if a loop
5626 was modulo scheduled we may want to prevent the later scheduling passes
5627 from changing its schedule, we use this option to control that.
5629 @item -fcaller-saves
5630 @opindex fcaller-saves
5631 Enable values to be allocated in registers that will be clobbered by
5632 function calls, by emitting extra instructions to save and restore the
5633 registers around such calls. Such allocation is done only when it
5634 seems to result in better code than would otherwise be produced.
5636 This option is always enabled by default on certain machines, usually
5637 those which have no call-preserved registers to use instead.
5639 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5641 @item -ftree-reassoc
5642 Perform Reassociation on trees This flag is enabled by default
5643 at @option{-O} and higher.
5646 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5647 enabled by default at @option{-O2} and @option{-O3}.
5650 Perform Full Redundancy Elimination (FRE) on trees. The difference
5651 between FRE and PRE is that FRE only considers expressions
5652 that are computed on all paths leading to the redundant computation.
5653 This analysis is faster than PRE, though it exposes fewer redundancies.
5654 This flag is enabled by default at @option{-O} and higher.
5656 @item -ftree-copy-prop
5657 Perform copy propagation on trees. This pass eliminates unnecessary
5658 copy operations. This flag is enabled by default at @option{-O} and
5661 @item -ftree-store-copy-prop
5662 Perform copy propagation of memory loads and stores. This pass
5663 eliminates unnecessary copy operations in memory references
5664 (structures, global variables, arrays, etc). This flag is enabled by
5665 default at @option{-O2} and higher.
5668 Perform structural alias analysis on trees. This flag
5669 is enabled by default at @option{-O} and higher.
5672 Perform interprocedural pointer analysis.
5675 Perform forward store motion on trees. This flag is
5676 enabled by default at @option{-O} and higher.
5679 Perform sparse conditional constant propagation (CCP) on trees. This
5680 pass only operates on local scalar variables and is enabled by default
5681 at @option{-O} and higher.
5683 @item -ftree-store-ccp
5684 Perform sparse conditional constant propagation (CCP) on trees. This
5685 pass operates on both local scalar variables and memory stores and
5686 loads (global variables, structures, arrays, etc). This flag is
5687 enabled by default at @option{-O2} and higher.
5690 Perform dead code elimination (DCE) on trees. This flag is enabled by
5691 default at @option{-O} and higher.
5693 @item -ftree-dominator-opts
5694 Perform a variety of simple scalar cleanups (constant/copy
5695 propagation, redundancy elimination, range propagation and expression
5696 simplification) based on a dominator tree traversal. This also
5697 performs jump threading (to reduce jumps to jumps). This flag is
5698 enabled by default at @option{-O} and higher.
5702 Perform dead store elimination (DSE) on trees. A dead store is a store into
5703 a memory location which will later be overwritten by another store without
5704 any intervening loads. In this case the earlier store can be deleted. This
5705 flag is enabled by default at @option{-O} and higher.
5708 Perform loop header copying on trees. This is beneficial since it increases
5709 effectiveness of code motion optimizations. It also saves one jump. This flag
5710 is enabled by default at @option{-O} and higher. It is not enabled
5711 for @option{-Os}, since it usually increases code size.
5713 @item -ftree-loop-optimize
5714 Perform loop optimizations on trees. This flag is enabled by default
5715 at @option{-O} and higher.
5717 @item -ftree-loop-linear
5718 Perform linear loop transformations on tree. This flag can improve cache
5719 performance and allow further loop optimizations to take place.
5721 @item -fcheck-data-deps
5722 Compare the results of several data dependence analyzers. This option
5723 is used for debugging the data dependence analyzers.
5725 @item -ftree-loop-im
5726 Perform loop invariant motion on trees. This pass moves only invariants that
5727 would be hard to handle at RTL level (function calls, operations that expand to
5728 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5729 operands of conditions that are invariant out of the loop, so that we can use
5730 just trivial invariantness analysis in loop unswitching. The pass also includes
5733 @item -ftree-loop-ivcanon
5734 Create a canonical counter for number of iterations in the loop for that
5735 determining number of iterations requires complicated analysis. Later
5736 optimizations then may determine the number easily. Useful especially
5737 in connection with unrolling.
5740 Perform induction variable optimizations (strength reduction, induction
5741 variable merging and induction variable elimination) on trees.
5744 Perform scalar replacement of aggregates. This pass replaces structure
5745 references with scalars to prevent committing structures to memory too
5746 early. This flag is enabled by default at @option{-O} and higher.
5748 @item -ftree-copyrename
5749 Perform copy renaming on trees. This pass attempts to rename compiler
5750 temporaries to other variables at copy locations, usually resulting in
5751 variable names which more closely resemble the original variables. This flag
5752 is enabled by default at @option{-O} and higher.
5755 Perform temporary expression replacement during the SSA->normal phase. Single
5756 use/single def temporaries are replaced at their use location with their
5757 defining expression. This results in non-GIMPLE code, but gives the expanders
5758 much more complex trees to work on resulting in better RTL generation. This is
5759 enabled by default at @option{-O} and higher.
5761 @item -ftree-vectorize
5762 Perform loop vectorization on trees.
5764 @item -ftree-vect-loop-version
5765 @opindex ftree-vect-loop-version
5766 Perform loop versioning when doing loop vectorization on trees. When a loop
5767 appears to be vectorizable except that data alignment or data dependence cannot
5768 be determined at compile time then vectorized and non-vectorized versions of
5769 the loop are generated along with runtime checks for alignment or dependence
5770 to control which version is executed. This option is enabled by default
5771 except at level @option{-Os} where it is disabled.
5773 @item -fvect-cost-model
5774 Enable cost model for vectorization.
5777 Perform Value Range Propagation on trees. This is similar to the
5778 constant propagation pass, but instead of values, ranges of values are
5779 propagated. This allows the optimizers to remove unnecessary range
5780 checks like array bound checks and null pointer checks. This is
5781 enabled by default at @option{-O2} and higher. Null pointer check
5782 elimination is only done if @option{-fdelete-null-pointer-checks} is
5787 Perform tail duplication to enlarge superblock size. This transformation
5788 simplifies the control flow of the function allowing other optimizations to do
5791 @item -funroll-loops
5792 @opindex funroll-loops
5793 Unroll loops whose number of iterations can be determined at compile
5794 time or upon entry to the loop. @option{-funroll-loops} implies
5795 @option{-frerun-cse-after-loop}. This option makes code larger,
5796 and may or may not make it run faster.
5798 @item -funroll-all-loops
5799 @opindex funroll-all-loops
5800 Unroll all loops, even if their number of iterations is uncertain when
5801 the loop is entered. This usually makes programs run more slowly.
5802 @option{-funroll-all-loops} implies the same options as
5803 @option{-funroll-loops},
5805 @item -fsplit-ivs-in-unroller
5806 @opindex fsplit-ivs-in-unroller
5807 Enables expressing of values of induction variables in later iterations
5808 of the unrolled loop using the value in the first iteration. This breaks
5809 long dependency chains, thus improving efficiency of the scheduling passes.
5811 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5812 same effect. However in cases the loop body is more complicated than
5813 a single basic block, this is not reliable. It also does not work at all
5814 on some of the architectures due to restrictions in the CSE pass.
5816 This optimization is enabled by default.
5818 @item -fvariable-expansion-in-unroller
5819 @opindex fvariable-expansion-in-unroller
5820 With this option, the compiler will create multiple copies of some
5821 local variables when unrolling a loop which can result in superior code.
5823 @item -fpredictive-commoning
5824 @opindex fpredictive-commoning
5825 Perform predictive commoning optimization, i.e., reusing computations
5826 (especially memory loads and stores) performed in previous
5827 iterations of loops.
5829 This option is enabled at level @option{-O3}.
5831 @item -fprefetch-loop-arrays
5832 @opindex fprefetch-loop-arrays
5833 If supported by the target machine, generate instructions to prefetch
5834 memory to improve the performance of loops that access large arrays.
5836 This option may generate better or worse code; results are highly
5837 dependent on the structure of loops within the source code.
5839 Disabled at level @option{-Os}.
5842 @itemx -fno-peephole2
5843 @opindex fno-peephole
5844 @opindex fno-peephole2
5845 Disable any machine-specific peephole optimizations. The difference
5846 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5847 are implemented in the compiler; some targets use one, some use the
5848 other, a few use both.
5850 @option{-fpeephole} is enabled by default.
5851 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5853 @item -fno-guess-branch-probability
5854 @opindex fno-guess-branch-probability
5855 Do not guess branch probabilities using heuristics.
5857 GCC will use heuristics to guess branch probabilities if they are
5858 not provided by profiling feedback (@option{-fprofile-arcs}). These
5859 heuristics are based on the control flow graph. If some branch probabilities
5860 are specified by @samp{__builtin_expect}, then the heuristics will be
5861 used to guess branch probabilities for the rest of the control flow graph,
5862 taking the @samp{__builtin_expect} info into account. The interactions
5863 between the heuristics and @samp{__builtin_expect} can be complex, and in
5864 some cases, it may be useful to disable the heuristics so that the effects
5865 of @samp{__builtin_expect} are easier to understand.
5867 The default is @option{-fguess-branch-probability} at levels
5868 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5870 @item -freorder-blocks
5871 @opindex freorder-blocks
5872 Reorder basic blocks in the compiled function in order to reduce number of
5873 taken branches and improve code locality.
5875 Enabled at levels @option{-O2}, @option{-O3}.
5877 @item -freorder-blocks-and-partition
5878 @opindex freorder-blocks-and-partition
5879 In addition to reordering basic blocks in the compiled function, in order
5880 to reduce number of taken branches, partitions hot and cold basic blocks
5881 into separate sections of the assembly and .o files, to improve
5882 paging and cache locality performance.
5884 This optimization is automatically turned off in the presence of
5885 exception handling, for linkonce sections, for functions with a user-defined
5886 section attribute and on any architecture that does not support named
5889 @item -freorder-functions
5890 @opindex freorder-functions
5891 Reorder functions in the object file in order to
5892 improve code locality. This is implemented by using special
5893 subsections @code{.text.hot} for most frequently executed functions and
5894 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5895 the linker so object file format must support named sections and linker must
5896 place them in a reasonable way.
5898 Also profile feedback must be available in to make this option effective. See
5899 @option{-fprofile-arcs} for details.
5901 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5903 @item -fstrict-aliasing
5904 @opindex fstrict-aliasing
5905 Allows the compiler to assume the strictest aliasing rules applicable to
5906 the language being compiled. For C (and C++), this activates
5907 optimizations based on the type of expressions. In particular, an
5908 object of one type is assumed never to reside at the same address as an
5909 object of a different type, unless the types are almost the same. For
5910 example, an @code{unsigned int} can alias an @code{int}, but not a
5911 @code{void*} or a @code{double}. A character type may alias any other
5914 Pay special attention to code like this:
5927 The practice of reading from a different union member than the one most
5928 recently written to (called ``type-punning'') is common. Even with
5929 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5930 is accessed through the union type. So, the code above will work as
5931 expected. However, this code might not:
5942 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5944 @item -fstrict-overflow
5945 @opindex fstrict-overflow
5946 Allow the compiler to assume strict signed overflow rules, depending
5947 on the language being compiled. For C (and C++) this means that
5948 overflow when doing arithmetic with signed numbers is undefined, which
5949 means that the compiler may assume that it will not happen. This
5950 permits various optimizations. For example, the compiler will assume
5951 that an expression like @code{i + 10 > i} will always be true for
5952 signed @code{i}. This assumption is only valid if signed overflow is
5953 undefined, as the expression is false if @code{i + 10} overflows when
5954 using twos complement arithmetic. When this option is in effect any
5955 attempt to determine whether an operation on signed numbers will
5956 overflow must be written carefully to not actually involve overflow.
5958 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5959 that signed overflow is fully defined: it wraps. When
5960 @option{-fwrapv} is used, there is no difference between
5961 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5962 @option{-fwrapv} certain types of overflow are permitted. For
5963 example, if the compiler gets an overflow when doing arithmetic on
5964 constants, the overflowed value can still be used with
5965 @option{-fwrapv}, but not otherwise.
5967 The @option{-fstrict-overflow} option is enabled at levels
5968 @option{-O2}, @option{-O3}, @option{-Os}.
5970 @item -falign-functions
5971 @itemx -falign-functions=@var{n}
5972 @opindex falign-functions
5973 Align the start of functions to the next power-of-two greater than
5974 @var{n}, skipping up to @var{n} bytes. For instance,
5975 @option{-falign-functions=32} aligns functions to the next 32-byte
5976 boundary, but @option{-falign-functions=24} would align to the next
5977 32-byte boundary only if this can be done by skipping 23 bytes or less.
5979 @option{-fno-align-functions} and @option{-falign-functions=1} are
5980 equivalent and mean that functions will not be aligned.
5982 Some assemblers only support this flag when @var{n} is a power of two;
5983 in that case, it is rounded up.
5985 If @var{n} is not specified or is zero, use a machine-dependent default.
5987 Enabled at levels @option{-O2}, @option{-O3}.
5989 @item -falign-labels
5990 @itemx -falign-labels=@var{n}
5991 @opindex falign-labels
5992 Align all branch targets to a power-of-two boundary, skipping up to
5993 @var{n} bytes like @option{-falign-functions}. This option can easily
5994 make code slower, because it must insert dummy operations for when the
5995 branch target is reached in the usual flow of the code.
5997 @option{-fno-align-labels} and @option{-falign-labels=1} are
5998 equivalent and mean that labels will not be aligned.
6000 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6001 are greater than this value, then their values are used instead.
6003 If @var{n} is not specified or is zero, use a machine-dependent default
6004 which is very likely to be @samp{1}, meaning no alignment.
6006 Enabled at levels @option{-O2}, @option{-O3}.
6009 @itemx -falign-loops=@var{n}
6010 @opindex falign-loops
6011 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6012 like @option{-falign-functions}. The hope is that the loop will be
6013 executed many times, which will make up for any execution of the dummy
6016 @option{-fno-align-loops} and @option{-falign-loops=1} are
6017 equivalent and mean that loops will not be aligned.
6019 If @var{n} is not specified or is zero, use a machine-dependent default.
6021 Enabled at levels @option{-O2}, @option{-O3}.
6024 @itemx -falign-jumps=@var{n}
6025 @opindex falign-jumps
6026 Align branch targets to a power-of-two boundary, for branch targets
6027 where the targets can only be reached by jumping, skipping up to @var{n}
6028 bytes like @option{-falign-functions}. In this case, no dummy operations
6031 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6032 equivalent and mean that loops will not be aligned.
6034 If @var{n} is not specified or is zero, use a machine-dependent default.
6036 Enabled at levels @option{-O2}, @option{-O3}.
6038 @item -funit-at-a-time
6039 @opindex funit-at-a-time
6040 Parse the whole compilation unit before starting to produce code.
6041 This allows some extra optimizations to take place but consumes
6042 more memory (in general). There are some compatibility issues
6043 with @emph{unit-at-a-time} mode:
6046 enabling @emph{unit-at-a-time} mode may change the order
6047 in which functions, variables, and top-level @code{asm} statements
6048 are emitted, and will likely break code relying on some particular
6049 ordering. The majority of such top-level @code{asm} statements,
6050 though, can be replaced by @code{section} attributes. The
6051 @option{fno-toplevel-reorder} option may be used to keep the ordering
6052 used in the input file, at the cost of some optimizations.
6055 @emph{unit-at-a-time} mode removes unreferenced static variables
6056 and functions. This may result in undefined references
6057 when an @code{asm} statement refers directly to variables or functions
6058 that are otherwise unused. In that case either the variable/function
6059 shall be listed as an operand of the @code{asm} statement operand or,
6060 in the case of top-level @code{asm} statements the attribute @code{used}
6061 shall be used on the declaration.
6064 Static functions now can use non-standard passing conventions that
6065 may break @code{asm} statements calling functions directly. Again,
6066 attribute @code{used} will prevent this behavior.
6069 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6070 but this scheme may not be supported by future releases of GCC@.
6072 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6074 @item -fno-toplevel-reorder
6075 Do not reorder top-level functions, variables, and @code{asm}
6076 statements. Output them in the same order that they appear in the
6077 input file. When this option is used, unreferenced static variables
6078 will not be removed. This option is intended to support existing code
6079 which relies on a particular ordering. For new code, it is better to
6084 Constructs webs as commonly used for register allocation purposes and assign
6085 each web individual pseudo register. This allows the register allocation pass
6086 to operate on pseudos directly, but also strengthens several other optimization
6087 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6088 however, make debugging impossible, since variables will no longer stay in a
6091 Enabled by default with @option{-funroll-loops}.
6093 @item -fwhole-program
6094 @opindex fwhole-program
6095 Assume that the current compilation unit represents whole program being
6096 compiled. All public functions and variables with the exception of @code{main}
6097 and those merged by attribute @code{externally_visible} become static functions
6098 and in a affect gets more aggressively optimized by interprocedural optimizers.
6099 While this option is equivalent to proper use of @code{static} keyword for
6100 programs consisting of single file, in combination with option
6101 @option{--combine} this flag can be used to compile most of smaller scale C
6102 programs since the functions and variables become local for the whole combined
6103 compilation unit, not for the single source file itself.
6106 @item -fno-cprop-registers
6107 @opindex fno-cprop-registers
6108 After register allocation and post-register allocation instruction splitting,
6109 we perform a copy-propagation pass to try to reduce scheduling dependencies
6110 and occasionally eliminate the copy.
6112 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6114 @item -fprofile-generate
6115 @opindex fprofile-generate
6117 Enable options usually used for instrumenting application to produce
6118 profile useful for later recompilation with profile feedback based
6119 optimization. You must use @option{-fprofile-generate} both when
6120 compiling and when linking your program.
6122 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6125 @opindex fprofile-use
6126 Enable profile feedback directed optimizations, and optimizations
6127 generally profitable only with profile feedback available.
6129 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6130 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6132 By default, GCC emits an error message if the feedback profiles do not
6133 match the source code. This error can be turned into a warning by using
6134 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6138 The following options control compiler behavior regarding floating
6139 point arithmetic. These options trade off between speed and
6140 correctness. All must be specifically enabled.
6144 @opindex ffloat-store
6145 Do not store floating point variables in registers, and inhibit other
6146 options that might change whether a floating point value is taken from a
6149 @cindex floating point precision
6150 This option prevents undesirable excess precision on machines such as
6151 the 68000 where the floating registers (of the 68881) keep more
6152 precision than a @code{double} is supposed to have. Similarly for the
6153 x86 architecture. For most programs, the excess precision does only
6154 good, but a few programs rely on the precise definition of IEEE floating
6155 point. Use @option{-ffloat-store} for such programs, after modifying
6156 them to store all pertinent intermediate computations into variables.
6160 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
6161 @option{-fno-trapping-math}, @option{-ffinite-math-only},
6162 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
6163 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
6165 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6167 This option is not turned on by any @option{-O} option since
6168 it can result in incorrect output for programs which depend on
6169 an exact implementation of IEEE or ISO rules/specifications for
6170 math functions. It may, however, yield faster code for programs
6171 that do not require the guarantees of these specifications.
6173 @item -fno-math-errno
6174 @opindex fno-math-errno
6175 Do not set ERRNO after calling math functions that are executed
6176 with a single instruction, e.g., sqrt. A program that relies on
6177 IEEE exceptions for math error handling may want to use this flag
6178 for speed while maintaining IEEE arithmetic compatibility.
6180 This option is not turned on by any @option{-O} option since
6181 it can result in incorrect output for programs which depend on
6182 an exact implementation of IEEE or ISO rules/specifications for
6183 math functions. It may, however, yield faster code for programs
6184 that do not require the guarantees of these specifications.
6186 The default is @option{-fmath-errno}.
6188 On Darwin systems, the math library never sets @code{errno}. There is
6189 therefore no reason for the compiler to consider the possibility that
6190 it might, and @option{-fno-math-errno} is the default.
6192 @item -funsafe-math-optimizations
6193 @opindex funsafe-math-optimizations
6195 Allow optimizations for floating-point arithmetic that (a) assume
6196 that arguments and results are valid and (b) may violate IEEE or
6197 ANSI standards. When used at link-time, it may include libraries
6198 or startup files that change the default FPU control word or other
6199 similar optimizations.
6201 This option is not turned on by any @option{-O} option since
6202 it can result in incorrect output for programs which depend on
6203 an exact implementation of IEEE or ISO rules/specifications for
6204 math functions. It may, however, yield faster code for programs
6205 that do not require the guarantees of these specifications.
6206 Enables @option{-freciprocal-math} and @option{-fassociative-math}.
6208 The default is @option{-fno-unsafe-math-optimizations}.
6210 @item -fassociative-math
6211 @opindex -fassociative-math
6213 Allow re-association of operands in series of floating-point operations.
6214 This violates the ISO C and C++ language standard by possibly changing
6215 computation result. NOTE: re-ordering may change the sign of zero as
6216 well as ignore NaNs and inhibit or create underflow or overflow (and
6217 thus cannot be used on a code which relies on rounding behavior like
6218 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6219 and thus may not be used when ordered comparisons are required.
6220 This flag doesn't make much sense without @option{-fno-signed-zeros}
6221 or @option{-fno-trapping-math} or with @option{-frounding-math}.
6223 The default is @option{-fno-associative-math}.
6225 @item -freciprocal-math
6226 @opindex -freciprocal-math
6228 Allow the reciprocal of a value to be used instead of dividing by
6229 the value if this enables optimizations. For example @code{x / y}
6230 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6231 is subject to common subexpression elimination. Note that this loses
6232 precision and increases the number of flops operating on the value.
6234 The default is @option{-fno-reciprocal-math}.
6236 @item -ffinite-math-only
6237 @opindex ffinite-math-only
6238 Allow optimizations for floating-point arithmetic that assume
6239 that arguments and results are not NaNs or +-Infs.
6241 This option is not turned on by any @option{-O} option since
6242 it can result in incorrect output for programs which depend on
6243 an exact implementation of IEEE or ISO rules/specifications for
6244 math functions. It may, however, yield faster code for programs
6245 that do not require the guarantees of these specifications.
6247 The default is @option{-fno-finite-math-only}.
6249 @item -fno-signed-zeros
6250 @opindex fno-signed-zeros
6251 Allow optimizations for floating point arithmetic that ignore the
6252 signedness of zero. IEEE arithmetic specifies the behavior of
6253 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6254 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6255 This option implies that the sign of a zero result isn't significant.
6257 The default is @option{-fsigned-zeros}.
6259 @item -fno-trapping-math
6260 @opindex fno-trapping-math
6261 Compile code assuming that floating-point operations cannot generate
6262 user-visible traps. These traps include division by zero, overflow,
6263 underflow, inexact result and invalid operation. This option implies
6264 @option{-fno-signaling-nans}. Setting this option may allow faster
6265 code if one relies on ``non-stop'' IEEE arithmetic, for example.
6267 This option should never be turned on by any @option{-O} option since
6268 it can result in incorrect output for programs which depend on
6269 an exact implementation of IEEE or ISO rules/specifications for
6272 The default is @option{-ftrapping-math}.
6274 @item -frounding-math
6275 @opindex frounding-math
6276 Disable transformations and optimizations that assume default floating
6277 point rounding behavior. This is round-to-zero for all floating point
6278 to integer conversions, and round-to-nearest for all other arithmetic
6279 truncations. This option should be specified for programs that change
6280 the FP rounding mode dynamically, or that may be executed with a
6281 non-default rounding mode. This option disables constant folding of
6282 floating point expressions at compile-time (which may be affected by
6283 rounding mode) and arithmetic transformations that are unsafe in the
6284 presence of sign-dependent rounding modes.
6286 The default is @option{-fno-rounding-math}.
6288 This option is experimental and does not currently guarantee to
6289 disable all GCC optimizations that are affected by rounding mode.
6290 Future versions of GCC may provide finer control of this setting
6291 using C99's @code{FENV_ACCESS} pragma. This command line option
6292 will be used to specify the default state for @code{FENV_ACCESS}.
6294 @item -frtl-abstract-sequences
6295 @opindex frtl-abstract-sequences
6296 It is a size optimization method. This option is to find identical
6297 sequences of code, which can be turned into pseudo-procedures and
6298 then replace all occurrences with calls to the newly created
6299 subroutine. It is kind of an opposite of @option{-finline-functions}.
6300 This optimization runs at RTL level.
6302 @item -fsignaling-nans
6303 @opindex fsignaling-nans
6304 Compile code assuming that IEEE signaling NaNs may generate user-visible
6305 traps during floating-point operations. Setting this option disables
6306 optimizations that may change the number of exceptions visible with
6307 signaling NaNs. This option implies @option{-ftrapping-math}.
6309 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6312 The default is @option{-fno-signaling-nans}.
6314 This option is experimental and does not currently guarantee to
6315 disable all GCC optimizations that affect signaling NaN behavior.
6317 @item -fsingle-precision-constant
6318 @opindex fsingle-precision-constant
6319 Treat floating point constant as single precision constant instead of
6320 implicitly converting it to double precision constant.
6322 @item -fcx-limited-range
6323 @itemx -fno-cx-limited-range
6324 @opindex fcx-limited-range
6325 @opindex fno-cx-limited-range
6326 When enabled, this option states that a range reduction step is not
6327 needed when performing complex division. The default is
6328 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6330 This option controls the default setting of the ISO C99
6331 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6336 The following options control optimizations that may improve
6337 performance, but are not enabled by any @option{-O} options. This
6338 section includes experimental options that may produce broken code.
6341 @item -fbranch-probabilities
6342 @opindex fbranch-probabilities
6343 After running a program compiled with @option{-fprofile-arcs}
6344 (@pxref{Debugging Options,, Options for Debugging Your Program or
6345 @command{gcc}}), you can compile it a second time using
6346 @option{-fbranch-probabilities}, to improve optimizations based on
6347 the number of times each branch was taken. When the program
6348 compiled with @option{-fprofile-arcs} exits it saves arc execution
6349 counts to a file called @file{@var{sourcename}.gcda} for each source
6350 file. The information in this data file is very dependent on the
6351 structure of the generated code, so you must use the same source code
6352 and the same optimization options for both compilations.
6354 With @option{-fbranch-probabilities}, GCC puts a
6355 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6356 These can be used to improve optimization. Currently, they are only
6357 used in one place: in @file{reorg.c}, instead of guessing which path a
6358 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6359 exactly determine which path is taken more often.
6361 @item -fprofile-values
6362 @opindex fprofile-values
6363 If combined with @option{-fprofile-arcs}, it adds code so that some
6364 data about values of expressions in the program is gathered.
6366 With @option{-fbranch-probabilities}, it reads back the data gathered
6367 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6368 notes to instructions for their later usage in optimizations.
6370 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6374 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6375 a code to gather information about values of expressions.
6377 With @option{-fbranch-probabilities}, it reads back the data gathered
6378 and actually performs the optimizations based on them.
6379 Currently the optimizations include specialization of division operation
6380 using the knowledge about the value of the denominator.
6382 @item -frename-registers
6383 @opindex frename-registers
6384 Attempt to avoid false dependencies in scheduled code by making use
6385 of registers left over after register allocation. This optimization
6386 will most benefit processors with lots of registers. Depending on the
6387 debug information format adopted by the target, however, it can
6388 make debugging impossible, since variables will no longer stay in
6389 a ``home register''.
6391 Enabled by default with @option{-funroll-loops}.
6395 Perform tail duplication to enlarge superblock size. This transformation
6396 simplifies the control flow of the function allowing other optimizations to do
6399 Enabled with @option{-fprofile-use}.
6401 @item -funroll-loops
6402 @opindex funroll-loops
6403 Unroll loops whose number of iterations can be determined at compile time or
6404 upon entry to the loop. @option{-funroll-loops} implies
6405 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6406 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6407 small constant number of iterations). This option makes code larger, and may
6408 or may not make it run faster.
6410 Enabled with @option{-fprofile-use}.
6412 @item -funroll-all-loops
6413 @opindex funroll-all-loops
6414 Unroll all loops, even if their number of iterations is uncertain when
6415 the loop is entered. This usually makes programs run more slowly.
6416 @option{-funroll-all-loops} implies the same options as
6417 @option{-funroll-loops}.
6420 @opindex fpeel-loops
6421 Peels the loops for that there is enough information that they do not
6422 roll much (from profile feedback). It also turns on complete loop peeling
6423 (i.e.@: complete removal of loops with small constant number of iterations).
6425 Enabled with @option{-fprofile-use}.
6427 @item -fmove-loop-invariants
6428 @opindex fmove-loop-invariants
6429 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6430 at level @option{-O1}
6432 @item -funswitch-loops
6433 @opindex funswitch-loops
6434 Move branches with loop invariant conditions out of the loop, with duplicates
6435 of the loop on both branches (modified according to result of the condition).
6437 @item -ffunction-sections
6438 @itemx -fdata-sections
6439 @opindex ffunction-sections
6440 @opindex fdata-sections
6441 Place each function or data item into its own section in the output
6442 file if the target supports arbitrary sections. The name of the
6443 function or the name of the data item determines the section's name
6446 Use these options on systems where the linker can perform optimizations
6447 to improve locality of reference in the instruction space. Most systems
6448 using the ELF object format and SPARC processors running Solaris 2 have
6449 linkers with such optimizations. AIX may have these optimizations in
6452 Only use these options when there are significant benefits from doing
6453 so. When you specify these options, the assembler and linker will
6454 create larger object and executable files and will also be slower.
6455 You will not be able to use @code{gprof} on all systems if you
6456 specify this option and you may have problems with debugging if
6457 you specify both this option and @option{-g}.
6459 @item -fbranch-target-load-optimize
6460 @opindex fbranch-target-load-optimize
6461 Perform branch target register load optimization before prologue / epilogue
6463 The use of target registers can typically be exposed only during reload,
6464 thus hoisting loads out of loops and doing inter-block scheduling needs
6465 a separate optimization pass.
6467 @item -fbranch-target-load-optimize2
6468 @opindex fbranch-target-load-optimize2
6469 Perform branch target register load optimization after prologue / epilogue
6472 @item -fbtr-bb-exclusive
6473 @opindex fbtr-bb-exclusive
6474 When performing branch target register load optimization, don't reuse
6475 branch target registers in within any basic block.
6477 @item -fstack-protector
6478 Emit extra code to check for buffer overflows, such as stack smashing
6479 attacks. This is done by adding a guard variable to functions with
6480 vulnerable objects. This includes functions that call alloca, and
6481 functions with buffers larger than 8 bytes. The guards are initialized
6482 when a function is entered and then checked when the function exits.
6483 If a guard check fails, an error message is printed and the program exits.
6485 @item -fstack-protector-all
6486 Like @option{-fstack-protector} except that all functions are protected.
6488 @item -fsection-anchors
6489 @opindex fsection-anchors
6490 Try to reduce the number of symbolic address calculations by using
6491 shared ``anchor'' symbols to address nearby objects. This transformation
6492 can help to reduce the number of GOT entries and GOT accesses on some
6495 For example, the implementation of the following function @code{foo}:
6499 int foo (void) @{ return a + b + c; @}
6502 would usually calculate the addresses of all three variables, but if you
6503 compile it with @option{-fsection-anchors}, it will access the variables
6504 from a common anchor point instead. The effect is similar to the
6505 following pseudocode (which isn't valid C):
6510 register int *xr = &x;
6511 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6515 Not all targets support this option.
6517 @item --param @var{name}=@var{value}
6519 In some places, GCC uses various constants to control the amount of
6520 optimization that is done. For example, GCC will not inline functions
6521 that contain more that a certain number of instructions. You can
6522 control some of these constants on the command-line using the
6523 @option{--param} option.
6525 The names of specific parameters, and the meaning of the values, are
6526 tied to the internals of the compiler, and are subject to change
6527 without notice in future releases.
6529 In each case, the @var{value} is an integer. The allowable choices for
6530 @var{name} are given in the following table:
6533 @item salias-max-implicit-fields
6534 The maximum number of fields in a variable without direct
6535 structure accesses for which structure aliasing will consider trying
6536 to track each field. The default is 5
6538 @item salias-max-array-elements
6539 The maximum number of elements an array can have and its elements
6540 still be tracked individually by structure aliasing. The default is 4
6542 @item sra-max-structure-size
6543 The maximum structure size, in bytes, at which the scalar replacement
6544 of aggregates (SRA) optimization will perform block copies. The
6545 default value, 0, implies that GCC will select the most appropriate
6548 @item sra-field-structure-ratio
6549 The threshold ratio (as a percentage) between instantiated fields and
6550 the complete structure size. We say that if the ratio of the number
6551 of bytes in instantiated fields to the number of bytes in the complete
6552 structure exceeds this parameter, then block copies are not used. The
6555 @item max-crossjump-edges
6556 The maximum number of incoming edges to consider for crossjumping.
6557 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6558 the number of edges incoming to each block. Increasing values mean
6559 more aggressive optimization, making the compile time increase with
6560 probably small improvement in executable size.
6562 @item min-crossjump-insns
6563 The minimum number of instructions which must be matched at the end
6564 of two blocks before crossjumping will be performed on them. This
6565 value is ignored in the case where all instructions in the block being
6566 crossjumped from are matched. The default value is 5.
6568 @item max-grow-copy-bb-insns
6569 The maximum code size expansion factor when copying basic blocks
6570 instead of jumping. The expansion is relative to a jump instruction.
6571 The default value is 8.
6573 @item max-goto-duplication-insns
6574 The maximum number of instructions to duplicate to a block that jumps
6575 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6576 passes, GCC factors computed gotos early in the compilation process,
6577 and unfactors them as late as possible. Only computed jumps at the
6578 end of a basic blocks with no more than max-goto-duplication-insns are
6579 unfactored. The default value is 8.
6581 @item max-delay-slot-insn-search
6582 The maximum number of instructions to consider when looking for an
6583 instruction to fill a delay slot. If more than this arbitrary number of
6584 instructions is searched, the time savings from filling the delay slot
6585 will be minimal so stop searching. Increasing values mean more
6586 aggressive optimization, making the compile time increase with probably
6587 small improvement in executable run time.
6589 @item max-delay-slot-live-search
6590 When trying to fill delay slots, the maximum number of instructions to
6591 consider when searching for a block with valid live register
6592 information. Increasing this arbitrarily chosen value means more
6593 aggressive optimization, increasing the compile time. This parameter
6594 should be removed when the delay slot code is rewritten to maintain the
6597 @item max-gcse-memory
6598 The approximate maximum amount of memory that will be allocated in
6599 order to perform the global common subexpression elimination
6600 optimization. If more memory than specified is required, the
6601 optimization will not be done.
6603 @item max-gcse-passes
6604 The maximum number of passes of GCSE to run. The default is 1.
6606 @item max-pending-list-length
6607 The maximum number of pending dependencies scheduling will allow
6608 before flushing the current state and starting over. Large functions
6609 with few branches or calls can create excessively large lists which
6610 needlessly consume memory and resources.
6612 @item max-inline-insns-single
6613 Several parameters control the tree inliner used in gcc.
6614 This number sets the maximum number of instructions (counted in GCC's
6615 internal representation) in a single function that the tree inliner
6616 will consider for inlining. This only affects functions declared
6617 inline and methods implemented in a class declaration (C++).
6618 The default value is 450.
6620 @item max-inline-insns-auto
6621 When you use @option{-finline-functions} (included in @option{-O3}),
6622 a lot of functions that would otherwise not be considered for inlining
6623 by the compiler will be investigated. To those functions, a different
6624 (more restrictive) limit compared to functions declared inline can
6626 The default value is 90.
6628 @item large-function-insns
6629 The limit specifying really large functions. For functions larger than this
6630 limit after inlining inlining is constrained by
6631 @option{--param large-function-growth}. This parameter is useful primarily
6632 to avoid extreme compilation time caused by non-linear algorithms used by the
6634 This parameter is ignored when @option{-funit-at-a-time} is not used.
6635 The default value is 2700.
6637 @item large-function-growth
6638 Specifies maximal growth of large function caused by inlining in percents.
6639 This parameter is ignored when @option{-funit-at-a-time} is not used.
6640 The default value is 100 which limits large function growth to 2.0 times
6643 @item large-unit-insns
6644 The limit specifying large translation unit. Growth caused by inlining of
6645 units larger than this limit is limited by @option{--param inline-unit-growth}.
6646 For small units this might be too tight (consider unit consisting of function A
6647 that is inline and B that just calls A three time. If B is small relative to
6648 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6649 large units consisting of small inlininable functions however the overall unit
6650 growth limit is needed to avoid exponential explosion of code size. Thus for
6651 smaller units, the size is increased to @option{--param large-unit-insns}
6652 before applying @option{--param inline-unit-growth}. The default is 10000
6654 @item inline-unit-growth
6655 Specifies maximal overall growth of the compilation unit caused by inlining.
6656 This parameter is ignored when @option{-funit-at-a-time} is not used.
6657 The default value is 30 which limits unit growth to 1.3 times the original
6660 @item large-stack-frame
6661 The limit specifying large stack frames. While inlining the algorithm is trying
6662 to not grow past this limit too much. Default value is 256 bytes.
6664 @item large-stack-frame-growth
6665 Specifies maximal growth of large stack frames caused by inlining in percents.
6666 The default value is 1000 which limits large stack frame growth to 11 times
6669 @item max-inline-insns-recursive
6670 @itemx max-inline-insns-recursive-auto
6671 Specifies maximum number of instructions out-of-line copy of self recursive inline
6672 function can grow into by performing recursive inlining.
6674 For functions declared inline @option{--param max-inline-insns-recursive} is
6675 taken into account. For function not declared inline, recursive inlining
6676 happens only when @option{-finline-functions} (included in @option{-O3}) is
6677 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6678 default value is 450.
6680 @item max-inline-recursive-depth
6681 @itemx max-inline-recursive-depth-auto
6682 Specifies maximum recursion depth used by the recursive inlining.
6684 For functions declared inline @option{--param max-inline-recursive-depth} is
6685 taken into account. For function not declared inline, recursive inlining
6686 happens only when @option{-finline-functions} (included in @option{-O3}) is
6687 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6688 default value is 450.
6690 @item min-inline-recursive-probability
6691 Recursive inlining is profitable only for function having deep recursion
6692 in average and can hurt for function having little recursion depth by
6693 increasing the prologue size or complexity of function body to other
6696 When profile feedback is available (see @option{-fprofile-generate}) the actual
6697 recursion depth can be guessed from probability that function will recurse via
6698 given call expression. This parameter limits inlining only to call expression
6699 whose probability exceeds given threshold (in percents). The default value is
6702 @item inline-call-cost
6703 Specify cost of call instruction relative to simple arithmetics operations
6704 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6705 functions and at the same time increases size of leaf function that is believed to
6706 reduce function size by being inlined. In effect it increases amount of
6707 inlining for code having large abstraction penalty (many functions that just
6708 pass the arguments to other functions) and decrease inlining for code with low
6709 abstraction penalty. The default value is 16.
6711 @item min-vect-loop-bound
6712 The minimum number of iterations under which a loop will not get vectorized
6713 when @option{-ftree-vectorize} is used. The number of iterations after
6714 vectorization needs to be greater than the value specified by this option
6715 to allow vectorization. The default value is 0.
6717 @item max-unrolled-insns
6718 The maximum number of instructions that a loop should have if that loop
6719 is unrolled, and if the loop is unrolled, it determines how many times
6720 the loop code is unrolled.
6722 @item max-average-unrolled-insns
6723 The maximum number of instructions biased by probabilities of their execution
6724 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6725 it determines how many times the loop code is unrolled.
6727 @item max-unroll-times
6728 The maximum number of unrollings of a single loop.
6730 @item max-peeled-insns
6731 The maximum number of instructions that a loop should have if that loop
6732 is peeled, and if the loop is peeled, it determines how many times
6733 the loop code is peeled.
6735 @item max-peel-times
6736 The maximum number of peelings of a single loop.
6738 @item max-completely-peeled-insns
6739 The maximum number of insns of a completely peeled loop.
6741 @item max-completely-peel-times
6742 The maximum number of iterations of a loop to be suitable for complete peeling.
6744 @item max-unswitch-insns
6745 The maximum number of insns of an unswitched loop.
6747 @item max-unswitch-level
6748 The maximum number of branches unswitched in a single loop.
6751 The minimum cost of an expensive expression in the loop invariant motion.
6753 @item iv-consider-all-candidates-bound
6754 Bound on number of candidates for induction variables below that
6755 all candidates are considered for each use in induction variable
6756 optimizations. Only the most relevant candidates are considered
6757 if there are more candidates, to avoid quadratic time complexity.
6759 @item iv-max-considered-uses
6760 The induction variable optimizations give up on loops that contain more
6761 induction variable uses.
6763 @item iv-always-prune-cand-set-bound
6764 If number of candidates in the set is smaller than this value,
6765 we always try to remove unnecessary ivs from the set during its
6766 optimization when a new iv is added to the set.
6768 @item scev-max-expr-size
6769 Bound on size of expressions used in the scalar evolutions analyzer.
6770 Large expressions slow the analyzer.
6772 @item omega-max-vars
6773 The maximum number of variables in an Omega constraint system.
6774 The default value is 128.
6776 @item omega-max-geqs
6777 The maximum number of inequalities in an Omega constraint system.
6778 The default value is 256.
6781 The maximum number of equalities in an Omega constraint system.
6782 The default value is 128.
6784 @item omega-max-wild-cards
6785 The maximum number of wildcard variables that the Omega solver will
6786 be able to insert. The default value is 18.
6788 @item omega-hash-table-size
6789 The size of the hash table in the Omega solver. The default value is
6792 @item omega-max-keys
6793 The maximal number of keys used by the Omega solver. The default
6796 @item omega-eliminate-redundant-constraints
6797 When set to 1, use expensive methods to eliminate all redundant
6798 constraints. The default value is 0.
6800 @item vect-max-version-for-alignment-checks
6801 The maximum number of runtime checks that can be performed when
6802 doing loop versioning for alignment in the vectorizer. See option
6803 ftree-vect-loop-version for more information.
6805 @item vect-max-version-for-alias-checks
6806 The maximum number of runtime checks that can be performed when
6807 doing loop versioning for alias in the vectorizer. See option
6808 ftree-vect-loop-version for more information.
6810 @item max-iterations-to-track
6812 The maximum number of iterations of a loop the brute force algorithm
6813 for analysis of # of iterations of the loop tries to evaluate.
6815 @item hot-bb-count-fraction
6816 Select fraction of the maximal count of repetitions of basic block in program
6817 given basic block needs to have to be considered hot.
6819 @item hot-bb-frequency-fraction
6820 Select fraction of the maximal frequency of executions of basic block in
6821 function given basic block needs to have to be considered hot
6823 @item max-predicted-iterations
6824 The maximum number of loop iterations we predict statically. This is useful
6825 in cases where function contain single loop with known bound and other loop
6826 with unknown. We predict the known number of iterations correctly, while
6827 the unknown number of iterations average to roughly 10. This means that the
6828 loop without bounds would appear artificially cold relative to the other one.
6830 @item tracer-dynamic-coverage
6831 @itemx tracer-dynamic-coverage-feedback
6833 This value is used to limit superblock formation once the given percentage of
6834 executed instructions is covered. This limits unnecessary code size
6837 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6838 feedback is available. The real profiles (as opposed to statically estimated
6839 ones) are much less balanced allowing the threshold to be larger value.
6841 @item tracer-max-code-growth
6842 Stop tail duplication once code growth has reached given percentage. This is
6843 rather hokey argument, as most of the duplicates will be eliminated later in
6844 cross jumping, so it may be set to much higher values than is the desired code
6847 @item tracer-min-branch-ratio
6849 Stop reverse growth when the reverse probability of best edge is less than this
6850 threshold (in percent).
6852 @item tracer-min-branch-ratio
6853 @itemx tracer-min-branch-ratio-feedback
6855 Stop forward growth if the best edge do have probability lower than this
6858 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6859 compilation for profile feedback and one for compilation without. The value
6860 for compilation with profile feedback needs to be more conservative (higher) in
6861 order to make tracer effective.
6863 @item max-cse-path-length
6865 Maximum number of basic blocks on path that cse considers. The default is 10.
6868 The maximum instructions CSE process before flushing. The default is 1000.
6870 @item max-aliased-vops
6872 Maximum number of virtual operands per function allowed to represent
6873 aliases before triggering the alias partitioning heuristic. Alias
6874 partitioning reduces compile times and memory consumption needed for
6875 aliasing at the expense of precision loss in alias information. The
6876 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
6879 Notice that if a function contains more memory statements than the
6880 value of this parameter, it is not really possible to achieve this
6881 reduction. In this case, the compiler will use the number of memory
6882 statements as the value for @option{max-aliased-vops}.
6884 @item avg-aliased-vops
6886 Average number of virtual operands per statement allowed to represent
6887 aliases before triggering the alias partitioning heuristic. This
6888 works in conjunction with @option{max-aliased-vops}. If a function
6889 contains more than @option{max-aliased-vops} virtual operators, then
6890 memory symbols will be grouped into memory partitions until either the
6891 total number of virtual operators is below @option{max-aliased-vops}
6892 or the average number of virtual operators per memory statement is
6893 below @option{avg-aliased-vops}. The default value for this parameter
6894 is 1 for -O1 and -O2, and 3 for -O3.
6896 @item ggc-min-expand
6898 GCC uses a garbage collector to manage its own memory allocation. This
6899 parameter specifies the minimum percentage by which the garbage
6900 collector's heap should be allowed to expand between collections.
6901 Tuning this may improve compilation speed; it has no effect on code
6904 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6905 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6906 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6907 GCC is not able to calculate RAM on a particular platform, the lower
6908 bound of 30% is used. Setting this parameter and
6909 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6910 every opportunity. This is extremely slow, but can be useful for
6913 @item ggc-min-heapsize
6915 Minimum size of the garbage collector's heap before it begins bothering
6916 to collect garbage. The first collection occurs after the heap expands
6917 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6918 tuning this may improve compilation speed, and has no effect on code
6921 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6922 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6923 with a lower bound of 4096 (four megabytes) and an upper bound of
6924 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6925 particular platform, the lower bound is used. Setting this parameter
6926 very large effectively disables garbage collection. Setting this
6927 parameter and @option{ggc-min-expand} to zero causes a full collection
6928 to occur at every opportunity.
6930 @item max-reload-search-insns
6931 The maximum number of instruction reload should look backward for equivalent
6932 register. Increasing values mean more aggressive optimization, making the
6933 compile time increase with probably slightly better performance. The default
6936 @item max-cselib-memory-locations
6937 The maximum number of memory locations cselib should take into account.
6938 Increasing values mean more aggressive optimization, making the compile time
6939 increase with probably slightly better performance. The default value is 500.
6941 @item max-flow-memory-locations
6942 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6943 The default value is 100.
6945 @item reorder-blocks-duplicate
6946 @itemx reorder-blocks-duplicate-feedback
6948 Used by basic block reordering pass to decide whether to use unconditional
6949 branch or duplicate the code on its destination. Code is duplicated when its
6950 estimated size is smaller than this value multiplied by the estimated size of
6951 unconditional jump in the hot spots of the program.
6953 The @option{reorder-block-duplicate-feedback} is used only when profile
6954 feedback is available and may be set to higher values than
6955 @option{reorder-block-duplicate} since information about the hot spots is more
6958 @item max-sched-ready-insns
6959 The maximum number of instructions ready to be issued the scheduler should
6960 consider at any given time during the first scheduling pass. Increasing
6961 values mean more thorough searches, making the compilation time increase
6962 with probably little benefit. The default value is 100.
6964 @item max-sched-region-blocks
6965 The maximum number of blocks in a region to be considered for
6966 interblock scheduling. The default value is 10.
6968 @item max-sched-region-insns
6969 The maximum number of insns in a region to be considered for
6970 interblock scheduling. The default value is 100.
6973 The minimum probability (in percents) of reaching a source block
6974 for interblock speculative scheduling. The default value is 40.
6976 @item max-sched-extend-regions-iters
6977 The maximum number of iterations through CFG to extend regions.
6978 0 - disable region extension,
6979 N - do at most N iterations.
6980 The default value is 0.
6982 @item max-sched-insn-conflict-delay
6983 The maximum conflict delay for an insn to be considered for speculative motion.
6984 The default value is 3.
6986 @item sched-spec-prob-cutoff
6987 The minimal probability of speculation success (in percents), so that
6988 speculative insn will be scheduled.
6989 The default value is 40.
6991 @item max-last-value-rtl
6993 The maximum size measured as number of RTLs that can be recorded in an expression
6994 in combiner for a pseudo register as last known value of that register. The default
6997 @item integer-share-limit
6998 Small integer constants can use a shared data structure, reducing the
6999 compiler's memory usage and increasing its speed. This sets the maximum
7000 value of a shared integer constant's. The default value is 256.
7002 @item min-virtual-mappings
7003 Specifies the minimum number of virtual mappings in the incremental
7004 SSA updater that should be registered to trigger the virtual mappings
7005 heuristic defined by virtual-mappings-ratio. The default value is
7008 @item virtual-mappings-ratio
7009 If the number of virtual mappings is virtual-mappings-ratio bigger
7010 than the number of virtual symbols to be updated, then the incremental
7011 SSA updater switches to a full update for those symbols. The default
7014 @item ssp-buffer-size
7015 The minimum size of buffers (i.e. arrays) that will receive stack smashing
7016 protection when @option{-fstack-protection} is used.
7018 @item max-jump-thread-duplication-stmts
7019 Maximum number of statements allowed in a block that needs to be
7020 duplicated when threading jumps.
7022 @item max-fields-for-field-sensitive
7023 Maximum number of fields in a structure we will treat in
7024 a field sensitive manner during pointer analysis.
7026 @item prefetch-latency
7027 Estimate on average number of instructions that are executed before
7028 prefetch finishes. The distance we prefetch ahead is proportional
7029 to this constant. Increasing this number may also lead to less
7030 streams being prefetched (see @option{simultaneous-prefetches}).
7032 @item simultaneous-prefetches
7033 Maximum number of prefetches that can run at the same time.
7035 @item l1-cache-line-size
7036 The size of cache line in L1 cache, in bytes.
7039 The size of L1 cache, in kilobytes.
7042 The size of L2 cache, in kilobytes.
7044 @item use-canonical-types
7045 Whether the compiler should use the ``canonical'' type system. By
7046 default, this should always be 1, which uses a more efficient internal
7047 mechanism for comparing types in C++ and Objective-C++. However, if
7048 bugs in the canonical type system are causing compilation failures,
7049 set this value to 0 to disable canonical types.
7054 @node Preprocessor Options
7055 @section Options Controlling the Preprocessor
7056 @cindex preprocessor options
7057 @cindex options, preprocessor
7059 These options control the C preprocessor, which is run on each C source
7060 file before actual compilation.
7062 If you use the @option{-E} option, nothing is done except preprocessing.
7063 Some of these options make sense only together with @option{-E} because
7064 they cause the preprocessor output to be unsuitable for actual
7069 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7070 and pass @var{option} directly through to the preprocessor. If
7071 @var{option} contains commas, it is split into multiple options at the
7072 commas. However, many options are modified, translated or interpreted
7073 by the compiler driver before being passed to the preprocessor, and
7074 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7075 interface is undocumented and subject to change, so whenever possible
7076 you should avoid using @option{-Wp} and let the driver handle the
7079 @item -Xpreprocessor @var{option}
7080 @opindex preprocessor
7081 Pass @var{option} as an option to the preprocessor. You can use this to
7082 supply system-specific preprocessor options which GCC does not know how to
7085 If you want to pass an option that takes an argument, you must use
7086 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7089 @include cppopts.texi
7091 @node Assembler Options
7092 @section Passing Options to the Assembler
7094 @c prevent bad page break with this line
7095 You can pass options to the assembler.
7098 @item -Wa,@var{option}
7100 Pass @var{option} as an option to the assembler. If @var{option}
7101 contains commas, it is split into multiple options at the commas.
7103 @item -Xassembler @var{option}
7105 Pass @var{option} as an option to the assembler. You can use this to
7106 supply system-specific assembler options which GCC does not know how to
7109 If you want to pass an option that takes an argument, you must use
7110 @option{-Xassembler} twice, once for the option and once for the argument.
7115 @section Options for Linking
7116 @cindex link options
7117 @cindex options, linking
7119 These options come into play when the compiler links object files into
7120 an executable output file. They are meaningless if the compiler is
7121 not doing a link step.
7125 @item @var{object-file-name}
7126 A file name that does not end in a special recognized suffix is
7127 considered to name an object file or library. (Object files are
7128 distinguished from libraries by the linker according to the file
7129 contents.) If linking is done, these object files are used as input
7138 If any of these options is used, then the linker is not run, and
7139 object file names should not be used as arguments. @xref{Overall
7143 @item -l@var{library}
7144 @itemx -l @var{library}
7146 Search the library named @var{library} when linking. (The second
7147 alternative with the library as a separate argument is only for
7148 POSIX compliance and is not recommended.)
7150 It makes a difference where in the command you write this option; the
7151 linker searches and processes libraries and object files in the order they
7152 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7153 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7154 to functions in @samp{z}, those functions may not be loaded.
7156 The linker searches a standard list of directories for the library,
7157 which is actually a file named @file{lib@var{library}.a}. The linker
7158 then uses this file as if it had been specified precisely by name.
7160 The directories searched include several standard system directories
7161 plus any that you specify with @option{-L}.
7163 Normally the files found this way are library files---archive files
7164 whose members are object files. The linker handles an archive file by
7165 scanning through it for members which define symbols that have so far
7166 been referenced but not defined. But if the file that is found is an
7167 ordinary object file, it is linked in the usual fashion. The only
7168 difference between using an @option{-l} option and specifying a file name
7169 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7170 and searches several directories.
7174 You need this special case of the @option{-l} option in order to
7175 link an Objective-C or Objective-C++ program.
7178 @opindex nostartfiles
7179 Do not use the standard system startup files when linking.
7180 The standard system libraries are used normally, unless @option{-nostdlib}
7181 or @option{-nodefaultlibs} is used.
7183 @item -nodefaultlibs
7184 @opindex nodefaultlibs
7185 Do not use the standard system libraries when linking.
7186 Only the libraries you specify will be passed to the linker.
7187 The standard startup files are used normally, unless @option{-nostartfiles}
7188 is used. The compiler may generate calls to @code{memcmp},
7189 @code{memset}, @code{memcpy} and @code{memmove}.
7190 These entries are usually resolved by entries in
7191 libc. These entry points should be supplied through some other
7192 mechanism when this option is specified.
7196 Do not use the standard system startup files or libraries when linking.
7197 No startup files and only the libraries you specify will be passed to
7198 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7199 @code{memcpy} and @code{memmove}.
7200 These entries are usually resolved by entries in
7201 libc. These entry points should be supplied through some other
7202 mechanism when this option is specified.
7204 @cindex @option{-lgcc}, use with @option{-nostdlib}
7205 @cindex @option{-nostdlib} and unresolved references
7206 @cindex unresolved references and @option{-nostdlib}
7207 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7208 @cindex @option{-nodefaultlibs} and unresolved references
7209 @cindex unresolved references and @option{-nodefaultlibs}
7210 One of the standard libraries bypassed by @option{-nostdlib} and
7211 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7212 that GCC uses to overcome shortcomings of particular machines, or special
7213 needs for some languages.
7214 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7215 Collection (GCC) Internals},
7216 for more discussion of @file{libgcc.a}.)
7217 In most cases, you need @file{libgcc.a} even when you want to avoid
7218 other standard libraries. In other words, when you specify @option{-nostdlib}
7219 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7220 This ensures that you have no unresolved references to internal GCC
7221 library subroutines. (For example, @samp{__main}, used to ensure C++
7222 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7223 GNU Compiler Collection (GCC) Internals}.)
7227 Produce a position independent executable on targets which support it.
7228 For predictable results, you must also specify the same set of options
7229 that were used to generate code (@option{-fpie}, @option{-fPIE},
7230 or model suboptions) when you specify this option.
7234 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7235 that support it. This instructs the linker to add all symbols, not
7236 only used ones, to the dynamic symbol table. This option is needed
7237 for some uses of @code{dlopen} or to allow obtaining backtraces
7238 from within a program.
7242 Remove all symbol table and relocation information from the executable.
7246 On systems that support dynamic linking, this prevents linking with the shared
7247 libraries. On other systems, this option has no effect.
7251 Produce a shared object which can then be linked with other objects to
7252 form an executable. Not all systems support this option. For predictable
7253 results, you must also specify the same set of options that were used to
7254 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7255 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7256 needs to build supplementary stub code for constructors to work. On
7257 multi-libbed systems, @samp{gcc -shared} must select the correct support
7258 libraries to link against. Failing to supply the correct flags may lead
7259 to subtle defects. Supplying them in cases where they are not necessary
7262 @item -shared-libgcc
7263 @itemx -static-libgcc
7264 @opindex shared-libgcc
7265 @opindex static-libgcc
7266 On systems that provide @file{libgcc} as a shared library, these options
7267 force the use of either the shared or static version respectively.
7268 If no shared version of @file{libgcc} was built when the compiler was
7269 configured, these options have no effect.
7271 There are several situations in which an application should use the
7272 shared @file{libgcc} instead of the static version. The most common
7273 of these is when the application wishes to throw and catch exceptions
7274 across different shared libraries. In that case, each of the libraries
7275 as well as the application itself should use the shared @file{libgcc}.
7277 Therefore, the G++ and GCJ drivers automatically add
7278 @option{-shared-libgcc} whenever you build a shared library or a main
7279 executable, because C++ and Java programs typically use exceptions, so
7280 this is the right thing to do.
7282 If, instead, you use the GCC driver to create shared libraries, you may
7283 find that they will not always be linked with the shared @file{libgcc}.
7284 If GCC finds, at its configuration time, that you have a non-GNU linker
7285 or a GNU linker that does not support option @option{--eh-frame-hdr},
7286 it will link the shared version of @file{libgcc} into shared libraries
7287 by default. Otherwise, it will take advantage of the linker and optimize
7288 away the linking with the shared version of @file{libgcc}, linking with
7289 the static version of libgcc by default. This allows exceptions to
7290 propagate through such shared libraries, without incurring relocation
7291 costs at library load time.
7293 However, if a library or main executable is supposed to throw or catch
7294 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7295 for the languages used in the program, or using the option
7296 @option{-shared-libgcc}, such that it is linked with the shared
7301 Bind references to global symbols when building a shared object. Warn
7302 about any unresolved references (unless overridden by the link editor
7303 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7306 @item -Xlinker @var{option}
7308 Pass @var{option} as an option to the linker. You can use this to
7309 supply system-specific linker options which GCC does not know how to
7312 If you want to pass an option that takes an argument, you must use
7313 @option{-Xlinker} twice, once for the option and once for the argument.
7314 For example, to pass @option{-assert definitions}, you must write
7315 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7316 @option{-Xlinker "-assert definitions"}, because this passes the entire
7317 string as a single argument, which is not what the linker expects.
7319 @item -Wl,@var{option}
7321 Pass @var{option} as an option to the linker. If @var{option} contains
7322 commas, it is split into multiple options at the commas.
7324 @item -u @var{symbol}
7326 Pretend the symbol @var{symbol} is undefined, to force linking of
7327 library modules to define it. You can use @option{-u} multiple times with
7328 different symbols to force loading of additional library modules.
7331 @node Directory Options
7332 @section Options for Directory Search
7333 @cindex directory options
7334 @cindex options, directory search
7337 These options specify directories to search for header files, for
7338 libraries and for parts of the compiler:
7343 Add the directory @var{dir} to the head of the list of directories to be
7344 searched for header files. This can be used to override a system header
7345 file, substituting your own version, since these directories are
7346 searched before the system header file directories. However, you should
7347 not use this option to add directories that contain vendor-supplied
7348 system header files (use @option{-isystem} for that). If you use more than
7349 one @option{-I} option, the directories are scanned in left-to-right
7350 order; the standard system directories come after.
7352 If a standard system include directory, or a directory specified with
7353 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7354 option will be ignored. The directory will still be searched but as a
7355 system directory at its normal position in the system include chain.
7356 This is to ensure that GCC's procedure to fix buggy system headers and
7357 the ordering for the include_next directive are not inadvertently changed.
7358 If you really need to change the search order for system directories,
7359 use the @option{-nostdinc} and/or @option{-isystem} options.
7361 @item -iquote@var{dir}
7363 Add the directory @var{dir} to the head of the list of directories to
7364 be searched for header files only for the case of @samp{#include
7365 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7366 otherwise just like @option{-I}.
7370 Add directory @var{dir} to the list of directories to be searched
7373 @item -B@var{prefix}
7375 This option specifies where to find the executables, libraries,
7376 include files, and data files of the compiler itself.
7378 The compiler driver program runs one or more of the subprograms
7379 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7380 @var{prefix} as a prefix for each program it tries to run, both with and
7381 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7383 For each subprogram to be run, the compiler driver first tries the
7384 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7385 was not specified, the driver tries two standard prefixes, which are
7386 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7387 those results in a file name that is found, the unmodified program
7388 name is searched for using the directories specified in your
7389 @env{PATH} environment variable.
7391 The compiler will check to see if the path provided by the @option{-B}
7392 refers to a directory, and if necessary it will add a directory
7393 separator character at the end of the path.
7395 @option{-B} prefixes that effectively specify directory names also apply
7396 to libraries in the linker, because the compiler translates these
7397 options into @option{-L} options for the linker. They also apply to
7398 includes files in the preprocessor, because the compiler translates these
7399 options into @option{-isystem} options for the preprocessor. In this case,
7400 the compiler appends @samp{include} to the prefix.
7402 The run-time support file @file{libgcc.a} can also be searched for using
7403 the @option{-B} prefix, if needed. If it is not found there, the two
7404 standard prefixes above are tried, and that is all. The file is left
7405 out of the link if it is not found by those means.
7407 Another way to specify a prefix much like the @option{-B} prefix is to use
7408 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7411 As a special kludge, if the path provided by @option{-B} is
7412 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7413 9, then it will be replaced by @file{[dir/]include}. This is to help
7414 with boot-strapping the compiler.
7416 @item -specs=@var{file}
7418 Process @var{file} after the compiler reads in the standard @file{specs}
7419 file, in order to override the defaults that the @file{gcc} driver
7420 program uses when determining what switches to pass to @file{cc1},
7421 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7422 @option{-specs=@var{file}} can be specified on the command line, and they
7423 are processed in order, from left to right.
7425 @item --sysroot=@var{dir}
7427 Use @var{dir} as the logical root directory for headers and libraries.
7428 For example, if the compiler would normally search for headers in
7429 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7430 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7432 If you use both this option and the @option{-isysroot} option, then
7433 the @option{--sysroot} option will apply to libraries, but the
7434 @option{-isysroot} option will apply to header files.
7436 The GNU linker (beginning with version 2.16) has the necessary support
7437 for this option. If your linker does not support this option, the
7438 header file aspect of @option{--sysroot} will still work, but the
7439 library aspect will not.
7443 This option has been deprecated. Please use @option{-iquote} instead for
7444 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7445 Any directories you specify with @option{-I} options before the @option{-I-}
7446 option are searched only for the case of @samp{#include "@var{file}"};
7447 they are not searched for @samp{#include <@var{file}>}.
7449 If additional directories are specified with @option{-I} options after
7450 the @option{-I-}, these directories are searched for all @samp{#include}
7451 directives. (Ordinarily @emph{all} @option{-I} directories are used
7454 In addition, the @option{-I-} option inhibits the use of the current
7455 directory (where the current input file came from) as the first search
7456 directory for @samp{#include "@var{file}"}. There is no way to
7457 override this effect of @option{-I-}. With @option{-I.} you can specify
7458 searching the directory which was current when the compiler was
7459 invoked. That is not exactly the same as what the preprocessor does
7460 by default, but it is often satisfactory.
7462 @option{-I-} does not inhibit the use of the standard system directories
7463 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7470 @section Specifying subprocesses and the switches to pass to them
7473 @command{gcc} is a driver program. It performs its job by invoking a
7474 sequence of other programs to do the work of compiling, assembling and
7475 linking. GCC interprets its command-line parameters and uses these to
7476 deduce which programs it should invoke, and which command-line options
7477 it ought to place on their command lines. This behavior is controlled
7478 by @dfn{spec strings}. In most cases there is one spec string for each
7479 program that GCC can invoke, but a few programs have multiple spec
7480 strings to control their behavior. The spec strings built into GCC can
7481 be overridden by using the @option{-specs=} command-line switch to specify
7484 @dfn{Spec files} are plaintext files that are used to construct spec
7485 strings. They consist of a sequence of directives separated by blank
7486 lines. The type of directive is determined by the first non-whitespace
7487 character on the line and it can be one of the following:
7490 @item %@var{command}
7491 Issues a @var{command} to the spec file processor. The commands that can
7495 @item %include <@var{file}>
7497 Search for @var{file} and insert its text at the current point in the
7500 @item %include_noerr <@var{file}>
7501 @cindex %include_noerr
7502 Just like @samp{%include}, but do not generate an error message if the include
7503 file cannot be found.
7505 @item %rename @var{old_name} @var{new_name}
7507 Rename the spec string @var{old_name} to @var{new_name}.
7511 @item *[@var{spec_name}]:
7512 This tells the compiler to create, override or delete the named spec
7513 string. All lines after this directive up to the next directive or
7514 blank line are considered to be the text for the spec string. If this
7515 results in an empty string then the spec will be deleted. (Or, if the
7516 spec did not exist, then nothing will happened.) Otherwise, if the spec
7517 does not currently exist a new spec will be created. If the spec does
7518 exist then its contents will be overridden by the text of this
7519 directive, unless the first character of that text is the @samp{+}
7520 character, in which case the text will be appended to the spec.
7522 @item [@var{suffix}]:
7523 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7524 and up to the next directive or blank line are considered to make up the
7525 spec string for the indicated suffix. When the compiler encounters an
7526 input file with the named suffix, it will processes the spec string in
7527 order to work out how to compile that file. For example:
7534 This says that any input file whose name ends in @samp{.ZZ} should be
7535 passed to the program @samp{z-compile}, which should be invoked with the
7536 command-line switch @option{-input} and with the result of performing the
7537 @samp{%i} substitution. (See below.)
7539 As an alternative to providing a spec string, the text that follows a
7540 suffix directive can be one of the following:
7543 @item @@@var{language}
7544 This says that the suffix is an alias for a known @var{language}. This is
7545 similar to using the @option{-x} command-line switch to GCC to specify a
7546 language explicitly. For example:
7553 Says that .ZZ files are, in fact, C++ source files.
7556 This causes an error messages saying:
7559 @var{name} compiler not installed on this system.
7563 GCC already has an extensive list of suffixes built into it.
7564 This directive will add an entry to the end of the list of suffixes, but
7565 since the list is searched from the end backwards, it is effectively
7566 possible to override earlier entries using this technique.
7570 GCC has the following spec strings built into it. Spec files can
7571 override these strings or create their own. Note that individual
7572 targets can also add their own spec strings to this list.
7575 asm Options to pass to the assembler
7576 asm_final Options to pass to the assembler post-processor
7577 cpp Options to pass to the C preprocessor
7578 cc1 Options to pass to the C compiler
7579 cc1plus Options to pass to the C++ compiler
7580 endfile Object files to include at the end of the link
7581 link Options to pass to the linker
7582 lib Libraries to include on the command line to the linker
7583 libgcc Decides which GCC support library to pass to the linker
7584 linker Sets the name of the linker
7585 predefines Defines to be passed to the C preprocessor
7586 signed_char Defines to pass to CPP to say whether @code{char} is signed
7588 startfile Object files to include at the start of the link
7591 Here is a small example of a spec file:
7597 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7600 This example renames the spec called @samp{lib} to @samp{old_lib} and
7601 then overrides the previous definition of @samp{lib} with a new one.
7602 The new definition adds in some extra command-line options before
7603 including the text of the old definition.
7605 @dfn{Spec strings} are a list of command-line options to be passed to their
7606 corresponding program. In addition, the spec strings can contain
7607 @samp{%}-prefixed sequences to substitute variable text or to
7608 conditionally insert text into the command line. Using these constructs
7609 it is possible to generate quite complex command lines.
7611 Here is a table of all defined @samp{%}-sequences for spec
7612 strings. Note that spaces are not generated automatically around the
7613 results of expanding these sequences. Therefore you can concatenate them
7614 together or combine them with constant text in a single argument.
7618 Substitute one @samp{%} into the program name or argument.
7621 Substitute the name of the input file being processed.
7624 Substitute the basename of the input file being processed.
7625 This is the substring up to (and not including) the last period
7626 and not including the directory.
7629 This is the same as @samp{%b}, but include the file suffix (text after
7633 Marks the argument containing or following the @samp{%d} as a
7634 temporary file name, so that that file will be deleted if GCC exits
7635 successfully. Unlike @samp{%g}, this contributes no text to the
7638 @item %g@var{suffix}
7639 Substitute a file name that has suffix @var{suffix} and is chosen
7640 once per compilation, and mark the argument in the same way as
7641 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7642 name is now chosen in a way that is hard to predict even when previously
7643 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7644 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7645 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7646 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7647 was simply substituted with a file name chosen once per compilation,
7648 without regard to any appended suffix (which was therefore treated
7649 just like ordinary text), making such attacks more likely to succeed.
7651 @item %u@var{suffix}
7652 Like @samp{%g}, but generates a new temporary file name even if
7653 @samp{%u@var{suffix}} was already seen.
7655 @item %U@var{suffix}
7656 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7657 new one if there is no such last file name. In the absence of any
7658 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7659 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7660 would involve the generation of two distinct file names, one
7661 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7662 simply substituted with a file name chosen for the previous @samp{%u},
7663 without regard to any appended suffix.
7665 @item %j@var{suffix}
7666 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7667 writable, and if save-temps is off; otherwise, substitute the name
7668 of a temporary file, just like @samp{%u}. This temporary file is not
7669 meant for communication between processes, but rather as a junk
7672 @item %|@var{suffix}
7673 @itemx %m@var{suffix}
7674 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7675 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7676 all. These are the two most common ways to instruct a program that it
7677 should read from standard input or write to standard output. If you
7678 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7679 construct: see for example @file{f/lang-specs.h}.
7681 @item %.@var{SUFFIX}
7682 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7683 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7684 terminated by the next space or %.
7687 Marks the argument containing or following the @samp{%w} as the
7688 designated output file of this compilation. This puts the argument
7689 into the sequence of arguments that @samp{%o} will substitute later.
7692 Substitutes the names of all the output files, with spaces
7693 automatically placed around them. You should write spaces
7694 around the @samp{%o} as well or the results are undefined.
7695 @samp{%o} is for use in the specs for running the linker.
7696 Input files whose names have no recognized suffix are not compiled
7697 at all, but they are included among the output files, so they will
7701 Substitutes the suffix for object files. Note that this is
7702 handled specially when it immediately follows @samp{%g, %u, or %U},
7703 because of the need for those to form complete file names. The
7704 handling is such that @samp{%O} is treated exactly as if it had already
7705 been substituted, except that @samp{%g, %u, and %U} do not currently
7706 support additional @var{suffix} characters following @samp{%O} as they would
7707 following, for example, @samp{.o}.
7710 Substitutes the standard macro predefinitions for the
7711 current target machine. Use this when running @code{cpp}.
7714 Like @samp{%p}, but puts @samp{__} before and after the name of each
7715 predefined macro, except for macros that start with @samp{__} or with
7716 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7720 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7721 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7722 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7723 and @option{-imultilib} as necessary.
7726 Current argument is the name of a library or startup file of some sort.
7727 Search for that file in a standard list of directories and substitute
7728 the full name found.
7731 Print @var{str} as an error message. @var{str} is terminated by a newline.
7732 Use this when inconsistent options are detected.
7735 Substitute the contents of spec string @var{name} at this point.
7738 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7740 @item %x@{@var{option}@}
7741 Accumulate an option for @samp{%X}.
7744 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7748 Output the accumulated assembler options specified by @option{-Wa}.
7751 Output the accumulated preprocessor options specified by @option{-Wp}.
7754 Process the @code{asm} spec. This is used to compute the
7755 switches to be passed to the assembler.
7758 Process the @code{asm_final} spec. This is a spec string for
7759 passing switches to an assembler post-processor, if such a program is
7763 Process the @code{link} spec. This is the spec for computing the
7764 command line passed to the linker. Typically it will make use of the
7765 @samp{%L %G %S %D and %E} sequences.
7768 Dump out a @option{-L} option for each directory that GCC believes might
7769 contain startup files. If the target supports multilibs then the
7770 current multilib directory will be prepended to each of these paths.
7773 Process the @code{lib} spec. This is a spec string for deciding which
7774 libraries should be included on the command line to the linker.
7777 Process the @code{libgcc} spec. This is a spec string for deciding
7778 which GCC support library should be included on the command line to the linker.
7781 Process the @code{startfile} spec. This is a spec for deciding which
7782 object files should be the first ones passed to the linker. Typically
7783 this might be a file named @file{crt0.o}.
7786 Process the @code{endfile} spec. This is a spec string that specifies
7787 the last object files that will be passed to the linker.
7790 Process the @code{cpp} spec. This is used to construct the arguments
7791 to be passed to the C preprocessor.
7794 Process the @code{cc1} spec. This is used to construct the options to be
7795 passed to the actual C compiler (@samp{cc1}).
7798 Process the @code{cc1plus} spec. This is used to construct the options to be
7799 passed to the actual C++ compiler (@samp{cc1plus}).
7802 Substitute the variable part of a matched option. See below.
7803 Note that each comma in the substituted string is replaced by
7807 Remove all occurrences of @code{-S} from the command line. Note---this
7808 command is position dependent. @samp{%} commands in the spec string
7809 before this one will see @code{-S}, @samp{%} commands in the spec string
7810 after this one will not.
7812 @item %:@var{function}(@var{args})
7813 Call the named function @var{function}, passing it @var{args}.
7814 @var{args} is first processed as a nested spec string, then split
7815 into an argument vector in the usual fashion. The function returns
7816 a string which is processed as if it had appeared literally as part
7817 of the current spec.
7819 The following built-in spec functions are provided:
7823 The @code{getenv} spec function takes two arguments: an environment
7824 variable name and a string. If the environment variable is not
7825 defined, a fatal error is issued. Otherwise, the return value is the
7826 value of the environment variable concatenated with the string. For
7827 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7830 %:getenv(TOPDIR /include)
7833 expands to @file{/path/to/top/include}.
7835 @item @code{if-exists}
7836 The @code{if-exists} spec function takes one argument, an absolute
7837 pathname to a file. If the file exists, @code{if-exists} returns the
7838 pathname. Here is a small example of its usage:
7842 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7845 @item @code{if-exists-else}
7846 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7847 spec function, except that it takes two arguments. The first argument is
7848 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7849 returns the pathname. If it does not exist, it returns the second argument.
7850 This way, @code{if-exists-else} can be used to select one file or another,
7851 based on the existence of the first. Here is a small example of its usage:
7855 crt0%O%s %:if-exists(crti%O%s) \
7856 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7859 @item @code{replace-outfile}
7860 The @code{replace-outfile} spec function takes two arguments. It looks for the
7861 first argument in the outfiles array and replaces it with the second argument. Here
7862 is a small example of its usage:
7865 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7868 @item @code{print-asm-header}
7869 The @code{print-asm-header} function takes no arguments and simply
7870 prints a banner like:
7876 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
7879 It is used to separate compiler options from assembler options
7880 in the @option{--target-help} output.
7884 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7885 If that switch was not specified, this substitutes nothing. Note that
7886 the leading dash is omitted when specifying this option, and it is
7887 automatically inserted if the substitution is performed. Thus the spec
7888 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7889 and would output the command line option @option{-foo}.
7891 @item %W@{@code{S}@}
7892 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7895 @item %@{@code{S}*@}
7896 Substitutes all the switches specified to GCC whose names start
7897 with @code{-S}, but which also take an argument. This is used for
7898 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7899 GCC considers @option{-o foo} as being
7900 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7901 text, including the space. Thus two arguments would be generated.
7903 @item %@{@code{S}*&@code{T}*@}
7904 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7905 (the order of @code{S} and @code{T} in the spec is not significant).
7906 There can be any number of ampersand-separated variables; for each the
7907 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7909 @item %@{@code{S}:@code{X}@}
7910 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7912 @item %@{!@code{S}:@code{X}@}
7913 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7915 @item %@{@code{S}*:@code{X}@}
7916 Substitutes @code{X} if one or more switches whose names start with
7917 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7918 once, no matter how many such switches appeared. However, if @code{%*}
7919 appears somewhere in @code{X}, then @code{X} will be substituted once
7920 for each matching switch, with the @code{%*} replaced by the part of
7921 that switch that matched the @code{*}.
7923 @item %@{.@code{S}:@code{X}@}
7924 Substitutes @code{X}, if processing a file with suffix @code{S}.
7926 @item %@{!.@code{S}:@code{X}@}
7927 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7929 @item %@{,@code{S}:@code{X}@}
7930 Substitutes @code{X}, if processing a file for language @code{S}.
7932 @item %@{!,@code{S}:@code{X}@}
7933 Substitutes @code{X}, if not processing a file for language @code{S}.
7935 @item %@{@code{S}|@code{P}:@code{X}@}
7936 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
7937 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
7938 @code{*} sequences as well, although they have a stronger binding than
7939 the @samp{|}. If @code{%*} appears in @code{X}, all of the
7940 alternatives must be starred, and only the first matching alternative
7943 For example, a spec string like this:
7946 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7949 will output the following command-line options from the following input
7950 command-line options:
7955 -d fred.c -foo -baz -boggle
7956 -d jim.d -bar -baz -boggle
7959 @item %@{S:X; T:Y; :D@}
7961 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7962 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7963 be as many clauses as you need. This may be combined with @code{.},
7964 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
7969 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7970 construct may contain other nested @samp{%} constructs or spaces, or
7971 even newlines. They are processed as usual, as described above.
7972 Trailing white space in @code{X} is ignored. White space may also
7973 appear anywhere on the left side of the colon in these constructs,
7974 except between @code{.} or @code{*} and the corresponding word.
7976 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7977 handled specifically in these constructs. If another value of
7978 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7979 @option{-W} switch is found later in the command line, the earlier
7980 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7981 just one letter, which passes all matching options.
7983 The character @samp{|} at the beginning of the predicate text is used to
7984 indicate that a command should be piped to the following command, but
7985 only if @option{-pipe} is specified.
7987 It is built into GCC which switches take arguments and which do not.
7988 (You might think it would be useful to generalize this to allow each
7989 compiler's spec to say which switches take arguments. But this cannot
7990 be done in a consistent fashion. GCC cannot even decide which input
7991 files have been specified without knowing which switches take arguments,
7992 and it must know which input files to compile in order to tell which
7995 GCC also knows implicitly that arguments starting in @option{-l} are to be
7996 treated as compiler output files, and passed to the linker in their
7997 proper position among the other output files.
7999 @c man begin OPTIONS
8001 @node Target Options
8002 @section Specifying Target Machine and Compiler Version
8003 @cindex target options
8004 @cindex cross compiling
8005 @cindex specifying machine version
8006 @cindex specifying compiler version and target machine
8007 @cindex compiler version, specifying
8008 @cindex target machine, specifying
8010 The usual way to run GCC is to run the executable called @file{gcc}, or
8011 @file{<machine>-gcc} when cross-compiling, or
8012 @file{<machine>-gcc-<version>} to run a version other than the one that
8013 was installed last. Sometimes this is inconvenient, so GCC provides
8014 options that will switch to another cross-compiler or version.
8017 @item -b @var{machine}
8019 The argument @var{machine} specifies the target machine for compilation.
8021 The value to use for @var{machine} is the same as was specified as the
8022 machine type when configuring GCC as a cross-compiler. For
8023 example, if a cross-compiler was configured with @samp{configure
8024 arm-elf}, meaning to compile for an arm processor with elf binaries,
8025 then you would specify @option{-b arm-elf} to run that cross compiler.
8026 Because there are other options beginning with @option{-b}, the
8027 configuration must contain a hyphen.
8029 @item -V @var{version}
8031 The argument @var{version} specifies which version of GCC to run.
8032 This is useful when multiple versions are installed. For example,
8033 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8036 The @option{-V} and @option{-b} options work by running the
8037 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8038 use them if you can just run that directly.
8040 @node Submodel Options
8041 @section Hardware Models and Configurations
8042 @cindex submodel options
8043 @cindex specifying hardware config
8044 @cindex hardware models and configurations, specifying
8045 @cindex machine dependent options
8047 Earlier we discussed the standard option @option{-b} which chooses among
8048 different installed compilers for completely different target
8049 machines, such as VAX vs.@: 68000 vs.@: 80386.
8051 In addition, each of these target machine types can have its own
8052 special options, starting with @samp{-m}, to choose among various
8053 hardware models or configurations---for example, 68010 vs 68020,
8054 floating coprocessor or none. A single installed version of the
8055 compiler can compile for any model or configuration, according to the
8058 Some configurations of the compiler also support additional special
8059 options, usually for compatibility with other compilers on the same
8062 @c This list is ordered alphanumerically by subsection name.
8063 @c It should be the same order and spelling as these options are listed
8064 @c in Machine Dependent Options
8070 * Blackfin Options::
8074 * DEC Alpha Options::
8075 * DEC Alpha/VMS Options::
8077 * GNU/Linux Options::
8080 * i386 and x86-64 Options::
8093 * RS/6000 and PowerPC Options::
8094 * S/390 and zSeries Options::
8099 * System V Options::
8100 * TMS320C3x/C4x Options::
8105 * Xstormy16 Options::
8111 @subsection ARC Options
8114 These options are defined for ARC implementations:
8119 Compile code for little endian mode. This is the default.
8123 Compile code for big endian mode.
8126 @opindex mmangle-cpu
8127 Prepend the name of the cpu to all public symbol names.
8128 In multiple-processor systems, there are many ARC variants with different
8129 instruction and register set characteristics. This flag prevents code
8130 compiled for one cpu to be linked with code compiled for another.
8131 No facility exists for handling variants that are ``almost identical''.
8132 This is an all or nothing option.
8134 @item -mcpu=@var{cpu}
8136 Compile code for ARC variant @var{cpu}.
8137 Which variants are supported depend on the configuration.
8138 All variants support @option{-mcpu=base}, this is the default.
8140 @item -mtext=@var{text-section}
8141 @itemx -mdata=@var{data-section}
8142 @itemx -mrodata=@var{readonly-data-section}
8146 Put functions, data, and readonly data in @var{text-section},
8147 @var{data-section}, and @var{readonly-data-section} respectively
8148 by default. This can be overridden with the @code{section} attribute.
8149 @xref{Variable Attributes}.
8154 @subsection ARM Options
8157 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8161 @item -mabi=@var{name}
8163 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8164 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8167 @opindex mapcs-frame
8168 Generate a stack frame that is compliant with the ARM Procedure Call
8169 Standard for all functions, even if this is not strictly necessary for
8170 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8171 with this option will cause the stack frames not to be generated for
8172 leaf functions. The default is @option{-mno-apcs-frame}.
8176 This is a synonym for @option{-mapcs-frame}.
8179 @c not currently implemented
8180 @item -mapcs-stack-check
8181 @opindex mapcs-stack-check
8182 Generate code to check the amount of stack space available upon entry to
8183 every function (that actually uses some stack space). If there is
8184 insufficient space available then either the function
8185 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8186 called, depending upon the amount of stack space required. The run time
8187 system is required to provide these functions. The default is
8188 @option{-mno-apcs-stack-check}, since this produces smaller code.
8190 @c not currently implemented
8192 @opindex mapcs-float
8193 Pass floating point arguments using the float point registers. This is
8194 one of the variants of the APCS@. This option is recommended if the
8195 target hardware has a floating point unit or if a lot of floating point
8196 arithmetic is going to be performed by the code. The default is
8197 @option{-mno-apcs-float}, since integer only code is slightly increased in
8198 size if @option{-mapcs-float} is used.
8200 @c not currently implemented
8201 @item -mapcs-reentrant
8202 @opindex mapcs-reentrant
8203 Generate reentrant, position independent code. The default is
8204 @option{-mno-apcs-reentrant}.
8207 @item -mthumb-interwork
8208 @opindex mthumb-interwork
8209 Generate code which supports calling between the ARM and Thumb
8210 instruction sets. Without this option the two instruction sets cannot
8211 be reliably used inside one program. The default is
8212 @option{-mno-thumb-interwork}, since slightly larger code is generated
8213 when @option{-mthumb-interwork} is specified.
8215 @item -mno-sched-prolog
8216 @opindex mno-sched-prolog
8217 Prevent the reordering of instructions in the function prolog, or the
8218 merging of those instruction with the instructions in the function's
8219 body. This means that all functions will start with a recognizable set
8220 of instructions (or in fact one of a choice from a small set of
8221 different function prologues), and this information can be used to
8222 locate the start if functions inside an executable piece of code. The
8223 default is @option{-msched-prolog}.
8226 @opindex mhard-float
8227 Generate output containing floating point instructions. This is the
8231 @opindex msoft-float
8232 Generate output containing library calls for floating point.
8233 @strong{Warning:} the requisite libraries are not available for all ARM
8234 targets. Normally the facilities of the machine's usual C compiler are
8235 used, but this cannot be done directly in cross-compilation. You must make
8236 your own arrangements to provide suitable library functions for
8239 @option{-msoft-float} changes the calling convention in the output file;
8240 therefore, it is only useful if you compile @emph{all} of a program with
8241 this option. In particular, you need to compile @file{libgcc.a}, the
8242 library that comes with GCC, with @option{-msoft-float} in order for
8245 @item -mfloat-abi=@var{name}
8247 Specifies which ABI to use for floating point values. Permissible values
8248 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8250 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8251 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8252 of floating point instructions, but still uses the soft-float calling
8255 @item -mlittle-endian
8256 @opindex mlittle-endian
8257 Generate code for a processor running in little-endian mode. This is
8258 the default for all standard configurations.
8261 @opindex mbig-endian
8262 Generate code for a processor running in big-endian mode; the default is
8263 to compile code for a little-endian processor.
8265 @item -mwords-little-endian
8266 @opindex mwords-little-endian
8267 This option only applies when generating code for big-endian processors.
8268 Generate code for a little-endian word order but a big-endian byte
8269 order. That is, a byte order of the form @samp{32107654}. Note: this
8270 option should only be used if you require compatibility with code for
8271 big-endian ARM processors generated by versions of the compiler prior to
8274 @item -mcpu=@var{name}
8276 This specifies the name of the target ARM processor. GCC uses this name
8277 to determine what kind of instructions it can emit when generating
8278 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8279 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8280 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8281 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8282 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8283 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8284 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8285 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8286 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8287 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8288 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8289 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8290 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8291 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8292 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8293 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8295 @itemx -mtune=@var{name}
8297 This option is very similar to the @option{-mcpu=} option, except that
8298 instead of specifying the actual target processor type, and hence
8299 restricting which instructions can be used, it specifies that GCC should
8300 tune the performance of the code as if the target were of the type
8301 specified in this option, but still choosing the instructions that it
8302 will generate based on the cpu specified by a @option{-mcpu=} option.
8303 For some ARM implementations better performance can be obtained by using
8306 @item -march=@var{name}
8308 This specifies the name of the target ARM architecture. GCC uses this
8309 name to determine what kind of instructions it can emit when generating
8310 assembly code. This option can be used in conjunction with or instead
8311 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8312 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8313 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8314 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8315 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8317 @item -mfpu=@var{name}
8318 @itemx -mfpe=@var{number}
8319 @itemx -mfp=@var{number}
8323 This specifies what floating point hardware (or hardware emulation) is
8324 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8325 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8326 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8327 with older versions of GCC@.
8329 If @option{-msoft-float} is specified this specifies the format of
8330 floating point values.
8332 @item -mstructure-size-boundary=@var{n}
8333 @opindex mstructure-size-boundary
8334 The size of all structures and unions will be rounded up to a multiple
8335 of the number of bits set by this option. Permissible values are 8, 32
8336 and 64. The default value varies for different toolchains. For the COFF
8337 targeted toolchain the default value is 8. A value of 64 is only allowed
8338 if the underlying ABI supports it.
8340 Specifying the larger number can produce faster, more efficient code, but
8341 can also increase the size of the program. Different values are potentially
8342 incompatible. Code compiled with one value cannot necessarily expect to
8343 work with code or libraries compiled with another value, if they exchange
8344 information using structures or unions.
8346 @item -mabort-on-noreturn
8347 @opindex mabort-on-noreturn
8348 Generate a call to the function @code{abort} at the end of a
8349 @code{noreturn} function. It will be executed if the function tries to
8353 @itemx -mno-long-calls
8354 @opindex mlong-calls
8355 @opindex mno-long-calls
8356 Tells the compiler to perform function calls by first loading the
8357 address of the function into a register and then performing a subroutine
8358 call on this register. This switch is needed if the target function
8359 will lie outside of the 64 megabyte addressing range of the offset based
8360 version of subroutine call instruction.
8362 Even if this switch is enabled, not all function calls will be turned
8363 into long calls. The heuristic is that static functions, functions
8364 which have the @samp{short-call} attribute, functions that are inside
8365 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8366 definitions have already been compiled within the current compilation
8367 unit, will not be turned into long calls. The exception to this rule is
8368 that weak function definitions, functions with the @samp{long-call}
8369 attribute or the @samp{section} attribute, and functions that are within
8370 the scope of a @samp{#pragma long_calls} directive, will always be
8371 turned into long calls.
8373 This feature is not enabled by default. Specifying
8374 @option{-mno-long-calls} will restore the default behavior, as will
8375 placing the function calls within the scope of a @samp{#pragma
8376 long_calls_off} directive. Note these switches have no effect on how
8377 the compiler generates code to handle function calls via function
8380 @item -mnop-fun-dllimport
8381 @opindex mnop-fun-dllimport
8382 Disable support for the @code{dllimport} attribute.
8384 @item -msingle-pic-base
8385 @opindex msingle-pic-base
8386 Treat the register used for PIC addressing as read-only, rather than
8387 loading it in the prologue for each function. The run-time system is
8388 responsible for initializing this register with an appropriate value
8389 before execution begins.
8391 @item -mpic-register=@var{reg}
8392 @opindex mpic-register
8393 Specify the register to be used for PIC addressing. The default is R10
8394 unless stack-checking is enabled, when R9 is used.
8396 @item -mcirrus-fix-invalid-insns
8397 @opindex mcirrus-fix-invalid-insns
8398 @opindex mno-cirrus-fix-invalid-insns
8399 Insert NOPs into the instruction stream to in order to work around
8400 problems with invalid Maverick instruction combinations. This option
8401 is only valid if the @option{-mcpu=ep9312} option has been used to
8402 enable generation of instructions for the Cirrus Maverick floating
8403 point co-processor. This option is not enabled by default, since the
8404 problem is only present in older Maverick implementations. The default
8405 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8408 @item -mpoke-function-name
8409 @opindex mpoke-function-name
8410 Write the name of each function into the text section, directly
8411 preceding the function prologue. The generated code is similar to this:
8415 .ascii "arm_poke_function_name", 0
8418 .word 0xff000000 + (t1 - t0)
8419 arm_poke_function_name
8421 stmfd sp!, @{fp, ip, lr, pc@}
8425 When performing a stack backtrace, code can inspect the value of
8426 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8427 location @code{pc - 12} and the top 8 bits are set, then we know that
8428 there is a function name embedded immediately preceding this location
8429 and has length @code{((pc[-3]) & 0xff000000)}.
8433 Generate code for the Thumb instruction set. The default is to
8434 use the 32-bit ARM instruction set.
8435 This option automatically enables either 16-bit Thumb-1 or
8436 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8437 and @option{-march=@var{name}} options.
8440 @opindex mtpcs-frame
8441 Generate a stack frame that is compliant with the Thumb Procedure Call
8442 Standard for all non-leaf functions. (A leaf function is one that does
8443 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8445 @item -mtpcs-leaf-frame
8446 @opindex mtpcs-leaf-frame
8447 Generate a stack frame that is compliant with the Thumb Procedure Call
8448 Standard for all leaf functions. (A leaf function is one that does
8449 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8451 @item -mcallee-super-interworking
8452 @opindex mcallee-super-interworking
8453 Gives all externally visible functions in the file being compiled an ARM
8454 instruction set header which switches to Thumb mode before executing the
8455 rest of the function. This allows these functions to be called from
8456 non-interworking code.
8458 @item -mcaller-super-interworking
8459 @opindex mcaller-super-interworking
8460 Allows calls via function pointers (including virtual functions) to
8461 execute correctly regardless of whether the target code has been
8462 compiled for interworking or not. There is a small overhead in the cost
8463 of executing a function pointer if this option is enabled.
8465 @item -mtp=@var{name}
8467 Specify the access model for the thread local storage pointer. The valid
8468 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8469 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8470 (supported in the arm6k architecture), and @option{auto}, which uses the
8471 best available method for the selected processor. The default setting is
8477 @subsection AVR Options
8480 These options are defined for AVR implementations:
8483 @item -mmcu=@var{mcu}
8485 Specify ATMEL AVR instruction set or MCU type.
8487 Instruction set avr1 is for the minimal AVR core, not supported by the C
8488 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8489 attiny11, attiny12, attiny15, attiny28).
8491 Instruction set avr2 (default) is for the classic AVR core with up to
8492 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8493 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8494 at90c8534, at90s8535).
8496 Instruction set avr3 is for the classic AVR core with up to 128K program
8497 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8499 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8500 memory space (MCU types: atmega8, atmega83, atmega85).
8502 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8503 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8504 atmega64, atmega128, at43usb355, at94k).
8508 Output instruction sizes to the asm file.
8510 @item -minit-stack=@var{N}
8511 @opindex minit-stack
8512 Specify the initial stack address, which may be a symbol or numeric value,
8513 @samp{__stack} is the default.
8515 @item -mno-interrupts
8516 @opindex mno-interrupts
8517 Generated code is not compatible with hardware interrupts.
8518 Code size will be smaller.
8520 @item -mcall-prologues
8521 @opindex mcall-prologues
8522 Functions prologues/epilogues expanded as call to appropriate
8523 subroutines. Code size will be smaller.
8525 @item -mno-tablejump
8526 @opindex mno-tablejump
8527 Do not generate tablejump insns which sometimes increase code size.
8530 @opindex mtiny-stack
8531 Change only the low 8 bits of the stack pointer.
8535 Assume int to be 8 bit integer. This affects the sizes of all types: A
8536 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8537 and long long will be 4 bytes. Please note that this option does not
8538 comply to the C standards, but it will provide you with smaller code
8542 @node Blackfin Options
8543 @subsection Blackfin Options
8544 @cindex Blackfin Options
8547 @item -mcpu=@var{cpu}
8549 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8550 can be one of @samp{bf531}, @samp{bf532}, @samp{bf533},
8551 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf561}.
8552 Without this option, @samp{bf532} is used as the processor by default.
8553 The corresponding predefined processor macros for @var{cpu} is to
8554 be defined. For the @samp{bfin-elf} toolchain, this causes the hardware
8555 BSP provided by libgloss to be linked in if @samp{-msim} is not given.
8556 Support for @samp{bf561} is incomplete; only the processor macro is defined.
8560 Specifies that the program will be run on the simulator. This causes
8561 the simulator BSP provided by libgloss to be linked in. This option
8562 has effect only for @samp{bfin-elf} toolchain.
8564 @item -momit-leaf-frame-pointer
8565 @opindex momit-leaf-frame-pointer
8566 Don't keep the frame pointer in a register for leaf functions. This
8567 avoids the instructions to save, set up and restore frame pointers and
8568 makes an extra register available in leaf functions. The option
8569 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8570 which might make debugging harder.
8572 @item -mspecld-anomaly
8573 @opindex mspecld-anomaly
8574 When enabled, the compiler will ensure that the generated code does not
8575 contain speculative loads after jump instructions. This option is enabled
8578 @item -mno-specld-anomaly
8579 @opindex mno-specld-anomaly
8580 Don't generate extra code to prevent speculative loads from occurring.
8582 @item -mcsync-anomaly
8583 @opindex mcsync-anomaly
8584 When enabled, the compiler will ensure that the generated code does not
8585 contain CSYNC or SSYNC instructions too soon after conditional branches.
8586 This option is enabled by default.
8588 @item -mno-csync-anomaly
8589 @opindex mno-csync-anomaly
8590 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8591 occurring too soon after a conditional branch.
8595 When enabled, the compiler is free to take advantage of the knowledge that
8596 the entire program fits into the low 64k of memory.
8599 @opindex mno-low-64k
8600 Assume that the program is arbitrarily large. This is the default.
8602 @item -mstack-check-l1
8603 @opindex mstack-check-l1
8604 Do stack checking using information placed into L1 scratchpad memory by the
8607 @item -mid-shared-library
8608 @opindex mid-shared-library
8609 Generate code that supports shared libraries via the library ID method.
8610 This allows for execute in place and shared libraries in an environment
8611 without virtual memory management. This option implies @option{-fPIC}.
8613 @item -mno-id-shared-library
8614 @opindex mno-id-shared-library
8615 Generate code that doesn't assume ID based shared libraries are being used.
8616 This is the default.
8618 @item -mleaf-id-shared-library
8619 @opindex mleaf-id-shared-library
8620 Generate code that supports shared libraries via the library ID method,
8621 but assumes that this library or executable won't link against any other
8622 ID shared libraries. That allows the compiler to use faster code for jumps
8625 @item -mno-leaf-id-shared-library
8626 @opindex mno-leaf-id-shared-library
8627 Do not assume that the code being compiled won't link against any ID shared
8628 libraries. Slower code will be generated for jump and call insns.
8630 @item -mshared-library-id=n
8631 @opindex mshared-library-id
8632 Specified the identification number of the ID based shared library being
8633 compiled. Specifying a value of 0 will generate more compact code, specifying
8634 other values will force the allocation of that number to the current
8635 library but is no more space or time efficient than omitting this option.
8639 Generate code that allows the data segment to be located in a different
8640 area of memory from the text segment. This allows for execute in place in
8641 an environment without virtual memory management by eliminating relocations
8642 against the text section.
8645 @opindex mno-sep-data
8646 Generate code that assumes that the data segment follows the text segment.
8647 This is the default.
8650 @itemx -mno-long-calls
8651 @opindex mlong-calls
8652 @opindex mno-long-calls
8653 Tells the compiler to perform function calls by first loading the
8654 address of the function into a register and then performing a subroutine
8655 call on this register. This switch is needed if the target function
8656 will lie outside of the 24 bit addressing range of the offset based
8657 version of subroutine call instruction.
8659 This feature is not enabled by default. Specifying
8660 @option{-mno-long-calls} will restore the default behavior. Note these
8661 switches have no effect on how the compiler generates code to handle
8662 function calls via function pointers.
8666 Link with the fast floating-point library. This library relaxes some of
8667 the IEEE floating-point standard's rules for checking inputs against
8668 Not-a-Number (NAN), in the interest of performance.
8671 @opindex minline-plt
8672 Enable inlining of PLT entries in function calls to functions that are
8673 not known to bind locally. It has no effect without @option{-mfdpic}.
8677 @subsection CRIS Options
8678 @cindex CRIS Options
8680 These options are defined specifically for the CRIS ports.
8683 @item -march=@var{architecture-type}
8684 @itemx -mcpu=@var{architecture-type}
8687 Generate code for the specified architecture. The choices for
8688 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8689 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8690 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8693 @item -mtune=@var{architecture-type}
8695 Tune to @var{architecture-type} everything applicable about the generated
8696 code, except for the ABI and the set of available instructions. The
8697 choices for @var{architecture-type} are the same as for
8698 @option{-march=@var{architecture-type}}.
8700 @item -mmax-stack-frame=@var{n}
8701 @opindex mmax-stack-frame
8702 Warn when the stack frame of a function exceeds @var{n} bytes.
8704 @item -melinux-stacksize=@var{n}
8705 @opindex melinux-stacksize
8706 Only available with the @samp{cris-axis-aout} target. Arranges for
8707 indications in the program to the kernel loader that the stack of the
8708 program should be set to @var{n} bytes.
8714 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8715 @option{-march=v3} and @option{-march=v8} respectively.
8717 @item -mmul-bug-workaround
8718 @itemx -mno-mul-bug-workaround
8719 @opindex mmul-bug-workaround
8720 @opindex mno-mul-bug-workaround
8721 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8722 models where it applies. This option is active by default.
8726 Enable CRIS-specific verbose debug-related information in the assembly
8727 code. This option also has the effect to turn off the @samp{#NO_APP}
8728 formatted-code indicator to the assembler at the beginning of the
8733 Do not use condition-code results from previous instruction; always emit
8734 compare and test instructions before use of condition codes.
8736 @item -mno-side-effects
8737 @opindex mno-side-effects
8738 Do not emit instructions with side-effects in addressing modes other than
8742 @itemx -mno-stack-align
8744 @itemx -mno-data-align
8745 @itemx -mconst-align
8746 @itemx -mno-const-align
8747 @opindex mstack-align
8748 @opindex mno-stack-align
8749 @opindex mdata-align
8750 @opindex mno-data-align
8751 @opindex mconst-align
8752 @opindex mno-const-align
8753 These options (no-options) arranges (eliminate arrangements) for the
8754 stack-frame, individual data and constants to be aligned for the maximum
8755 single data access size for the chosen CPU model. The default is to
8756 arrange for 32-bit alignment. ABI details such as structure layout are
8757 not affected by these options.
8765 Similar to the stack- data- and const-align options above, these options
8766 arrange for stack-frame, writable data and constants to all be 32-bit,
8767 16-bit or 8-bit aligned. The default is 32-bit alignment.
8769 @item -mno-prologue-epilogue
8770 @itemx -mprologue-epilogue
8771 @opindex mno-prologue-epilogue
8772 @opindex mprologue-epilogue
8773 With @option{-mno-prologue-epilogue}, the normal function prologue and
8774 epilogue that sets up the stack-frame are omitted and no return
8775 instructions or return sequences are generated in the code. Use this
8776 option only together with visual inspection of the compiled code: no
8777 warnings or errors are generated when call-saved registers must be saved,
8778 or storage for local variable needs to be allocated.
8784 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8785 instruction sequences that load addresses for functions from the PLT part
8786 of the GOT rather than (traditional on other architectures) calls to the
8787 PLT@. The default is @option{-mgotplt}.
8791 Legacy no-op option only recognized with the cris-axis-aout target.
8795 Legacy no-op option only recognized with the cris-axis-elf and
8796 cris-axis-linux-gnu targets.
8800 Only recognized with the cris-axis-aout target, where it selects a
8801 GNU/linux-like multilib, include files and instruction set for
8806 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8810 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8811 to link with input-output functions from a simulator library. Code,
8812 initialized data and zero-initialized data are allocated consecutively.
8816 Like @option{-sim}, but pass linker options to locate initialized data at
8817 0x40000000 and zero-initialized data at 0x80000000.
8821 @subsection CRX Options
8824 These options are defined specifically for the CRX ports.
8830 Enable the use of multiply-accumulate instructions. Disabled by default.
8834 Push instructions will be used to pass outgoing arguments when functions
8835 are called. Enabled by default.
8838 @node Darwin Options
8839 @subsection Darwin Options
8840 @cindex Darwin options
8842 These options are defined for all architectures running the Darwin operating
8845 FSF GCC on Darwin does not create ``fat'' object files; it will create
8846 an object file for the single architecture that it was built to
8847 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8848 @option{-arch} options are used; it does so by running the compiler or
8849 linker multiple times and joining the results together with
8852 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8853 @samp{i686}) is determined by the flags that specify the ISA
8854 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8855 @option{-force_cpusubtype_ALL} option can be used to override this.
8857 The Darwin tools vary in their behavior when presented with an ISA
8858 mismatch. The assembler, @file{as}, will only permit instructions to
8859 be used that are valid for the subtype of the file it is generating,
8860 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8861 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8862 and print an error if asked to create a shared library with a less
8863 restrictive subtype than its input files (for instance, trying to put
8864 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8865 for executables, @file{ld}, will quietly give the executable the most
8866 restrictive subtype of any of its input files.
8871 Add the framework directory @var{dir} to the head of the list of
8872 directories to be searched for header files. These directories are
8873 interleaved with those specified by @option{-I} options and are
8874 scanned in a left-to-right order.
8876 A framework directory is a directory with frameworks in it. A
8877 framework is a directory with a @samp{"Headers"} and/or
8878 @samp{"PrivateHeaders"} directory contained directly in it that ends
8879 in @samp{".framework"}. The name of a framework is the name of this
8880 directory excluding the @samp{".framework"}. Headers associated with
8881 the framework are found in one of those two directories, with
8882 @samp{"Headers"} being searched first. A subframework is a framework
8883 directory that is in a framework's @samp{"Frameworks"} directory.
8884 Includes of subframework headers can only appear in a header of a
8885 framework that contains the subframework, or in a sibling subframework
8886 header. Two subframeworks are siblings if they occur in the same
8887 framework. A subframework should not have the same name as a
8888 framework, a warning will be issued if this is violated. Currently a
8889 subframework cannot have subframeworks, in the future, the mechanism
8890 may be extended to support this. The standard frameworks can be found
8891 in @samp{"/System/Library/Frameworks"} and
8892 @samp{"/Library/Frameworks"}. An example include looks like
8893 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8894 the name of the framework and header.h is found in the
8895 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8897 @item -iframework@var{dir}
8899 Like @option{-F} except the directory is a treated as a system
8900 directory. The main difference between this @option{-iframework} and
8901 @option{-F} is that with @option{-iframework} the compiler does not
8902 warn about constructs contained within header files found via
8903 @var{dir}. This option is valid only for the C family of languages.
8907 Emit debugging information for symbols that are used. For STABS
8908 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8909 This is by default ON@.
8913 Emit debugging information for all symbols and types.
8915 @item -mmacosx-version-min=@var{version}
8916 The earliest version of MacOS X that this executable will run on
8917 is @var{version}. Typical values of @var{version} include @code{10.1},
8918 @code{10.2}, and @code{10.3.9}.
8920 If the compiler was built to use the system's headers by default,
8921 then the default for this option is the system version on which the
8922 compiler is running, otherwise the default is to make choices which
8923 are compatible with as many systems and code bases as possible.
8927 Enable kernel development mode. The @option{-mkernel} option sets
8928 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8929 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8930 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8931 applicable. This mode also sets @option{-mno-altivec},
8932 @option{-msoft-float}, @option{-fno-builtin} and
8933 @option{-mlong-branch} for PowerPC targets.
8935 @item -mone-byte-bool
8936 @opindex mone-byte-bool
8937 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8938 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8939 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8940 option has no effect on x86.
8942 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8943 to generate code that is not binary compatible with code generated
8944 without that switch. Using this switch may require recompiling all
8945 other modules in a program, including system libraries. Use this
8946 switch to conform to a non-default data model.
8948 @item -mfix-and-continue
8949 @itemx -ffix-and-continue
8950 @itemx -findirect-data
8951 @opindex mfix-and-continue
8952 @opindex ffix-and-continue
8953 @opindex findirect-data
8954 Generate code suitable for fast turn around development. Needed to
8955 enable gdb to dynamically load @code{.o} files into already running
8956 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8957 are provided for backwards compatibility.
8961 Loads all members of static archive libraries.
8962 See man ld(1) for more information.
8964 @item -arch_errors_fatal
8965 @opindex arch_errors_fatal
8966 Cause the errors having to do with files that have the wrong architecture
8970 @opindex bind_at_load
8971 Causes the output file to be marked such that the dynamic linker will
8972 bind all undefined references when the file is loaded or launched.
8976 Produce a Mach-o bundle format file.
8977 See man ld(1) for more information.
8979 @item -bundle_loader @var{executable}
8980 @opindex bundle_loader
8981 This option specifies the @var{executable} that will be loading the build
8982 output file being linked. See man ld(1) for more information.
8986 When passed this option, GCC will produce a dynamic library instead of
8987 an executable when linking, using the Darwin @file{libtool} command.
8989 @item -force_cpusubtype_ALL
8990 @opindex force_cpusubtype_ALL
8991 This causes GCC's output file to have the @var{ALL} subtype, instead of
8992 one controlled by the @option{-mcpu} or @option{-march} option.
8994 @item -allowable_client @var{client_name}
8996 @itemx -compatibility_version
8997 @itemx -current_version
8999 @itemx -dependency-file
9001 @itemx -dylinker_install_name
9003 @itemx -exported_symbols_list
9005 @itemx -flat_namespace
9006 @itemx -force_flat_namespace
9007 @itemx -headerpad_max_install_names
9010 @itemx -install_name
9011 @itemx -keep_private_externs
9012 @itemx -multi_module
9013 @itemx -multiply_defined
9014 @itemx -multiply_defined_unused
9016 @itemx -no_dead_strip_inits_and_terms
9017 @itemx -nofixprebinding
9020 @itemx -noseglinkedit
9021 @itemx -pagezero_size
9023 @itemx -prebind_all_twolevel_modules
9024 @itemx -private_bundle
9025 @itemx -read_only_relocs
9027 @itemx -sectobjectsymbols
9031 @itemx -sectobjectsymbols
9034 @itemx -segs_read_only_addr
9035 @itemx -segs_read_write_addr
9036 @itemx -seg_addr_table
9037 @itemx -seg_addr_table_filename
9040 @itemx -segs_read_only_addr
9041 @itemx -segs_read_write_addr
9042 @itemx -single_module
9045 @itemx -sub_umbrella
9046 @itemx -twolevel_namespace
9049 @itemx -unexported_symbols_list
9050 @itemx -weak_reference_mismatches
9053 @opindex allowable_client
9054 @opindex client_name
9055 @opindex compatibility_version
9056 @opindex current_version
9058 @opindex dependency-file
9060 @opindex dylinker_install_name
9062 @opindex exported_symbols_list
9064 @opindex flat_namespace
9065 @opindex force_flat_namespace
9066 @opindex headerpad_max_install_names
9069 @opindex install_name
9070 @opindex keep_private_externs
9071 @opindex multi_module
9072 @opindex multiply_defined
9073 @opindex multiply_defined_unused
9075 @opindex no_dead_strip_inits_and_terms
9076 @opindex nofixprebinding
9077 @opindex nomultidefs
9079 @opindex noseglinkedit
9080 @opindex pagezero_size
9082 @opindex prebind_all_twolevel_modules
9083 @opindex private_bundle
9084 @opindex read_only_relocs
9086 @opindex sectobjectsymbols
9090 @opindex sectobjectsymbols
9093 @opindex segs_read_only_addr
9094 @opindex segs_read_write_addr
9095 @opindex seg_addr_table
9096 @opindex seg_addr_table_filename
9097 @opindex seglinkedit
9099 @opindex segs_read_only_addr
9100 @opindex segs_read_write_addr
9101 @opindex single_module
9103 @opindex sub_library
9104 @opindex sub_umbrella
9105 @opindex twolevel_namespace
9108 @opindex unexported_symbols_list
9109 @opindex weak_reference_mismatches
9110 @opindex whatsloaded
9112 These options are passed to the Darwin linker. The Darwin linker man page
9113 describes them in detail.
9116 @node DEC Alpha Options
9117 @subsection DEC Alpha Options
9119 These @samp{-m} options are defined for the DEC Alpha implementations:
9122 @item -mno-soft-float
9124 @opindex mno-soft-float
9125 @opindex msoft-float
9126 Use (do not use) the hardware floating-point instructions for
9127 floating-point operations. When @option{-msoft-float} is specified,
9128 functions in @file{libgcc.a} will be used to perform floating-point
9129 operations. Unless they are replaced by routines that emulate the
9130 floating-point operations, or compiled in such a way as to call such
9131 emulations routines, these routines will issue floating-point
9132 operations. If you are compiling for an Alpha without floating-point
9133 operations, you must ensure that the library is built so as not to call
9136 Note that Alpha implementations without floating-point operations are
9137 required to have floating-point registers.
9142 @opindex mno-fp-regs
9143 Generate code that uses (does not use) the floating-point register set.
9144 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9145 register set is not used, floating point operands are passed in integer
9146 registers as if they were integers and floating-point results are passed
9147 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9148 so any function with a floating-point argument or return value called by code
9149 compiled with @option{-mno-fp-regs} must also be compiled with that
9152 A typical use of this option is building a kernel that does not use,
9153 and hence need not save and restore, any floating-point registers.
9157 The Alpha architecture implements floating-point hardware optimized for
9158 maximum performance. It is mostly compliant with the IEEE floating
9159 point standard. However, for full compliance, software assistance is
9160 required. This option generates code fully IEEE compliant code
9161 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9162 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9163 defined during compilation. The resulting code is less efficient but is
9164 able to correctly support denormalized numbers and exceptional IEEE
9165 values such as not-a-number and plus/minus infinity. Other Alpha
9166 compilers call this option @option{-ieee_with_no_inexact}.
9168 @item -mieee-with-inexact
9169 @opindex mieee-with-inexact
9170 This is like @option{-mieee} except the generated code also maintains
9171 the IEEE @var{inexact-flag}. Turning on this option causes the
9172 generated code to implement fully-compliant IEEE math. In addition to
9173 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9174 macro. On some Alpha implementations the resulting code may execute
9175 significantly slower than the code generated by default. Since there is
9176 very little code that depends on the @var{inexact-flag}, you should
9177 normally not specify this option. Other Alpha compilers call this
9178 option @option{-ieee_with_inexact}.
9180 @item -mfp-trap-mode=@var{trap-mode}
9181 @opindex mfp-trap-mode
9182 This option controls what floating-point related traps are enabled.
9183 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9184 The trap mode can be set to one of four values:
9188 This is the default (normal) setting. The only traps that are enabled
9189 are the ones that cannot be disabled in software (e.g., division by zero
9193 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9197 Like @samp{u}, but the instructions are marked to be safe for software
9198 completion (see Alpha architecture manual for details).
9201 Like @samp{su}, but inexact traps are enabled as well.
9204 @item -mfp-rounding-mode=@var{rounding-mode}
9205 @opindex mfp-rounding-mode
9206 Selects the IEEE rounding mode. Other Alpha compilers call this option
9207 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9212 Normal IEEE rounding mode. Floating point numbers are rounded towards
9213 the nearest machine number or towards the even machine number in case
9217 Round towards minus infinity.
9220 Chopped rounding mode. Floating point numbers are rounded towards zero.
9223 Dynamic rounding mode. A field in the floating point control register
9224 (@var{fpcr}, see Alpha architecture reference manual) controls the
9225 rounding mode in effect. The C library initializes this register for
9226 rounding towards plus infinity. Thus, unless your program modifies the
9227 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9230 @item -mtrap-precision=@var{trap-precision}
9231 @opindex mtrap-precision
9232 In the Alpha architecture, floating point traps are imprecise. This
9233 means without software assistance it is impossible to recover from a
9234 floating trap and program execution normally needs to be terminated.
9235 GCC can generate code that can assist operating system trap handlers
9236 in determining the exact location that caused a floating point trap.
9237 Depending on the requirements of an application, different levels of
9238 precisions can be selected:
9242 Program precision. This option is the default and means a trap handler
9243 can only identify which program caused a floating point exception.
9246 Function precision. The trap handler can determine the function that
9247 caused a floating point exception.
9250 Instruction precision. The trap handler can determine the exact
9251 instruction that caused a floating point exception.
9254 Other Alpha compilers provide the equivalent options called
9255 @option{-scope_safe} and @option{-resumption_safe}.
9257 @item -mieee-conformant
9258 @opindex mieee-conformant
9259 This option marks the generated code as IEEE conformant. You must not
9260 use this option unless you also specify @option{-mtrap-precision=i} and either
9261 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9262 is to emit the line @samp{.eflag 48} in the function prologue of the
9263 generated assembly file. Under DEC Unix, this has the effect that
9264 IEEE-conformant math library routines will be linked in.
9266 @item -mbuild-constants
9267 @opindex mbuild-constants
9268 Normally GCC examines a 32- or 64-bit integer constant to
9269 see if it can construct it from smaller constants in two or three
9270 instructions. If it cannot, it will output the constant as a literal and
9271 generate code to load it from the data segment at runtime.
9273 Use this option to require GCC to construct @emph{all} integer constants
9274 using code, even if it takes more instructions (the maximum is six).
9276 You would typically use this option to build a shared library dynamic
9277 loader. Itself a shared library, it must relocate itself in memory
9278 before it can find the variables and constants in its own data segment.
9284 Select whether to generate code to be assembled by the vendor-supplied
9285 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9303 Indicate whether GCC should generate code to use the optional BWX,
9304 CIX, FIX and MAX instruction sets. The default is to use the instruction
9305 sets supported by the CPU type specified via @option{-mcpu=} option or that
9306 of the CPU on which GCC was built if none was specified.
9311 @opindex mfloat-ieee
9312 Generate code that uses (does not use) VAX F and G floating point
9313 arithmetic instead of IEEE single and double precision.
9315 @item -mexplicit-relocs
9316 @itemx -mno-explicit-relocs
9317 @opindex mexplicit-relocs
9318 @opindex mno-explicit-relocs
9319 Older Alpha assemblers provided no way to generate symbol relocations
9320 except via assembler macros. Use of these macros does not allow
9321 optimal instruction scheduling. GNU binutils as of version 2.12
9322 supports a new syntax that allows the compiler to explicitly mark
9323 which relocations should apply to which instructions. This option
9324 is mostly useful for debugging, as GCC detects the capabilities of
9325 the assembler when it is built and sets the default accordingly.
9329 @opindex msmall-data
9330 @opindex mlarge-data
9331 When @option{-mexplicit-relocs} is in effect, static data is
9332 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9333 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9334 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9335 16-bit relocations off of the @code{$gp} register. This limits the
9336 size of the small data area to 64KB, but allows the variables to be
9337 directly accessed via a single instruction.
9339 The default is @option{-mlarge-data}. With this option the data area
9340 is limited to just below 2GB@. Programs that require more than 2GB of
9341 data must use @code{malloc} or @code{mmap} to allocate the data in the
9342 heap instead of in the program's data segment.
9344 When generating code for shared libraries, @option{-fpic} implies
9345 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9349 @opindex msmall-text
9350 @opindex mlarge-text
9351 When @option{-msmall-text} is used, the compiler assumes that the
9352 code of the entire program (or shared library) fits in 4MB, and is
9353 thus reachable with a branch instruction. When @option{-msmall-data}
9354 is used, the compiler can assume that all local symbols share the
9355 same @code{$gp} value, and thus reduce the number of instructions
9356 required for a function call from 4 to 1.
9358 The default is @option{-mlarge-text}.
9360 @item -mcpu=@var{cpu_type}
9362 Set the instruction set and instruction scheduling parameters for
9363 machine type @var{cpu_type}. You can specify either the @samp{EV}
9364 style name or the corresponding chip number. GCC supports scheduling
9365 parameters for the EV4, EV5 and EV6 family of processors and will
9366 choose the default values for the instruction set from the processor
9367 you specify. If you do not specify a processor type, GCC will default
9368 to the processor on which the compiler was built.
9370 Supported values for @var{cpu_type} are
9376 Schedules as an EV4 and has no instruction set extensions.
9380 Schedules as an EV5 and has no instruction set extensions.
9384 Schedules as an EV5 and supports the BWX extension.
9389 Schedules as an EV5 and supports the BWX and MAX extensions.
9393 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9397 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9400 @item -mtune=@var{cpu_type}
9402 Set only the instruction scheduling parameters for machine type
9403 @var{cpu_type}. The instruction set is not changed.
9405 @item -mmemory-latency=@var{time}
9406 @opindex mmemory-latency
9407 Sets the latency the scheduler should assume for typical memory
9408 references as seen by the application. This number is highly
9409 dependent on the memory access patterns used by the application
9410 and the size of the external cache on the machine.
9412 Valid options for @var{time} are
9416 A decimal number representing clock cycles.
9422 The compiler contains estimates of the number of clock cycles for
9423 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9424 (also called Dcache, Scache, and Bcache), as well as to main memory.
9425 Note that L3 is only valid for EV5.
9430 @node DEC Alpha/VMS Options
9431 @subsection DEC Alpha/VMS Options
9433 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9436 @item -mvms-return-codes
9437 @opindex mvms-return-codes
9438 Return VMS condition codes from main. The default is to return POSIX
9439 style condition (e.g.@: error) codes.
9443 @subsection FRV Options
9450 Only use the first 32 general purpose registers.
9455 Use all 64 general purpose registers.
9460 Use only the first 32 floating point registers.
9465 Use all 64 floating point registers
9468 @opindex mhard-float
9470 Use hardware instructions for floating point operations.
9473 @opindex msoft-float
9475 Use library routines for floating point operations.
9480 Dynamically allocate condition code registers.
9485 Do not try to dynamically allocate condition code registers, only
9486 use @code{icc0} and @code{fcc0}.
9491 Change ABI to use double word insns.
9496 Do not use double word instructions.
9501 Use floating point double instructions.
9506 Do not use floating point double instructions.
9511 Use media instructions.
9516 Do not use media instructions.
9521 Use multiply and add/subtract instructions.
9526 Do not use multiply and add/subtract instructions.
9531 Select the FDPIC ABI, that uses function descriptors to represent
9532 pointers to functions. Without any PIC/PIE-related options, it
9533 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9534 assumes GOT entries and small data are within a 12-bit range from the
9535 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9536 are computed with 32 bits.
9539 @opindex minline-plt
9541 Enable inlining of PLT entries in function calls to functions that are
9542 not known to bind locally. It has no effect without @option{-mfdpic}.
9543 It's enabled by default if optimizing for speed and compiling for
9544 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9545 optimization option such as @option{-O3} or above is present in the
9551 Assume a large TLS segment when generating thread-local code.
9556 Do not assume a large TLS segment when generating thread-local code.
9561 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9562 that is known to be in read-only sections. It's enabled by default,
9563 except for @option{-fpic} or @option{-fpie}: even though it may help
9564 make the global offset table smaller, it trades 1 instruction for 4.
9565 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9566 one of which may be shared by multiple symbols, and it avoids the need
9567 for a GOT entry for the referenced symbol, so it's more likely to be a
9568 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9570 @item -multilib-library-pic
9571 @opindex multilib-library-pic
9573 Link with the (library, not FD) pic libraries. It's implied by
9574 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9575 @option{-fpic} without @option{-mfdpic}. You should never have to use
9581 Follow the EABI requirement of always creating a frame pointer whenever
9582 a stack frame is allocated. This option is enabled by default and can
9583 be disabled with @option{-mno-linked-fp}.
9586 @opindex mlong-calls
9588 Use indirect addressing to call functions outside the current
9589 compilation unit. This allows the functions to be placed anywhere
9590 within the 32-bit address space.
9592 @item -malign-labels
9593 @opindex malign-labels
9595 Try to align labels to an 8-byte boundary by inserting nops into the
9596 previous packet. This option only has an effect when VLIW packing
9597 is enabled. It doesn't create new packets; it merely adds nops to
9601 @opindex mlibrary-pic
9603 Generate position-independent EABI code.
9608 Use only the first four media accumulator registers.
9613 Use all eight media accumulator registers.
9618 Pack VLIW instructions.
9623 Do not pack VLIW instructions.
9628 Do not mark ABI switches in e_flags.
9633 Enable the use of conditional-move instructions (default).
9635 This switch is mainly for debugging the compiler and will likely be removed
9636 in a future version.
9638 @item -mno-cond-move
9639 @opindex mno-cond-move
9641 Disable the use of conditional-move instructions.
9643 This switch is mainly for debugging the compiler and will likely be removed
9644 in a future version.
9649 Enable the use of conditional set instructions (default).
9651 This switch is mainly for debugging the compiler and will likely be removed
9652 in a future version.
9657 Disable the use of conditional set instructions.
9659 This switch is mainly for debugging the compiler and will likely be removed
9660 in a future version.
9665 Enable the use of conditional execution (default).
9667 This switch is mainly for debugging the compiler and will likely be removed
9668 in a future version.
9670 @item -mno-cond-exec
9671 @opindex mno-cond-exec
9673 Disable the use of conditional execution.
9675 This switch is mainly for debugging the compiler and will likely be removed
9676 in a future version.
9679 @opindex mvliw-branch
9681 Run a pass to pack branches into VLIW instructions (default).
9683 This switch is mainly for debugging the compiler and will likely be removed
9684 in a future version.
9686 @item -mno-vliw-branch
9687 @opindex mno-vliw-branch
9689 Do not run a pass to pack branches into VLIW instructions.
9691 This switch is mainly for debugging the compiler and will likely be removed
9692 in a future version.
9694 @item -mmulti-cond-exec
9695 @opindex mmulti-cond-exec
9697 Enable optimization of @code{&&} and @code{||} in conditional execution
9700 This switch is mainly for debugging the compiler and will likely be removed
9701 in a future version.
9703 @item -mno-multi-cond-exec
9704 @opindex mno-multi-cond-exec
9706 Disable optimization of @code{&&} and @code{||} in conditional execution.
9708 This switch is mainly for debugging the compiler and will likely be removed
9709 in a future version.
9711 @item -mnested-cond-exec
9712 @opindex mnested-cond-exec
9714 Enable nested conditional execution optimizations (default).
9716 This switch is mainly for debugging the compiler and will likely be removed
9717 in a future version.
9719 @item -mno-nested-cond-exec
9720 @opindex mno-nested-cond-exec
9722 Disable nested conditional execution optimizations.
9724 This switch is mainly for debugging the compiler and will likely be removed
9725 in a future version.
9727 @item -moptimize-membar
9728 @opindex moptimize-membar
9730 This switch removes redundant @code{membar} instructions from the
9731 compiler generated code. It is enabled by default.
9733 @item -mno-optimize-membar
9734 @opindex mno-optimize-membar
9736 This switch disables the automatic removal of redundant @code{membar}
9737 instructions from the generated code.
9739 @item -mtomcat-stats
9740 @opindex mtomcat-stats
9742 Cause gas to print out tomcat statistics.
9744 @item -mcpu=@var{cpu}
9747 Select the processor type for which to generate code. Possible values are
9748 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9749 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9753 @node GNU/Linux Options
9754 @subsection GNU/Linux Options
9756 These @samp{-m} options are defined for GNU/Linux targets:
9761 Use the GNU C library instead of uClibc. This is the default except
9762 on @samp{*-*-linux-*uclibc*} targets.
9766 Use uClibc instead of the GNU C library. This is the default on
9767 @samp{*-*-linux-*uclibc*} targets.
9770 @node H8/300 Options
9771 @subsection H8/300 Options
9773 These @samp{-m} options are defined for the H8/300 implementations:
9778 Shorten some address references at link time, when possible; uses the
9779 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9780 ld, Using ld}, for a fuller description.
9784 Generate code for the H8/300H@.
9788 Generate code for the H8S@.
9792 Generate code for the H8S and H8/300H in the normal mode. This switch
9793 must be used either with @option{-mh} or @option{-ms}.
9797 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9801 Make @code{int} data 32 bits by default.
9805 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9806 The default for the H8/300H and H8S is to align longs and floats on 4
9808 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9809 This option has no effect on the H8/300.
9813 @subsection HPPA Options
9814 @cindex HPPA Options
9816 These @samp{-m} options are defined for the HPPA family of computers:
9819 @item -march=@var{architecture-type}
9821 Generate code for the specified architecture. The choices for
9822 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9823 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9824 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9825 architecture option for your machine. Code compiled for lower numbered
9826 architectures will run on higher numbered architectures, but not the
9830 @itemx -mpa-risc-1-1
9831 @itemx -mpa-risc-2-0
9832 @opindex mpa-risc-1-0
9833 @opindex mpa-risc-1-1
9834 @opindex mpa-risc-2-0
9835 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9838 @opindex mbig-switch
9839 Generate code suitable for big switch tables. Use this option only if
9840 the assembler/linker complain about out of range branches within a switch
9843 @item -mjump-in-delay
9844 @opindex mjump-in-delay
9845 Fill delay slots of function calls with unconditional jump instructions
9846 by modifying the return pointer for the function call to be the target
9847 of the conditional jump.
9849 @item -mdisable-fpregs
9850 @opindex mdisable-fpregs
9851 Prevent floating point registers from being used in any manner. This is
9852 necessary for compiling kernels which perform lazy context switching of
9853 floating point registers. If you use this option and attempt to perform
9854 floating point operations, the compiler will abort.
9856 @item -mdisable-indexing
9857 @opindex mdisable-indexing
9858 Prevent the compiler from using indexing address modes. This avoids some
9859 rather obscure problems when compiling MIG generated code under MACH@.
9861 @item -mno-space-regs
9862 @opindex mno-space-regs
9863 Generate code that assumes the target has no space registers. This allows
9864 GCC to generate faster indirect calls and use unscaled index address modes.
9866 Such code is suitable for level 0 PA systems and kernels.
9868 @item -mfast-indirect-calls
9869 @opindex mfast-indirect-calls
9870 Generate code that assumes calls never cross space boundaries. This
9871 allows GCC to emit code which performs faster indirect calls.
9873 This option will not work in the presence of shared libraries or nested
9876 @item -mfixed-range=@var{register-range}
9877 @opindex mfixed-range
9878 Generate code treating the given register range as fixed registers.
9879 A fixed register is one that the register allocator can not use. This is
9880 useful when compiling kernel code. A register range is specified as
9881 two registers separated by a dash. Multiple register ranges can be
9882 specified separated by a comma.
9884 @item -mlong-load-store
9885 @opindex mlong-load-store
9886 Generate 3-instruction load and store sequences as sometimes required by
9887 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9890 @item -mportable-runtime
9891 @opindex mportable-runtime
9892 Use the portable calling conventions proposed by HP for ELF systems.
9896 Enable the use of assembler directives only GAS understands.
9898 @item -mschedule=@var{cpu-type}
9900 Schedule code according to the constraints for the machine type
9901 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9902 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9903 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9904 proper scheduling option for your machine. The default scheduling is
9908 @opindex mlinker-opt
9909 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9910 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9911 linkers in which they give bogus error messages when linking some programs.
9914 @opindex msoft-float
9915 Generate output containing library calls for floating point.
9916 @strong{Warning:} the requisite libraries are not available for all HPPA
9917 targets. Normally the facilities of the machine's usual C compiler are
9918 used, but this cannot be done directly in cross-compilation. You must make
9919 your own arrangements to provide suitable library functions for
9920 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9921 does provide software floating point support.
9923 @option{-msoft-float} changes the calling convention in the output file;
9924 therefore, it is only useful if you compile @emph{all} of a program with
9925 this option. In particular, you need to compile @file{libgcc.a}, the
9926 library that comes with GCC, with @option{-msoft-float} in order for
9931 Generate the predefine, @code{_SIO}, for server IO@. The default is
9932 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9933 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9934 options are available under HP-UX and HI-UX@.
9938 Use GNU ld specific options. This passes @option{-shared} to ld when
9939 building a shared library. It is the default when GCC is configured,
9940 explicitly or implicitly, with the GNU linker. This option does not
9941 have any affect on which ld is called, it only changes what parameters
9942 are passed to that ld. The ld that is called is determined by the
9943 @option{--with-ld} configure option, GCC's program search path, and
9944 finally by the user's @env{PATH}. The linker used by GCC can be printed
9945 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9946 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9950 Use HP ld specific options. This passes @option{-b} to ld when building
9951 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9952 links. It is the default when GCC is configured, explicitly or
9953 implicitly, with the HP linker. This option does not have any affect on
9954 which ld is called, it only changes what parameters are passed to that
9955 ld. The ld that is called is determined by the @option{--with-ld}
9956 configure option, GCC's program search path, and finally by the user's
9957 @env{PATH}. The linker used by GCC can be printed using @samp{which
9958 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9959 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9962 @opindex mno-long-calls
9963 Generate code that uses long call sequences. This ensures that a call
9964 is always able to reach linker generated stubs. The default is to generate
9965 long calls only when the distance from the call site to the beginning
9966 of the function or translation unit, as the case may be, exceeds a
9967 predefined limit set by the branch type being used. The limits for
9968 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9969 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9972 Distances are measured from the beginning of functions when using the
9973 @option{-ffunction-sections} option, or when using the @option{-mgas}
9974 and @option{-mno-portable-runtime} options together under HP-UX with
9977 It is normally not desirable to use this option as it will degrade
9978 performance. However, it may be useful in large applications,
9979 particularly when partial linking is used to build the application.
9981 The types of long calls used depends on the capabilities of the
9982 assembler and linker, and the type of code being generated. The
9983 impact on systems that support long absolute calls, and long pic
9984 symbol-difference or pc-relative calls should be relatively small.
9985 However, an indirect call is used on 32-bit ELF systems in pic code
9986 and it is quite long.
9988 @item -munix=@var{unix-std}
9990 Generate compiler predefines and select a startfile for the specified
9991 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9992 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9993 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9994 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9995 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9998 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9999 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10000 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10001 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10002 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10003 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10005 It is @emph{important} to note that this option changes the interfaces
10006 for various library routines. It also affects the operational behavior
10007 of the C library. Thus, @emph{extreme} care is needed in using this
10010 Library code that is intended to operate with more than one UNIX
10011 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10012 as appropriate. Most GNU software doesn't provide this capability.
10016 Suppress the generation of link options to search libdld.sl when the
10017 @option{-static} option is specified on HP-UX 10 and later.
10021 The HP-UX implementation of setlocale in libc has a dependency on
10022 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10023 when the @option{-static} option is specified, special link options
10024 are needed to resolve this dependency.
10026 On HP-UX 10 and later, the GCC driver adds the necessary options to
10027 link with libdld.sl when the @option{-static} option is specified.
10028 This causes the resulting binary to be dynamic. On the 64-bit port,
10029 the linkers generate dynamic binaries by default in any case. The
10030 @option{-nolibdld} option can be used to prevent the GCC driver from
10031 adding these link options.
10035 Add support for multithreading with the @dfn{dce thread} library
10036 under HP-UX@. This option sets flags for both the preprocessor and
10040 @node i386 and x86-64 Options
10041 @subsection Intel 386 and AMD x86-64 Options
10042 @cindex i386 Options
10043 @cindex x86-64 Options
10044 @cindex Intel 386 Options
10045 @cindex AMD x86-64 Options
10047 These @samp{-m} options are defined for the i386 and x86-64 family of
10051 @item -mtune=@var{cpu-type}
10053 Tune to @var{cpu-type} everything applicable about the generated code, except
10054 for the ABI and the set of available instructions. The choices for
10055 @var{cpu-type} are:
10058 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10059 If you know the CPU on which your code will run, then you should use
10060 the corresponding @option{-mtune} option instead of
10061 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10062 of your application will have, then you should use this option.
10064 As new processors are deployed in the marketplace, the behavior of this
10065 option will change. Therefore, if you upgrade to a newer version of
10066 GCC, the code generated option will change to reflect the processors
10067 that were most common when that version of GCC was released.
10069 There is no @option{-march=generic} option because @option{-march}
10070 indicates the instruction set the compiler can use, and there is no
10071 generic instruction set applicable to all processors. In contrast,
10072 @option{-mtune} indicates the processor (or, in this case, collection of
10073 processors) for which the code is optimized.
10075 This selects the CPU to tune for at compilation time by determining
10076 the processor type of the compiling machine. Using @option{-mtune=native}
10077 will produce code optimized for the local machine under the constraints
10078 of the selected instruction set. Using @option{-march=native} will
10079 enable all instruction subsets supported by the local machine (hence
10080 the result might not run on different machines).
10082 Original Intel's i386 CPU@.
10084 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10085 @item i586, pentium
10086 Intel Pentium CPU with no MMX support.
10088 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10090 Intel PentiumPro CPU@.
10092 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10093 instruction set will be used, so the code will run on all i686 family chips.
10095 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10096 @item pentium3, pentium3m
10097 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10100 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10101 support. Used by Centrino notebooks.
10102 @item pentium4, pentium4m
10103 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10105 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10108 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10109 SSE2 and SSE3 instruction set support.
10111 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10112 instruction set support.
10114 AMD K6 CPU with MMX instruction set support.
10116 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
10117 @item athlon, athlon-tbird
10118 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
10120 @item athlon-4, athlon-xp, athlon-mp
10121 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
10122 instruction set support.
10123 @item k8, opteron, athlon64, athlon-fx
10124 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10125 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
10126 @item k8-sse3, opteron-sse3, athlon64-sse3
10127 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10128 @item amdfam10, barcelona
10129 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10130 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10131 instruction set extensions.)
10133 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10136 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
10137 instruction set support.
10139 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
10140 implemented for this chip.)
10142 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10143 implemented for this chip.)
10145 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10148 While picking a specific @var{cpu-type} will schedule things appropriately
10149 for that particular chip, the compiler will not generate any code that
10150 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10153 @item -march=@var{cpu-type}
10155 Generate instructions for the machine type @var{cpu-type}. The choices
10156 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10157 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10159 @item -mcpu=@var{cpu-type}
10161 A deprecated synonym for @option{-mtune}.
10163 @item -mfpmath=@var{unit}
10165 Generate floating point arithmetics for selected unit @var{unit}. The choices
10166 for @var{unit} are:
10170 Use the standard 387 floating point coprocessor present majority of chips and
10171 emulated otherwise. Code compiled with this option will run almost everywhere.
10172 The temporary results are computed in 80bit precision instead of precision
10173 specified by the type resulting in slightly different results compared to most
10174 of other chips. See @option{-ffloat-store} for more detailed description.
10176 This is the default choice for i386 compiler.
10179 Use scalar floating point instructions present in the SSE instruction set.
10180 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10181 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10182 instruction set supports only single precision arithmetics, thus the double and
10183 extended precision arithmetics is still done using 387. Later version, present
10184 only in Pentium4 and the future AMD x86-64 chips supports double precision
10187 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10188 or @option{-msse2} switches to enable SSE extensions and make this option
10189 effective. For the x86-64 compiler, these extensions are enabled by default.
10191 The resulting code should be considerably faster in the majority of cases and avoid
10192 the numerical instability problems of 387 code, but may break some existing
10193 code that expects temporaries to be 80bit.
10195 This is the default choice for the x86-64 compiler.
10198 Attempt to utilize both instruction sets at once. This effectively double the
10199 amount of available registers and on chips with separate execution units for
10200 387 and SSE the execution resources too. Use this option with care, as it is
10201 still experimental, because the GCC register allocator does not model separate
10202 functional units well resulting in instable performance.
10205 @item -masm=@var{dialect}
10206 @opindex masm=@var{dialect}
10207 Output asm instructions using selected @var{dialect}. Supported
10208 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10209 not support @samp{intel}.
10212 @itemx -mno-ieee-fp
10214 @opindex mno-ieee-fp
10215 Control whether or not the compiler uses IEEE floating point
10216 comparisons. These handle correctly the case where the result of a
10217 comparison is unordered.
10220 @opindex msoft-float
10221 Generate output containing library calls for floating point.
10222 @strong{Warning:} the requisite libraries are not part of GCC@.
10223 Normally the facilities of the machine's usual C compiler are used, but
10224 this can't be done directly in cross-compilation. You must make your
10225 own arrangements to provide suitable library functions for
10228 On machines where a function returns floating point results in the 80387
10229 register stack, some floating point opcodes may be emitted even if
10230 @option{-msoft-float} is used.
10232 @item -mno-fp-ret-in-387
10233 @opindex mno-fp-ret-in-387
10234 Do not use the FPU registers for return values of functions.
10236 The usual calling convention has functions return values of types
10237 @code{float} and @code{double} in an FPU register, even if there
10238 is no FPU@. The idea is that the operating system should emulate
10241 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10242 in ordinary CPU registers instead.
10244 @item -mno-fancy-math-387
10245 @opindex mno-fancy-math-387
10246 Some 387 emulators do not support the @code{sin}, @code{cos} and
10247 @code{sqrt} instructions for the 387. Specify this option to avoid
10248 generating those instructions. This option is the default on FreeBSD,
10249 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10250 indicates that the target cpu will always have an FPU and so the
10251 instruction will not need emulation. As of revision 2.6.1, these
10252 instructions are not generated unless you also use the
10253 @option{-funsafe-math-optimizations} switch.
10255 @item -malign-double
10256 @itemx -mno-align-double
10257 @opindex malign-double
10258 @opindex mno-align-double
10259 Control whether GCC aligns @code{double}, @code{long double}, and
10260 @code{long long} variables on a two word boundary or a one word
10261 boundary. Aligning @code{double} variables on a two word boundary will
10262 produce code that runs somewhat faster on a @samp{Pentium} at the
10263 expense of more memory.
10265 On x86-64, @option{-malign-double} is enabled by default.
10267 @strong{Warning:} if you use the @option{-malign-double} switch,
10268 structures containing the above types will be aligned differently than
10269 the published application binary interface specifications for the 386
10270 and will not be binary compatible with structures in code compiled
10271 without that switch.
10273 @item -m96bit-long-double
10274 @itemx -m128bit-long-double
10275 @opindex m96bit-long-double
10276 @opindex m128bit-long-double
10277 These switches control the size of @code{long double} type. The i386
10278 application binary interface specifies the size to be 96 bits,
10279 so @option{-m96bit-long-double} is the default in 32 bit mode.
10281 Modern architectures (Pentium and newer) would prefer @code{long double}
10282 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10283 conforming to the ABI, this would not be possible. So specifying a
10284 @option{-m128bit-long-double} will align @code{long double}
10285 to a 16 byte boundary by padding the @code{long double} with an additional
10288 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10289 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10291 Notice that neither of these options enable any extra precision over the x87
10292 standard of 80 bits for a @code{long double}.
10294 @strong{Warning:} if you override the default value for your target ABI, the
10295 structures and arrays containing @code{long double} variables will change
10296 their size as well as function calling convention for function taking
10297 @code{long double} will be modified. Hence they will not be binary
10298 compatible with arrays or structures in code compiled without that switch.
10300 @item -mmlarge-data-threshold=@var{number}
10301 @opindex mlarge-data-threshold=@var{number}
10302 When @option{-mcmodel=medium} is specified, the data greater than
10303 @var{threshold} are placed in large data section. This value must be the
10304 same across all object linked into the binary and defaults to 65535.
10308 Use a different function-calling convention, in which functions that
10309 take a fixed number of arguments return with the @code{ret} @var{num}
10310 instruction, which pops their arguments while returning. This saves one
10311 instruction in the caller since there is no need to pop the arguments
10314 You can specify that an individual function is called with this calling
10315 sequence with the function attribute @samp{stdcall}. You can also
10316 override the @option{-mrtd} option by using the function attribute
10317 @samp{cdecl}. @xref{Function Attributes}.
10319 @strong{Warning:} this calling convention is incompatible with the one
10320 normally used on Unix, so you cannot use it if you need to call
10321 libraries compiled with the Unix compiler.
10323 Also, you must provide function prototypes for all functions that
10324 take variable numbers of arguments (including @code{printf});
10325 otherwise incorrect code will be generated for calls to those
10328 In addition, seriously incorrect code will result if you call a
10329 function with too many arguments. (Normally, extra arguments are
10330 harmlessly ignored.)
10332 @item -mregparm=@var{num}
10334 Control how many registers are used to pass integer arguments. By
10335 default, no registers are used to pass arguments, and at most 3
10336 registers can be used. You can control this behavior for a specific
10337 function by using the function attribute @samp{regparm}.
10338 @xref{Function Attributes}.
10340 @strong{Warning:} if you use this switch, and
10341 @var{num} is nonzero, then you must build all modules with the same
10342 value, including any libraries. This includes the system libraries and
10346 @opindex msseregparm
10347 Use SSE register passing conventions for float and double arguments
10348 and return values. You can control this behavior for a specific
10349 function by using the function attribute @samp{sseregparm}.
10350 @xref{Function Attributes}.
10352 @strong{Warning:} if you use this switch then you must build all
10353 modules with the same value, including any libraries. This includes
10354 the system libraries and startup modules.
10363 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10364 is specified, the significands of results of floating-point operations are
10365 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10366 significands of results of floating-point operations to 53 bits (double
10367 precision) and @option{-mpc80} rounds the significands of results of
10368 floating-point operations to 64 bits (extended double precision), which is
10369 the default. When this option is used, floating-point operations in higher
10370 precisions are not available to the programmer without setting the FPU
10371 control word explicitly.
10373 Setting the rounding of floating-point operations to less than the default
10374 80 bits can speed some programs by 2% or more. Note that some mathematical
10375 libraries assume that extended precision (80 bit) floating-point operations
10376 are enabled by default; routines in such libraries could suffer significant
10377 loss of accuracy, typically through so-called "catastrophic cancellation",
10378 when this option is used to set the precision to less than extended precision.
10380 @item -mstackrealign
10381 @opindex mstackrealign
10382 Realign the stack at entry. On the Intel x86, the
10383 @option{-mstackrealign} option will generate an alternate prologue and
10384 epilogue that realigns the runtime stack. This supports mixing legacy
10385 codes that keep a 4-byte aligned stack with modern codes that keep a
10386 16-byte stack for SSE compatibility. The alternate prologue and
10387 epilogue are slower and bigger than the regular ones, and the
10388 alternate prologue requires an extra scratch register; this lowers the
10389 number of registers available if used in conjunction with the
10390 @code{regparm} attribute. The @option{-mstackrealign} option is
10391 incompatible with the nested function prologue; this is considered a
10392 hard error. See also the attribute @code{force_align_arg_pointer},
10393 applicable to individual functions.
10395 @item -mpreferred-stack-boundary=@var{num}
10396 @opindex mpreferred-stack-boundary
10397 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10398 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10399 the default is 4 (16 bytes or 128 bits).
10401 On Pentium and PentiumPro, @code{double} and @code{long double} values
10402 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10403 suffer significant run time performance penalties. On Pentium III, the
10404 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10405 properly if it is not 16 byte aligned.
10407 To ensure proper alignment of this values on the stack, the stack boundary
10408 must be as aligned as that required by any value stored on the stack.
10409 Further, every function must be generated such that it keeps the stack
10410 aligned. Thus calling a function compiled with a higher preferred
10411 stack boundary from a function compiled with a lower preferred stack
10412 boundary will most likely misalign the stack. It is recommended that
10413 libraries that use callbacks always use the default setting.
10415 This extra alignment does consume extra stack space, and generally
10416 increases code size. Code that is sensitive to stack space usage, such
10417 as embedded systems and operating system kernels, may want to reduce the
10418 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10450 These switches enable or disable the use of instructions in the MMX,
10451 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, ABM or 3DNow! extended
10453 These extensions are also available as built-in functions: see
10454 @ref{X86 Built-in Functions}, for details of the functions enabled and
10455 disabled by these switches.
10457 To have SSE/SSE2 instructions generated automatically from floating-point
10458 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10460 These options will enable GCC to use these extended instructions in
10461 generated code, even without @option{-mfpmath=sse}. Applications which
10462 perform runtime CPU detection must compile separate files for each
10463 supported architecture, using the appropriate flags. In particular,
10464 the file containing the CPU detection code should be compiled without
10469 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10470 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10471 data types. This is useful for high resolution counters that could be updated
10472 by multiple processors (or cores). This instruction is generated as part of
10473 atomic built-in functions: see @ref{Atomic Builtins} for details.
10477 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10478 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10479 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10480 SAHF are load and store instructions, respectively, for certain status flags.
10481 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10482 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10486 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10487 vectorized variants RCPPS and RSQRTPS) instead of DIVSS and SQRTSS (and their
10488 vectorized variants). These instructions will be generated only when
10489 @option{-funsafe-math-optimizations} is enabled.
10491 @item -mveclibabi=@var{type}
10492 @opindex mveclibabi
10493 Specifies the ABI type to use for vectorizing intrinsics using an
10494 external library. Supported types are @code{acml} for the AMD
10495 math core library style of interfacing. GCC will currently emit
10496 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10497 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10498 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10499 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10500 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10501 is enabled. A ACML ABI compatible library will have to be specified
10505 @itemx -mno-push-args
10506 @opindex mpush-args
10507 @opindex mno-push-args
10508 Use PUSH operations to store outgoing parameters. This method is shorter
10509 and usually equally fast as method using SUB/MOV operations and is enabled
10510 by default. In some cases disabling it may improve performance because of
10511 improved scheduling and reduced dependencies.
10513 @item -maccumulate-outgoing-args
10514 @opindex maccumulate-outgoing-args
10515 If enabled, the maximum amount of space required for outgoing arguments will be
10516 computed in the function prologue. This is faster on most modern CPUs
10517 because of reduced dependencies, improved scheduling and reduced stack usage
10518 when preferred stack boundary is not equal to 2. The drawback is a notable
10519 increase in code size. This switch implies @option{-mno-push-args}.
10523 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10524 on thread-safe exception handling must compile and link all code with the
10525 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10526 @option{-D_MT}; when linking, it links in a special thread helper library
10527 @option{-lmingwthrd} which cleans up per thread exception handling data.
10529 @item -mno-align-stringops
10530 @opindex mno-align-stringops
10531 Do not align destination of inlined string operations. This switch reduces
10532 code size and improves performance in case the destination is already aligned,
10533 but GCC doesn't know about it.
10535 @item -minline-all-stringops
10536 @opindex minline-all-stringops
10537 By default GCC inlines string operations only when destination is known to be
10538 aligned at least to 4 byte boundary. This enables more inlining, increase code
10539 size, but may improve performance of code that depends on fast memcpy, strlen
10540 and memset for short lengths.
10542 @item -minline-stringops-dynamically
10543 @opindex minline-stringops-dynamically
10544 For string operation of unknown size, inline runtime checks so for small
10545 blocks inline code is used, while for large blocks library call is used.
10547 @item -mstringop-strategy=@var{alg}
10548 @opindex mstringop-strategy=@var{alg}
10549 Overwrite internal decision heuristic about particular algorithm to inline
10550 string operation with. The allowed values are @code{rep_byte},
10551 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10552 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10553 expanding inline loop, @code{libcall} for always expanding library call.
10555 @item -momit-leaf-frame-pointer
10556 @opindex momit-leaf-frame-pointer
10557 Don't keep the frame pointer in a register for leaf functions. This
10558 avoids the instructions to save, set up and restore frame pointers and
10559 makes an extra register available in leaf functions. The option
10560 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10561 which might make debugging harder.
10563 @item -mtls-direct-seg-refs
10564 @itemx -mno-tls-direct-seg-refs
10565 @opindex mtls-direct-seg-refs
10566 Controls whether TLS variables may be accessed with offsets from the
10567 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10568 or whether the thread base pointer must be added. Whether or not this
10569 is legal depends on the operating system, and whether it maps the
10570 segment to cover the entire TLS area.
10572 For systems that use GNU libc, the default is on.
10575 These @samp{-m} switches are supported in addition to the above
10576 on AMD x86-64 processors in 64-bit environments.
10583 Generate code for a 32-bit or 64-bit environment.
10584 The 32-bit environment sets int, long and pointer to 32 bits and
10585 generates code that runs on any i386 system.
10586 The 64-bit environment sets int to 32 bits and long and pointer
10587 to 64 bits and generates code for AMD's x86-64 architecture. For
10588 darwin only the -m64 option turns off the @option{-fno-pic} and
10589 @option{-mdynamic-no-pic} options.
10591 @item -mno-red-zone
10592 @opindex no-red-zone
10593 Do not use a so called red zone for x86-64 code. The red zone is mandated
10594 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10595 stack pointer that will not be modified by signal or interrupt handlers
10596 and therefore can be used for temporary data without adjusting the stack
10597 pointer. The flag @option{-mno-red-zone} disables this red zone.
10599 @item -mcmodel=small
10600 @opindex mcmodel=small
10601 Generate code for the small code model: the program and its symbols must
10602 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10603 Programs can be statically or dynamically linked. This is the default
10606 @item -mcmodel=kernel
10607 @opindex mcmodel=kernel
10608 Generate code for the kernel code model. The kernel runs in the
10609 negative 2 GB of the address space.
10610 This model has to be used for Linux kernel code.
10612 @item -mcmodel=medium
10613 @opindex mcmodel=medium
10614 Generate code for the medium model: The program is linked in the lower 2
10615 GB of the address space but symbols can be located anywhere in the
10616 address space. Programs can be statically or dynamically linked, but
10617 building of shared libraries are not supported with the medium model.
10619 @item -mcmodel=large
10620 @opindex mcmodel=large
10621 Generate code for the large model: This model makes no assumptions
10622 about addresses and sizes of sections.
10625 @node IA-64 Options
10626 @subsection IA-64 Options
10627 @cindex IA-64 Options
10629 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10633 @opindex mbig-endian
10634 Generate code for a big endian target. This is the default for HP-UX@.
10636 @item -mlittle-endian
10637 @opindex mlittle-endian
10638 Generate code for a little endian target. This is the default for AIX5
10644 @opindex mno-gnu-as
10645 Generate (or don't) code for the GNU assembler. This is the default.
10646 @c Also, this is the default if the configure option @option{--with-gnu-as}
10652 @opindex mno-gnu-ld
10653 Generate (or don't) code for the GNU linker. This is the default.
10654 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10659 Generate code that does not use a global pointer register. The result
10660 is not position independent code, and violates the IA-64 ABI@.
10662 @item -mvolatile-asm-stop
10663 @itemx -mno-volatile-asm-stop
10664 @opindex mvolatile-asm-stop
10665 @opindex mno-volatile-asm-stop
10666 Generate (or don't) a stop bit immediately before and after volatile asm
10669 @item -mregister-names
10670 @itemx -mno-register-names
10671 @opindex mregister-names
10672 @opindex mno-register-names
10673 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10674 the stacked registers. This may make assembler output more readable.
10680 Disable (or enable) optimizations that use the small data section. This may
10681 be useful for working around optimizer bugs.
10683 @item -mconstant-gp
10684 @opindex mconstant-gp
10685 Generate code that uses a single constant global pointer value. This is
10686 useful when compiling kernel code.
10690 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10691 This is useful when compiling firmware code.
10693 @item -minline-float-divide-min-latency
10694 @opindex minline-float-divide-min-latency
10695 Generate code for inline divides of floating point values
10696 using the minimum latency algorithm.
10698 @item -minline-float-divide-max-throughput
10699 @opindex minline-float-divide-max-throughput
10700 Generate code for inline divides of floating point values
10701 using the maximum throughput algorithm.
10703 @item -minline-int-divide-min-latency
10704 @opindex minline-int-divide-min-latency
10705 Generate code for inline divides of integer values
10706 using the minimum latency algorithm.
10708 @item -minline-int-divide-max-throughput
10709 @opindex minline-int-divide-max-throughput
10710 Generate code for inline divides of integer values
10711 using the maximum throughput algorithm.
10713 @item -minline-sqrt-min-latency
10714 @opindex minline-sqrt-min-latency
10715 Generate code for inline square roots
10716 using the minimum latency algorithm.
10718 @item -minline-sqrt-max-throughput
10719 @opindex minline-sqrt-max-throughput
10720 Generate code for inline square roots
10721 using the maximum throughput algorithm.
10723 @item -mno-dwarf2-asm
10724 @itemx -mdwarf2-asm
10725 @opindex mno-dwarf2-asm
10726 @opindex mdwarf2-asm
10727 Don't (or do) generate assembler code for the DWARF2 line number debugging
10728 info. This may be useful when not using the GNU assembler.
10730 @item -mearly-stop-bits
10731 @itemx -mno-early-stop-bits
10732 @opindex mearly-stop-bits
10733 @opindex mno-early-stop-bits
10734 Allow stop bits to be placed earlier than immediately preceding the
10735 instruction that triggered the stop bit. This can improve instruction
10736 scheduling, but does not always do so.
10738 @item -mfixed-range=@var{register-range}
10739 @opindex mfixed-range
10740 Generate code treating the given register range as fixed registers.
10741 A fixed register is one that the register allocator can not use. This is
10742 useful when compiling kernel code. A register range is specified as
10743 two registers separated by a dash. Multiple register ranges can be
10744 specified separated by a comma.
10746 @item -mtls-size=@var{tls-size}
10748 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10751 @item -mtune=@var{cpu-type}
10753 Tune the instruction scheduling for a particular CPU, Valid values are
10754 itanium, itanium1, merced, itanium2, and mckinley.
10760 Add support for multithreading using the POSIX threads library. This
10761 option sets flags for both the preprocessor and linker. It does
10762 not affect the thread safety of object code produced by the compiler or
10763 that of libraries supplied with it. These are HP-UX specific flags.
10769 Generate code for a 32-bit or 64-bit environment.
10770 The 32-bit environment sets int, long and pointer to 32 bits.
10771 The 64-bit environment sets int to 32 bits and long and pointer
10772 to 64 bits. These are HP-UX specific flags.
10774 @item -mno-sched-br-data-spec
10775 @itemx -msched-br-data-spec
10776 @opindex mno-sched-br-data-spec
10777 @opindex msched-br-data-spec
10778 (Dis/En)able data speculative scheduling before reload.
10779 This will result in generation of the ld.a instructions and
10780 the corresponding check instructions (ld.c / chk.a).
10781 The default is 'disable'.
10783 @item -msched-ar-data-spec
10784 @itemx -mno-sched-ar-data-spec
10785 @opindex msched-ar-data-spec
10786 @opindex mno-sched-ar-data-spec
10787 (En/Dis)able data speculative scheduling after reload.
10788 This will result in generation of the ld.a instructions and
10789 the corresponding check instructions (ld.c / chk.a).
10790 The default is 'enable'.
10792 @item -mno-sched-control-spec
10793 @itemx -msched-control-spec
10794 @opindex mno-sched-control-spec
10795 @opindex msched-control-spec
10796 (Dis/En)able control speculative scheduling. This feature is
10797 available only during region scheduling (i.e. before reload).
10798 This will result in generation of the ld.s instructions and
10799 the corresponding check instructions chk.s .
10800 The default is 'disable'.
10802 @item -msched-br-in-data-spec
10803 @itemx -mno-sched-br-in-data-spec
10804 @opindex msched-br-in-data-spec
10805 @opindex mno-sched-br-in-data-spec
10806 (En/Dis)able speculative scheduling of the instructions that
10807 are dependent on the data speculative loads before reload.
10808 This is effective only with @option{-msched-br-data-spec} enabled.
10809 The default is 'enable'.
10811 @item -msched-ar-in-data-spec
10812 @itemx -mno-sched-ar-in-data-spec
10813 @opindex msched-ar-in-data-spec
10814 @opindex mno-sched-ar-in-data-spec
10815 (En/Dis)able speculative scheduling of the instructions that
10816 are dependent on the data speculative loads after reload.
10817 This is effective only with @option{-msched-ar-data-spec} enabled.
10818 The default is 'enable'.
10820 @item -msched-in-control-spec
10821 @itemx -mno-sched-in-control-spec
10822 @opindex msched-in-control-spec
10823 @opindex mno-sched-in-control-spec
10824 (En/Dis)able speculative scheduling of the instructions that
10825 are dependent on the control speculative loads.
10826 This is effective only with @option{-msched-control-spec} enabled.
10827 The default is 'enable'.
10830 @itemx -mno-sched-ldc
10831 @opindex msched-ldc
10832 @opindex mno-sched-ldc
10833 (En/Dis)able use of simple data speculation checks ld.c .
10834 If disabled, only chk.a instructions will be emitted to check
10835 data speculative loads.
10836 The default is 'enable'.
10838 @item -mno-sched-control-ldc
10839 @itemx -msched-control-ldc
10840 @opindex mno-sched-control-ldc
10841 @opindex msched-control-ldc
10842 (Dis/En)able use of ld.c instructions to check control speculative loads.
10843 If enabled, in case of control speculative load with no speculatively
10844 scheduled dependent instructions this load will be emitted as ld.sa and
10845 ld.c will be used to check it.
10846 The default is 'disable'.
10848 @item -mno-sched-spec-verbose
10849 @itemx -msched-spec-verbose
10850 @opindex mno-sched-spec-verbose
10851 @opindex msched-spec-verbose
10852 (Dis/En)able printing of the information about speculative motions.
10854 @item -mno-sched-prefer-non-data-spec-insns
10855 @itemx -msched-prefer-non-data-spec-insns
10856 @opindex mno-sched-prefer-non-data-spec-insns
10857 @opindex msched-prefer-non-data-spec-insns
10858 If enabled, data speculative instructions will be chosen for schedule
10859 only if there are no other choices at the moment. This will make
10860 the use of the data speculation much more conservative.
10861 The default is 'disable'.
10863 @item -mno-sched-prefer-non-control-spec-insns
10864 @itemx -msched-prefer-non-control-spec-insns
10865 @opindex mno-sched-prefer-non-control-spec-insns
10866 @opindex msched-prefer-non-control-spec-insns
10867 If enabled, control speculative instructions will be chosen for schedule
10868 only if there are no other choices at the moment. This will make
10869 the use of the control speculation much more conservative.
10870 The default is 'disable'.
10872 @item -mno-sched-count-spec-in-critical-path
10873 @itemx -msched-count-spec-in-critical-path
10874 @opindex mno-sched-count-spec-in-critical-path
10875 @opindex msched-count-spec-in-critical-path
10876 If enabled, speculative dependencies will be considered during
10877 computation of the instructions priorities. This will make the use of the
10878 speculation a bit more conservative.
10879 The default is 'disable'.
10884 @subsection M32C Options
10885 @cindex M32C options
10888 @item -mcpu=@var{name}
10890 Select the CPU for which code is generated. @var{name} may be one of
10891 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10892 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10893 the M32C/80 series.
10897 Specifies that the program will be run on the simulator. This causes
10898 an alternate runtime library to be linked in which supports, for
10899 example, file I/O. You must not use this option when generating
10900 programs that will run on real hardware; you must provide your own
10901 runtime library for whatever I/O functions are needed.
10903 @item -memregs=@var{number}
10905 Specifies the number of memory-based pseudo-registers GCC will use
10906 during code generation. These pseudo-registers will be used like real
10907 registers, so there is a tradeoff between GCC's ability to fit the
10908 code into available registers, and the performance penalty of using
10909 memory instead of registers. Note that all modules in a program must
10910 be compiled with the same value for this option. Because of that, you
10911 must not use this option with the default runtime libraries gcc
10916 @node M32R/D Options
10917 @subsection M32R/D Options
10918 @cindex M32R/D options
10920 These @option{-m} options are defined for Renesas M32R/D architectures:
10925 Generate code for the M32R/2@.
10929 Generate code for the M32R/X@.
10933 Generate code for the M32R@. This is the default.
10935 @item -mmodel=small
10936 @opindex mmodel=small
10937 Assume all objects live in the lower 16MB of memory (so that their addresses
10938 can be loaded with the @code{ld24} instruction), and assume all subroutines
10939 are reachable with the @code{bl} instruction.
10940 This is the default.
10942 The addressability of a particular object can be set with the
10943 @code{model} attribute.
10945 @item -mmodel=medium
10946 @opindex mmodel=medium
10947 Assume objects may be anywhere in the 32-bit address space (the compiler
10948 will generate @code{seth/add3} instructions to load their addresses), and
10949 assume all subroutines are reachable with the @code{bl} instruction.
10951 @item -mmodel=large
10952 @opindex mmodel=large
10953 Assume objects may be anywhere in the 32-bit address space (the compiler
10954 will generate @code{seth/add3} instructions to load their addresses), and
10955 assume subroutines may not be reachable with the @code{bl} instruction
10956 (the compiler will generate the much slower @code{seth/add3/jl}
10957 instruction sequence).
10960 @opindex msdata=none
10961 Disable use of the small data area. Variables will be put into
10962 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10963 @code{section} attribute has been specified).
10964 This is the default.
10966 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10967 Objects may be explicitly put in the small data area with the
10968 @code{section} attribute using one of these sections.
10970 @item -msdata=sdata
10971 @opindex msdata=sdata
10972 Put small global and static data in the small data area, but do not
10973 generate special code to reference them.
10976 @opindex msdata=use
10977 Put small global and static data in the small data area, and generate
10978 special instructions to reference them.
10982 @cindex smaller data references
10983 Put global and static objects less than or equal to @var{num} bytes
10984 into the small data or bss sections instead of the normal data or bss
10985 sections. The default value of @var{num} is 8.
10986 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10987 for this option to have any effect.
10989 All modules should be compiled with the same @option{-G @var{num}} value.
10990 Compiling with different values of @var{num} may or may not work; if it
10991 doesn't the linker will give an error message---incorrect code will not be
10996 Makes the M32R specific code in the compiler display some statistics
10997 that might help in debugging programs.
10999 @item -malign-loops
11000 @opindex malign-loops
11001 Align all loops to a 32-byte boundary.
11003 @item -mno-align-loops
11004 @opindex mno-align-loops
11005 Do not enforce a 32-byte alignment for loops. This is the default.
11007 @item -missue-rate=@var{number}
11008 @opindex missue-rate=@var{number}
11009 Issue @var{number} instructions per cycle. @var{number} can only be 1
11012 @item -mbranch-cost=@var{number}
11013 @opindex mbranch-cost=@var{number}
11014 @var{number} can only be 1 or 2. If it is 1 then branches will be
11015 preferred over conditional code, if it is 2, then the opposite will
11018 @item -mflush-trap=@var{number}
11019 @opindex mflush-trap=@var{number}
11020 Specifies the trap number to use to flush the cache. The default is
11021 12. Valid numbers are between 0 and 15 inclusive.
11023 @item -mno-flush-trap
11024 @opindex mno-flush-trap
11025 Specifies that the cache cannot be flushed by using a trap.
11027 @item -mflush-func=@var{name}
11028 @opindex mflush-func=@var{name}
11029 Specifies the name of the operating system function to call to flush
11030 the cache. The default is @emph{_flush_cache}, but a function call
11031 will only be used if a trap is not available.
11033 @item -mno-flush-func
11034 @opindex mno-flush-func
11035 Indicates that there is no OS function for flushing the cache.
11039 @node M680x0 Options
11040 @subsection M680x0 Options
11041 @cindex M680x0 options
11043 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11044 The default settings depend on which architecture was selected when
11045 the compiler was configured; the defaults for the most common choices
11049 @item -march=@var{arch}
11051 Generate code for a specific M680x0 or ColdFire instruction set
11052 architecture. Permissible values of @var{arch} for M680x0
11053 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11054 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11055 architectures are selected according to Freescale's ISA classification
11056 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11057 @samp{isab} and @samp{isac}.
11059 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11060 code for a ColdFire target. The @var{arch} in this macro is one of the
11061 @option{-march} arguments given above.
11063 When used together, @option{-march} and @option{-mtune} select code
11064 that runs on a family of similar processors but that is optimized
11065 for a particular microarchitecture.
11067 @item -mcpu=@var{cpu}
11069 Generate code for a specific M680x0 or ColdFire processor.
11070 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11071 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11072 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11073 below, which also classifies the CPUs into families:
11075 @multitable @columnfractions 0.20 0.80
11076 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11077 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11078 @item @samp{5206e} @tab @samp{5206e}
11079 @item @samp{5208} @tab @samp{5207} @samp{5208}
11080 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11081 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11082 @item @samp{5216} @tab @samp{5214} @samp{5216}
11083 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11084 @item @samp{5225} @tab @samp{5224} @samp{5225}
11085 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11086 @item @samp{5249} @tab @samp{5249}
11087 @item @samp{5250} @tab @samp{5250}
11088 @item @samp{5271} @tab @samp{5270} @samp{5271}
11089 @item @samp{5272} @tab @samp{5272}
11090 @item @samp{5275} @tab @samp{5274} @samp{5275}
11091 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11092 @item @samp{5307} @tab @samp{5307}
11093 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11094 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11095 @item @samp{5407} @tab @samp{5407}
11096 @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}
11099 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11100 @var{arch} is compatible with @var{cpu}. Other combinations of
11101 @option{-mcpu} and @option{-march} are rejected.
11103 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11104 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11105 where the value of @var{family} is given by the table above.
11107 @item -mtune=@var{tune}
11109 Tune the code for a particular microarchitecture, within the
11110 constraints set by @option{-march} and @option{-mcpu}.
11111 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11112 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11113 and @samp{cpu32}. The ColdFire microarchitectures
11114 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11116 You can also use @option{-mtune=68020-40} for code that needs
11117 to run relatively well on 68020, 68030 and 68040 targets.
11118 @option{-mtune=68020-60} is similar but includes 68060 targets
11119 as well. These two options select the same tuning decisions as
11120 @option{-m68020-40} and @option{-m68020-60} respectively.
11122 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11123 when tuning for 680x0 architecture @var{arch}. It also defines
11124 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11125 option is used. If gcc is tuning for a range of architectures,
11126 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11127 it defines the macros for every architecture in the range.
11129 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11130 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11131 of the arguments given above.
11137 Generate output for a 68000. This is the default
11138 when the compiler is configured for 68000-based systems.
11139 It is equivalent to @option{-march=68000}.
11141 Use this option for microcontrollers with a 68000 or EC000 core,
11142 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11146 Generate output for a 68010. This is the default
11147 when the compiler is configured for 68010-based systems.
11148 It is equivalent to @option{-march=68010}.
11154 Generate output for a 68020. This is the default
11155 when the compiler is configured for 68020-based systems.
11156 It is equivalent to @option{-march=68020}.
11160 Generate output for a 68030. This is the default when the compiler is
11161 configured for 68030-based systems. It is equivalent to
11162 @option{-march=68030}.
11166 Generate output for a 68040. This is the default when the compiler is
11167 configured for 68040-based systems. It is equivalent to
11168 @option{-march=68040}.
11170 This option inhibits the use of 68881/68882 instructions that have to be
11171 emulated by software on the 68040. Use this option if your 68040 does not
11172 have code to emulate those instructions.
11176 Generate output for a 68060. This is the default when the compiler is
11177 configured for 68060-based systems. It is equivalent to
11178 @option{-march=68060}.
11180 This option inhibits the use of 68020 and 68881/68882 instructions that
11181 have to be emulated by software on the 68060. Use this option if your 68060
11182 does not have code to emulate those instructions.
11186 Generate output for a CPU32. This is the default
11187 when the compiler is configured for CPU32-based systems.
11188 It is equivalent to @option{-march=cpu32}.
11190 Use this option for microcontrollers with a
11191 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11192 68336, 68340, 68341, 68349 and 68360.
11196 Generate output for a 520X ColdFire CPU. This is the default
11197 when the compiler is configured for 520X-based systems.
11198 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11199 in favor of that option.
11201 Use this option for microcontroller with a 5200 core, including
11202 the MCF5202, MCF5203, MCF5204 and MCF5206.
11206 Generate output for a 5206e ColdFire CPU. The option is now
11207 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11211 Generate output for a member of the ColdFire 528X family.
11212 The option is now deprecated in favor of the equivalent
11213 @option{-mcpu=528x}.
11217 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11218 in favor of the equivalent @option{-mcpu=5307}.
11222 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11223 in favor of the equivalent @option{-mcpu=5407}.
11227 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11228 This includes use of hardware floating point instructions.
11229 The option is equivalent to @option{-mcpu=547x}, and is now
11230 deprecated in favor of that option.
11234 Generate output for a 68040, without using any of the new instructions.
11235 This results in code which can run relatively efficiently on either a
11236 68020/68881 or a 68030 or a 68040. The generated code does use the
11237 68881 instructions that are emulated on the 68040.
11239 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11243 Generate output for a 68060, without using any of the new instructions.
11244 This results in code which can run relatively efficiently on either a
11245 68020/68881 or a 68030 or a 68040. The generated code does use the
11246 68881 instructions that are emulated on the 68060.
11248 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11252 @opindex mhard-float
11254 Generate floating-point instructions. This is the default for 68020
11255 and above, and for ColdFire devices that have an FPU. It defines the
11256 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11257 on ColdFire targets.
11260 @opindex msoft-float
11261 Do not generate floating-point instructions; use library calls instead.
11262 This is the default for 68000, 68010, and 68832 targets. It is also
11263 the default for ColdFire devices that have no FPU.
11269 Generate (do not generate) ColdFire hardware divide and remainder
11270 instructions. If @option{-march} is used without @option{-mcpu},
11271 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11272 architectures. Otherwise, the default is taken from the target CPU
11273 (either the default CPU, or the one specified by @option{-mcpu}). For
11274 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11275 @option{-mcpu=5206e}.
11277 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11281 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11282 Additionally, parameters passed on the stack are also aligned to a
11283 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11287 Do not consider type @code{int} to be 16 bits wide. This is the default.
11290 @itemx -mno-bitfield
11291 @opindex mnobitfield
11292 @opindex mno-bitfield
11293 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11294 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11298 Do use the bit-field instructions. The @option{-m68020} option implies
11299 @option{-mbitfield}. This is the default if you use a configuration
11300 designed for a 68020.
11304 Use a different function-calling convention, in which functions
11305 that take a fixed number of arguments return with the @code{rtd}
11306 instruction, which pops their arguments while returning. This
11307 saves one instruction in the caller since there is no need to pop
11308 the arguments there.
11310 This calling convention is incompatible with the one normally
11311 used on Unix, so you cannot use it if you need to call libraries
11312 compiled with the Unix compiler.
11314 Also, you must provide function prototypes for all functions that
11315 take variable numbers of arguments (including @code{printf});
11316 otherwise incorrect code will be generated for calls to those
11319 In addition, seriously incorrect code will result if you call a
11320 function with too many arguments. (Normally, extra arguments are
11321 harmlessly ignored.)
11323 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11324 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11328 Do not use the calling conventions selected by @option{-mrtd}.
11329 This is the default.
11332 @itemx -mno-align-int
11333 @opindex malign-int
11334 @opindex mno-align-int
11335 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11336 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11337 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11338 Aligning variables on 32-bit boundaries produces code that runs somewhat
11339 faster on processors with 32-bit busses at the expense of more memory.
11341 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11342 align structures containing the above types differently than
11343 most published application binary interface specifications for the m68k.
11347 Use the pc-relative addressing mode of the 68000 directly, instead of
11348 using a global offset table. At present, this option implies @option{-fpic},
11349 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11350 not presently supported with @option{-mpcrel}, though this could be supported for
11351 68020 and higher processors.
11353 @item -mno-strict-align
11354 @itemx -mstrict-align
11355 @opindex mno-strict-align
11356 @opindex mstrict-align
11357 Do not (do) assume that unaligned memory references will be handled by
11361 Generate code that allows the data segment to be located in a different
11362 area of memory from the text segment. This allows for execute in place in
11363 an environment without virtual memory management. This option implies
11366 @item -mno-sep-data
11367 Generate code that assumes that the data segment follows the text segment.
11368 This is the default.
11370 @item -mid-shared-library
11371 Generate code that supports shared libraries via the library ID method.
11372 This allows for execute in place and shared libraries in an environment
11373 without virtual memory management. This option implies @option{-fPIC}.
11375 @item -mno-id-shared-library
11376 Generate code that doesn't assume ID based shared libraries are being used.
11377 This is the default.
11379 @item -mshared-library-id=n
11380 Specified the identification number of the ID based shared library being
11381 compiled. Specifying a value of 0 will generate more compact code, specifying
11382 other values will force the allocation of that number to the current
11383 library but is no more space or time efficient than omitting this option.
11387 @node M68hc1x Options
11388 @subsection M68hc1x Options
11389 @cindex M68hc1x options
11391 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11392 microcontrollers. The default values for these options depends on
11393 which style of microcontroller was selected when the compiler was configured;
11394 the defaults for the most common choices are given below.
11401 Generate output for a 68HC11. This is the default
11402 when the compiler is configured for 68HC11-based systems.
11408 Generate output for a 68HC12. This is the default
11409 when the compiler is configured for 68HC12-based systems.
11415 Generate output for a 68HCS12.
11417 @item -mauto-incdec
11418 @opindex mauto-incdec
11419 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11426 Enable the use of 68HC12 min and max instructions.
11429 @itemx -mno-long-calls
11430 @opindex mlong-calls
11431 @opindex mno-long-calls
11432 Treat all calls as being far away (near). If calls are assumed to be
11433 far away, the compiler will use the @code{call} instruction to
11434 call a function and the @code{rtc} instruction for returning.
11438 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11440 @item -msoft-reg-count=@var{count}
11441 @opindex msoft-reg-count
11442 Specify the number of pseudo-soft registers which are used for the
11443 code generation. The maximum number is 32. Using more pseudo-soft
11444 register may or may not result in better code depending on the program.
11445 The default is 4 for 68HC11 and 2 for 68HC12.
11449 @node MCore Options
11450 @subsection MCore Options
11451 @cindex MCore options
11453 These are the @samp{-m} options defined for the Motorola M*Core
11459 @itemx -mno-hardlit
11461 @opindex mno-hardlit
11462 Inline constants into the code stream if it can be done in two
11463 instructions or less.
11469 Use the divide instruction. (Enabled by default).
11471 @item -mrelax-immediate
11472 @itemx -mno-relax-immediate
11473 @opindex mrelax-immediate
11474 @opindex mno-relax-immediate
11475 Allow arbitrary sized immediates in bit operations.
11477 @item -mwide-bitfields
11478 @itemx -mno-wide-bitfields
11479 @opindex mwide-bitfields
11480 @opindex mno-wide-bitfields
11481 Always treat bit-fields as int-sized.
11483 @item -m4byte-functions
11484 @itemx -mno-4byte-functions
11485 @opindex m4byte-functions
11486 @opindex mno-4byte-functions
11487 Force all functions to be aligned to a four byte boundary.
11489 @item -mcallgraph-data
11490 @itemx -mno-callgraph-data
11491 @opindex mcallgraph-data
11492 @opindex mno-callgraph-data
11493 Emit callgraph information.
11496 @itemx -mno-slow-bytes
11497 @opindex mslow-bytes
11498 @opindex mno-slow-bytes
11499 Prefer word access when reading byte quantities.
11501 @item -mlittle-endian
11502 @itemx -mbig-endian
11503 @opindex mlittle-endian
11504 @opindex mbig-endian
11505 Generate code for a little endian target.
11511 Generate code for the 210 processor.
11515 @subsection MIPS Options
11516 @cindex MIPS options
11522 Generate big-endian code.
11526 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11529 @item -march=@var{arch}
11531 Generate code that will run on @var{arch}, which can be the name of a
11532 generic MIPS ISA, or the name of a particular processor.
11534 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11535 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11536 The processor names are:
11537 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11538 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11539 @samp{5kc}, @samp{5kf},
11541 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11542 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11543 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11544 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11547 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11548 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11549 @samp{rm7000}, @samp{rm9000},
11552 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11553 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11554 The special value @samp{from-abi} selects the
11555 most compatible architecture for the selected ABI (that is,
11556 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11558 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11559 (for example, @samp{-march=r2k}). Prefixes are optional, and
11560 @samp{vr} may be written @samp{r}.
11562 Names of the form @samp{@var{n}f2_1} refer to processors with
11563 FPUs clocked at half the rate of the core, names of the form
11564 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11565 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11566 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11567 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11568 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11569 accepted as synonyms for @samp{@var{n}f1_1}.
11571 GCC defines two macros based on the value of this option. The first
11572 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11573 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11574 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11575 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11576 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11578 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11579 above. In other words, it will have the full prefix and will not
11580 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11581 the macro names the resolved architecture (either @samp{"mips1"} or
11582 @samp{"mips3"}). It names the default architecture when no
11583 @option{-march} option is given.
11585 @item -mtune=@var{arch}
11587 Optimize for @var{arch}. Among other things, this option controls
11588 the way instructions are scheduled, and the perceived cost of arithmetic
11589 operations. The list of @var{arch} values is the same as for
11592 When this option is not used, GCC will optimize for the processor
11593 specified by @option{-march}. By using @option{-march} and
11594 @option{-mtune} together, it is possible to generate code that will
11595 run on a family of processors, but optimize the code for one
11596 particular member of that family.
11598 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11599 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11600 @samp{-march} ones described above.
11604 Equivalent to @samp{-march=mips1}.
11608 Equivalent to @samp{-march=mips2}.
11612 Equivalent to @samp{-march=mips3}.
11616 Equivalent to @samp{-march=mips4}.
11620 Equivalent to @samp{-march=mips32}.
11624 Equivalent to @samp{-march=mips32r2}.
11628 Equivalent to @samp{-march=mips64}.
11633 @opindex mno-mips16
11634 Generate (do not generate) MIPS16 code. If GCC is targetting a
11635 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11637 MIPS16 code generation can also be controlled on a per-function basis
11638 by means of @code{mips16} and @code{nomips16} attributes.
11639 @xref{Function Attributes}, for more information.
11641 @item -mflip-mips16
11642 @opindex mflip-mips16
11643 Generate MIPS16 code on alternating functions. This option is provided
11644 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11645 not intended for ordinary use in compiling user code.
11657 Generate code for the given ABI@.
11659 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11660 generates 64-bit code when you select a 64-bit architecture, but you
11661 can use @option{-mgp32} to get 32-bit code instead.
11663 For information about the O64 ABI, see
11664 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11666 GCC supports a variant of the o32 ABI in which floating-point registers
11667 are 64 rather than 32 bits wide. You can select this combination with
11668 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11669 and @samp{mfhc1} instructions and is therefore only supported for
11670 MIPS32R2 processors.
11672 The register assignments for arguments and return values remain the
11673 same, but each scalar value is passed in a single 64-bit register
11674 rather than a pair of 32-bit registers. For example, scalar
11675 floating-point values are returned in @samp{$f0} only, not a
11676 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11677 remains the same, but all 64 bits are saved.
11680 @itemx -mno-abicalls
11682 @opindex mno-abicalls
11683 Generate (do not generate) code that is suitable for SVR4-style
11684 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11689 Generate (do not generate) code that is fully position-independent,
11690 and that can therefore be linked into shared libraries. This option
11691 only affects @option{-mabicalls}.
11693 All @option{-mabicalls} code has traditionally been position-independent,
11694 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11695 as an extension, the GNU toolchain allows executables to use absolute
11696 accesses for locally-binding symbols. It can also use shorter GP
11697 initialization sequences and generate direct calls to locally-defined
11698 functions. This mode is selected by @option{-mno-shared}.
11700 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11701 objects that can only be linked by the GNU linker. However, the option
11702 does not affect the ABI of the final executable; it only affects the ABI
11703 of relocatable objects. Using @option{-mno-shared} will generally make
11704 executables both smaller and quicker.
11706 @option{-mshared} is the default.
11712 Lift (do not lift) the usual restrictions on the size of the global
11715 GCC normally uses a single instruction to load values from the GOT@.
11716 While this is relatively efficient, it will only work if the GOT
11717 is smaller than about 64k. Anything larger will cause the linker
11718 to report an error such as:
11720 @cindex relocation truncated to fit (MIPS)
11722 relocation truncated to fit: R_MIPS_GOT16 foobar
11725 If this happens, you should recompile your code with @option{-mxgot}.
11726 It should then work with very large GOTs, although it will also be
11727 less efficient, since it will take three instructions to fetch the
11728 value of a global symbol.
11730 Note that some linkers can create multiple GOTs. If you have such a
11731 linker, you should only need to use @option{-mxgot} when a single object
11732 file accesses more than 64k's worth of GOT entries. Very few do.
11734 These options have no effect unless GCC is generating position
11739 Assume that general-purpose registers are 32 bits wide.
11743 Assume that general-purpose registers are 64 bits wide.
11747 Assume that floating-point registers are 32 bits wide.
11751 Assume that floating-point registers are 64 bits wide.
11754 @opindex mhard-float
11755 Use floating-point coprocessor instructions.
11758 @opindex msoft-float
11759 Do not use floating-point coprocessor instructions. Implement
11760 floating-point calculations using library calls instead.
11762 @item -msingle-float
11763 @opindex msingle-float
11764 Assume that the floating-point coprocessor only supports single-precision
11767 @item -mdouble-float
11768 @opindex mdouble-float
11769 Assume that the floating-point coprocessor supports double-precision
11770 operations. This is the default.
11776 Use (do not use) revision 1 of the MIPS DSP ASE.
11777 @xref{MIPS DSP Built-in Functions}. This option defines the
11778 preprocessor macro @samp{__mips_dsp}. It also defines
11779 @samp{__mips_dsp_rev} to 1.
11785 Use (do not use) revision 2 of the MIPS DSP ASE.
11786 @xref{MIPS DSP Built-in Functions}. This option defines the
11787 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
11788 It also defines @samp{__mips_dsp_rev} to 2.
11791 @itemx -mno-smartmips
11792 @opindex msmartmips
11793 @opindex mno-smartmips
11794 Use (do not use) the MIPS SmartMIPS ASE.
11796 @item -mpaired-single
11797 @itemx -mno-paired-single
11798 @opindex mpaired-single
11799 @opindex mno-paired-single
11800 Use (do not use) paired-single floating-point instructions.
11801 @xref{MIPS Paired-Single Support}. This option can only be used
11802 when generating 64-bit code and requires hardware floating-point
11803 support to be enabled.
11809 Use (do not use) MIPS Digital Media Extension instructions.
11810 This option can only be used when generating 64-bit code and requires
11811 hardware floating-point support to be enabled.
11816 @opindex mno-mips3d
11817 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11818 The option @option{-mips3d} implies @option{-mpaired-single}.
11824 Use (do not use) MT Multithreading instructions.
11828 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11829 an explanation of the default and the way that the pointer size is
11834 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11836 The default size of @code{int}s, @code{long}s and pointers depends on
11837 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11838 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11839 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11840 or the same size as integer registers, whichever is smaller.
11846 Assume (do not assume) that all symbols have 32-bit values, regardless
11847 of the selected ABI@. This option is useful in combination with
11848 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11849 to generate shorter and faster references to symbolic addresses.
11853 @cindex smaller data references (MIPS)
11854 @cindex gp-relative references (MIPS)
11855 Put global and static items less than or equal to @var{num} bytes into
11856 the small data or bss section instead of the normal data or bss section.
11857 This allows the data to be accessed using a single instruction.
11859 All modules should be compiled with the same @option{-G @var{num}}
11862 @item -membedded-data
11863 @itemx -mno-embedded-data
11864 @opindex membedded-data
11865 @opindex mno-embedded-data
11866 Allocate variables to the read-only data section first if possible, then
11867 next in the small data section if possible, otherwise in data. This gives
11868 slightly slower code than the default, but reduces the amount of RAM required
11869 when executing, and thus may be preferred for some embedded systems.
11871 @item -muninit-const-in-rodata
11872 @itemx -mno-uninit-const-in-rodata
11873 @opindex muninit-const-in-rodata
11874 @opindex mno-uninit-const-in-rodata
11875 Put uninitialized @code{const} variables in the read-only data section.
11876 This option is only meaningful in conjunction with @option{-membedded-data}.
11878 @item -mcode-readable=@var{setting}
11879 @opindex mcode-readable
11880 Specify whether GCC may generate code that reads from executable sections.
11881 There are three possible settings:
11884 @item -mcode-readable=yes
11885 Instructions may freely access executable sections. This is the
11888 @item -mcode-readable=pcrel
11889 MIPS16 PC-relative load instructions can access executable sections,
11890 but other instructions must not do so. This option is useful on 4KSc
11891 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
11892 It is also useful on processors that can be configured to have a dual
11893 instruction/data SRAM interface and that, like the M4K, automatically
11894 redirect PC-relative loads to the instruction RAM.
11896 @item -mcode-readable=no
11897 Instructions must not access executable sections. This option can be
11898 useful on targets that are configured to have a dual instruction/data
11899 SRAM interface but that (unlike the M4K) do not automatically redirect
11900 PC-relative loads to the instruction RAM.
11903 @item -msplit-addresses
11904 @itemx -mno-split-addresses
11905 @opindex msplit-addresses
11906 @opindex mno-split-addresses
11907 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11908 relocation operators. This option has been superseded by
11909 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11911 @item -mexplicit-relocs
11912 @itemx -mno-explicit-relocs
11913 @opindex mexplicit-relocs
11914 @opindex mno-explicit-relocs
11915 Use (do not use) assembler relocation operators when dealing with symbolic
11916 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11917 is to use assembler macros instead.
11919 @option{-mexplicit-relocs} is the default if GCC was configured
11920 to use an assembler that supports relocation operators.
11922 @item -mcheck-zero-division
11923 @itemx -mno-check-zero-division
11924 @opindex mcheck-zero-division
11925 @opindex mno-check-zero-division
11926 Trap (do not trap) on integer division by zero.
11928 The default is @option{-mcheck-zero-division}.
11930 @item -mdivide-traps
11931 @itemx -mdivide-breaks
11932 @opindex mdivide-traps
11933 @opindex mdivide-breaks
11934 MIPS systems check for division by zero by generating either a
11935 conditional trap or a break instruction. Using traps results in
11936 smaller code, but is only supported on MIPS II and later. Also, some
11937 versions of the Linux kernel have a bug that prevents trap from
11938 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11939 allow conditional traps on architectures that support them and
11940 @option{-mdivide-breaks} to force the use of breaks.
11942 The default is usually @option{-mdivide-traps}, but this can be
11943 overridden at configure time using @option{--with-divide=breaks}.
11944 Divide-by-zero checks can be completely disabled using
11945 @option{-mno-check-zero-division}.
11950 @opindex mno-memcpy
11951 Force (do not force) the use of @code{memcpy()} for non-trivial block
11952 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
11953 most constant-sized copies.
11956 @itemx -mno-long-calls
11957 @opindex mlong-calls
11958 @opindex mno-long-calls
11959 Disable (do not disable) use of the @code{jal} instruction. Calling
11960 functions using @code{jal} is more efficient but requires the caller
11961 and callee to be in the same 256 megabyte segment.
11963 This option has no effect on abicalls code. The default is
11964 @option{-mno-long-calls}.
11970 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
11971 instructions, as provided by the R4650 ISA@.
11974 @itemx -mno-fused-madd
11975 @opindex mfused-madd
11976 @opindex mno-fused-madd
11977 Enable (disable) use of the floating point multiply-accumulate
11978 instructions, when they are available. The default is
11979 @option{-mfused-madd}.
11981 When multiply-accumulate instructions are used, the intermediate
11982 product is calculated to infinite precision and is not subject to
11983 the FCSR Flush to Zero bit. This may be undesirable in some
11988 Tell the MIPS assembler to not run its preprocessor over user
11989 assembler files (with a @samp{.s} suffix) when assembling them.
11992 @itemx -mno-fix-r4000
11993 @opindex mfix-r4000
11994 @opindex mno-fix-r4000
11995 Work around certain R4000 CPU errata:
11998 A double-word or a variable shift may give an incorrect result if executed
11999 immediately after starting an integer division.
12001 A double-word or a variable shift may give an incorrect result if executed
12002 while an integer multiplication is in progress.
12004 An integer division may give an incorrect result if started in a delay slot
12005 of a taken branch or a jump.
12009 @itemx -mno-fix-r4400
12010 @opindex mfix-r4400
12011 @opindex mno-fix-r4400
12012 Work around certain R4400 CPU errata:
12015 A double-word or a variable shift may give an incorrect result if executed
12016 immediately after starting an integer division.
12020 @itemx -mno-fix-vr4120
12021 @opindex mfix-vr4120
12022 Work around certain VR4120 errata:
12025 @code{dmultu} does not always produce the correct result.
12027 @code{div} and @code{ddiv} do not always produce the correct result if one
12028 of the operands is negative.
12030 The workarounds for the division errata rely on special functions in
12031 @file{libgcc.a}. At present, these functions are only provided by
12032 the @code{mips64vr*-elf} configurations.
12034 Other VR4120 errata require a nop to be inserted between certain pairs of
12035 instructions. These errata are handled by the assembler, not by GCC itself.
12038 @opindex mfix-vr4130
12039 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12040 workarounds are implemented by the assembler rather than by GCC,
12041 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12042 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12043 instructions are available instead.
12046 @itemx -mno-fix-sb1
12048 Work around certain SB-1 CPU core errata.
12049 (This flag currently works around the SB-1 revision 2
12050 ``F1'' and ``F2'' floating point errata.)
12052 @item -mflush-func=@var{func}
12053 @itemx -mno-flush-func
12054 @opindex mflush-func
12055 Specifies the function to call to flush the I and D caches, or to not
12056 call any such function. If called, the function must take the same
12057 arguments as the common @code{_flush_func()}, that is, the address of the
12058 memory range for which the cache is being flushed, the size of the
12059 memory range, and the number 3 (to flush both caches). The default
12060 depends on the target GCC was configured for, but commonly is either
12061 @samp{_flush_func} or @samp{__cpu_flush}.
12063 @item mbranch-cost=@var{num}
12064 @opindex mbranch-cost
12065 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12066 This cost is only a heuristic and is not guaranteed to produce
12067 consistent results across releases. A zero cost redundantly selects
12068 the default, which is based on the @option{-mtune} setting.
12070 @item -mbranch-likely
12071 @itemx -mno-branch-likely
12072 @opindex mbranch-likely
12073 @opindex mno-branch-likely
12074 Enable or disable use of Branch Likely instructions, regardless of the
12075 default for the selected architecture. By default, Branch Likely
12076 instructions may be generated if they are supported by the selected
12077 architecture. An exception is for the MIPS32 and MIPS64 architectures
12078 and processors which implement those architectures; for those, Branch
12079 Likely instructions will not be generated by default because the MIPS32
12080 and MIPS64 architectures specifically deprecate their use.
12082 @item -mfp-exceptions
12083 @itemx -mno-fp-exceptions
12084 @opindex mfp-exceptions
12085 Specifies whether FP exceptions are enabled. This affects how we schedule
12086 FP instructions for some processors. The default is that FP exceptions are
12089 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12090 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12093 @item -mvr4130-align
12094 @itemx -mno-vr4130-align
12095 @opindex mvr4130-align
12096 The VR4130 pipeline is two-way superscalar, but can only issue two
12097 instructions together if the first one is 8-byte aligned. When this
12098 option is enabled, GCC will align pairs of instructions that it
12099 thinks should execute in parallel.
12101 This option only has an effect when optimizing for the VR4130.
12102 It normally makes code faster, but at the expense of making it bigger.
12103 It is enabled by default at optimization level @option{-O3}.
12107 @subsection MMIX Options
12108 @cindex MMIX Options
12110 These options are defined for the MMIX:
12114 @itemx -mno-libfuncs
12116 @opindex mno-libfuncs
12117 Specify that intrinsic library functions are being compiled, passing all
12118 values in registers, no matter the size.
12121 @itemx -mno-epsilon
12123 @opindex mno-epsilon
12124 Generate floating-point comparison instructions that compare with respect
12125 to the @code{rE} epsilon register.
12127 @item -mabi=mmixware
12129 @opindex mabi-mmixware
12131 Generate code that passes function parameters and return values that (in
12132 the called function) are seen as registers @code{$0} and up, as opposed to
12133 the GNU ABI which uses global registers @code{$231} and up.
12135 @item -mzero-extend
12136 @itemx -mno-zero-extend
12137 @opindex mzero-extend
12138 @opindex mno-zero-extend
12139 When reading data from memory in sizes shorter than 64 bits, use (do not
12140 use) zero-extending load instructions by default, rather than
12141 sign-extending ones.
12144 @itemx -mno-knuthdiv
12146 @opindex mno-knuthdiv
12147 Make the result of a division yielding a remainder have the same sign as
12148 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12149 remainder follows the sign of the dividend. Both methods are
12150 arithmetically valid, the latter being almost exclusively used.
12152 @item -mtoplevel-symbols
12153 @itemx -mno-toplevel-symbols
12154 @opindex mtoplevel-symbols
12155 @opindex mno-toplevel-symbols
12156 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12157 code can be used with the @code{PREFIX} assembly directive.
12161 Generate an executable in the ELF format, rather than the default
12162 @samp{mmo} format used by the @command{mmix} simulator.
12164 @item -mbranch-predict
12165 @itemx -mno-branch-predict
12166 @opindex mbranch-predict
12167 @opindex mno-branch-predict
12168 Use (do not use) the probable-branch instructions, when static branch
12169 prediction indicates a probable branch.
12171 @item -mbase-addresses
12172 @itemx -mno-base-addresses
12173 @opindex mbase-addresses
12174 @opindex mno-base-addresses
12175 Generate (do not generate) code that uses @emph{base addresses}. Using a
12176 base address automatically generates a request (handled by the assembler
12177 and the linker) for a constant to be set up in a global register. The
12178 register is used for one or more base address requests within the range 0
12179 to 255 from the value held in the register. The generally leads to short
12180 and fast code, but the number of different data items that can be
12181 addressed is limited. This means that a program that uses lots of static
12182 data may require @option{-mno-base-addresses}.
12184 @item -msingle-exit
12185 @itemx -mno-single-exit
12186 @opindex msingle-exit
12187 @opindex mno-single-exit
12188 Force (do not force) generated code to have a single exit point in each
12192 @node MN10300 Options
12193 @subsection MN10300 Options
12194 @cindex MN10300 options
12196 These @option{-m} options are defined for Matsushita MN10300 architectures:
12201 Generate code to avoid bugs in the multiply instructions for the MN10300
12202 processors. This is the default.
12204 @item -mno-mult-bug
12205 @opindex mno-mult-bug
12206 Do not generate code to avoid bugs in the multiply instructions for the
12207 MN10300 processors.
12211 Generate code which uses features specific to the AM33 processor.
12215 Do not generate code which uses features specific to the AM33 processor. This
12218 @item -mreturn-pointer-on-d0
12219 @opindex mreturn-pointer-on-d0
12220 When generating a function which returns a pointer, return the pointer
12221 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12222 only in a0, and attempts to call such functions without a prototype
12223 would result in errors. Note that this option is on by default; use
12224 @option{-mno-return-pointer-on-d0} to disable it.
12228 Do not link in the C run-time initialization object file.
12232 Indicate to the linker that it should perform a relaxation optimization pass
12233 to shorten branches, calls and absolute memory addresses. This option only
12234 has an effect when used on the command line for the final link step.
12236 This option makes symbolic debugging impossible.
12240 @subsection MT Options
12243 These @option{-m} options are defined for Morpho MT architectures:
12247 @item -march=@var{cpu-type}
12249 Generate code that will run on @var{cpu-type}, which is the name of a system
12250 representing a certain processor type. Possible values for
12251 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12252 @samp{ms1-16-003} and @samp{ms2}.
12254 When this option is not used, the default is @option{-march=ms1-16-002}.
12258 Use byte loads and stores when generating code.
12262 Do not use byte loads and stores when generating code.
12266 Use simulator runtime
12270 Do not link in the C run-time initialization object file
12271 @file{crti.o}. Other run-time initialization and termination files
12272 such as @file{startup.o} and @file{exit.o} are still included on the
12273 linker command line.
12277 @node PDP-11 Options
12278 @subsection PDP-11 Options
12279 @cindex PDP-11 Options
12281 These options are defined for the PDP-11:
12286 Use hardware FPP floating point. This is the default. (FIS floating
12287 point on the PDP-11/40 is not supported.)
12290 @opindex msoft-float
12291 Do not use hardware floating point.
12295 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12299 Return floating-point results in memory. This is the default.
12303 Generate code for a PDP-11/40.
12307 Generate code for a PDP-11/45. This is the default.
12311 Generate code for a PDP-11/10.
12313 @item -mbcopy-builtin
12314 @opindex bcopy-builtin
12315 Use inline @code{movmemhi} patterns for copying memory. This is the
12320 Do not use inline @code{movmemhi} patterns for copying memory.
12326 Use 16-bit @code{int}. This is the default.
12332 Use 32-bit @code{int}.
12335 @itemx -mno-float32
12337 @opindex mno-float32
12338 Use 64-bit @code{float}. This is the default.
12341 @itemx -mno-float64
12343 @opindex mno-float64
12344 Use 32-bit @code{float}.
12348 Use @code{abshi2} pattern. This is the default.
12352 Do not use @code{abshi2} pattern.
12354 @item -mbranch-expensive
12355 @opindex mbranch-expensive
12356 Pretend that branches are expensive. This is for experimenting with
12357 code generation only.
12359 @item -mbranch-cheap
12360 @opindex mbranch-cheap
12361 Do not pretend that branches are expensive. This is the default.
12365 Generate code for a system with split I&D@.
12369 Generate code for a system without split I&D@. This is the default.
12373 Use Unix assembler syntax. This is the default when configured for
12374 @samp{pdp11-*-bsd}.
12378 Use DEC assembler syntax. This is the default when configured for any
12379 PDP-11 target other than @samp{pdp11-*-bsd}.
12382 @node PowerPC Options
12383 @subsection PowerPC Options
12384 @cindex PowerPC options
12386 These are listed under @xref{RS/6000 and PowerPC Options}.
12388 @node RS/6000 and PowerPC Options
12389 @subsection IBM RS/6000 and PowerPC Options
12390 @cindex RS/6000 and PowerPC Options
12391 @cindex IBM RS/6000 and PowerPC Options
12393 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12400 @itemx -mno-powerpc
12401 @itemx -mpowerpc-gpopt
12402 @itemx -mno-powerpc-gpopt
12403 @itemx -mpowerpc-gfxopt
12404 @itemx -mno-powerpc-gfxopt
12406 @itemx -mno-powerpc64
12410 @itemx -mno-popcntb
12422 @opindex mno-power2
12424 @opindex mno-powerpc
12425 @opindex mpowerpc-gpopt
12426 @opindex mno-powerpc-gpopt
12427 @opindex mpowerpc-gfxopt
12428 @opindex mno-powerpc-gfxopt
12429 @opindex mpowerpc64
12430 @opindex mno-powerpc64
12434 @opindex mno-popcntb
12440 @opindex mno-mfpgpr
12443 GCC supports two related instruction set architectures for the
12444 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12445 instructions supported by the @samp{rios} chip set used in the original
12446 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12447 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12448 the IBM 4xx, 6xx, and follow-on microprocessors.
12450 Neither architecture is a subset of the other. However there is a
12451 large common subset of instructions supported by both. An MQ
12452 register is included in processors supporting the POWER architecture.
12454 You use these options to specify which instructions are available on the
12455 processor you are using. The default value of these options is
12456 determined when configuring GCC@. Specifying the
12457 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12458 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12459 rather than the options listed above.
12461 The @option{-mpower} option allows GCC to generate instructions that
12462 are found only in the POWER architecture and to use the MQ register.
12463 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12464 to generate instructions that are present in the POWER2 architecture but
12465 not the original POWER architecture.
12467 The @option{-mpowerpc} option allows GCC to generate instructions that
12468 are found only in the 32-bit subset of the PowerPC architecture.
12469 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12470 GCC to use the optional PowerPC architecture instructions in the
12471 General Purpose group, including floating-point square root. Specifying
12472 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12473 use the optional PowerPC architecture instructions in the Graphics
12474 group, including floating-point select.
12476 The @option{-mmfcrf} option allows GCC to generate the move from
12477 condition register field instruction implemented on the POWER4
12478 processor and other processors that support the PowerPC V2.01
12480 The @option{-mpopcntb} option allows GCC to generate the popcount and
12481 double precision FP reciprocal estimate instruction implemented on the
12482 POWER5 processor and other processors that support the PowerPC V2.02
12484 The @option{-mfprnd} option allows GCC to generate the FP round to
12485 integer instructions implemented on the POWER5+ processor and other
12486 processors that support the PowerPC V2.03 architecture.
12487 The @option{-mcmpb} option allows GCC to generate the compare bytes
12488 instruction implemented on the POWER6 processor and other processors
12489 that support the PowerPC V2.05 architecture.
12490 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12491 general purpose register instructions implemented on the POWER6X
12492 processor and other processors that support the extended PowerPC V2.05
12494 The @option{-mdfp} option allows GCC to generate the decimal floating
12495 point instructions implemented on some POWER processors.
12497 The @option{-mpowerpc64} option allows GCC to generate the additional
12498 64-bit instructions that are found in the full PowerPC64 architecture
12499 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12500 @option{-mno-powerpc64}.
12502 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12503 will use only the instructions in the common subset of both
12504 architectures plus some special AIX common-mode calls, and will not use
12505 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12506 permits GCC to use any instruction from either architecture and to
12507 allow use of the MQ register; specify this for the Motorola MPC601.
12509 @item -mnew-mnemonics
12510 @itemx -mold-mnemonics
12511 @opindex mnew-mnemonics
12512 @opindex mold-mnemonics
12513 Select which mnemonics to use in the generated assembler code. With
12514 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12515 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12516 assembler mnemonics defined for the POWER architecture. Instructions
12517 defined in only one architecture have only one mnemonic; GCC uses that
12518 mnemonic irrespective of which of these options is specified.
12520 GCC defaults to the mnemonics appropriate for the architecture in
12521 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12522 value of these option. Unless you are building a cross-compiler, you
12523 should normally not specify either @option{-mnew-mnemonics} or
12524 @option{-mold-mnemonics}, but should instead accept the default.
12526 @item -mcpu=@var{cpu_type}
12528 Set architecture type, register usage, choice of mnemonics, and
12529 instruction scheduling parameters for machine type @var{cpu_type}.
12530 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12531 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12532 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12533 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12534 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12535 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12536 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12537 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12538 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12539 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12541 @option{-mcpu=common} selects a completely generic processor. Code
12542 generated under this option will run on any POWER or PowerPC processor.
12543 GCC will use only the instructions in the common subset of both
12544 architectures, and will not use the MQ register. GCC assumes a generic
12545 processor model for scheduling purposes.
12547 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12548 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12549 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12550 types, with an appropriate, generic processor model assumed for
12551 scheduling purposes.
12553 The other options specify a specific processor. Code generated under
12554 those options will run best on that processor, and may not run at all on
12557 The @option{-mcpu} options automatically enable or disable the
12560 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12561 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12562 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12564 The particular options set for any particular CPU will vary between
12565 compiler versions, depending on what setting seems to produce optimal
12566 code for that CPU; it doesn't necessarily reflect the actual hardware's
12567 capabilities. If you wish to set an individual option to a particular
12568 value, you may specify it after the @option{-mcpu} option, like
12569 @samp{-mcpu=970 -mno-altivec}.
12571 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12572 not enabled or disabled by the @option{-mcpu} option at present because
12573 AIX does not have full support for these options. You may still
12574 enable or disable them individually if you're sure it'll work in your
12577 @item -mtune=@var{cpu_type}
12579 Set the instruction scheduling parameters for machine type
12580 @var{cpu_type}, but do not set the architecture type, register usage, or
12581 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12582 values for @var{cpu_type} are used for @option{-mtune} as for
12583 @option{-mcpu}. If both are specified, the code generated will use the
12584 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12585 scheduling parameters set by @option{-mtune}.
12591 Generate code to compute division as reciprocal estimate and iterative
12592 refinement, creating opportunities for increased throughput. This
12593 feature requires: optional PowerPC Graphics instruction set for single
12594 precision and FRE instruction for double precision, assuming divides
12595 cannot generate user-visible traps, and the domain values not include
12596 Infinities, denormals or zero denominator.
12599 @itemx -mno-altivec
12601 @opindex mno-altivec
12602 Generate code that uses (does not use) AltiVec instructions, and also
12603 enable the use of built-in functions that allow more direct access to
12604 the AltiVec instruction set. You may also need to set
12605 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12611 @opindex mno-vrsave
12612 Generate VRSAVE instructions when generating AltiVec code.
12615 @opindex msecure-plt
12616 Generate code that allows ld and ld.so to build executables and shared
12617 libraries with non-exec .plt and .got sections. This is a PowerPC
12618 32-bit SYSV ABI option.
12622 Generate code that uses a BSS .plt section that ld.so fills in, and
12623 requires .plt and .got sections that are both writable and executable.
12624 This is a PowerPC 32-bit SYSV ABI option.
12630 This switch enables or disables the generation of ISEL instructions.
12632 @item -misel=@var{yes/no}
12633 This switch has been deprecated. Use @option{-misel} and
12634 @option{-mno-isel} instead.
12640 This switch enables or disables the generation of SPE simd
12646 @opindex mno-paired
12647 This switch enables or disables the generation of PAIRED simd
12650 @item -mspe=@var{yes/no}
12651 This option has been deprecated. Use @option{-mspe} and
12652 @option{-mno-spe} instead.
12654 @item -mfloat-gprs=@var{yes/single/double/no}
12655 @itemx -mfloat-gprs
12656 @opindex mfloat-gprs
12657 This switch enables or disables the generation of floating point
12658 operations on the general purpose registers for architectures that
12661 The argument @var{yes} or @var{single} enables the use of
12662 single-precision floating point operations.
12664 The argument @var{double} enables the use of single and
12665 double-precision floating point operations.
12667 The argument @var{no} disables floating point operations on the
12668 general purpose registers.
12670 This option is currently only available on the MPC854x.
12676 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12677 targets (including GNU/Linux). The 32-bit environment sets int, long
12678 and pointer to 32 bits and generates code that runs on any PowerPC
12679 variant. The 64-bit environment sets int to 32 bits and long and
12680 pointer to 64 bits, and generates code for PowerPC64, as for
12681 @option{-mpowerpc64}.
12684 @itemx -mno-fp-in-toc
12685 @itemx -mno-sum-in-toc
12686 @itemx -mminimal-toc
12688 @opindex mno-fp-in-toc
12689 @opindex mno-sum-in-toc
12690 @opindex mminimal-toc
12691 Modify generation of the TOC (Table Of Contents), which is created for
12692 every executable file. The @option{-mfull-toc} option is selected by
12693 default. In that case, GCC will allocate at least one TOC entry for
12694 each unique non-automatic variable reference in your program. GCC
12695 will also place floating-point constants in the TOC@. However, only
12696 16,384 entries are available in the TOC@.
12698 If you receive a linker error message that saying you have overflowed
12699 the available TOC space, you can reduce the amount of TOC space used
12700 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12701 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12702 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12703 generate code to calculate the sum of an address and a constant at
12704 run-time instead of putting that sum into the TOC@. You may specify one
12705 or both of these options. Each causes GCC to produce very slightly
12706 slower and larger code at the expense of conserving TOC space.
12708 If you still run out of space in the TOC even when you specify both of
12709 these options, specify @option{-mminimal-toc} instead. This option causes
12710 GCC to make only one TOC entry for every file. When you specify this
12711 option, GCC will produce code that is slower and larger but which
12712 uses extremely little TOC space. You may wish to use this option
12713 only on files that contain less frequently executed code.
12719 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12720 @code{long} type, and the infrastructure needed to support them.
12721 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12722 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12723 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12726 @itemx -mno-xl-compat
12727 @opindex mxl-compat
12728 @opindex mno-xl-compat
12729 Produce code that conforms more closely to IBM XL compiler semantics
12730 when using AIX-compatible ABI. Pass floating-point arguments to
12731 prototyped functions beyond the register save area (RSA) on the stack
12732 in addition to argument FPRs. Do not assume that most significant
12733 double in 128-bit long double value is properly rounded when comparing
12734 values and converting to double. Use XL symbol names for long double
12737 The AIX calling convention was extended but not initially documented to
12738 handle an obscure K&R C case of calling a function that takes the
12739 address of its arguments with fewer arguments than declared. IBM XL
12740 compilers access floating point arguments which do not fit in the
12741 RSA from the stack when a subroutine is compiled without
12742 optimization. Because always storing floating-point arguments on the
12743 stack is inefficient and rarely needed, this option is not enabled by
12744 default and only is necessary when calling subroutines compiled by IBM
12745 XL compilers without optimization.
12749 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12750 application written to use message passing with special startup code to
12751 enable the application to run. The system must have PE installed in the
12752 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12753 must be overridden with the @option{-specs=} option to specify the
12754 appropriate directory location. The Parallel Environment does not
12755 support threads, so the @option{-mpe} option and the @option{-pthread}
12756 option are incompatible.
12758 @item -malign-natural
12759 @itemx -malign-power
12760 @opindex malign-natural
12761 @opindex malign-power
12762 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12763 @option{-malign-natural} overrides the ABI-defined alignment of larger
12764 types, such as floating-point doubles, on their natural size-based boundary.
12765 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12766 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12768 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12772 @itemx -mhard-float
12773 @opindex msoft-float
12774 @opindex mhard-float
12775 Generate code that does not use (uses) the floating-point register set.
12776 Software floating point emulation is provided if you use the
12777 @option{-msoft-float} option, and pass the option to GCC when linking.
12780 @itemx -mno-multiple
12782 @opindex mno-multiple
12783 Generate code that uses (does not use) the load multiple word
12784 instructions and the store multiple word instructions. These
12785 instructions are generated by default on POWER systems, and not
12786 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12787 endian PowerPC systems, since those instructions do not work when the
12788 processor is in little endian mode. The exceptions are PPC740 and
12789 PPC750 which permit the instructions usage in little endian mode.
12794 @opindex mno-string
12795 Generate code that uses (does not use) the load string instructions
12796 and the store string word instructions to save multiple registers and
12797 do small block moves. These instructions are generated by default on
12798 POWER systems, and not generated on PowerPC systems. Do not use
12799 @option{-mstring} on little endian PowerPC systems, since those
12800 instructions do not work when the processor is in little endian mode.
12801 The exceptions are PPC740 and PPC750 which permit the instructions
12802 usage in little endian mode.
12807 @opindex mno-update
12808 Generate code that uses (does not use) the load or store instructions
12809 that update the base register to the address of the calculated memory
12810 location. These instructions are generated by default. If you use
12811 @option{-mno-update}, there is a small window between the time that the
12812 stack pointer is updated and the address of the previous frame is
12813 stored, which means code that walks the stack frame across interrupts or
12814 signals may get corrupted data.
12817 @itemx -mno-fused-madd
12818 @opindex mfused-madd
12819 @opindex mno-fused-madd
12820 Generate code that uses (does not use) the floating point multiply and
12821 accumulate instructions. These instructions are generated by default if
12822 hardware floating is used.
12828 Generate code that uses (does not use) the half-word multiply and
12829 multiply-accumulate instructions on the IBM 405 and 440 processors.
12830 These instructions are generated by default when targetting those
12837 Generate code that uses (does not use) the string-search @samp{dlmzb}
12838 instruction on the IBM 405 and 440 processors. This instruction is
12839 generated by default when targetting those processors.
12841 @item -mno-bit-align
12843 @opindex mno-bit-align
12844 @opindex mbit-align
12845 On System V.4 and embedded PowerPC systems do not (do) force structures
12846 and unions that contain bit-fields to be aligned to the base type of the
12849 For example, by default a structure containing nothing but 8
12850 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12851 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12852 the structure would be aligned to a 1 byte boundary and be one byte in
12855 @item -mno-strict-align
12856 @itemx -mstrict-align
12857 @opindex mno-strict-align
12858 @opindex mstrict-align
12859 On System V.4 and embedded PowerPC systems do not (do) assume that
12860 unaligned memory references will be handled by the system.
12862 @item -mrelocatable
12863 @itemx -mno-relocatable
12864 @opindex mrelocatable
12865 @opindex mno-relocatable
12866 On embedded PowerPC systems generate code that allows (does not allow)
12867 the program to be relocated to a different address at runtime. If you
12868 use @option{-mrelocatable} on any module, all objects linked together must
12869 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12871 @item -mrelocatable-lib
12872 @itemx -mno-relocatable-lib
12873 @opindex mrelocatable-lib
12874 @opindex mno-relocatable-lib
12875 On embedded PowerPC systems generate code that allows (does not allow)
12876 the program to be relocated to a different address at runtime. Modules
12877 compiled with @option{-mrelocatable-lib} can be linked with either modules
12878 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12879 with modules compiled with the @option{-mrelocatable} options.
12885 On System V.4 and embedded PowerPC systems do not (do) assume that
12886 register 2 contains a pointer to a global area pointing to the addresses
12887 used in the program.
12890 @itemx -mlittle-endian
12892 @opindex mlittle-endian
12893 On System V.4 and embedded PowerPC systems compile code for the
12894 processor in little endian mode. The @option{-mlittle-endian} option is
12895 the same as @option{-mlittle}.
12898 @itemx -mbig-endian
12900 @opindex mbig-endian
12901 On System V.4 and embedded PowerPC systems compile code for the
12902 processor in big endian mode. The @option{-mbig-endian} option is
12903 the same as @option{-mbig}.
12905 @item -mdynamic-no-pic
12906 @opindex mdynamic-no-pic
12907 On Darwin and Mac OS X systems, compile code so that it is not
12908 relocatable, but that its external references are relocatable. The
12909 resulting code is suitable for applications, but not shared
12912 @item -mprioritize-restricted-insns=@var{priority}
12913 @opindex mprioritize-restricted-insns
12914 This option controls the priority that is assigned to
12915 dispatch-slot restricted instructions during the second scheduling
12916 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12917 @var{no/highest/second-highest} priority to dispatch slot restricted
12920 @item -msched-costly-dep=@var{dependence_type}
12921 @opindex msched-costly-dep
12922 This option controls which dependences are considered costly
12923 by the target during instruction scheduling. The argument
12924 @var{dependence_type} takes one of the following values:
12925 @var{no}: no dependence is costly,
12926 @var{all}: all dependences are costly,
12927 @var{true_store_to_load}: a true dependence from store to load is costly,
12928 @var{store_to_load}: any dependence from store to load is costly,
12929 @var{number}: any dependence which latency >= @var{number} is costly.
12931 @item -minsert-sched-nops=@var{scheme}
12932 @opindex minsert-sched-nops
12933 This option controls which nop insertion scheme will be used during
12934 the second scheduling pass. The argument @var{scheme} takes one of the
12936 @var{no}: Don't insert nops.
12937 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12938 according to the scheduler's grouping.
12939 @var{regroup_exact}: Insert nops to force costly dependent insns into
12940 separate groups. Insert exactly as many nops as needed to force an insn
12941 to a new group, according to the estimated processor grouping.
12942 @var{number}: Insert nops to force costly dependent insns into
12943 separate groups. Insert @var{number} nops to force an insn to a new group.
12946 @opindex mcall-sysv
12947 On System V.4 and embedded PowerPC systems compile code using calling
12948 conventions that adheres to the March 1995 draft of the System V
12949 Application Binary Interface, PowerPC processor supplement. This is the
12950 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
12952 @item -mcall-sysv-eabi
12953 @opindex mcall-sysv-eabi
12954 Specify both @option{-mcall-sysv} and @option{-meabi} options.
12956 @item -mcall-sysv-noeabi
12957 @opindex mcall-sysv-noeabi
12958 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
12960 @item -mcall-solaris
12961 @opindex mcall-solaris
12962 On System V.4 and embedded PowerPC systems compile code for the Solaris
12966 @opindex mcall-linux
12967 On System V.4 and embedded PowerPC systems compile code for the
12968 Linux-based GNU system.
12972 On System V.4 and embedded PowerPC systems compile code for the
12973 Hurd-based GNU system.
12975 @item -mcall-netbsd
12976 @opindex mcall-netbsd
12977 On System V.4 and embedded PowerPC systems compile code for the
12978 NetBSD operating system.
12980 @item -maix-struct-return
12981 @opindex maix-struct-return
12982 Return all structures in memory (as specified by the AIX ABI)@.
12984 @item -msvr4-struct-return
12985 @opindex msvr4-struct-return
12986 Return structures smaller than 8 bytes in registers (as specified by the
12989 @item -mabi=@var{abi-type}
12991 Extend the current ABI with a particular extension, or remove such extension.
12992 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
12993 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
12997 Extend the current ABI with SPE ABI extensions. This does not change
12998 the default ABI, instead it adds the SPE ABI extensions to the current
13002 @opindex mabi=no-spe
13003 Disable Booke SPE ABI extensions for the current ABI@.
13005 @item -mabi=ibmlongdouble
13006 @opindex mabi=ibmlongdouble
13007 Change the current ABI to use IBM extended precision long double.
13008 This is a PowerPC 32-bit SYSV ABI option.
13010 @item -mabi=ieeelongdouble
13011 @opindex mabi=ieeelongdouble
13012 Change the current ABI to use IEEE extended precision long double.
13013 This is a PowerPC 32-bit Linux ABI option.
13016 @itemx -mno-prototype
13017 @opindex mprototype
13018 @opindex mno-prototype
13019 On System V.4 and embedded PowerPC systems assume that all calls to
13020 variable argument functions are properly prototyped. Otherwise, the
13021 compiler must insert an instruction before every non prototyped call to
13022 set or clear bit 6 of the condition code register (@var{CR}) to
13023 indicate whether floating point values were passed in the floating point
13024 registers in case the function takes a variable arguments. With
13025 @option{-mprototype}, only calls to prototyped variable argument functions
13026 will set or clear the bit.
13030 On embedded PowerPC systems, assume that the startup module is called
13031 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13032 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
13037 On embedded PowerPC systems, assume that the startup module is called
13038 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13043 On embedded PowerPC systems, assume that the startup module is called
13044 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13047 @item -myellowknife
13048 @opindex myellowknife
13049 On embedded PowerPC systems, assume that the startup module is called
13050 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13055 On System V.4 and embedded PowerPC systems, specify that you are
13056 compiling for a VxWorks system.
13060 Specify that you are compiling for the WindISS simulation environment.
13064 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13065 header to indicate that @samp{eabi} extended relocations are used.
13071 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13072 Embedded Applications Binary Interface (eabi) which is a set of
13073 modifications to the System V.4 specifications. Selecting @option{-meabi}
13074 means that the stack is aligned to an 8 byte boundary, a function
13075 @code{__eabi} is called to from @code{main} to set up the eabi
13076 environment, and the @option{-msdata} option can use both @code{r2} and
13077 @code{r13} to point to two separate small data areas. Selecting
13078 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13079 do not call an initialization function from @code{main}, and the
13080 @option{-msdata} option will only use @code{r13} to point to a single
13081 small data area. The @option{-meabi} option is on by default if you
13082 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13085 @opindex msdata=eabi
13086 On System V.4 and embedded PowerPC systems, put small initialized
13087 @code{const} global and static data in the @samp{.sdata2} section, which
13088 is pointed to by register @code{r2}. Put small initialized
13089 non-@code{const} global and static data in the @samp{.sdata} section,
13090 which is pointed to by register @code{r13}. Put small uninitialized
13091 global and static data in the @samp{.sbss} section, which is adjacent to
13092 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13093 incompatible with the @option{-mrelocatable} option. The
13094 @option{-msdata=eabi} option also sets the @option{-memb} option.
13097 @opindex msdata=sysv
13098 On System V.4 and embedded PowerPC systems, put small global and static
13099 data in the @samp{.sdata} section, which is pointed to by register
13100 @code{r13}. Put small uninitialized global and static data in the
13101 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13102 The @option{-msdata=sysv} option is incompatible with the
13103 @option{-mrelocatable} option.
13105 @item -msdata=default
13107 @opindex msdata=default
13109 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13110 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13111 same as @option{-msdata=sysv}.
13114 @opindex msdata-data
13115 On System V.4 and embedded PowerPC systems, put small global
13116 data in the @samp{.sdata} section. Put small uninitialized global
13117 data in the @samp{.sbss} section. Do not use register @code{r13}
13118 to address small data however. This is the default behavior unless
13119 other @option{-msdata} options are used.
13123 @opindex msdata=none
13125 On embedded PowerPC systems, put all initialized global and static data
13126 in the @samp{.data} section, and all uninitialized data in the
13127 @samp{.bss} section.
13131 @cindex smaller data references (PowerPC)
13132 @cindex .sdata/.sdata2 references (PowerPC)
13133 On embedded PowerPC systems, put global and static items less than or
13134 equal to @var{num} bytes into the small data or bss sections instead of
13135 the normal data or bss section. By default, @var{num} is 8. The
13136 @option{-G @var{num}} switch is also passed to the linker.
13137 All modules should be compiled with the same @option{-G @var{num}} value.
13140 @itemx -mno-regnames
13142 @opindex mno-regnames
13143 On System V.4 and embedded PowerPC systems do (do not) emit register
13144 names in the assembly language output using symbolic forms.
13147 @itemx -mno-longcall
13149 @opindex mno-longcall
13150 By default assume that all calls are far away so that a longer more
13151 expensive calling sequence is required. This is required for calls
13152 further than 32 megabytes (33,554,432 bytes) from the current location.
13153 A short call will be generated if the compiler knows
13154 the call cannot be that far away. This setting can be overridden by
13155 the @code{shortcall} function attribute, or by @code{#pragma
13158 Some linkers are capable of detecting out-of-range calls and generating
13159 glue code on the fly. On these systems, long calls are unnecessary and
13160 generate slower code. As of this writing, the AIX linker can do this,
13161 as can the GNU linker for PowerPC/64. It is planned to add this feature
13162 to the GNU linker for 32-bit PowerPC systems as well.
13164 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13165 callee, L42'', plus a ``branch island'' (glue code). The two target
13166 addresses represent the callee and the ``branch island''. The
13167 Darwin/PPC linker will prefer the first address and generate a ``bl
13168 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13169 otherwise, the linker will generate ``bl L42'' to call the ``branch
13170 island''. The ``branch island'' is appended to the body of the
13171 calling function; it computes the full 32-bit address of the callee
13174 On Mach-O (Darwin) systems, this option directs the compiler emit to
13175 the glue for every direct call, and the Darwin linker decides whether
13176 to use or discard it.
13178 In the future, we may cause GCC to ignore all longcall specifications
13179 when the linker is known to generate glue.
13183 Adds support for multithreading with the @dfn{pthreads} library.
13184 This option sets flags for both the preprocessor and linker.
13188 @node S/390 and zSeries Options
13189 @subsection S/390 and zSeries Options
13190 @cindex S/390 and zSeries Options
13192 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13196 @itemx -msoft-float
13197 @opindex mhard-float
13198 @opindex msoft-float
13199 Use (do not use) the hardware floating-point instructions and registers
13200 for floating-point operations. When @option{-msoft-float} is specified,
13201 functions in @file{libgcc.a} will be used to perform floating-point
13202 operations. When @option{-mhard-float} is specified, the compiler
13203 generates IEEE floating-point instructions. This is the default.
13205 @item -mlong-double-64
13206 @itemx -mlong-double-128
13207 @opindex mlong-double-64
13208 @opindex mlong-double-128
13209 These switches control the size of @code{long double} type. A size
13210 of 64bit makes the @code{long double} type equivalent to the @code{double}
13211 type. This is the default.
13214 @itemx -mno-backchain
13215 @opindex mbackchain
13216 @opindex mno-backchain
13217 Store (do not store) the address of the caller's frame as backchain pointer
13218 into the callee's stack frame.
13219 A backchain may be needed to allow debugging using tools that do not understand
13220 DWARF-2 call frame information.
13221 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13222 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13223 the backchain is placed into the topmost word of the 96/160 byte register
13226 In general, code compiled with @option{-mbackchain} is call-compatible with
13227 code compiled with @option{-mmo-backchain}; however, use of the backchain
13228 for debugging purposes usually requires that the whole binary is built with
13229 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13230 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13231 to build a linux kernel use @option{-msoft-float}.
13233 The default is to not maintain the backchain.
13235 @item -mpacked-stack
13236 @item -mno-packed-stack
13237 @opindex mpacked-stack
13238 @opindex mno-packed-stack
13239 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13240 specified, the compiler uses the all fields of the 96/160 byte register save
13241 area only for their default purpose; unused fields still take up stack space.
13242 When @option{-mpacked-stack} is specified, register save slots are densely
13243 packed at the top of the register save area; unused space is reused for other
13244 purposes, allowing for more efficient use of the available stack space.
13245 However, when @option{-mbackchain} is also in effect, the topmost word of
13246 the save area is always used to store the backchain, and the return address
13247 register is always saved two words below the backchain.
13249 As long as the stack frame backchain is not used, code generated with
13250 @option{-mpacked-stack} is call-compatible with code generated with
13251 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13252 S/390 or zSeries generated code that uses the stack frame backchain at run
13253 time, not just for debugging purposes. Such code is not call-compatible
13254 with code compiled with @option{-mpacked-stack}. Also, note that the
13255 combination of @option{-mbackchain},
13256 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13257 to build a linux kernel use @option{-msoft-float}.
13259 The default is to not use the packed stack layout.
13262 @itemx -mno-small-exec
13263 @opindex msmall-exec
13264 @opindex mno-small-exec
13265 Generate (or do not generate) code using the @code{bras} instruction
13266 to do subroutine calls.
13267 This only works reliably if the total executable size does not
13268 exceed 64k. The default is to use the @code{basr} instruction instead,
13269 which does not have this limitation.
13275 When @option{-m31} is specified, generate code compliant to the
13276 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13277 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13278 particular to generate 64-bit instructions. For the @samp{s390}
13279 targets, the default is @option{-m31}, while the @samp{s390x}
13280 targets default to @option{-m64}.
13286 When @option{-mzarch} is specified, generate code using the
13287 instructions available on z/Architecture.
13288 When @option{-mesa} is specified, generate code using the
13289 instructions available on ESA/390. Note that @option{-mesa} is
13290 not possible with @option{-m64}.
13291 When generating code compliant to the GNU/Linux for S/390 ABI,
13292 the default is @option{-mesa}. When generating code compliant
13293 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13299 Generate (or do not generate) code using the @code{mvcle} instruction
13300 to perform block moves. When @option{-mno-mvcle} is specified,
13301 use a @code{mvc} loop instead. This is the default unless optimizing for
13308 Print (or do not print) additional debug information when compiling.
13309 The default is to not print debug information.
13311 @item -march=@var{cpu-type}
13313 Generate code that will run on @var{cpu-type}, which is the name of a system
13314 representing a certain processor type. Possible values for
13315 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13316 When generating code using the instructions available on z/Architecture,
13317 the default is @option{-march=z900}. Otherwise, the default is
13318 @option{-march=g5}.
13320 @item -mtune=@var{cpu-type}
13322 Tune to @var{cpu-type} everything applicable about the generated code,
13323 except for the ABI and the set of available instructions.
13324 The list of @var{cpu-type} values is the same as for @option{-march}.
13325 The default is the value used for @option{-march}.
13328 @itemx -mno-tpf-trace
13329 @opindex mtpf-trace
13330 @opindex mno-tpf-trace
13331 Generate code that adds (does not add) in TPF OS specific branches to trace
13332 routines in the operating system. This option is off by default, even
13333 when compiling for the TPF OS@.
13336 @itemx -mno-fused-madd
13337 @opindex mfused-madd
13338 @opindex mno-fused-madd
13339 Generate code that uses (does not use) the floating point multiply and
13340 accumulate instructions. These instructions are generated by default if
13341 hardware floating point is used.
13343 @item -mwarn-framesize=@var{framesize}
13344 @opindex mwarn-framesize
13345 Emit a warning if the current function exceeds the given frame size. Because
13346 this is a compile time check it doesn't need to be a real problem when the program
13347 runs. It is intended to identify functions which most probably cause
13348 a stack overflow. It is useful to be used in an environment with limited stack
13349 size e.g.@: the linux kernel.
13351 @item -mwarn-dynamicstack
13352 @opindex mwarn-dynamicstack
13353 Emit a warning if the function calls alloca or uses dynamically
13354 sized arrays. This is generally a bad idea with a limited stack size.
13356 @item -mstack-guard=@var{stack-guard}
13357 @item -mstack-size=@var{stack-size}
13358 @opindex mstack-guard
13359 @opindex mstack-size
13360 If these options are provided the s390 back end emits additional instructions in
13361 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13362 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13363 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13364 the frame size of the compiled function is chosen.
13365 These options are intended to be used to help debugging stack overflow problems.
13366 The additionally emitted code causes only little overhead and hence can also be
13367 used in production like systems without greater performance degradation. The given
13368 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13369 @var{stack-guard} without exceeding 64k.
13370 In order to be efficient the extra code makes the assumption that the stack starts
13371 at an address aligned to the value given by @var{stack-size}.
13372 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13375 @node Score Options
13376 @subsection Score Options
13377 @cindex Score Options
13379 These options are defined for Score implementations:
13384 Compile code for big endian mode. This is the default.
13388 Compile code for little endian mode.
13392 Disable generate bcnz instruction.
13396 Enable generate unaligned load and store instruction.
13400 Enable the use of multiply-accumulate instructions. Disabled by default.
13404 Specify the SCORE5 as the target architecture.
13408 Specify the SCORE5U of the target architecture.
13412 Specify the SCORE7 as the target architecture. This is the default.
13416 Specify the SCORE7D as the target architecture.
13420 @subsection SH Options
13422 These @samp{-m} options are defined for the SH implementations:
13427 Generate code for the SH1.
13431 Generate code for the SH2.
13434 Generate code for the SH2e.
13438 Generate code for the SH3.
13442 Generate code for the SH3e.
13446 Generate code for the SH4 without a floating-point unit.
13448 @item -m4-single-only
13449 @opindex m4-single-only
13450 Generate code for the SH4 with a floating-point unit that only
13451 supports single-precision arithmetic.
13455 Generate code for the SH4 assuming the floating-point unit is in
13456 single-precision mode by default.
13460 Generate code for the SH4.
13464 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13465 floating-point unit is not used.
13467 @item -m4a-single-only
13468 @opindex m4a-single-only
13469 Generate code for the SH4a, in such a way that no double-precision
13470 floating point operations are used.
13473 @opindex m4a-single
13474 Generate code for the SH4a assuming the floating-point unit is in
13475 single-precision mode by default.
13479 Generate code for the SH4a.
13483 Same as @option{-m4a-nofpu}, except that it implicitly passes
13484 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13485 instructions at the moment.
13489 Compile code for the processor in big endian mode.
13493 Compile code for the processor in little endian mode.
13497 Align doubles at 64-bit boundaries. Note that this changes the calling
13498 conventions, and thus some functions from the standard C library will
13499 not work unless you recompile it first with @option{-mdalign}.
13503 Shorten some address references at link time, when possible; uses the
13504 linker option @option{-relax}.
13508 Use 32-bit offsets in @code{switch} tables. The default is to use
13513 Enable the use of the instruction @code{fmovd}.
13517 Comply with the calling conventions defined by Renesas.
13521 Comply with the calling conventions defined by Renesas.
13525 Comply with the calling conventions defined for GCC before the Renesas
13526 conventions were available. This option is the default for all
13527 targets of the SH toolchain except for @samp{sh-symbianelf}.
13530 @opindex mnomacsave
13531 Mark the @code{MAC} register as call-clobbered, even if
13532 @option{-mhitachi} is given.
13536 Increase IEEE-compliance of floating-point code.
13537 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13538 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13539 comparisons of NANs / infinities incurs extra overhead in every
13540 floating point comparison, therefore the default is set to
13541 @option{-ffinite-math-only}.
13543 @item -minline-ic_invalidate
13544 @opindex minline-ic_invalidate
13545 Inline code to invalidate instruction cache entries after setting up
13546 nested function trampolines.
13547 This option has no effect if -musermode is in effect and the selected
13548 code generation option (e.g. -m4) does not allow the use of the icbi
13550 If the selected code generation option does not allow the use of the icbi
13551 instruction, and -musermode is not in effect, the inlined code will
13552 manipulate the instruction cache address array directly with an associative
13553 write. This not only requires privileged mode, but it will also
13554 fail if the cache line had been mapped via the TLB and has become unmapped.
13558 Dump instruction size and location in the assembly code.
13561 @opindex mpadstruct
13562 This option is deprecated. It pads structures to multiple of 4 bytes,
13563 which is incompatible with the SH ABI@.
13567 Optimize for space instead of speed. Implied by @option{-Os}.
13570 @opindex mprefergot
13571 When generating position-independent code, emit function calls using
13572 the Global Offset Table instead of the Procedure Linkage Table.
13576 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13577 if the inlined code would not work in user mode.
13578 This is the default when the target is @code{sh-*-linux*}.
13580 @item -multcost=@var{number}
13581 @opindex multcost=@var{number}
13582 Set the cost to assume for a multiply insn.
13584 @item -mdiv=@var{strategy}
13585 @opindex mdiv=@var{strategy}
13586 Set the division strategy to use for SHmedia code. @var{strategy} must be
13587 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13588 inv:call2, inv:fp .
13589 "fp" performs the operation in floating point. This has a very high latency,
13590 but needs only a few instructions, so it might be a good choice if
13591 your code has enough easily exploitable ILP to allow the compiler to
13592 schedule the floating point instructions together with other instructions.
13593 Division by zero causes a floating point exception.
13594 "inv" uses integer operations to calculate the inverse of the divisor,
13595 and then multiplies the dividend with the inverse. This strategy allows
13596 cse and hoisting of the inverse calculation. Division by zero calculates
13597 an unspecified result, but does not trap.
13598 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13599 have been found, or if the entire operation has been hoisted to the same
13600 place, the last stages of the inverse calculation are intertwined with the
13601 final multiply to reduce the overall latency, at the expense of using a few
13602 more instructions, and thus offering fewer scheduling opportunities with
13604 "call" calls a library function that usually implements the inv:minlat
13606 This gives high code density for m5-*media-nofpu compilations.
13607 "call2" uses a different entry point of the same library function, where it
13608 assumes that a pointer to a lookup table has already been set up, which
13609 exposes the pointer load to cse / code hoisting optimizations.
13610 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13611 code generation, but if the code stays unoptimized, revert to the "call",
13612 "call2", or "fp" strategies, respectively. Note that the
13613 potentially-trapping side effect of division by zero is carried by a
13614 separate instruction, so it is possible that all the integer instructions
13615 are hoisted out, but the marker for the side effect stays where it is.
13616 A recombination to fp operations or a call is not possible in that case.
13617 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13618 that the inverse calculation was nor separated from the multiply, they speed
13619 up division where the dividend fits into 20 bits (plus sign where applicable),
13620 by inserting a test to skip a number of operations in this case; this test
13621 slows down the case of larger dividends. inv20u assumes the case of a such
13622 a small dividend to be unlikely, and inv20l assumes it to be likely.
13624 @item -mdivsi3_libfunc=@var{name}
13625 @opindex mdivsi3_libfunc=@var{name}
13626 Set the name of the library function used for 32 bit signed division to
13627 @var{name}. This only affect the name used in the call and inv:call
13628 division strategies, and the compiler will still expect the same
13629 sets of input/output/clobbered registers as if this option was not present.
13631 @item -madjust-unroll
13632 @opindex madjust-unroll
13633 Throttle unrolling to avoid thrashing target registers.
13634 This option only has an effect if the gcc code base supports the
13635 TARGET_ADJUST_UNROLL_MAX target hook.
13637 @item -mindexed-addressing
13638 @opindex mindexed-addressing
13639 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13640 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13641 semantics for the indexed addressing mode. The architecture allows the
13642 implementation of processors with 64 bit MMU, which the OS could use to
13643 get 32 bit addressing, but since no current hardware implementation supports
13644 this or any other way to make the indexed addressing mode safe to use in
13645 the 32 bit ABI, the default is -mno-indexed-addressing.
13647 @item -mgettrcost=@var{number}
13648 @opindex mgettrcost=@var{number}
13649 Set the cost assumed for the gettr instruction to @var{number}.
13650 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13654 Assume pt* instructions won't trap. This will generally generate better
13655 scheduled code, but is unsafe on current hardware. The current architecture
13656 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13657 This has the unintentional effect of making it unsafe to schedule ptabs /
13658 ptrel before a branch, or hoist it out of a loop. For example,
13659 __do_global_ctors, a part of libgcc that runs constructors at program
13660 startup, calls functions in a list which is delimited by @minus{}1. With the
13661 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13662 That means that all the constructors will be run a bit quicker, but when
13663 the loop comes to the end of the list, the program crashes because ptabs
13664 loads @minus{}1 into a target register. Since this option is unsafe for any
13665 hardware implementing the current architecture specification, the default
13666 is -mno-pt-fixed. Unless the user specifies a specific cost with
13667 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13668 this deters register allocation using target registers for storing
13671 @item -minvalid-symbols
13672 @opindex minvalid-symbols
13673 Assume symbols might be invalid. Ordinary function symbols generated by
13674 the compiler will always be valid to load with movi/shori/ptabs or
13675 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13676 to generate symbols that will cause ptabs / ptrel to trap.
13677 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13678 It will then prevent cross-basic-block cse, hoisting and most scheduling
13679 of symbol loads. The default is @option{-mno-invalid-symbols}.
13682 @node SPARC Options
13683 @subsection SPARC Options
13684 @cindex SPARC options
13686 These @samp{-m} options are supported on the SPARC:
13689 @item -mno-app-regs
13691 @opindex mno-app-regs
13693 Specify @option{-mapp-regs} to generate output using the global registers
13694 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13697 To be fully SVR4 ABI compliant at the cost of some performance loss,
13698 specify @option{-mno-app-regs}. You should compile libraries and system
13699 software with this option.
13702 @itemx -mhard-float
13704 @opindex mhard-float
13705 Generate output containing floating point instructions. This is the
13709 @itemx -msoft-float
13711 @opindex msoft-float
13712 Generate output containing library calls for floating point.
13713 @strong{Warning:} the requisite libraries are not available for all SPARC
13714 targets. Normally the facilities of the machine's usual C compiler are
13715 used, but this cannot be done directly in cross-compilation. You must make
13716 your own arrangements to provide suitable library functions for
13717 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13718 @samp{sparclite-*-*} do provide software floating point support.
13720 @option{-msoft-float} changes the calling convention in the output file;
13721 therefore, it is only useful if you compile @emph{all} of a program with
13722 this option. In particular, you need to compile @file{libgcc.a}, the
13723 library that comes with GCC, with @option{-msoft-float} in order for
13726 @item -mhard-quad-float
13727 @opindex mhard-quad-float
13728 Generate output containing quad-word (long double) floating point
13731 @item -msoft-quad-float
13732 @opindex msoft-quad-float
13733 Generate output containing library calls for quad-word (long double)
13734 floating point instructions. The functions called are those specified
13735 in the SPARC ABI@. This is the default.
13737 As of this writing, there are no SPARC implementations that have hardware
13738 support for the quad-word floating point instructions. They all invoke
13739 a trap handler for one of these instructions, and then the trap handler
13740 emulates the effect of the instruction. Because of the trap handler overhead,
13741 this is much slower than calling the ABI library routines. Thus the
13742 @option{-msoft-quad-float} option is the default.
13744 @item -mno-unaligned-doubles
13745 @itemx -munaligned-doubles
13746 @opindex mno-unaligned-doubles
13747 @opindex munaligned-doubles
13748 Assume that doubles have 8 byte alignment. This is the default.
13750 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13751 alignment only if they are contained in another type, or if they have an
13752 absolute address. Otherwise, it assumes they have 4 byte alignment.
13753 Specifying this option avoids some rare compatibility problems with code
13754 generated by other compilers. It is not the default because it results
13755 in a performance loss, especially for floating point code.
13757 @item -mno-faster-structs
13758 @itemx -mfaster-structs
13759 @opindex mno-faster-structs
13760 @opindex mfaster-structs
13761 With @option{-mfaster-structs}, the compiler assumes that structures
13762 should have 8 byte alignment. This enables the use of pairs of
13763 @code{ldd} and @code{std} instructions for copies in structure
13764 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13765 However, the use of this changed alignment directly violates the SPARC
13766 ABI@. Thus, it's intended only for use on targets where the developer
13767 acknowledges that their resulting code will not be directly in line with
13768 the rules of the ABI@.
13770 @item -mimpure-text
13771 @opindex mimpure-text
13772 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13773 the compiler to not pass @option{-z text} to the linker when linking a
13774 shared object. Using this option, you can link position-dependent
13775 code into a shared object.
13777 @option{-mimpure-text} suppresses the ``relocations remain against
13778 allocatable but non-writable sections'' linker error message.
13779 However, the necessary relocations will trigger copy-on-write, and the
13780 shared object is not actually shared across processes. Instead of
13781 using @option{-mimpure-text}, you should compile all source code with
13782 @option{-fpic} or @option{-fPIC}.
13784 This option is only available on SunOS and Solaris.
13786 @item -mcpu=@var{cpu_type}
13788 Set the instruction set, register set, and instruction scheduling parameters
13789 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13790 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13791 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13792 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13793 @samp{ultrasparc3}, and @samp{niagara}.
13795 Default instruction scheduling parameters are used for values that select
13796 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13797 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
13799 Here is a list of each supported architecture and their supported
13804 v8: supersparc, hypersparc
13805 sparclite: f930, f934, sparclite86x
13807 v9: ultrasparc, ultrasparc3, niagara
13810 By default (unless configured otherwise), GCC generates code for the V7
13811 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
13812 additionally optimizes it for the Cypress CY7C602 chip, as used in the
13813 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
13814 SPARCStation 1, 2, IPX etc.
13816 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
13817 architecture. The only difference from V7 code is that the compiler emits
13818 the integer multiply and integer divide instructions which exist in SPARC-V8
13819 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
13820 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
13823 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
13824 the SPARC architecture. This adds the integer multiply, integer divide step
13825 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
13826 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
13827 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
13828 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
13829 MB86934 chip, which is the more recent SPARClite with FPU@.
13831 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
13832 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
13833 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
13834 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
13835 optimizes it for the TEMIC SPARClet chip.
13837 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
13838 architecture. This adds 64-bit integer and floating-point move instructions,
13839 3 additional floating-point condition code registers and conditional move
13840 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
13841 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
13842 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
13843 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
13844 @option{-mcpu=niagara}, the compiler additionally optimizes it for
13845 Sun UltraSPARC T1 chips.
13847 @item -mtune=@var{cpu_type}
13849 Set the instruction scheduling parameters for machine type
13850 @var{cpu_type}, but do not set the instruction set or register set that the
13851 option @option{-mcpu=@var{cpu_type}} would.
13853 The same values for @option{-mcpu=@var{cpu_type}} can be used for
13854 @option{-mtune=@var{cpu_type}}, but the only useful values are those
13855 that select a particular cpu implementation. Those are @samp{cypress},
13856 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
13857 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
13858 @samp{ultrasparc3}, and @samp{niagara}.
13863 @opindex mno-v8plus
13864 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
13865 difference from the V8 ABI is that the global and out registers are
13866 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
13867 mode for all SPARC-V9 processors.
13873 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13874 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13877 These @samp{-m} options are supported in addition to the above
13878 on SPARC-V9 processors in 64-bit environments:
13881 @item -mlittle-endian
13882 @opindex mlittle-endian
13883 Generate code for a processor running in little-endian mode. It is only
13884 available for a few configurations and most notably not on Solaris and Linux.
13890 Generate code for a 32-bit or 64-bit environment.
13891 The 32-bit environment sets int, long and pointer to 32 bits.
13892 The 64-bit environment sets int to 32 bits and long and pointer
13895 @item -mcmodel=medlow
13896 @opindex mcmodel=medlow
13897 Generate code for the Medium/Low code model: 64-bit addresses, programs
13898 must be linked in the low 32 bits of memory. Programs can be statically
13899 or dynamically linked.
13901 @item -mcmodel=medmid
13902 @opindex mcmodel=medmid
13903 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13904 must be linked in the low 44 bits of memory, the text and data segments must
13905 be less than 2GB in size and the data segment must be located within 2GB of
13908 @item -mcmodel=medany
13909 @opindex mcmodel=medany
13910 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13911 may be linked anywhere in memory, the text and data segments must be less
13912 than 2GB in size and the data segment must be located within 2GB of the
13915 @item -mcmodel=embmedany
13916 @opindex mcmodel=embmedany
13917 Generate code for the Medium/Anywhere code model for embedded systems:
13918 64-bit addresses, the text and data segments must be less than 2GB in
13919 size, both starting anywhere in memory (determined at link time). The
13920 global register %g4 points to the base of the data segment. Programs
13921 are statically linked and PIC is not supported.
13924 @itemx -mno-stack-bias
13925 @opindex mstack-bias
13926 @opindex mno-stack-bias
13927 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13928 frame pointer if present, are offset by @minus{}2047 which must be added back
13929 when making stack frame references. This is the default in 64-bit mode.
13930 Otherwise, assume no such offset is present.
13933 These switches are supported in addition to the above on Solaris:
13938 Add support for multithreading using the Solaris threads library. This
13939 option sets flags for both the preprocessor and linker. This option does
13940 not affect the thread safety of object code produced by the compiler or
13941 that of libraries supplied with it.
13945 Add support for multithreading using the POSIX threads library. This
13946 option sets flags for both the preprocessor and linker. This option does
13947 not affect the thread safety of object code produced by the compiler or
13948 that of libraries supplied with it.
13952 This is a synonym for @option{-pthreads}.
13956 @subsection SPU Options
13957 @cindex SPU options
13959 These @samp{-m} options are supported on the SPU:
13963 @itemx -merror-reloc
13964 @opindex mwarn-reloc
13965 @opindex merror-reloc
13967 The loader for SPU does not handle dynamic relocations. By default, GCC
13968 will give an error when it generates code that requires a dynamic
13969 relocation. @option{-mno-error-reloc} disables the error,
13970 @option{-mwarn-reloc} will generate a warning instead.
13973 @itemx -munsafe-dma
13975 @opindex munsafe-dma
13977 Instructions which initiate or test completion of DMA must not be
13978 reordered with respect to loads and stores of the memory which is being
13979 accessed. Users typically address this problem using the volatile
13980 keyword, but that can lead to inefficient code in places where the
13981 memory is known to not change. Rather than mark the memory as volatile
13982 we treat the DMA instructions as potentially effecting all memory. With
13983 @option{-munsafe-dma} users must use the volatile keyword to protect
13986 @item -mbranch-hints
13987 @opindex mbranch-hints
13989 By default, GCC will generate a branch hint instruction to avoid
13990 pipeline stalls for always taken or probably taken branches. A hint
13991 will not be generated closer than 8 instructions away from its branch.
13992 There is little reason to disable them, except for debugging purposes,
13993 or to make an object a little bit smaller.
13997 @opindex msmall-mem
13998 @opindex mlarge-mem
14000 By default, GCC generates code assuming that addresses are never larger
14001 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14002 a full 32 bit address.
14007 By default, GCC links against startup code that assumes the SPU-style
14008 main function interface (which has an unconventional parameter list).
14009 With @option{-mstdmain}, GCC will link your program against startup
14010 code that assumes a C99-style interface to @code{main}, including a
14011 local copy of @code{argv} strings.
14013 @item -mfixed-range=@var{register-range}
14014 @opindex mfixed-range
14015 Generate code treating the given register range as fixed registers.
14016 A fixed register is one that the register allocator can not use. This is
14017 useful when compiling kernel code. A register range is specified as
14018 two registers separated by a dash. Multiple register ranges can be
14019 specified separated by a comma.
14023 @node System V Options
14024 @subsection Options for System V
14026 These additional options are available on System V Release 4 for
14027 compatibility with other compilers on those systems:
14032 Create a shared object.
14033 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14037 Identify the versions of each tool used by the compiler, in a
14038 @code{.ident} assembler directive in the output.
14042 Refrain from adding @code{.ident} directives to the output file (this is
14045 @item -YP,@var{dirs}
14047 Search the directories @var{dirs}, and no others, for libraries
14048 specified with @option{-l}.
14050 @item -Ym,@var{dir}
14052 Look in the directory @var{dir} to find the M4 preprocessor.
14053 The assembler uses this option.
14054 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14055 @c the generic assembler that comes with Solaris takes just -Ym.
14058 @node TMS320C3x/C4x Options
14059 @subsection TMS320C3x/C4x Options
14060 @cindex TMS320C3x/C4x Options
14062 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
14066 @item -mcpu=@var{cpu_type}
14068 Set the instruction set, register set, and instruction scheduling
14069 parameters for machine type @var{cpu_type}. Supported values for
14070 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
14071 @samp{c44}. The default is @samp{c40} to generate code for the
14076 @itemx -msmall-memory
14078 @opindex mbig-memory
14080 @opindex msmall-memory
14082 Generates code for the big or small memory model. The small memory
14083 model assumed that all data fits into one 64K word page. At run-time
14084 the data page (DP) register must be set to point to the 64K page
14085 containing the .bss and .data program sections. The big memory model is
14086 the default and requires reloading of the DP register for every direct
14093 Allow (disallow) allocation of general integer operands into the block
14094 count register BK@.
14100 Enable (disable) generation of code using decrement and branch,
14101 DBcond(D), instructions. This is enabled by default for the C4x. To be
14102 on the safe side, this is disabled for the C3x, since the maximum
14103 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
14104 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
14105 that it can utilize the decrement and branch instruction, but will give
14106 up if there is more than one memory reference in the loop. Thus a loop
14107 where the loop counter is decremented can generate slightly more
14108 efficient code, in cases where the RPTB instruction cannot be utilized.
14110 @item -mdp-isr-reload
14112 @opindex mdp-isr-reload
14114 Force the DP register to be saved on entry to an interrupt service
14115 routine (ISR), reloaded to point to the data section, and restored on
14116 exit from the ISR@. This should not be required unless someone has
14117 violated the small memory model by modifying the DP register, say within
14124 For the C3x use the 24-bit MPYI instruction for integer multiplies
14125 instead of a library call to guarantee 32-bit results. Note that if one
14126 of the operands is a constant, then the multiplication will be performed
14127 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
14128 then squaring operations are performed inline instead of a library call.
14131 @itemx -mno-fast-fix
14133 @opindex mno-fast-fix
14134 The C3x/C4x FIX instruction to convert a floating point value to an
14135 integer value chooses the nearest integer less than or equal to the
14136 floating point value rather than to the nearest integer. Thus if the
14137 floating point number is negative, the result will be incorrectly
14138 truncated an additional code is necessary to detect and correct this
14139 case. This option can be used to disable generation of the additional
14140 code required to correct the result.
14146 Enable (disable) generation of repeat block sequences using the RPTB
14147 instruction for zero overhead looping. The RPTB construct is only used
14148 for innermost loops that do not call functions or jump across the loop
14149 boundaries. There is no advantage having nested RPTB loops due to the
14150 overhead required to save and restore the RC, RS, and RE registers.
14151 This is enabled by default with @option{-O2}.
14153 @item -mrpts=@var{count}
14157 Enable (disable) the use of the single instruction repeat instruction
14158 RPTS@. If a repeat block contains a single instruction, and the loop
14159 count can be guaranteed to be less than the value @var{count}, GCC will
14160 emit a RPTS instruction instead of a RPTB@. If no value is specified,
14161 then a RPTS will be emitted even if the loop count cannot be determined
14162 at compile time. Note that the repeated instruction following RPTS does
14163 not have to be reloaded from memory each iteration, thus freeing up the
14164 CPU buses for operands. However, since interrupts are blocked by this
14165 instruction, it is disabled by default.
14167 @item -mloop-unsigned
14168 @itemx -mno-loop-unsigned
14169 @opindex mloop-unsigned
14170 @opindex mno-loop-unsigned
14171 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
14172 is @math{2^{31} + 1} since these instructions test if the iteration count is
14173 negative to terminate the loop. If the iteration count is unsigned
14174 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
14175 exceeded. This switch allows an unsigned iteration count.
14179 Try to emit an assembler syntax that the TI assembler (asm30) is happy
14180 with. This also enforces compatibility with the API employed by the TI
14181 C3x C compiler. For example, long doubles are passed as structures
14182 rather than in floating point registers.
14188 Generate code that uses registers (stack) for passing arguments to functions.
14189 By default, arguments are passed in registers where possible rather
14190 than by pushing arguments on to the stack.
14192 @item -mparallel-insns
14193 @itemx -mno-parallel-insns
14194 @opindex mparallel-insns
14195 @opindex mno-parallel-insns
14196 Allow the generation of parallel instructions. This is enabled by
14197 default with @option{-O2}.
14199 @item -mparallel-mpy
14200 @itemx -mno-parallel-mpy
14201 @opindex mparallel-mpy
14202 @opindex mno-parallel-mpy
14203 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
14204 provided @option{-mparallel-insns} is also specified. These instructions have
14205 tight register constraints which can pessimize the code generation
14206 of large functions.
14211 @subsection V850 Options
14212 @cindex V850 Options
14214 These @samp{-m} options are defined for V850 implementations:
14218 @itemx -mno-long-calls
14219 @opindex mlong-calls
14220 @opindex mno-long-calls
14221 Treat all calls as being far away (near). If calls are assumed to be
14222 far away, the compiler will always load the functions address up into a
14223 register, and call indirect through the pointer.
14229 Do not optimize (do optimize) basic blocks that use the same index
14230 pointer 4 or more times to copy pointer into the @code{ep} register, and
14231 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14232 option is on by default if you optimize.
14234 @item -mno-prolog-function
14235 @itemx -mprolog-function
14236 @opindex mno-prolog-function
14237 @opindex mprolog-function
14238 Do not use (do use) external functions to save and restore registers
14239 at the prologue and epilogue of a function. The external functions
14240 are slower, but use less code space if more than one function saves
14241 the same number of registers. The @option{-mprolog-function} option
14242 is on by default if you optimize.
14246 Try to make the code as small as possible. At present, this just turns
14247 on the @option{-mep} and @option{-mprolog-function} options.
14249 @item -mtda=@var{n}
14251 Put static or global variables whose size is @var{n} bytes or less into
14252 the tiny data area that register @code{ep} points to. The tiny data
14253 area can hold up to 256 bytes in total (128 bytes for byte references).
14255 @item -msda=@var{n}
14257 Put static or global variables whose size is @var{n} bytes or less into
14258 the small data area that register @code{gp} points to. The small data
14259 area can hold up to 64 kilobytes.
14261 @item -mzda=@var{n}
14263 Put static or global variables whose size is @var{n} bytes or less into
14264 the first 32 kilobytes of memory.
14268 Specify that the target processor is the V850.
14271 @opindex mbig-switch
14272 Generate code suitable for big switch tables. Use this option only if
14273 the assembler/linker complain about out of range branches within a switch
14278 This option will cause r2 and r5 to be used in the code generated by
14279 the compiler. This setting is the default.
14281 @item -mno-app-regs
14282 @opindex mno-app-regs
14283 This option will cause r2 and r5 to be treated as fixed registers.
14287 Specify that the target processor is the V850E1. The preprocessor
14288 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14289 this option is used.
14293 Specify that the target processor is the V850E@. The preprocessor
14294 constant @samp{__v850e__} will be defined if this option is used.
14296 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14297 are defined then a default target processor will be chosen and the
14298 relevant @samp{__v850*__} preprocessor constant will be defined.
14300 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14301 defined, regardless of which processor variant is the target.
14303 @item -mdisable-callt
14304 @opindex mdisable-callt
14305 This option will suppress generation of the CALLT instruction for the
14306 v850e and v850e1 flavors of the v850 architecture. The default is
14307 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14312 @subsection VAX Options
14313 @cindex VAX options
14315 These @samp{-m} options are defined for the VAX:
14320 Do not output certain jump instructions (@code{aobleq} and so on)
14321 that the Unix assembler for the VAX cannot handle across long
14326 Do output those jump instructions, on the assumption that you
14327 will assemble with the GNU assembler.
14331 Output code for g-format floating point numbers instead of d-format.
14334 @node VxWorks Options
14335 @subsection VxWorks Options
14336 @cindex VxWorks Options
14338 The options in this section are defined for all VxWorks targets.
14339 Options specific to the target hardware are listed with the other
14340 options for that target.
14345 GCC can generate code for both VxWorks kernels and real time processes
14346 (RTPs). This option switches from the former to the latter. It also
14347 defines the preprocessor macro @code{__RTP__}.
14350 @opindex non-static
14351 Link an RTP executable against shared libraries rather than static
14352 libraries. The options @option{-static} and @option{-shared} can
14353 also be used for RTPs (@pxref{Link Options}); @option{-static}
14360 These options are passed down to the linker. They are defined for
14361 compatibility with Diab.
14364 @opindex Xbind-lazy
14365 Enable lazy binding of function calls. This option is equivalent to
14366 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14370 Disable lazy binding of function calls. This option is the default and
14371 is defined for compatibility with Diab.
14374 @node x86-64 Options
14375 @subsection x86-64 Options
14376 @cindex x86-64 options
14378 These are listed under @xref{i386 and x86-64 Options}.
14380 @node Xstormy16 Options
14381 @subsection Xstormy16 Options
14382 @cindex Xstormy16 Options
14384 These options are defined for Xstormy16:
14389 Choose startup files and linker script suitable for the simulator.
14392 @node Xtensa Options
14393 @subsection Xtensa Options
14394 @cindex Xtensa Options
14396 These options are supported for Xtensa targets:
14400 @itemx -mno-const16
14402 @opindex mno-const16
14403 Enable or disable use of @code{CONST16} instructions for loading
14404 constant values. The @code{CONST16} instruction is currently not a
14405 standard option from Tensilica. When enabled, @code{CONST16}
14406 instructions are always used in place of the standard @code{L32R}
14407 instructions. The use of @code{CONST16} is enabled by default only if
14408 the @code{L32R} instruction is not available.
14411 @itemx -mno-fused-madd
14412 @opindex mfused-madd
14413 @opindex mno-fused-madd
14414 Enable or disable use of fused multiply/add and multiply/subtract
14415 instructions in the floating-point option. This has no effect if the
14416 floating-point option is not also enabled. Disabling fused multiply/add
14417 and multiply/subtract instructions forces the compiler to use separate
14418 instructions for the multiply and add/subtract operations. This may be
14419 desirable in some cases where strict IEEE 754-compliant results are
14420 required: the fused multiply add/subtract instructions do not round the
14421 intermediate result, thereby producing results with @emph{more} bits of
14422 precision than specified by the IEEE standard. Disabling fused multiply
14423 add/subtract instructions also ensures that the program output is not
14424 sensitive to the compiler's ability to combine multiply and add/subtract
14427 @item -mtext-section-literals
14428 @itemx -mno-text-section-literals
14429 @opindex mtext-section-literals
14430 @opindex mno-text-section-literals
14431 Control the treatment of literal pools. The default is
14432 @option{-mno-text-section-literals}, which places literals in a separate
14433 section in the output file. This allows the literal pool to be placed
14434 in a data RAM/ROM, and it also allows the linker to combine literal
14435 pools from separate object files to remove redundant literals and
14436 improve code size. With @option{-mtext-section-literals}, the literals
14437 are interspersed in the text section in order to keep them as close as
14438 possible to their references. This may be necessary for large assembly
14441 @item -mtarget-align
14442 @itemx -mno-target-align
14443 @opindex mtarget-align
14444 @opindex mno-target-align
14445 When this option is enabled, GCC instructs the assembler to
14446 automatically align instructions to reduce branch penalties at the
14447 expense of some code density. The assembler attempts to widen density
14448 instructions to align branch targets and the instructions following call
14449 instructions. If there are not enough preceding safe density
14450 instructions to align a target, no widening will be performed. The
14451 default is @option{-mtarget-align}. These options do not affect the
14452 treatment of auto-aligned instructions like @code{LOOP}, which the
14453 assembler will always align, either by widening density instructions or
14454 by inserting no-op instructions.
14457 @itemx -mno-longcalls
14458 @opindex mlongcalls
14459 @opindex mno-longcalls
14460 When this option is enabled, GCC instructs the assembler to translate
14461 direct calls to indirect calls unless it can determine that the target
14462 of a direct call is in the range allowed by the call instruction. This
14463 translation typically occurs for calls to functions in other source
14464 files. Specifically, the assembler translates a direct @code{CALL}
14465 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14466 The default is @option{-mno-longcalls}. This option should be used in
14467 programs where the call target can potentially be out of range. This
14468 option is implemented in the assembler, not the compiler, so the
14469 assembly code generated by GCC will still show direct call
14470 instructions---look at the disassembled object code to see the actual
14471 instructions. Note that the assembler will use an indirect call for
14472 every cross-file call, not just those that really will be out of range.
14475 @node zSeries Options
14476 @subsection zSeries Options
14477 @cindex zSeries options
14479 These are listed under @xref{S/390 and zSeries Options}.
14481 @node Code Gen Options
14482 @section Options for Code Generation Conventions
14483 @cindex code generation conventions
14484 @cindex options, code generation
14485 @cindex run-time options
14487 These machine-independent options control the interface conventions
14488 used in code generation.
14490 Most of them have both positive and negative forms; the negative form
14491 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14492 one of the forms is listed---the one which is not the default. You
14493 can figure out the other form by either removing @samp{no-} or adding
14497 @item -fbounds-check
14498 @opindex fbounds-check
14499 For front-ends that support it, generate additional code to check that
14500 indices used to access arrays are within the declared range. This is
14501 currently only supported by the Java and Fortran front-ends, where
14502 this option defaults to true and false respectively.
14506 This option generates traps for signed overflow on addition, subtraction,
14507 multiplication operations.
14511 This option instructs the compiler to assume that signed arithmetic
14512 overflow of addition, subtraction and multiplication wraps around
14513 using twos-complement representation. This flag enables some optimizations
14514 and disables others. This option is enabled by default for the Java
14515 front-end, as required by the Java language specification.
14518 @opindex fexceptions
14519 Enable exception handling. Generates extra code needed to propagate
14520 exceptions. For some targets, this implies GCC will generate frame
14521 unwind information for all functions, which can produce significant data
14522 size overhead, although it does not affect execution. If you do not
14523 specify this option, GCC will enable it by default for languages like
14524 C++ which normally require exception handling, and disable it for
14525 languages like C that do not normally require it. However, you may need
14526 to enable this option when compiling C code that needs to interoperate
14527 properly with exception handlers written in C++. You may also wish to
14528 disable this option if you are compiling older C++ programs that don't
14529 use exception handling.
14531 @item -fnon-call-exceptions
14532 @opindex fnon-call-exceptions
14533 Generate code that allows trapping instructions to throw exceptions.
14534 Note that this requires platform-specific runtime support that does
14535 not exist everywhere. Moreover, it only allows @emph{trapping}
14536 instructions to throw exceptions, i.e.@: memory references or floating
14537 point instructions. It does not allow exceptions to be thrown from
14538 arbitrary signal handlers such as @code{SIGALRM}.
14540 @item -funwind-tables
14541 @opindex funwind-tables
14542 Similar to @option{-fexceptions}, except that it will just generate any needed
14543 static data, but will not affect the generated code in any other way.
14544 You will normally not enable this option; instead, a language processor
14545 that needs this handling would enable it on your behalf.
14547 @item -fasynchronous-unwind-tables
14548 @opindex fasynchronous-unwind-tables
14549 Generate unwind table in dwarf2 format, if supported by target machine. The
14550 table is exact at each instruction boundary, so it can be used for stack
14551 unwinding from asynchronous events (such as debugger or garbage collector).
14553 @item -fpcc-struct-return
14554 @opindex fpcc-struct-return
14555 Return ``short'' @code{struct} and @code{union} values in memory like
14556 longer ones, rather than in registers. This convention is less
14557 efficient, but it has the advantage of allowing intercallability between
14558 GCC-compiled files and files compiled with other compilers, particularly
14559 the Portable C Compiler (pcc).
14561 The precise convention for returning structures in memory depends
14562 on the target configuration macros.
14564 Short structures and unions are those whose size and alignment match
14565 that of some integer type.
14567 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14568 switch is not binary compatible with code compiled with the
14569 @option{-freg-struct-return} switch.
14570 Use it to conform to a non-default application binary interface.
14572 @item -freg-struct-return
14573 @opindex freg-struct-return
14574 Return @code{struct} and @code{union} values in registers when possible.
14575 This is more efficient for small structures than
14576 @option{-fpcc-struct-return}.
14578 If you specify neither @option{-fpcc-struct-return} nor
14579 @option{-freg-struct-return}, GCC defaults to whichever convention is
14580 standard for the target. If there is no standard convention, GCC
14581 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14582 the principal compiler. In those cases, we can choose the standard, and
14583 we chose the more efficient register return alternative.
14585 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14586 switch is not binary compatible with code compiled with the
14587 @option{-fpcc-struct-return} switch.
14588 Use it to conform to a non-default application binary interface.
14590 @item -fshort-enums
14591 @opindex fshort-enums
14592 Allocate to an @code{enum} type only as many bytes as it needs for the
14593 declared range of possible values. Specifically, the @code{enum} type
14594 will be equivalent to the smallest integer type which has enough room.
14596 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14597 code that is not binary compatible with code generated without that switch.
14598 Use it to conform to a non-default application binary interface.
14600 @item -fshort-double
14601 @opindex fshort-double
14602 Use the same size for @code{double} as for @code{float}.
14604 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14605 code that is not binary compatible with code generated without that switch.
14606 Use it to conform to a non-default application binary interface.
14608 @item -fshort-wchar
14609 @opindex fshort-wchar
14610 Override the underlying type for @samp{wchar_t} to be @samp{short
14611 unsigned int} instead of the default for the target. This option is
14612 useful for building programs to run under WINE@.
14614 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14615 code that is not binary compatible with code generated without that switch.
14616 Use it to conform to a non-default application binary interface.
14619 @opindex fno-common
14620 In C, allocate even uninitialized global variables in the data section of the
14621 object file, rather than generating them as common blocks. This has the
14622 effect that if the same variable is declared (without @code{extern}) in
14623 two different compilations, you will get an error when you link them.
14624 The only reason this might be useful is if you wish to verify that the
14625 program will work on other systems which always work this way.
14629 Ignore the @samp{#ident} directive.
14631 @item -finhibit-size-directive
14632 @opindex finhibit-size-directive
14633 Don't output a @code{.size} assembler directive, or anything else that
14634 would cause trouble if the function is split in the middle, and the
14635 two halves are placed at locations far apart in memory. This option is
14636 used when compiling @file{crtstuff.c}; you should not need to use it
14639 @item -fverbose-asm
14640 @opindex fverbose-asm
14641 Put extra commentary information in the generated assembly code to
14642 make it more readable. This option is generally only of use to those
14643 who actually need to read the generated assembly code (perhaps while
14644 debugging the compiler itself).
14646 @option{-fno-verbose-asm}, the default, causes the
14647 extra information to be omitted and is useful when comparing two assembler
14650 @item -frecord-gcc-switches
14651 @opindex frecord-gcc-switches
14652 This switch causes the command line that was used to invoke the
14653 compiler to be recorded into the object file that is being created.
14654 This switch is only implemented on some targets and the exact format
14655 of the recording is target and binary file format dependent, but it
14656 usually takes the form of a section containing ASCII text. This
14657 switch is related to the @option{-fverbose-asm} switch, but that
14658 switch only records information in the assembler output file as
14659 comments, so it never reaches the object file.
14663 @cindex global offset table
14665 Generate position-independent code (PIC) suitable for use in a shared
14666 library, if supported for the target machine. Such code accesses all
14667 constant addresses through a global offset table (GOT)@. The dynamic
14668 loader resolves the GOT entries when the program starts (the dynamic
14669 loader is not part of GCC; it is part of the operating system). If
14670 the GOT size for the linked executable exceeds a machine-specific
14671 maximum size, you get an error message from the linker indicating that
14672 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14673 instead. (These maximums are 8k on the SPARC and 32k
14674 on the m68k and RS/6000. The 386 has no such limit.)
14676 Position-independent code requires special support, and therefore works
14677 only on certain machines. For the 386, GCC supports PIC for System V
14678 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14679 position-independent.
14681 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14686 If supported for the target machine, emit position-independent code,
14687 suitable for dynamic linking and avoiding any limit on the size of the
14688 global offset table. This option makes a difference on the m68k,
14689 PowerPC and SPARC@.
14691 Position-independent code requires special support, and therefore works
14692 only on certain machines.
14694 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14701 These options are similar to @option{-fpic} and @option{-fPIC}, but
14702 generated position independent code can be only linked into executables.
14703 Usually these options are used when @option{-pie} GCC option will be
14704 used during linking.
14706 @option{-fpie} and @option{-fPIE} both define the macros
14707 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14708 for @option{-fpie} and 2 for @option{-fPIE}.
14710 @item -fno-jump-tables
14711 @opindex fno-jump-tables
14712 Do not use jump tables for switch statements even where it would be
14713 more efficient than other code generation strategies. This option is
14714 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14715 building code which forms part of a dynamic linker and cannot
14716 reference the address of a jump table. On some targets, jump tables
14717 do not require a GOT and this option is not needed.
14719 @item -ffixed-@var{reg}
14721 Treat the register named @var{reg} as a fixed register; generated code
14722 should never refer to it (except perhaps as a stack pointer, frame
14723 pointer or in some other fixed role).
14725 @var{reg} must be the name of a register. The register names accepted
14726 are machine-specific and are defined in the @code{REGISTER_NAMES}
14727 macro in the machine description macro file.
14729 This flag does not have a negative form, because it specifies a
14732 @item -fcall-used-@var{reg}
14733 @opindex fcall-used
14734 Treat the register named @var{reg} as an allocable register that is
14735 clobbered by function calls. It may be allocated for temporaries or
14736 variables that do not live across a call. Functions compiled this way
14737 will not save and restore the register @var{reg}.
14739 It is an error to used this flag with the frame pointer or stack pointer.
14740 Use of this flag for other registers that have fixed pervasive roles in
14741 the machine's execution model will produce disastrous results.
14743 This flag does not have a negative form, because it specifies a
14746 @item -fcall-saved-@var{reg}
14747 @opindex fcall-saved
14748 Treat the register named @var{reg} as an allocable register saved by
14749 functions. It may be allocated even for temporaries or variables that
14750 live across a call. Functions compiled this way will save and restore
14751 the register @var{reg} if they use it.
14753 It is an error to used this flag with the frame pointer or stack pointer.
14754 Use of this flag for other registers that have fixed pervasive roles in
14755 the machine's execution model will produce disastrous results.
14757 A different sort of disaster will result from the use of this flag for
14758 a register in which function values may be returned.
14760 This flag does not have a negative form, because it specifies a
14763 @item -fpack-struct[=@var{n}]
14764 @opindex fpack-struct
14765 Without a value specified, pack all structure members together without
14766 holes. When a value is specified (which must be a small power of two), pack
14767 structure members according to this value, representing the maximum
14768 alignment (that is, objects with default alignment requirements larger than
14769 this will be output potentially unaligned at the next fitting location.
14771 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14772 code that is not binary compatible with code generated without that switch.
14773 Additionally, it makes the code suboptimal.
14774 Use it to conform to a non-default application binary interface.
14776 @item -finstrument-functions
14777 @opindex finstrument-functions
14778 Generate instrumentation calls for entry and exit to functions. Just
14779 after function entry and just before function exit, the following
14780 profiling functions will be called with the address of the current
14781 function and its call site. (On some platforms,
14782 @code{__builtin_return_address} does not work beyond the current
14783 function, so the call site information may not be available to the
14784 profiling functions otherwise.)
14787 void __cyg_profile_func_enter (void *this_fn,
14789 void __cyg_profile_func_exit (void *this_fn,
14793 The first argument is the address of the start of the current function,
14794 which may be looked up exactly in the symbol table.
14796 This instrumentation is also done for functions expanded inline in other
14797 functions. The profiling calls will indicate where, conceptually, the
14798 inline function is entered and exited. This means that addressable
14799 versions of such functions must be available. If all your uses of a
14800 function are expanded inline, this may mean an additional expansion of
14801 code size. If you use @samp{extern inline} in your C code, an
14802 addressable version of such functions must be provided. (This is
14803 normally the case anyways, but if you get lucky and the optimizer always
14804 expands the functions inline, you might have gotten away without
14805 providing static copies.)
14807 A function may be given the attribute @code{no_instrument_function}, in
14808 which case this instrumentation will not be done. This can be used, for
14809 example, for the profiling functions listed above, high-priority
14810 interrupt routines, and any functions from which the profiling functions
14811 cannot safely be called (perhaps signal handlers, if the profiling
14812 routines generate output or allocate memory).
14814 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
14815 @opindex finstrument-functions-exclude-file-list
14817 Set the list of functions that are excluded from instrumentation (see
14818 the description of @code{-finstrument-functions}). If the file that
14819 contains a function definition matches with one of @var{file}, then
14820 that function is not instrumented. The match is done on substrings:
14821 if the @var{file} parameter is a substring of the file name, it is
14822 considered to be a match.
14825 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
14826 will exclude any inline function defined in files whose pathnames
14827 contain @code{/bits/stl} or @code{include/sys}.
14829 If, for some reason, you want to include letter @code{','} in one of
14830 @var{sym}, write @code{'\,'}. For example,
14831 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
14832 (note the single quote surrounding the option).
14834 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
14835 @opindex finstrument-functions-exclude-function-list
14837 This is similar to @code{-finstrument-functions-exclude-file-list},
14838 but this option sets the list of function names to be excluded from
14839 instrumentation. The function name to be matched is its user-visible
14840 name, such as @code{vector<int> blah(const vector<int> &)}, not the
14841 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
14842 match is done on substrings: if the @var{sym} parameter is a substring
14843 of the function name, it is considered to be a match.
14845 @item -fstack-check
14846 @opindex fstack-check
14847 Generate code to verify that you do not go beyond the boundary of the
14848 stack. You should specify this flag if you are running in an
14849 environment with multiple threads, but only rarely need to specify it in
14850 a single-threaded environment since stack overflow is automatically
14851 detected on nearly all systems if there is only one stack.
14853 Note that this switch does not actually cause checking to be done; the
14854 operating system must do that. The switch causes generation of code
14855 to ensure that the operating system sees the stack being extended.
14857 @item -fstack-limit-register=@var{reg}
14858 @itemx -fstack-limit-symbol=@var{sym}
14859 @itemx -fno-stack-limit
14860 @opindex fstack-limit-register
14861 @opindex fstack-limit-symbol
14862 @opindex fno-stack-limit
14863 Generate code to ensure that the stack does not grow beyond a certain value,
14864 either the value of a register or the address of a symbol. If the stack
14865 would grow beyond the value, a signal is raised. For most targets,
14866 the signal is raised before the stack overruns the boundary, so
14867 it is possible to catch the signal without taking special precautions.
14869 For instance, if the stack starts at absolute address @samp{0x80000000}
14870 and grows downwards, you can use the flags
14871 @option{-fstack-limit-symbol=__stack_limit} and
14872 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14873 of 128KB@. Note that this may only work with the GNU linker.
14875 @cindex aliasing of parameters
14876 @cindex parameters, aliased
14877 @item -fargument-alias
14878 @itemx -fargument-noalias
14879 @itemx -fargument-noalias-global
14880 @itemx -fargument-noalias-anything
14881 @opindex fargument-alias
14882 @opindex fargument-noalias
14883 @opindex fargument-noalias-global
14884 @opindex fargument-noalias-anything
14885 Specify the possible relationships among parameters and between
14886 parameters and global data.
14888 @option{-fargument-alias} specifies that arguments (parameters) may
14889 alias each other and may alias global storage.@*
14890 @option{-fargument-noalias} specifies that arguments do not alias
14891 each other, but may alias global storage.@*
14892 @option{-fargument-noalias-global} specifies that arguments do not
14893 alias each other and do not alias global storage.
14894 @option{-fargument-noalias-anything} specifies that arguments do not
14895 alias any other storage.
14897 Each language will automatically use whatever option is required by
14898 the language standard. You should not need to use these options yourself.
14900 @item -fleading-underscore
14901 @opindex fleading-underscore
14902 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14903 change the way C symbols are represented in the object file. One use
14904 is to help link with legacy assembly code.
14906 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14907 generate code that is not binary compatible with code generated without that
14908 switch. Use it to conform to a non-default application binary interface.
14909 Not all targets provide complete support for this switch.
14911 @item -ftls-model=@var{model}
14912 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14913 The @var{model} argument should be one of @code{global-dynamic},
14914 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
14916 The default without @option{-fpic} is @code{initial-exec}; with
14917 @option{-fpic} the default is @code{global-dynamic}.
14919 @item -fvisibility=@var{default|internal|hidden|protected}
14920 @opindex fvisibility
14921 Set the default ELF image symbol visibility to the specified option---all
14922 symbols will be marked with this unless overridden within the code.
14923 Using this feature can very substantially improve linking and
14924 load times of shared object libraries, produce more optimized
14925 code, provide near-perfect API export and prevent symbol clashes.
14926 It is @strong{strongly} recommended that you use this in any shared objects
14929 Despite the nomenclature, @code{default} always means public ie;
14930 available to be linked against from outside the shared object.
14931 @code{protected} and @code{internal} are pretty useless in real-world
14932 usage so the only other commonly used option will be @code{hidden}.
14933 The default if @option{-fvisibility} isn't specified is
14934 @code{default}, i.e., make every
14935 symbol public---this causes the same behavior as previous versions of
14938 A good explanation of the benefits offered by ensuring ELF
14939 symbols have the correct visibility is given by ``How To Write
14940 Shared Libraries'' by Ulrich Drepper (which can be found at
14941 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
14942 solution made possible by this option to marking things hidden when
14943 the default is public is to make the default hidden and mark things
14944 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
14945 and @code{__attribute__ ((visibility("default")))} instead of
14946 @code{__declspec(dllexport)} you get almost identical semantics with
14947 identical syntax. This is a great boon to those working with
14948 cross-platform projects.
14950 For those adding visibility support to existing code, you may find
14951 @samp{#pragma GCC visibility} of use. This works by you enclosing
14952 the declarations you wish to set visibility for with (for example)
14953 @samp{#pragma GCC visibility push(hidden)} and
14954 @samp{#pragma GCC visibility pop}.
14955 Bear in mind that symbol visibility should be viewed @strong{as
14956 part of the API interface contract} and thus all new code should
14957 always specify visibility when it is not the default ie; declarations
14958 only for use within the local DSO should @strong{always} be marked explicitly
14959 as hidden as so to avoid PLT indirection overheads---making this
14960 abundantly clear also aids readability and self-documentation of the code.
14961 Note that due to ISO C++ specification requirements, operator new and
14962 operator delete must always be of default visibility.
14964 Be aware that headers from outside your project, in particular system
14965 headers and headers from any other library you use, may not be
14966 expecting to be compiled with visibility other than the default. You
14967 may need to explicitly say @samp{#pragma GCC visibility push(default)}
14968 before including any such headers.
14970 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
14971 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
14972 no modifications. However, this means that calls to @samp{extern}
14973 functions with no explicit visibility will use the PLT, so it is more
14974 effective to use @samp{__attribute ((visibility))} and/or
14975 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
14976 declarations should be treated as hidden.
14978 Note that @samp{-fvisibility} does affect C++ vague linkage
14979 entities. This means that, for instance, an exception class that will
14980 be thrown between DSOs must be explicitly marked with default
14981 visibility so that the @samp{type_info} nodes will be unified between
14984 An overview of these techniques, their benefits and how to use them
14985 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
14991 @node Environment Variables
14992 @section Environment Variables Affecting GCC
14993 @cindex environment variables
14995 @c man begin ENVIRONMENT
14996 This section describes several environment variables that affect how GCC
14997 operates. Some of them work by specifying directories or prefixes to use
14998 when searching for various kinds of files. Some are used to specify other
14999 aspects of the compilation environment.
15001 Note that you can also specify places to search using options such as
15002 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15003 take precedence over places specified using environment variables, which
15004 in turn take precedence over those specified by the configuration of GCC@.
15005 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15006 GNU Compiler Collection (GCC) Internals}.
15011 @c @itemx LC_COLLATE
15013 @c @itemx LC_MONETARY
15014 @c @itemx LC_NUMERIC
15019 @c @findex LC_COLLATE
15020 @findex LC_MESSAGES
15021 @c @findex LC_MONETARY
15022 @c @findex LC_NUMERIC
15026 These environment variables control the way that GCC uses
15027 localization information that allow GCC to work with different
15028 national conventions. GCC inspects the locale categories
15029 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15030 so. These locale categories can be set to any value supported by your
15031 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15032 Kingdom encoded in UTF-8.
15034 The @env{LC_CTYPE} environment variable specifies character
15035 classification. GCC uses it to determine the character boundaries in
15036 a string; this is needed for some multibyte encodings that contain quote
15037 and escape characters that would otherwise be interpreted as a string
15040 The @env{LC_MESSAGES} environment variable specifies the language to
15041 use in diagnostic messages.
15043 If the @env{LC_ALL} environment variable is set, it overrides the value
15044 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15045 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15046 environment variable. If none of these variables are set, GCC
15047 defaults to traditional C English behavior.
15051 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15052 files. GCC uses temporary files to hold the output of one stage of
15053 compilation which is to be used as input to the next stage: for example,
15054 the output of the preprocessor, which is the input to the compiler
15057 @item GCC_EXEC_PREFIX
15058 @findex GCC_EXEC_PREFIX
15059 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15060 names of the subprograms executed by the compiler. No slash is added
15061 when this prefix is combined with the name of a subprogram, but you can
15062 specify a prefix that ends with a slash if you wish.
15064 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15065 an appropriate prefix to use based on the pathname it was invoked with.
15067 If GCC cannot find the subprogram using the specified prefix, it
15068 tries looking in the usual places for the subprogram.
15070 The default value of @env{GCC_EXEC_PREFIX} is
15071 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15072 the installed compiler. In many cases @var{prefix} is the value
15073 of @code{prefix} when you ran the @file{configure} script.
15075 Other prefixes specified with @option{-B} take precedence over this prefix.
15077 This prefix is also used for finding files such as @file{crt0.o} that are
15080 In addition, the prefix is used in an unusual way in finding the
15081 directories to search for header files. For each of the standard
15082 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15083 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15084 replacing that beginning with the specified prefix to produce an
15085 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15086 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15087 These alternate directories are searched first; the standard directories
15088 come next. If a standard directory begins with the configured
15089 @var{prefix} then the value of @var{prefix} is replaced by
15090 @env{GCC_EXEC_PREFIX} when looking for header files.
15092 @item COMPILER_PATH
15093 @findex COMPILER_PATH
15094 The value of @env{COMPILER_PATH} is a colon-separated list of
15095 directories, much like @env{PATH}. GCC tries the directories thus
15096 specified when searching for subprograms, if it can't find the
15097 subprograms using @env{GCC_EXEC_PREFIX}.
15100 @findex LIBRARY_PATH
15101 The value of @env{LIBRARY_PATH} is a colon-separated list of
15102 directories, much like @env{PATH}. When configured as a native compiler,
15103 GCC tries the directories thus specified when searching for special
15104 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15105 using GCC also uses these directories when searching for ordinary
15106 libraries for the @option{-l} option (but directories specified with
15107 @option{-L} come first).
15111 @cindex locale definition
15112 This variable is used to pass locale information to the compiler. One way in
15113 which this information is used is to determine the character set to be used
15114 when character literals, string literals and comments are parsed in C and C++.
15115 When the compiler is configured to allow multibyte characters,
15116 the following values for @env{LANG} are recognized:
15120 Recognize JIS characters.
15122 Recognize SJIS characters.
15124 Recognize EUCJP characters.
15127 If @env{LANG} is not defined, or if it has some other value, then the
15128 compiler will use mblen and mbtowc as defined by the default locale to
15129 recognize and translate multibyte characters.
15133 Some additional environments variables affect the behavior of the
15136 @include cppenv.texi
15140 @node Precompiled Headers
15141 @section Using Precompiled Headers
15142 @cindex precompiled headers
15143 @cindex speed of compilation
15145 Often large projects have many header files that are included in every
15146 source file. The time the compiler takes to process these header files
15147 over and over again can account for nearly all of the time required to
15148 build the project. To make builds faster, GCC allows users to
15149 `precompile' a header file; then, if builds can use the precompiled
15150 header file they will be much faster.
15152 To create a precompiled header file, simply compile it as you would any
15153 other file, if necessary using the @option{-x} option to make the driver
15154 treat it as a C or C++ header file. You will probably want to use a
15155 tool like @command{make} to keep the precompiled header up-to-date when
15156 the headers it contains change.
15158 A precompiled header file will be searched for when @code{#include} is
15159 seen in the compilation. As it searches for the included file
15160 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15161 compiler looks for a precompiled header in each directory just before it
15162 looks for the include file in that directory. The name searched for is
15163 the name specified in the @code{#include} with @samp{.gch} appended. If
15164 the precompiled header file can't be used, it is ignored.
15166 For instance, if you have @code{#include "all.h"}, and you have
15167 @file{all.h.gch} in the same directory as @file{all.h}, then the
15168 precompiled header file will be used if possible, and the original
15169 header will be used otherwise.
15171 Alternatively, you might decide to put the precompiled header file in a
15172 directory and use @option{-I} to ensure that directory is searched
15173 before (or instead of) the directory containing the original header.
15174 Then, if you want to check that the precompiled header file is always
15175 used, you can put a file of the same name as the original header in this
15176 directory containing an @code{#error} command.
15178 This also works with @option{-include}. So yet another way to use
15179 precompiled headers, good for projects not designed with precompiled
15180 header files in mind, is to simply take most of the header files used by
15181 a project, include them from another header file, precompile that header
15182 file, and @option{-include} the precompiled header. If the header files
15183 have guards against multiple inclusion, they will be skipped because
15184 they've already been included (in the precompiled header).
15186 If you need to precompile the same header file for different
15187 languages, targets, or compiler options, you can instead make a
15188 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15189 header in the directory, perhaps using @option{-o}. It doesn't matter
15190 what you call the files in the directory, every precompiled header in
15191 the directory will be considered. The first precompiled header
15192 encountered in the directory that is valid for this compilation will
15193 be used; they're searched in no particular order.
15195 There are many other possibilities, limited only by your imagination,
15196 good sense, and the constraints of your build system.
15198 A precompiled header file can be used only when these conditions apply:
15202 Only one precompiled header can be used in a particular compilation.
15205 A precompiled header can't be used once the first C token is seen. You
15206 can have preprocessor directives before a precompiled header; you can
15207 even include a precompiled header from inside another header, so long as
15208 there are no C tokens before the @code{#include}.
15211 The precompiled header file must be produced for the same language as
15212 the current compilation. You can't use a C precompiled header for a C++
15216 The precompiled header file must have been produced by the same compiler
15217 binary as the current compilation is using.
15220 Any macros defined before the precompiled header is included must
15221 either be defined in the same way as when the precompiled header was
15222 generated, or must not affect the precompiled header, which usually
15223 means that they don't appear in the precompiled header at all.
15225 The @option{-D} option is one way to define a macro before a
15226 precompiled header is included; using a @code{#define} can also do it.
15227 There are also some options that define macros implicitly, like
15228 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15231 @item If debugging information is output when using the precompiled
15232 header, using @option{-g} or similar, the same kind of debugging information
15233 must have been output when building the precompiled header. However,
15234 a precompiled header built using @option{-g} can be used in a compilation
15235 when no debugging information is being output.
15237 @item The same @option{-m} options must generally be used when building
15238 and using the precompiled header. @xref{Submodel Options},
15239 for any cases where this rule is relaxed.
15241 @item Each of the following options must be the same when building and using
15242 the precompiled header:
15244 @gccoptlist{-fexceptions -funit-at-a-time}
15247 Some other command-line options starting with @option{-f},
15248 @option{-p}, or @option{-O} must be defined in the same way as when
15249 the precompiled header was generated. At present, it's not clear
15250 which options are safe to change and which are not; the safest choice
15251 is to use exactly the same options when generating and using the
15252 precompiled header. The following are known to be safe:
15254 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15255 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15256 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15261 For all of these except the last, the compiler will automatically
15262 ignore the precompiled header if the conditions aren't met. If you
15263 find an option combination that doesn't work and doesn't cause the
15264 precompiled header to be ignored, please consider filing a bug report,
15267 If you do use differing options when generating and using the
15268 precompiled header, the actual behavior will be a mixture of the
15269 behavior for the options. For instance, if you use @option{-g} to
15270 generate the precompiled header but not when using it, you may or may
15271 not get debugging information for routines in the precompiled header.
15273 @node Running Protoize
15274 @section Running Protoize
15276 The program @code{protoize} is an optional part of GCC@. You can use
15277 it to add prototypes to a program, thus converting the program to ISO
15278 C in one respect. The companion program @code{unprotoize} does the
15279 reverse: it removes argument types from any prototypes that are found.
15281 When you run these programs, you must specify a set of source files as
15282 command line arguments. The conversion programs start out by compiling
15283 these files to see what functions they define. The information gathered
15284 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15286 After scanning comes actual conversion. The specified files are all
15287 eligible to be converted; any files they include (whether sources or
15288 just headers) are eligible as well.
15290 But not all the eligible files are converted. By default,
15291 @code{protoize} and @code{unprotoize} convert only source and header
15292 files in the current directory. You can specify additional directories
15293 whose files should be converted with the @option{-d @var{directory}}
15294 option. You can also specify particular files to exclude with the
15295 @option{-x @var{file}} option. A file is converted if it is eligible, its
15296 directory name matches one of the specified directory names, and its
15297 name within the directory has not been excluded.
15299 Basic conversion with @code{protoize} consists of rewriting most
15300 function definitions and function declarations to specify the types of
15301 the arguments. The only ones not rewritten are those for varargs
15304 @code{protoize} optionally inserts prototype declarations at the
15305 beginning of the source file, to make them available for any calls that
15306 precede the function's definition. Or it can insert prototype
15307 declarations with block scope in the blocks where undeclared functions
15310 Basic conversion with @code{unprotoize} consists of rewriting most
15311 function declarations to remove any argument types, and rewriting
15312 function definitions to the old-style pre-ISO form.
15314 Both conversion programs print a warning for any function declaration or
15315 definition that they can't convert. You can suppress these warnings
15318 The output from @code{protoize} or @code{unprotoize} replaces the
15319 original source file. The original file is renamed to a name ending
15320 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15321 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15322 for DOS) file already exists, then the source file is simply discarded.
15324 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15325 scan the program and collect information about the functions it uses.
15326 So neither of these programs will work until GCC is installed.
15328 Here is a table of the options you can use with @code{protoize} and
15329 @code{unprotoize}. Each option works with both programs unless
15333 @item -B @var{directory}
15334 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15335 usual directory (normally @file{/usr/local/lib}). This file contains
15336 prototype information about standard system functions. This option
15337 applies only to @code{protoize}.
15339 @item -c @var{compilation-options}
15340 Use @var{compilation-options} as the options when running @command{gcc} to
15341 produce the @samp{.X} files. The special option @option{-aux-info} is
15342 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15344 Note that the compilation options must be given as a single argument to
15345 @code{protoize} or @code{unprotoize}. If you want to specify several
15346 @command{gcc} options, you must quote the entire set of compilation options
15347 to make them a single word in the shell.
15349 There are certain @command{gcc} arguments that you cannot use, because they
15350 would produce the wrong kind of output. These include @option{-g},
15351 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15352 the @var{compilation-options}, they are ignored.
15355 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15356 systems) instead of @samp{.c}. This is convenient if you are converting
15357 a C program to C++. This option applies only to @code{protoize}.
15360 Add explicit global declarations. This means inserting explicit
15361 declarations at the beginning of each source file for each function
15362 that is called in the file and was not declared. These declarations
15363 precede the first function definition that contains a call to an
15364 undeclared function. This option applies only to @code{protoize}.
15366 @item -i @var{string}
15367 Indent old-style parameter declarations with the string @var{string}.
15368 This option applies only to @code{protoize}.
15370 @code{unprotoize} converts prototyped function definitions to old-style
15371 function definitions, where the arguments are declared between the
15372 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15373 uses five spaces as the indentation. If you want to indent with just
15374 one space instead, use @option{-i " "}.
15377 Keep the @samp{.X} files. Normally, they are deleted after conversion
15381 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15382 a prototype declaration for each function in each block which calls the
15383 function without any declaration. This option applies only to
15387 Make no real changes. This mode just prints information about the conversions
15388 that would have been done without @option{-n}.
15391 Make no @samp{.save} files. The original files are simply deleted.
15392 Use this option with caution.
15394 @item -p @var{program}
15395 Use the program @var{program} as the compiler. Normally, the name
15396 @file{gcc} is used.
15399 Work quietly. Most warnings are suppressed.
15402 Print the version number, just like @option{-v} for @command{gcc}.
15405 If you need special compiler options to compile one of your program's
15406 source files, then you should generate that file's @samp{.X} file
15407 specially, by running @command{gcc} on that source file with the
15408 appropriate options and the option @option{-aux-info}. Then run
15409 @code{protoize} on the entire set of files. @code{protoize} will use
15410 the existing @samp{.X} file because it is newer than the source file.
15414 gcc -Dfoo=bar file1.c -aux-info file1.X
15419 You need to include the special files along with the rest in the
15420 @code{protoize} command, even though their @samp{.X} files already
15421 exist, because otherwise they won't get converted.
15423 @xref{Protoize Caveats}, for more information on how to use
15424 @code{protoize} successfully.