1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version @@@var{file}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 -fvisibility-ms-compat @gol
196 -Wabi -Wctor-dtor-privacy @gol
197 -Wnon-virtual-dtor -Wreorder @gol
198 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
199 -Wno-non-template-friend -Wold-style-cast @gol
200 -Woverloaded-virtual -Wno-pmf-conversions @gol
203 @item Objective-C and Objective-C++ Language Options
204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 Objective-C and Objective-C++ Dialects}.
206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 -fgnu-runtime -fnext-runtime @gol
208 -fno-nil-receivers @gol
209 -fobjc-call-cxx-cdtors @gol
210 -fobjc-direct-dispatch @gol
211 -fobjc-exceptions @gol
213 -freplace-objc-classes @gol
216 -Wassign-intercept @gol
217 -Wno-protocol -Wselector @gol
218 -Wstrict-selector-match @gol
219 -Wundeclared-selector}
221 @item Language Independent Options
222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 @gccoptlist{-fmessage-length=@var{n} @gol
224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 -fdiagnostics-show-option}
227 @item Warning Options
228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
232 -Wchar-subscripts -Wclobbered -Wcomment @gol
233 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
234 -Wdisabled-optimization -Wno-div-by-zero @gol
235 -Wempty-body -Wno-endif-labels @gol
236 -Werror -Werror=* @gol
237 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
238 -Wno-format-extra-args -Wformat-nonliteral @gol
239 -Wformat-security -Wformat-y2k @gol
240 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
241 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
242 -Wimport -Wno-import -Winit-self -Winline @gol
243 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
244 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
245 -Wlogical-op -Wlong-long @gol
246 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
247 -Wmissing-format-attribute -Wmissing-include-dirs @gol
248 -Wmissing-noreturn @gol
249 -Wno-multichar -Wnonnull -Wno-overflow @gol
250 -Woverlength-strings -Wpacked -Wpadded @gol
251 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
252 -Wredundant-decls @gol
253 -Wreturn-type -Wsequence-point -Wshadow @gol
254 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
255 -Wstrict-aliasing -Wstrict-aliasing=n @gol
256 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
257 -Wswitch -Wswitch-default -Wswitch-enum @gol
258 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
259 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
260 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
261 -Wunused-value -Wunused-variable @gol
262 -Wvariadic-macros -Wvla @gol
263 -Wvolatile-register-var -Wwrite-strings}
265 @item C and Objective-C-only Warning Options
266 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
267 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
268 -Wold-style-declaration -Wold-style-definition @gol
269 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
270 -Wdeclaration-after-statement -Wpointer-sign}
272 @item Debugging Options
273 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
274 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
275 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
276 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
277 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
278 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
280 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
284 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-nrv -fdump-tree-vect @gol
293 -fdump-tree-sink @gol
294 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-salias @gol
296 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
298 -ftree-vectorizer-verbose=@var{n} @gol
299 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
300 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
301 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
302 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
303 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
304 -ftest-coverage -ftime-report -fvar-tracking @gol
305 -g -g@var{level} -gcoff -gdwarf-2 @gol
306 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
307 -fno-merge-debug-strings -fdebug-prefix-map=@var{old}=@var{new} @gol
308 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
309 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
310 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
311 -print-multi-directory -print-multi-lib @gol
312 -print-prog-name=@var{program} -print-search-dirs -Q @gol
313 -print-sysroot-headers-suffix @gol
316 @item Optimization Options
317 @xref{Optimize Options,,Options that Control Optimization}.
319 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
320 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
321 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
322 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
323 -fcheck-data-deps -fcprop-registers -fcrossjumping -fcse-follow-jumps @gol
324 -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
325 -fdata-sections -fdce -fdce @gol
326 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
327 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
328 -ffinite-math-only -ffloat-store -fforward-propagate @gol
329 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
330 -fgcse-sm -fif-conversion -fif-conversion2 -finline-functions @gol
331 -finline-functions-called-once -finline-limit=@var{n} @gol
332 -finline-small-functions -fipa-cp -fipa-marix-reorg -fipa-pta @gol
333 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
334 -fipa-type-escape -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
335 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
336 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
337 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
338 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
339 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
340 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
341 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
342 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
343 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
344 -fprofile-generate -fprofile-use -fprofile-values -freciprocal-math @gol
345 -fregmove -frename-registers -freorder-blocks @gol
346 -freorder-blocks-and-partition -freorder-functions @gol
347 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
348 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
349 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
350 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
351 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
352 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
353 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
354 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer -ftree-ccp @gol
355 -ftree-ch -ftree-copy-prop -ftree-copyrename -ftree-dce @gol
356 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
357 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
358 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc -ftree-salias @gol
359 -ftree-sink -ftree-sra -ftree-store-ccp -ftree-ter @gol
360 -ftree-vect-loop-version -ftree-vectorize -ftree-vrp -funit-at-a-time @gol
361 -funroll-all-loops -funroll-loops -funsafe-loop-optimizations @gol
362 -funsafe-math-optimizations -funswitch-loops @gol
363 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
365 --param @var{name}=@var{value}
366 -O -O0 -O1 -O2 -O3 -Os}
368 @item Preprocessor Options
369 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
370 @gccoptlist{-A@var{question}=@var{answer} @gol
371 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
372 -C -dD -dI -dM -dN @gol
373 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
374 -idirafter @var{dir} @gol
375 -include @var{file} -imacros @var{file} @gol
376 -iprefix @var{file} -iwithprefix @var{dir} @gol
377 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
378 -imultilib @var{dir} -isysroot @var{dir} @gol
379 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
380 -P -fworking-directory -remap @gol
381 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
382 -Xpreprocessor @var{option}}
384 @item Assembler Option
385 @xref{Assembler Options,,Passing Options to the Assembler}.
386 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
389 @xref{Link Options,,Options for Linking}.
390 @gccoptlist{@var{object-file-name} -l@var{library} @gol
391 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
392 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
393 -Wl,@var{option} -Xlinker @var{option} @gol
396 @item Directory Options
397 @xref{Directory Options,,Options for Directory Search}.
398 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
399 -specs=@var{file} -I- --sysroot=@var{dir}}
402 @c I wrote this xref this way to avoid overfull hbox. -- rms
403 @xref{Target Options}.
404 @gccoptlist{-V @var{version} -b @var{machine}}
406 @item Machine Dependent Options
407 @xref{Submodel Options,,Hardware Models and Configurations}.
408 @c This list is ordered alphanumerically by subsection name.
409 @c Try and put the significant identifier (CPU or system) first,
410 @c so users have a clue at guessing where the ones they want will be.
413 @gccoptlist{-EB -EL @gol
414 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
415 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
418 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
419 -mabi=@var{name} @gol
420 -mapcs-stack-check -mno-apcs-stack-check @gol
421 -mapcs-float -mno-apcs-float @gol
422 -mapcs-reentrant -mno-apcs-reentrant @gol
423 -msched-prolog -mno-sched-prolog @gol
424 -mlittle-endian -mbig-endian -mwords-little-endian @gol
425 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
426 -mthumb-interwork -mno-thumb-interwork @gol
427 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
428 -mstructure-size-boundary=@var{n} @gol
429 -mabort-on-noreturn @gol
430 -mlong-calls -mno-long-calls @gol
431 -msingle-pic-base -mno-single-pic-base @gol
432 -mpic-register=@var{reg} @gol
433 -mnop-fun-dllimport @gol
434 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
435 -mpoke-function-name @gol
437 -mtpcs-frame -mtpcs-leaf-frame @gol
438 -mcaller-super-interworking -mcallee-super-interworking @gol
442 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
443 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
445 @emph{Blackfin Options}
446 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
447 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
448 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
449 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
450 -mno-id-shared-library -mshared-library-id=@var{n} @gol
451 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
452 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
453 -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 -msse5 @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} @gol
564 -mfused-madd -mno-fused-madd}
567 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
568 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
569 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
570 -minline-float-divide-max-throughput @gol
571 -minline-int-divide-min-latency @gol
572 -minline-int-divide-max-throughput @gol
573 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
574 -mno-dwarf2-asm -mearly-stop-bits @gol
575 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
576 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
577 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
578 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
579 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
580 -mno-sched-prefer-non-data-spec-insns @gol
581 -mno-sched-prefer-non-control-spec-insns @gol
582 -mno-sched-count-spec-in-critical-path}
584 @emph{M32R/D Options}
585 @gccoptlist{-m32r2 -m32rx -m32r @gol
587 -malign-loops -mno-align-loops @gol
588 -missue-rate=@var{number} @gol
589 -mbranch-cost=@var{number} @gol
590 -mmodel=@var{code-size-model-type} @gol
591 -msdata=@var{sdata-type} @gol
592 -mno-flush-func -mflush-func=@var{name} @gol
593 -mno-flush-trap -mflush-trap=@var{number} @gol
597 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
599 @emph{M680x0 Options}
600 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
601 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
602 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
603 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
604 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
605 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
606 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
607 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
609 @emph{M68hc1x Options}
610 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
611 -mauto-incdec -minmax -mlong-calls -mshort @gol
612 -msoft-reg-count=@var{count}}
615 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
616 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
617 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
618 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
619 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
622 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
623 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
624 -mips16 -mno-mips16 -mflip-mips16 @gol
625 -minterlink-mips16 -mno-interlink-mips16 @gol
626 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
627 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
628 -mfp32 -mfp64 -mhard-float -msoft-float @gol
629 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
630 -msmartmips -mno-smartmips @gol
631 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
632 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
633 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
634 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
635 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
636 -membedded-data -mno-embedded-data @gol
637 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
638 -mcode-readable=@var{setting} @gol
639 -msplit-addresses -mno-split-addresses @gol
640 -mexplicit-relocs -mno-explicit-relocs @gol
641 -mcheck-zero-division -mno-check-zero-division @gol
642 -mdivide-traps -mdivide-breaks @gol
643 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
644 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
645 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
646 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
647 -mfix-sb1 -mno-fix-sb1 @gol
648 -mflush-func=@var{func} -mno-flush-func @gol
649 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
650 -mfp-exceptions -mno-fp-exceptions @gol
651 -mvr4130-align -mno-vr4130-align}
654 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
655 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
656 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
657 -mno-base-addresses -msingle-exit -mno-single-exit}
659 @emph{MN10300 Options}
660 @gccoptlist{-mmult-bug -mno-mult-bug @gol
661 -mam33 -mno-am33 @gol
662 -mam33-2 -mno-am33-2 @gol
663 -mreturn-pointer-on-d0 @gol
667 @gccoptlist{-mno-crt0 -mbacc -msim @gol
668 -march=@var{cpu-type} }
670 @emph{PDP-11 Options}
671 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
672 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
673 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
674 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
675 -mbranch-expensive -mbranch-cheap @gol
676 -msplit -mno-split -munix-asm -mdec-asm}
678 @emph{PowerPC Options}
679 See RS/6000 and PowerPC Options.
681 @emph{RS/6000 and PowerPC Options}
682 @gccoptlist{-mcpu=@var{cpu-type} @gol
683 -mtune=@var{cpu-type} @gol
684 -mpower -mno-power -mpower2 -mno-power2 @gol
685 -mpowerpc -mpowerpc64 -mno-powerpc @gol
686 -maltivec -mno-altivec @gol
687 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
688 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
689 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
690 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
691 -mnew-mnemonics -mold-mnemonics @gol
692 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
693 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
694 -malign-power -malign-natural @gol
695 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
696 -mstring -mno-string -mupdate -mno-update @gol
697 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
698 -mstrict-align -mno-strict-align -mrelocatable @gol
699 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
700 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
701 -mdynamic-no-pic -maltivec -mswdiv @gol
702 -mprioritize-restricted-insns=@var{priority} @gol
703 -msched-costly-dep=@var{dependence_type} @gol
704 -minsert-sched-nops=@var{scheme} @gol
705 -mcall-sysv -mcall-netbsd @gol
706 -maix-struct-return -msvr4-struct-return @gol
707 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
708 -misel -mno-isel @gol
709 -misel=yes -misel=no @gol
711 -mspe=yes -mspe=no @gol
713 -mvrsave -mno-vrsave @gol
714 -mmulhw -mno-mulhw @gol
715 -mdlmzb -mno-dlmzb @gol
716 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
717 -mprototype -mno-prototype @gol
718 -msim -mmvme -mads -myellowknife -memb -msdata @gol
719 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
721 @emph{S/390 and zSeries Options}
722 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
723 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
724 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
725 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
726 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
727 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
728 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
731 @gccoptlist{-meb -mel @gol
735 -mscore5 -mscore5u -mscore7 -mscore7d}
738 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
739 -m4-nofpu -m4-single-only -m4-single -m4 @gol
740 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
741 -m5-64media -m5-64media-nofpu @gol
742 -m5-32media -m5-32media-nofpu @gol
743 -m5-compact -m5-compact-nofpu @gol
744 -mb -ml -mdalign -mrelax @gol
745 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
746 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
747 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
748 -mdivsi3_libfunc=@var{name} @gol
749 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
753 @gccoptlist{-mcpu=@var{cpu-type} @gol
754 -mtune=@var{cpu-type} @gol
755 -mcmodel=@var{code-model} @gol
756 -m32 -m64 -mapp-regs -mno-app-regs @gol
757 -mfaster-structs -mno-faster-structs @gol
758 -mfpu -mno-fpu -mhard-float -msoft-float @gol
759 -mhard-quad-float -msoft-quad-float @gol
760 -mimpure-text -mno-impure-text -mlittle-endian @gol
761 -mstack-bias -mno-stack-bias @gol
762 -munaligned-doubles -mno-unaligned-doubles @gol
763 -mv8plus -mno-v8plus -mvis -mno-vis
764 -threads -pthreads -pthread}
767 @gccoptlist{-mwarn-reloc -merror-reloc @gol
768 -msafe-dma -munsafe-dma @gol
770 -msmall-mem -mlarge-mem -mstdmain @gol
771 -mfixed-range=@var{register-range}}
773 @emph{System V Options}
774 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
777 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
778 -mprolog-function -mno-prolog-function -mspace @gol
779 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
780 -mapp-regs -mno-app-regs @gol
781 -mdisable-callt -mno-disable-callt @gol
787 @gccoptlist{-mg -mgnu -munix}
789 @emph{VxWorks Options}
790 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
791 -Xbind-lazy -Xbind-now}
793 @emph{x86-64 Options}
794 See i386 and x86-64 Options.
796 @emph{Xstormy16 Options}
799 @emph{Xtensa Options}
800 @gccoptlist{-mconst16 -mno-const16 @gol
801 -mfused-madd -mno-fused-madd @gol
802 -mtext-section-literals -mno-text-section-literals @gol
803 -mtarget-align -mno-target-align @gol
804 -mlongcalls -mno-longcalls}
806 @emph{zSeries Options}
807 See S/390 and zSeries Options.
809 @item Code Generation Options
810 @xref{Code Gen Options,,Options for Code Generation Conventions}.
811 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
812 -ffixed-@var{reg} -fexceptions @gol
813 -fnon-call-exceptions -funwind-tables @gol
814 -fasynchronous-unwind-tables @gol
815 -finhibit-size-directive -finstrument-functions @gol
816 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
817 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
818 -fno-common -fno-ident @gol
819 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
820 -fno-jump-tables @gol
821 -frecord-gcc-switches @gol
822 -freg-struct-return -fshort-enums @gol
823 -fshort-double -fshort-wchar @gol
824 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
825 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
826 -fno-stack-limit -fargument-alias -fargument-noalias @gol
827 -fargument-noalias-global -fargument-noalias-anything @gol
828 -fleading-underscore -ftls-model=@var{model} @gol
829 -ftrapv -fwrapv -fbounds-check @gol
834 * Overall Options:: Controlling the kind of output:
835 an executable, object files, assembler files,
836 or preprocessed source.
837 * C Dialect Options:: Controlling the variant of C language compiled.
838 * C++ Dialect Options:: Variations on C++.
839 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
841 * Language Independent Options:: Controlling how diagnostics should be
843 * Warning Options:: How picky should the compiler be?
844 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
845 * Optimize Options:: How much optimization?
846 * Preprocessor Options:: Controlling header files and macro definitions.
847 Also, getting dependency information for Make.
848 * Assembler Options:: Passing options to the assembler.
849 * Link Options:: Specifying libraries and so on.
850 * Directory Options:: Where to find header files and libraries.
851 Where to find the compiler executable files.
852 * Spec Files:: How to pass switches to sub-processes.
853 * Target Options:: Running a cross-compiler, or an old version of GCC.
856 @node Overall Options
857 @section Options Controlling the Kind of Output
859 Compilation can involve up to four stages: preprocessing, compilation
860 proper, assembly and linking, always in that order. GCC is capable of
861 preprocessing and compiling several files either into several
862 assembler input files, or into one assembler input file; then each
863 assembler input file produces an object file, and linking combines all
864 the object files (those newly compiled, and those specified as input)
865 into an executable file.
867 @cindex file name suffix
868 For any given input file, the file name suffix determines what kind of
873 C source code which must be preprocessed.
876 C source code which should not be preprocessed.
879 C++ source code which should not be preprocessed.
882 Objective-C source code. Note that you must link with the @file{libobjc}
883 library to make an Objective-C program work.
886 Objective-C source code which should not be preprocessed.
890 Objective-C++ source code. Note that you must link with the @file{libobjc}
891 library to make an Objective-C++ program work. Note that @samp{.M} refers
892 to a literal capital M@.
895 Objective-C++ source code which should not be preprocessed.
898 C, C++, Objective-C or Objective-C++ header file to be turned into a
903 @itemx @var{file}.cxx
904 @itemx @var{file}.cpp
905 @itemx @var{file}.CPP
906 @itemx @var{file}.c++
908 C++ source code which must be preprocessed. Note that in @samp{.cxx},
909 the last two letters must both be literally @samp{x}. Likewise,
910 @samp{.C} refers to a literal capital C@.
914 Objective-C++ source code which must be preprocessed.
917 Objective-C++ source code which should not be preprocessed.
922 @itemx @var{file}.hxx
923 @itemx @var{file}.hpp
924 @itemx @var{file}.HPP
925 @itemx @var{file}.h++
926 @itemx @var{file}.tcc
927 C++ header file to be turned into a precompiled header.
930 @itemx @var{file}.for
931 @itemx @var{file}.FOR
932 Fixed form Fortran source code which should not be preprocessed.
935 @itemx @var{file}.fpp
936 @itemx @var{file}.FPP
937 Fixed form Fortran source code which must be preprocessed (with the traditional
941 @itemx @var{file}.f95
942 Free form Fortran source code which should not be preprocessed.
945 @itemx @var{file}.F95
946 Free form Fortran source code which must be preprocessed (with the
947 traditional preprocessor).
949 @c FIXME: Descriptions of Java file types.
956 Ada source code file which contains a library unit declaration (a
957 declaration of a package, subprogram, or generic, or a generic
958 instantiation), or a library unit renaming declaration (a package,
959 generic, or subprogram renaming declaration). Such files are also
962 @itemx @var{file}.adb
963 Ada source code file containing a library unit body (a subprogram or
964 package body). Such files are also called @dfn{bodies}.
966 @c GCC also knows about some suffixes for languages not yet included:
978 Assembler code which must be preprocessed.
981 An object file to be fed straight into linking.
982 Any file name with no recognized suffix is treated this way.
986 You can specify the input language explicitly with the @option{-x} option:
989 @item -x @var{language}
990 Specify explicitly the @var{language} for the following input files
991 (rather than letting the compiler choose a default based on the file
992 name suffix). This option applies to all following input files until
993 the next @option{-x} option. Possible values for @var{language} are:
995 c c-header c-cpp-output
996 c++ c++-header c++-cpp-output
997 objective-c objective-c-header objective-c-cpp-output
998 objective-c++ objective-c++-header objective-c++-cpp-output
999 assembler assembler-with-cpp
1007 Turn off any specification of a language, so that subsequent files are
1008 handled according to their file name suffixes (as they are if @option{-x}
1009 has not been used at all).
1011 @item -pass-exit-codes
1012 @opindex pass-exit-codes
1013 Normally the @command{gcc} program will exit with the code of 1 if any
1014 phase of the compiler returns a non-success return code. If you specify
1015 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1016 numerically highest error produced by any phase that returned an error
1017 indication. The C, C++, and Fortran frontends return 4, if an internal
1018 compiler error is encountered.
1021 If you only want some of the stages of compilation, you can use
1022 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1023 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1024 @command{gcc} is to stop. Note that some combinations (for example,
1025 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1030 Compile or assemble the source files, but do not link. The linking
1031 stage simply is not done. The ultimate output is in the form of an
1032 object file for each source file.
1034 By default, the object file name for a source file is made by replacing
1035 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1037 Unrecognized input files, not requiring compilation or assembly, are
1042 Stop after the stage of compilation proper; do not assemble. The output
1043 is in the form of an assembler code file for each non-assembler input
1046 By default, the assembler file name for a source file is made by
1047 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1049 Input files that don't require compilation are ignored.
1053 Stop after the preprocessing stage; do not run the compiler proper. The
1054 output is in the form of preprocessed source code, which is sent to the
1057 Input files which don't require preprocessing are ignored.
1059 @cindex output file option
1062 Place output in file @var{file}. This applies regardless to whatever
1063 sort of output is being produced, whether it be an executable file,
1064 an object file, an assembler file or preprocessed C code.
1066 If @option{-o} is not specified, the default is to put an executable
1067 file in @file{a.out}, the object file for
1068 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1069 assembler file in @file{@var{source}.s}, a precompiled header file in
1070 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1075 Print (on standard error output) the commands executed to run the stages
1076 of compilation. Also print the version number of the compiler driver
1077 program and of the preprocessor and the compiler proper.
1081 Like @option{-v} except the commands are not executed and all command
1082 arguments are quoted. This is useful for shell scripts to capture the
1083 driver-generated command lines.
1087 Use pipes rather than temporary files for communication between the
1088 various stages of compilation. This fails to work on some systems where
1089 the assembler is unable to read from a pipe; but the GNU assembler has
1094 If you are compiling multiple source files, this option tells the driver
1095 to pass all the source files to the compiler at once (for those
1096 languages for which the compiler can handle this). This will allow
1097 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1098 language for which this is supported is C@. If you pass source files for
1099 multiple languages to the driver, using this option, the driver will invoke
1100 the compiler(s) that support IMA once each, passing each compiler all the
1101 source files appropriate for it. For those languages that do not support
1102 IMA this option will be ignored, and the compiler will be invoked once for
1103 each source file in that language. If you use this option in conjunction
1104 with @option{-save-temps}, the compiler will generate multiple
1106 (one for each source file), but only one (combined) @file{.o} or
1111 Print (on the standard output) a description of the command line options
1112 understood by @command{gcc}. If the @option{-v} option is also specified
1113 then @option{--help} will also be passed on to the various processes
1114 invoked by @command{gcc}, so that they can display the command line options
1115 they accept. If the @option{-Wextra} option has also been specified
1116 (prior to the @option{--help} option), then command line options which
1117 have no documentation associated with them will also be displayed.
1120 @opindex target-help
1121 Print (on the standard output) a description of target-specific command
1122 line options for each tool. For some targets extra target-specific
1123 information may also be printed.
1125 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1126 Print (on the standard output) a description of the command line
1127 options understood by the compiler that fit into a specific class.
1128 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1129 @samp{params}, or @var{language}:
1132 @item @samp{optimizers}
1133 This will display all of the optimization options supported by the
1136 @item @samp{warnings}
1137 This will display all of the options controlling warning messages
1138 produced by the compiler.
1141 This will display target-specific options. Unlike the
1142 @option{--target-help} option however, target-specific options of the
1143 linker and assembler will not be displayed. This is because those
1144 tools do not currently support the extended @option{--help=} syntax.
1147 This will display the values recognized by the @option{--param}
1150 @item @var{language}
1151 This will display the options supported for @var{language}, where
1152 @var{language} is the name of one of the languages supported in this
1156 This will display the options that are common to all languages.
1159 It is possible to further refine the output of the @option{--help=}
1160 option by adding a comma separated list of qualifiers after the
1161 class. These can be any from the following list:
1164 @item @samp{undocumented}
1165 Display only those options which are undocumented.
1168 Display options which take an argument that appears after an equal
1169 sign in the same continuous piece of text, such as:
1170 @samp{--help=target}.
1172 @item @samp{separate}
1173 Display options which take an argument that appears as a separate word
1174 following the original option, such as: @samp{-o output-file}.
1177 Thus for example to display all the undocumented target-specific
1178 switches supported by the compiler the following can be used:
1181 --help=target,undocumented
1184 The sense of a qualifier can be inverted by prefixing it with the
1185 @var{^} character, so for example to display all binary warning
1186 options (i.e., ones that are either on or off and that do not take an
1187 argument), which have a description the following can be used:
1190 --help=warnings,^joined,^undocumented
1193 A class can also be used as a qualifier, although this usually
1194 restricts the output by so much that there is nothing to display. One
1195 case where it does work however is when one of the classes is
1196 @var{target}. So for example to display all the target-specific
1197 optimization options the following can be used:
1200 --help=target,optimizers
1203 The @option{--help=} option can be repeated on the command line. Each
1204 successive use will display its requested class of options, skipping
1205 those that have already been displayed.
1207 If the @option{-Q} option appears on the command line before the
1208 @option{--help=} option, then the descriptive text displayed by
1209 @option{--help=} is changed. Instead of describing the displayed
1210 options, an indication is given as to whether the option is enabled,
1211 disabled or set to a specific value (assuming that the compiler
1212 knows this at the point where the @option{--help=} option is used).
1214 Here is a truncated example from the ARM port of @command{gcc}:
1217 % gcc -Q -mabi=2 --help=target -c
1218 The following options are target specific:
1220 -mabort-on-noreturn [disabled]
1224 The output is sensitive to the effects of previous command line
1225 options, so for example it is possible to find out which optimizations
1226 are enabled at @option{-O2} by using:
1229 -O2 --help=optimizers
1232 Alternatively you can discover which binary optimizations are enabled
1233 by @option{-O3} by using:
1236 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1237 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1238 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1243 Display the version number and copyrights of the invoked GCC@.
1245 @include @value{srcdir}/../libiberty/at-file.texi
1249 @section Compiling C++ Programs
1251 @cindex suffixes for C++ source
1252 @cindex C++ source file suffixes
1253 C++ source files conventionally use one of the suffixes @samp{.C},
1254 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1255 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1256 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1257 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1258 files with these names and compiles them as C++ programs even if you
1259 call the compiler the same way as for compiling C programs (usually
1260 with the name @command{gcc}).
1264 However, the use of @command{gcc} does not add the C++ library.
1265 @command{g++} is a program that calls GCC and treats @samp{.c},
1266 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1267 files unless @option{-x} is used, and automatically specifies linking
1268 against the C++ library. This program is also useful when
1269 precompiling a C header file with a @samp{.h} extension for use in C++
1270 compilations. On many systems, @command{g++} is also installed with
1271 the name @command{c++}.
1273 @cindex invoking @command{g++}
1274 When you compile C++ programs, you may specify many of the same
1275 command-line options that you use for compiling programs in any
1276 language; or command-line options meaningful for C and related
1277 languages; or options that are meaningful only for C++ programs.
1278 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1279 explanations of options for languages related to C@.
1280 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1281 explanations of options that are meaningful only for C++ programs.
1283 @node C Dialect Options
1284 @section Options Controlling C Dialect
1285 @cindex dialect options
1286 @cindex language dialect options
1287 @cindex options, dialect
1289 The following options control the dialect of C (or languages derived
1290 from C, such as C++, Objective-C and Objective-C++) that the compiler
1294 @cindex ANSI support
1298 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1299 equivalent to @samp{-std=c++98}.
1301 This turns off certain features of GCC that are incompatible with ISO
1302 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1303 such as the @code{asm} and @code{typeof} keywords, and
1304 predefined macros such as @code{unix} and @code{vax} that identify the
1305 type of system you are using. It also enables the undesirable and
1306 rarely used ISO trigraph feature. For the C compiler,
1307 it disables recognition of C++ style @samp{//} comments as well as
1308 the @code{inline} keyword.
1310 The alternate keywords @code{__asm__}, @code{__extension__},
1311 @code{__inline__} and @code{__typeof__} continue to work despite
1312 @option{-ansi}. You would not want to use them in an ISO C program, of
1313 course, but it is useful to put them in header files that might be included
1314 in compilations done with @option{-ansi}. Alternate predefined macros
1315 such as @code{__unix__} and @code{__vax__} are also available, with or
1316 without @option{-ansi}.
1318 The @option{-ansi} option does not cause non-ISO programs to be
1319 rejected gratuitously. For that, @option{-pedantic} is required in
1320 addition to @option{-ansi}. @xref{Warning Options}.
1322 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1323 option is used. Some header files may notice this macro and refrain
1324 from declaring certain functions or defining certain macros that the
1325 ISO standard doesn't call for; this is to avoid interfering with any
1326 programs that might use these names for other things.
1328 Functions that would normally be built in but do not have semantics
1329 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1330 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1331 built-in functions provided by GCC}, for details of the functions
1336 Determine the language standard. @xref{Standards,,Language Standards
1337 Supported by GCC}, for details of these standard versions. This option
1338 is currently only supported when compiling C or C++.
1340 The compiler can accept several base standards, such as @samp{c89} or
1341 @samp{c++98}, and GNU dialects of those standards, such as
1342 @samp{gnu89} or @samp{gnu++98}. By specifing a base standard, the
1343 compiler will accept all programs following that standard and those
1344 using GNU extensions that do not contradict it. For example,
1345 @samp{-std=c89} turns off certain features of GCC that are
1346 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1347 keywords, but not other GNU extensions that do not have a meaning in
1348 ISO C90, such as omitting the middle term of a @code{?:}
1349 expression. On the other hand, by specifing a GNU dialect of a
1350 standard, all features the compiler support are enabled, even when
1351 those features change the meaning of the base standard and some
1352 strict-conforming programs may be rejected. The particular standard
1353 is used by @option{-pedantic} to identify which features are GNU
1354 extensions given that version of the standard. For example
1355 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1356 comments, while @samp{-std=gnu99 -pedantic} would not.
1358 A value for this option must be provided; possible values are
1363 Support all ISO C90 programs (certain GNU extensions that conflict
1364 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1366 @item iso9899:199409
1367 ISO C90 as modified in amendment 1.
1373 ISO C99. Note that this standard is not yet fully supported; see
1374 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1375 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1378 GNU dialect of ISO C90 (including some C99 features). This
1379 is the default for C code.
1383 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1384 this will become the default. The name @samp{gnu9x} is deprecated.
1387 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1391 GNU dialect of @option{-std=c++98}. This is the default for
1395 The working draft of the upcoming ISO C++0x standard. This option
1396 enables experimental features that are likely to be included in
1397 C++0x. The working draft is constantly changing, and any feature that is
1398 enabled by this flag may be removed from future versions of GCC if it is
1399 not part of the C++0x standard.
1402 GNU dialect of @option{-std=c++0x}. This option enables
1403 experimental features that may be removed in future versions of GCC.
1406 @item -fgnu89-inline
1407 @opindex fgnu89-inline
1408 The option @option{-fgnu89-inline} tells GCC to use the traditional
1409 GNU semantics for @code{inline} functions when in C99 mode.
1410 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1411 is accepted and ignored by GCC versions 4.1.3 up to but not including
1412 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1413 C99 mode. Using this option is roughly equivalent to adding the
1414 @code{gnu_inline} function attribute to all inline functions
1415 (@pxref{Function Attributes}).
1417 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1418 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1419 specifies the default behavior). This option was first supported in
1420 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1422 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1423 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1424 in effect for @code{inline} functions. @xref{Common Predefined
1425 Macros,,,cpp,The C Preprocessor}.
1427 @item -aux-info @var{filename}
1429 Output to the given filename prototyped declarations for all functions
1430 declared and/or defined in a translation unit, including those in header
1431 files. This option is silently ignored in any language other than C@.
1433 Besides declarations, the file indicates, in comments, the origin of
1434 each declaration (source file and line), whether the declaration was
1435 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1436 @samp{O} for old, respectively, in the first character after the line
1437 number and the colon), and whether it came from a declaration or a
1438 definition (@samp{C} or @samp{F}, respectively, in the following
1439 character). In the case of function definitions, a K&R-style list of
1440 arguments followed by their declarations is also provided, inside
1441 comments, after the declaration.
1445 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1446 keyword, so that code can use these words as identifiers. You can use
1447 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1448 instead. @option{-ansi} implies @option{-fno-asm}.
1450 In C++, this switch only affects the @code{typeof} keyword, since
1451 @code{asm} and @code{inline} are standard keywords. You may want to
1452 use the @option{-fno-gnu-keywords} flag instead, which has the same
1453 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1454 switch only affects the @code{asm} and @code{typeof} keywords, since
1455 @code{inline} is a standard keyword in ISO C99.
1458 @itemx -fno-builtin-@var{function}
1459 @opindex fno-builtin
1460 @cindex built-in functions
1461 Don't recognize built-in functions that do not begin with
1462 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1463 functions provided by GCC}, for details of the functions affected,
1464 including those which are not built-in functions when @option{-ansi} or
1465 @option{-std} options for strict ISO C conformance are used because they
1466 do not have an ISO standard meaning.
1468 GCC normally generates special code to handle certain built-in functions
1469 more efficiently; for instance, calls to @code{alloca} may become single
1470 instructions that adjust the stack directly, and calls to @code{memcpy}
1471 may become inline copy loops. The resulting code is often both smaller
1472 and faster, but since the function calls no longer appear as such, you
1473 cannot set a breakpoint on those calls, nor can you change the behavior
1474 of the functions by linking with a different library. In addition,
1475 when a function is recognized as a built-in function, GCC may use
1476 information about that function to warn about problems with calls to
1477 that function, or to generate more efficient code, even if the
1478 resulting code still contains calls to that function. For example,
1479 warnings are given with @option{-Wformat} for bad calls to
1480 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1481 known not to modify global memory.
1483 With the @option{-fno-builtin-@var{function}} option
1484 only the built-in function @var{function} is
1485 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1486 function is named this is not built-in in this version of GCC, this
1487 option is ignored. There is no corresponding
1488 @option{-fbuiltin-@var{function}} option; if you wish to enable
1489 built-in functions selectively when using @option{-fno-builtin} or
1490 @option{-ffreestanding}, you may define macros such as:
1493 #define abs(n) __builtin_abs ((n))
1494 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1499 @cindex hosted environment
1501 Assert that compilation takes place in a hosted environment. This implies
1502 @option{-fbuiltin}. A hosted environment is one in which the
1503 entire standard library is available, and in which @code{main} has a return
1504 type of @code{int}. Examples are nearly everything except a kernel.
1505 This is equivalent to @option{-fno-freestanding}.
1507 @item -ffreestanding
1508 @opindex ffreestanding
1509 @cindex hosted environment
1511 Assert that compilation takes place in a freestanding environment. This
1512 implies @option{-fno-builtin}. A freestanding environment
1513 is one in which the standard library may not exist, and program startup may
1514 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1515 This is equivalent to @option{-fno-hosted}.
1517 @xref{Standards,,Language Standards Supported by GCC}, for details of
1518 freestanding and hosted environments.
1522 @cindex openmp parallel
1523 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1524 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1525 compiler generates parallel code according to the OpenMP Application
1526 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1527 implies @option{-pthread}, and thus is only supported on targets that
1528 have support for @option{-pthread}.
1530 @item -fms-extensions
1531 @opindex fms-extensions
1532 Accept some non-standard constructs used in Microsoft header files.
1534 Some cases of unnamed fields in structures and unions are only
1535 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1536 fields within structs/unions}, for details.
1540 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1541 options for strict ISO C conformance) implies @option{-trigraphs}.
1543 @item -no-integrated-cpp
1544 @opindex no-integrated-cpp
1545 Performs a compilation in two passes: preprocessing and compiling. This
1546 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1547 @option{-B} option. The user supplied compilation step can then add in
1548 an additional preprocessing step after normal preprocessing but before
1549 compiling. The default is to use the integrated cpp (internal cpp)
1551 The semantics of this option will change if "cc1", "cc1plus", and
1552 "cc1obj" are merged.
1554 @cindex traditional C language
1555 @cindex C language, traditional
1557 @itemx -traditional-cpp
1558 @opindex traditional-cpp
1559 @opindex traditional
1560 Formerly, these options caused GCC to attempt to emulate a pre-standard
1561 C compiler. They are now only supported with the @option{-E} switch.
1562 The preprocessor continues to support a pre-standard mode. See the GNU
1563 CPP manual for details.
1565 @item -fcond-mismatch
1566 @opindex fcond-mismatch
1567 Allow conditional expressions with mismatched types in the second and
1568 third arguments. The value of such an expression is void. This option
1569 is not supported for C++.
1571 @item -flax-vector-conversions
1572 @opindex flax-vector-conversions
1573 Allow implicit conversions between vectors with differing numbers of
1574 elements and/or incompatible element types. This option should not be
1577 @item -funsigned-char
1578 @opindex funsigned-char
1579 Let the type @code{char} be unsigned, like @code{unsigned char}.
1581 Each kind of machine has a default for what @code{char} should
1582 be. It is either like @code{unsigned char} by default or like
1583 @code{signed char} by default.
1585 Ideally, a portable program should always use @code{signed char} or
1586 @code{unsigned char} when it depends on the signedness of an object.
1587 But many programs have been written to use plain @code{char} and
1588 expect it to be signed, or expect it to be unsigned, depending on the
1589 machines they were written for. This option, and its inverse, let you
1590 make such a program work with the opposite default.
1592 The type @code{char} is always a distinct type from each of
1593 @code{signed char} or @code{unsigned char}, even though its behavior
1594 is always just like one of those two.
1597 @opindex fsigned-char
1598 Let the type @code{char} be signed, like @code{signed char}.
1600 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1601 the negative form of @option{-funsigned-char}. Likewise, the option
1602 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1604 @item -fsigned-bitfields
1605 @itemx -funsigned-bitfields
1606 @itemx -fno-signed-bitfields
1607 @itemx -fno-unsigned-bitfields
1608 @opindex fsigned-bitfields
1609 @opindex funsigned-bitfields
1610 @opindex fno-signed-bitfields
1611 @opindex fno-unsigned-bitfields
1612 These options control whether a bit-field is signed or unsigned, when the
1613 declaration does not use either @code{signed} or @code{unsigned}. By
1614 default, such a bit-field is signed, because this is consistent: the
1615 basic integer types such as @code{int} are signed types.
1618 @node C++ Dialect Options
1619 @section Options Controlling C++ Dialect
1621 @cindex compiler options, C++
1622 @cindex C++ options, command line
1623 @cindex options, C++
1624 This section describes the command-line options that are only meaningful
1625 for C++ programs; but you can also use most of the GNU compiler options
1626 regardless of what language your program is in. For example, you
1627 might compile a file @code{firstClass.C} like this:
1630 g++ -g -frepo -O -c firstClass.C
1634 In this example, only @option{-frepo} is an option meant
1635 only for C++ programs; you can use the other options with any
1636 language supported by GCC@.
1638 Here is a list of options that are @emph{only} for compiling C++ programs:
1642 @item -fabi-version=@var{n}
1643 @opindex fabi-version
1644 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1645 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1646 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1647 the version that conforms most closely to the C++ ABI specification.
1648 Therefore, the ABI obtained using version 0 will change as ABI bugs
1651 The default is version 2.
1653 @item -fno-access-control
1654 @opindex fno-access-control
1655 Turn off all access checking. This switch is mainly useful for working
1656 around bugs in the access control code.
1660 Check that the pointer returned by @code{operator new} is non-null
1661 before attempting to modify the storage allocated. This check is
1662 normally unnecessary because the C++ standard specifies that
1663 @code{operator new} will only return @code{0} if it is declared
1664 @samp{throw()}, in which case the compiler will always check the
1665 return value even without this option. In all other cases, when
1666 @code{operator new} has a non-empty exception specification, memory
1667 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1668 @samp{new (nothrow)}.
1670 @item -fconserve-space
1671 @opindex fconserve-space
1672 Put uninitialized or runtime-initialized global variables into the
1673 common segment, as C does. This saves space in the executable at the
1674 cost of not diagnosing duplicate definitions. If you compile with this
1675 flag and your program mysteriously crashes after @code{main()} has
1676 completed, you may have an object that is being destroyed twice because
1677 two definitions were merged.
1679 This option is no longer useful on most targets, now that support has
1680 been added for putting variables into BSS without making them common.
1682 @item -ffriend-injection
1683 @opindex ffriend-injection
1684 Inject friend functions into the enclosing namespace, so that they are
1685 visible outside the scope of the class in which they are declared.
1686 Friend functions were documented to work this way in the old Annotated
1687 C++ Reference Manual, and versions of G++ before 4.1 always worked
1688 that way. However, in ISO C++ a friend function which is not declared
1689 in an enclosing scope can only be found using argument dependent
1690 lookup. This option causes friends to be injected as they were in
1693 This option is for compatibility, and may be removed in a future
1696 @item -fno-elide-constructors
1697 @opindex fno-elide-constructors
1698 The C++ standard allows an implementation to omit creating a temporary
1699 which is only used to initialize another object of the same type.
1700 Specifying this option disables that optimization, and forces G++ to
1701 call the copy constructor in all cases.
1703 @item -fno-enforce-eh-specs
1704 @opindex fno-enforce-eh-specs
1705 Don't generate code to check for violation of exception specifications
1706 at runtime. This option violates the C++ standard, but may be useful
1707 for reducing code size in production builds, much like defining
1708 @samp{NDEBUG}. This does not give user code permission to throw
1709 exceptions in violation of the exception specifications; the compiler
1710 will still optimize based on the specifications, so throwing an
1711 unexpected exception will result in undefined behavior.
1714 @itemx -fno-for-scope
1716 @opindex fno-for-scope
1717 If @option{-ffor-scope} is specified, the scope of variables declared in
1718 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1719 as specified by the C++ standard.
1720 If @option{-fno-for-scope} is specified, the scope of variables declared in
1721 a @i{for-init-statement} extends to the end of the enclosing scope,
1722 as was the case in old versions of G++, and other (traditional)
1723 implementations of C++.
1725 The default if neither flag is given to follow the standard,
1726 but to allow and give a warning for old-style code that would
1727 otherwise be invalid, or have different behavior.
1729 @item -fno-gnu-keywords
1730 @opindex fno-gnu-keywords
1731 Do not recognize @code{typeof} as a keyword, so that code can use this
1732 word as an identifier. You can use the keyword @code{__typeof__} instead.
1733 @option{-ansi} implies @option{-fno-gnu-keywords}.
1735 @item -fno-implicit-templates
1736 @opindex fno-implicit-templates
1737 Never emit code for non-inline templates which are instantiated
1738 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1739 @xref{Template Instantiation}, for more information.
1741 @item -fno-implicit-inline-templates
1742 @opindex fno-implicit-inline-templates
1743 Don't emit code for implicit instantiations of inline templates, either.
1744 The default is to handle inlines differently so that compiles with and
1745 without optimization will need the same set of explicit instantiations.
1747 @item -fno-implement-inlines
1748 @opindex fno-implement-inlines
1749 To save space, do not emit out-of-line copies of inline functions
1750 controlled by @samp{#pragma implementation}. This will cause linker
1751 errors if these functions are not inlined everywhere they are called.
1753 @item -fms-extensions
1754 @opindex fms-extensions
1755 Disable pedantic warnings about constructs used in MFC, such as implicit
1756 int and getting a pointer to member function via non-standard syntax.
1758 @item -fno-nonansi-builtins
1759 @opindex fno-nonansi-builtins
1760 Disable built-in declarations of functions that are not mandated by
1761 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1762 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1764 @item -fno-operator-names
1765 @opindex fno-operator-names
1766 Do not treat the operator name keywords @code{and}, @code{bitand},
1767 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1768 synonyms as keywords.
1770 @item -fno-optional-diags
1771 @opindex fno-optional-diags
1772 Disable diagnostics that the standard says a compiler does not need to
1773 issue. Currently, the only such diagnostic issued by G++ is the one for
1774 a name having multiple meanings within a class.
1777 @opindex fpermissive
1778 Downgrade some diagnostics about nonconformant code from errors to
1779 warnings. Thus, using @option{-fpermissive} will allow some
1780 nonconforming code to compile.
1784 Enable automatic template instantiation at link time. This option also
1785 implies @option{-fno-implicit-templates}. @xref{Template
1786 Instantiation}, for more information.
1790 Disable generation of information about every class with virtual
1791 functions for use by the C++ runtime type identification features
1792 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1793 of the language, you can save some space by using this flag. Note that
1794 exception handling uses the same information, but it will generate it as
1795 needed. The @samp{dynamic_cast} operator can still be used for casts that
1796 do not require runtime type information, i.e.@: casts to @code{void *} or to
1797 unambiguous base classes.
1801 Emit statistics about front-end processing at the end of the compilation.
1802 This information is generally only useful to the G++ development team.
1804 @item -ftemplate-depth-@var{n}
1805 @opindex ftemplate-depth
1806 Set the maximum instantiation depth for template classes to @var{n}.
1807 A limit on the template instantiation depth is needed to detect
1808 endless recursions during template class instantiation. ANSI/ISO C++
1809 conforming programs must not rely on a maximum depth greater than 17.
1811 @item -fno-threadsafe-statics
1812 @opindex fno-threadsafe-statics
1813 Do not emit the extra code to use the routines specified in the C++
1814 ABI for thread-safe initialization of local statics. You can use this
1815 option to reduce code size slightly in code that doesn't need to be
1818 @item -fuse-cxa-atexit
1819 @opindex fuse-cxa-atexit
1820 Register destructors for objects with static storage duration with the
1821 @code{__cxa_atexit} function rather than the @code{atexit} function.
1822 This option is required for fully standards-compliant handling of static
1823 destructors, but will only work if your C library supports
1824 @code{__cxa_atexit}.
1826 @item -fno-use-cxa-get-exception-ptr
1827 @opindex fno-use-cxa-get-exception-ptr
1828 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1829 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1830 if the runtime routine is not available.
1832 @item -fvisibility-inlines-hidden
1833 @opindex fvisibility-inlines-hidden
1834 This switch declares that the user does not attempt to compare
1835 pointers to inline methods where the addresses of the two functions
1836 were taken in different shared objects.
1838 The effect of this is that GCC may, effectively, mark inline methods with
1839 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1840 appear in the export table of a DSO and do not require a PLT indirection
1841 when used within the DSO@. Enabling this option can have a dramatic effect
1842 on load and link times of a DSO as it massively reduces the size of the
1843 dynamic export table when the library makes heavy use of templates.
1845 The behavior of this switch is not quite the same as marking the
1846 methods as hidden directly, because it does not affect static variables
1847 local to the function or cause the compiler to deduce that
1848 the function is defined in only one shared object.
1850 You may mark a method as having a visibility explicitly to negate the
1851 effect of the switch for that method. For example, if you do want to
1852 compare pointers to a particular inline method, you might mark it as
1853 having default visibility. Marking the enclosing class with explicit
1854 visibility will have no effect.
1856 Explicitly instantiated inline methods are unaffected by this option
1857 as their linkage might otherwise cross a shared library boundary.
1858 @xref{Template Instantiation}.
1860 @item -fvisibility-ms-compat
1861 @opindex fvisibility-ms-compat
1862 This flag attempts to use visibility settings to make GCC's C++
1863 linkage model compatible with that of Microsoft Visual Studio.
1865 The flag makes these changes to GCC's linkage model:
1869 It sets the default visibility to @code{hidden}, like
1870 @option{-fvisibility=hidden}.
1873 Types, but not their members, are not hidden by default.
1876 The One Definition Rule is relaxed for types without explicit
1877 visibility specifications which are defined in more than one different
1878 shared object: those declarations are permitted if they would have
1879 been permitted when this option was not used.
1882 In new code it is better to use @option{-fvisibility=hidden} and
1883 export those classes which are intended to be externally visible.
1884 Unfortunately it is possible for code to rely, perhaps accidentally,
1885 on the Visual Studio behavior.
1887 Among the consequences of these changes are that static data members
1888 of the same type with the same name but defined in different shared
1889 objects will be different, so changing one will not change the other;
1890 and that pointers to function members defined in different shared
1891 objects may not compare equal. When this flag is given, it is a
1892 violation of the ODR to define types with the same name differently.
1896 Do not use weak symbol support, even if it is provided by the linker.
1897 By default, G++ will use weak symbols if they are available. This
1898 option exists only for testing, and should not be used by end-users;
1899 it will result in inferior code and has no benefits. This option may
1900 be removed in a future release of G++.
1904 Do not search for header files in the standard directories specific to
1905 C++, but do still search the other standard directories. (This option
1906 is used when building the C++ library.)
1909 In addition, these optimization, warning, and code generation options
1910 have meanings only for C++ programs:
1913 @item -fno-default-inline
1914 @opindex fno-default-inline
1915 Do not assume @samp{inline} for functions defined inside a class scope.
1916 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1917 functions will have linkage like inline functions; they just won't be
1920 @item -Wabi @r{(C++ and Objective-C++ only)}
1923 Warn when G++ generates code that is probably not compatible with the
1924 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1925 all such cases, there are probably some cases that are not warned about,
1926 even though G++ is generating incompatible code. There may also be
1927 cases where warnings are emitted even though the code that is generated
1930 You should rewrite your code to avoid these warnings if you are
1931 concerned about the fact that code generated by G++ may not be binary
1932 compatible with code generated by other compilers.
1934 The known incompatibilities at this point include:
1939 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1940 pack data into the same byte as a base class. For example:
1943 struct A @{ virtual void f(); int f1 : 1; @};
1944 struct B : public A @{ int f2 : 1; @};
1948 In this case, G++ will place @code{B::f2} into the same byte
1949 as@code{A::f1}; other compilers will not. You can avoid this problem
1950 by explicitly padding @code{A} so that its size is a multiple of the
1951 byte size on your platform; that will cause G++ and other compilers to
1952 layout @code{B} identically.
1955 Incorrect handling of tail-padding for virtual bases. G++ does not use
1956 tail padding when laying out virtual bases. For example:
1959 struct A @{ virtual void f(); char c1; @};
1960 struct B @{ B(); char c2; @};
1961 struct C : public A, public virtual B @{@};
1965 In this case, G++ will not place @code{B} into the tail-padding for
1966 @code{A}; other compilers will. You can avoid this problem by
1967 explicitly padding @code{A} so that its size is a multiple of its
1968 alignment (ignoring virtual base classes); that will cause G++ and other
1969 compilers to layout @code{C} identically.
1972 Incorrect handling of bit-fields with declared widths greater than that
1973 of their underlying types, when the bit-fields appear in a union. For
1977 union U @{ int i : 4096; @};
1981 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1982 union too small by the number of bits in an @code{int}.
1985 Empty classes can be placed at incorrect offsets. For example:
1995 struct C : public B, public A @{@};
1999 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2000 it should be placed at offset zero. G++ mistakenly believes that the
2001 @code{A} data member of @code{B} is already at offset zero.
2004 Names of template functions whose types involve @code{typename} or
2005 template template parameters can be mangled incorrectly.
2008 template <typename Q>
2009 void f(typename Q::X) @{@}
2011 template <template <typename> class Q>
2012 void f(typename Q<int>::X) @{@}
2016 Instantiations of these templates may be mangled incorrectly.
2020 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2021 @opindex Wctor-dtor-privacy
2022 @opindex Wno-ctor-dtor-privacy
2023 Warn when a class seems unusable because all the constructors or
2024 destructors in that class are private, and it has neither friends nor
2025 public static member functions.
2027 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2028 @opindex Wnon-virtual-dtor
2029 @opindex Wno-non-virtual-dtor
2030 Warn when a class has virtual functions and accessible non-virtual
2031 destructor, in which case it would be possible but unsafe to delete
2032 an instance of a derived class through a pointer to the base class.
2033 This warning is also enabled if -Weffc++ is specified.
2035 @item -Wreorder @r{(C++ and Objective-C++ only)}
2037 @opindex Wno-reorder
2038 @cindex reordering, warning
2039 @cindex warning for reordering of member initializers
2040 Warn when the order of member initializers given in the code does not
2041 match the order in which they must be executed. For instance:
2047 A(): j (0), i (1) @{ @}
2051 The compiler will rearrange the member initializers for @samp{i}
2052 and @samp{j} to match the declaration order of the members, emitting
2053 a warning to that effect. This warning is enabled by @option{-Wall}.
2056 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2059 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2062 Warn about violations of the following style guidelines from Scott Meyers'
2063 @cite{Effective C++} book:
2067 Item 11: Define a copy constructor and an assignment operator for classes
2068 with dynamically allocated memory.
2071 Item 12: Prefer initialization to assignment in constructors.
2074 Item 14: Make destructors virtual in base classes.
2077 Item 15: Have @code{operator=} return a reference to @code{*this}.
2080 Item 23: Don't try to return a reference when you must return an object.
2084 Also warn about violations of the following style guidelines from
2085 Scott Meyers' @cite{More Effective C++} book:
2089 Item 6: Distinguish between prefix and postfix forms of increment and
2090 decrement operators.
2093 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2097 When selecting this option, be aware that the standard library
2098 headers do not obey all of these guidelines; use @samp{grep -v}
2099 to filter out those warnings.
2101 @item -Wno-deprecated @r{(C++ and Objective-C++ only)}
2102 @opindex Wno-deprecated
2103 @opindex Wdeprecated
2104 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2106 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2107 @opindex Wstrict-null-sentinel
2108 @opindex Wno-strict-null-sentinel
2109 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2110 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2111 to @code{__null}. Although it is a null pointer constant not a null pointer,
2112 it is guaranteed to of the same size as a pointer. But this use is
2113 not portable across different compilers.
2115 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2116 @opindex Wno-non-template-friend
2117 @opindex Wnon-template-friend
2118 Disable warnings when non-templatized friend functions are declared
2119 within a template. Since the advent of explicit template specification
2120 support in G++, if the name of the friend is an unqualified-id (i.e.,
2121 @samp{friend foo(int)}), the C++ language specification demands that the
2122 friend declare or define an ordinary, nontemplate function. (Section
2123 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2124 could be interpreted as a particular specialization of a templatized
2125 function. Because this non-conforming behavior is no longer the default
2126 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2127 check existing code for potential trouble spots and is on by default.
2128 This new compiler behavior can be turned off with
2129 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2130 but disables the helpful warning.
2132 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2133 @opindex Wold-style-cast
2134 @opindex Wno-old-style-cast
2135 Warn if an old-style (C-style) cast to a non-void type is used within
2136 a C++ program. The new-style casts (@samp{dynamic_cast},
2137 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2138 less vulnerable to unintended effects and much easier to search for.
2140 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2141 @opindex Woverloaded-virtual
2142 @opindex Wno-overloaded-virtual
2143 @cindex overloaded virtual fn, warning
2144 @cindex warning for overloaded virtual fn
2145 Warn when a function declaration hides virtual functions from a
2146 base class. For example, in:
2153 struct B: public A @{
2158 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2166 will fail to compile.
2168 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2169 @opindex Wno-pmf-conversions
2170 @opindex Wpmf-conversions
2171 Disable the diagnostic for converting a bound pointer to member function
2174 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2175 @opindex Wsign-promo
2176 @opindex Wno-sign-promo
2177 Warn when overload resolution chooses a promotion from unsigned or
2178 enumerated type to a signed type, over a conversion to an unsigned type of
2179 the same size. Previous versions of G++ would try to preserve
2180 unsignedness, but the standard mandates the current behavior.
2185 A& operator = (int);
2195 In this example, G++ will synthesize a default @samp{A& operator =
2196 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2199 @node Objective-C and Objective-C++ Dialect Options
2200 @section Options Controlling Objective-C and Objective-C++ Dialects
2202 @cindex compiler options, Objective-C and Objective-C++
2203 @cindex Objective-C and Objective-C++ options, command line
2204 @cindex options, Objective-C and Objective-C++
2205 (NOTE: This manual does not describe the Objective-C and Objective-C++
2206 languages themselves. See @xref{Standards,,Language Standards
2207 Supported by GCC}, for references.)
2209 This section describes the command-line options that are only meaningful
2210 for Objective-C and Objective-C++ programs, but you can also use most of
2211 the language-independent GNU compiler options.
2212 For example, you might compile a file @code{some_class.m} like this:
2215 gcc -g -fgnu-runtime -O -c some_class.m
2219 In this example, @option{-fgnu-runtime} is an option meant only for
2220 Objective-C and Objective-C++ programs; you can use the other options with
2221 any language supported by GCC@.
2223 Note that since Objective-C is an extension of the C language, Objective-C
2224 compilations may also use options specific to the C front-end (e.g.,
2225 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2226 C++-specific options (e.g., @option{-Wabi}).
2228 Here is a list of options that are @emph{only} for compiling Objective-C
2229 and Objective-C++ programs:
2232 @item -fconstant-string-class=@var{class-name}
2233 @opindex fconstant-string-class
2234 Use @var{class-name} as the name of the class to instantiate for each
2235 literal string specified with the syntax @code{@@"@dots{}"}. The default
2236 class name is @code{NXConstantString} if the GNU runtime is being used, and
2237 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2238 @option{-fconstant-cfstrings} option, if also present, will override the
2239 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2240 to be laid out as constant CoreFoundation strings.
2243 @opindex fgnu-runtime
2244 Generate object code compatible with the standard GNU Objective-C
2245 runtime. This is the default for most types of systems.
2247 @item -fnext-runtime
2248 @opindex fnext-runtime
2249 Generate output compatible with the NeXT runtime. This is the default
2250 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2251 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2254 @item -fno-nil-receivers
2255 @opindex fno-nil-receivers
2256 Assume that all Objective-C message dispatches (e.g.,
2257 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2258 is not @code{nil}. This allows for more efficient entry points in the runtime
2259 to be used. Currently, this option is only available in conjunction with
2260 the NeXT runtime on Mac OS X 10.3 and later.
2262 @item -fobjc-call-cxx-cdtors
2263 @opindex fobjc-call-cxx-cdtors
2264 For each Objective-C class, check if any of its instance variables is a
2265 C++ object with a non-trivial default constructor. If so, synthesize a
2266 special @code{- (id) .cxx_construct} instance method that will run
2267 non-trivial default constructors on any such instance variables, in order,
2268 and then return @code{self}. Similarly, check if any instance variable
2269 is a C++ object with a non-trivial destructor, and if so, synthesize a
2270 special @code{- (void) .cxx_destruct} method that will run
2271 all such default destructors, in reverse order.
2273 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2274 thusly generated will only operate on instance variables declared in the
2275 current Objective-C class, and not those inherited from superclasses. It
2276 is the responsibility of the Objective-C runtime to invoke all such methods
2277 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2278 will be invoked by the runtime immediately after a new object
2279 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2280 be invoked immediately before the runtime deallocates an object instance.
2282 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2283 support for invoking the @code{- (id) .cxx_construct} and
2284 @code{- (void) .cxx_destruct} methods.
2286 @item -fobjc-direct-dispatch
2287 @opindex fobjc-direct-dispatch
2288 Allow fast jumps to the message dispatcher. On Darwin this is
2289 accomplished via the comm page.
2291 @item -fobjc-exceptions
2292 @opindex fobjc-exceptions
2293 Enable syntactic support for structured exception handling in Objective-C,
2294 similar to what is offered by C++ and Java. This option is
2295 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2304 @@catch (AnObjCClass *exc) @{
2311 @@catch (AnotherClass *exc) @{
2314 @@catch (id allOthers) @{
2324 The @code{@@throw} statement may appear anywhere in an Objective-C or
2325 Objective-C++ program; when used inside of a @code{@@catch} block, the
2326 @code{@@throw} may appear without an argument (as shown above), in which case
2327 the object caught by the @code{@@catch} will be rethrown.
2329 Note that only (pointers to) Objective-C objects may be thrown and
2330 caught using this scheme. When an object is thrown, it will be caught
2331 by the nearest @code{@@catch} clause capable of handling objects of that type,
2332 analogously to how @code{catch} blocks work in C++ and Java. A
2333 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2334 any and all Objective-C exceptions not caught by previous @code{@@catch}
2337 The @code{@@finally} clause, if present, will be executed upon exit from the
2338 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2339 regardless of whether any exceptions are thrown, caught or rethrown
2340 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2341 of the @code{finally} clause in Java.
2343 There are several caveats to using the new exception mechanism:
2347 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2348 idioms provided by the @code{NSException} class, the new
2349 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2350 systems, due to additional functionality needed in the (NeXT) Objective-C
2354 As mentioned above, the new exceptions do not support handling
2355 types other than Objective-C objects. Furthermore, when used from
2356 Objective-C++, the Objective-C exception model does not interoperate with C++
2357 exceptions at this time. This means you cannot @code{@@throw} an exception
2358 from Objective-C and @code{catch} it in C++, or vice versa
2359 (i.e., @code{throw @dots{} @@catch}).
2362 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2363 blocks for thread-safe execution:
2366 @@synchronized (ObjCClass *guard) @{
2371 Upon entering the @code{@@synchronized} block, a thread of execution shall
2372 first check whether a lock has been placed on the corresponding @code{guard}
2373 object by another thread. If it has, the current thread shall wait until
2374 the other thread relinquishes its lock. Once @code{guard} becomes available,
2375 the current thread will place its own lock on it, execute the code contained in
2376 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2377 making @code{guard} available to other threads).
2379 Unlike Java, Objective-C does not allow for entire methods to be marked
2380 @code{@@synchronized}. Note that throwing exceptions out of
2381 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2382 to be unlocked properly.
2386 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2388 @item -freplace-objc-classes
2389 @opindex freplace-objc-classes
2390 Emit a special marker instructing @command{ld(1)} not to statically link in
2391 the resulting object file, and allow @command{dyld(1)} to load it in at
2392 run time instead. This is used in conjunction with the Fix-and-Continue
2393 debugging mode, where the object file in question may be recompiled and
2394 dynamically reloaded in the course of program execution, without the need
2395 to restart the program itself. Currently, Fix-and-Continue functionality
2396 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2401 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2402 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2403 compile time) with static class references that get initialized at load time,
2404 which improves run-time performance. Specifying the @option{-fzero-link} flag
2405 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2406 to be retained. This is useful in Zero-Link debugging mode, since it allows
2407 for individual class implementations to be modified during program execution.
2411 Dump interface declarations for all classes seen in the source file to a
2412 file named @file{@var{sourcename}.decl}.
2414 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2415 @opindex Wassign-intercept
2416 @opindex Wno-assign-intercept
2417 Warn whenever an Objective-C assignment is being intercepted by the
2420 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2421 @opindex Wno-protocol
2423 If a class is declared to implement a protocol, a warning is issued for
2424 every method in the protocol that is not implemented by the class. The
2425 default behavior is to issue a warning for every method not explicitly
2426 implemented in the class, even if a method implementation is inherited
2427 from the superclass. If you use the @option{-Wno-protocol} option, then
2428 methods inherited from the superclass are considered to be implemented,
2429 and no warning is issued for them.
2431 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2433 @opindex Wno-selector
2434 Warn if multiple methods of different types for the same selector are
2435 found during compilation. The check is performed on the list of methods
2436 in the final stage of compilation. Additionally, a check is performed
2437 for each selector appearing in a @code{@@selector(@dots{})}
2438 expression, and a corresponding method for that selector has been found
2439 during compilation. Because these checks scan the method table only at
2440 the end of compilation, these warnings are not produced if the final
2441 stage of compilation is not reached, for example because an error is
2442 found during compilation, or because the @option{-fsyntax-only} option is
2445 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2446 @opindex Wstrict-selector-match
2447 @opindex Wno-strict-selector-match
2448 Warn if multiple methods with differing argument and/or return types are
2449 found for a given selector when attempting to send a message using this
2450 selector to a receiver of type @code{id} or @code{Class}. When this flag
2451 is off (which is the default behavior), the compiler will omit such warnings
2452 if any differences found are confined to types which share the same size
2455 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2456 @opindex Wundeclared-selector
2457 @opindex Wno-undeclared-selector
2458 Warn if a @code{@@selector(@dots{})} expression referring to an
2459 undeclared selector is found. A selector is considered undeclared if no
2460 method with that name has been declared before the
2461 @code{@@selector(@dots{})} expression, either explicitly in an
2462 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2463 an @code{@@implementation} section. This option always performs its
2464 checks as soon as a @code{@@selector(@dots{})} expression is found,
2465 while @option{-Wselector} only performs its checks in the final stage of
2466 compilation. This also enforces the coding style convention
2467 that methods and selectors must be declared before being used.
2469 @item -print-objc-runtime-info
2470 @opindex print-objc-runtime-info
2471 Generate C header describing the largest structure that is passed by
2476 @node Language Independent Options
2477 @section Options to Control Diagnostic Messages Formatting
2478 @cindex options to control diagnostics formatting
2479 @cindex diagnostic messages
2480 @cindex message formatting
2482 Traditionally, diagnostic messages have been formatted irrespective of
2483 the output device's aspect (e.g.@: its width, @dots{}). The options described
2484 below can be used to control the diagnostic messages formatting
2485 algorithm, e.g.@: how many characters per line, how often source location
2486 information should be reported. Right now, only the C++ front end can
2487 honor these options. However it is expected, in the near future, that
2488 the remaining front ends would be able to digest them correctly.
2491 @item -fmessage-length=@var{n}
2492 @opindex fmessage-length
2493 Try to format error messages so that they fit on lines of about @var{n}
2494 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2495 the front ends supported by GCC@. If @var{n} is zero, then no
2496 line-wrapping will be done; each error message will appear on a single
2499 @opindex fdiagnostics-show-location
2500 @item -fdiagnostics-show-location=once
2501 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2502 reporter to emit @emph{once} source location information; that is, in
2503 case the message is too long to fit on a single physical line and has to
2504 be wrapped, the source location won't be emitted (as prefix) again,
2505 over and over, in subsequent continuation lines. This is the default
2508 @item -fdiagnostics-show-location=every-line
2509 Only meaningful in line-wrapping mode. Instructs the diagnostic
2510 messages reporter to emit the same source location information (as
2511 prefix) for physical lines that result from the process of breaking
2512 a message which is too long to fit on a single line.
2514 @item -fdiagnostics-show-option
2515 @opindex fdiagnostics-show-option
2516 This option instructs the diagnostic machinery to add text to each
2517 diagnostic emitted, which indicates which command line option directly
2518 controls that diagnostic, when such an option is known to the
2519 diagnostic machinery.
2521 @item -Wcoverage-mismatch
2522 @opindex Wcoverage-mismatch
2523 Warn if feedback profiles do not match when using the
2524 @option{-fprofile-use} option.
2525 If a source file was changed between @option{-fprofile-gen} and
2526 @option{-fprofile-use}, the files with the profile feedback can fail
2527 to match the source file and GCC can not use the profile feedback
2528 information. By default, GCC emits an error message in this case.
2529 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2530 error. GCC does not use appropriate feedback profiles, so using this
2531 option can result in poorly optimized code. This option is useful
2532 only in the case of very minor changes such as bug fixes to an
2537 @node Warning Options
2538 @section Options to Request or Suppress Warnings
2539 @cindex options to control warnings
2540 @cindex warning messages
2541 @cindex messages, warning
2542 @cindex suppressing warnings
2544 Warnings are diagnostic messages that report constructions which
2545 are not inherently erroneous but which are risky or suggest there
2546 may have been an error.
2548 The following language-independent options do not enable specific
2549 warnings but control the kinds of diagnostics produced by GCC.
2552 @cindex syntax checking
2554 @opindex fsyntax-only
2555 Check the code for syntax errors, but don't do anything beyond that.
2559 Inhibit all warning messages.
2564 Make all warnings into errors.
2569 Make the specified warning into an error. The specifier for a warning
2570 is appended, for example @option{-Werror=switch} turns the warnings
2571 controlled by @option{-Wswitch} into errors. This switch takes a
2572 negative form, to be used to negate @option{-Werror} for specific
2573 warnings, for example @option{-Wno-error=switch} makes
2574 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2575 is in effect. You can use the @option{-fdiagnostics-show-option}
2576 option to have each controllable warning amended with the option which
2577 controls it, to determine what to use with this option.
2579 Note that specifying @option{-Werror=}@var{foo} automatically implies
2580 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2583 @item -Wfatal-errors
2584 @opindex Wfatal-errors
2585 @opindex Wno-fatal-errors
2586 This option causes the compiler to abort compilation on the first error
2587 occurred rather than trying to keep going and printing further error
2592 You can request many specific warnings with options beginning
2593 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2594 implicit declarations. Each of these specific warning options also
2595 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2596 example, @option{-Wno-implicit}. This manual lists only one of the
2597 two forms, whichever is not the default. For further,
2598 language-specific options also refer to @ref{C++ Dialect Options} and
2599 @ref{Objective-C and Objective-C++ Dialect Options}.
2604 Issue all the warnings demanded by strict ISO C and ISO C++;
2605 reject all programs that use forbidden extensions, and some other
2606 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2607 version of the ISO C standard specified by any @option{-std} option used.
2609 Valid ISO C and ISO C++ programs should compile properly with or without
2610 this option (though a rare few will require @option{-ansi} or a
2611 @option{-std} option specifying the required version of ISO C)@. However,
2612 without this option, certain GNU extensions and traditional C and C++
2613 features are supported as well. With this option, they are rejected.
2615 @option{-pedantic} does not cause warning messages for use of the
2616 alternate keywords whose names begin and end with @samp{__}. Pedantic
2617 warnings are also disabled in the expression that follows
2618 @code{__extension__}. However, only system header files should use
2619 these escape routes; application programs should avoid them.
2620 @xref{Alternate Keywords}.
2622 Some users try to use @option{-pedantic} to check programs for strict ISO
2623 C conformance. They soon find that it does not do quite what they want:
2624 it finds some non-ISO practices, but not all---only those for which
2625 ISO C @emph{requires} a diagnostic, and some others for which
2626 diagnostics have been added.
2628 A feature to report any failure to conform to ISO C might be useful in
2629 some instances, but would require considerable additional work and would
2630 be quite different from @option{-pedantic}. We don't have plans to
2631 support such a feature in the near future.
2633 Where the standard specified with @option{-std} represents a GNU
2634 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2635 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2636 extended dialect is based. Warnings from @option{-pedantic} are given
2637 where they are required by the base standard. (It would not make sense
2638 for such warnings to be given only for features not in the specified GNU
2639 C dialect, since by definition the GNU dialects of C include all
2640 features the compiler supports with the given option, and there would be
2641 nothing to warn about.)
2643 @item -pedantic-errors
2644 @opindex pedantic-errors
2645 Like @option{-pedantic}, except that errors are produced rather than
2651 This enables all the warnings about constructions that some users
2652 consider questionable, and that are easy to avoid (or modify to
2653 prevent the warning), even in conjunction with macros. This also
2654 enables some language-specific warnings described in @ref{C++ Dialect
2655 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2657 @option{-Wall} turns on the following warning flags:
2659 @gccoptlist{-Waddress @gol
2660 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2662 -Wchar-subscripts @gol
2664 -Wimplicit-function-declaration @gol
2667 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2668 -Wmissing-braces @gol
2674 -Wsequence-point @gol
2675 -Wsign-compare @r{(only in C++)} @gol
2676 -Wstrict-aliasing @gol
2677 -Wstrict-overflow=1 @gol
2680 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2681 -Wunknown-pragmas @gol
2682 -Wunused-function @gol
2685 -Wunused-variable @gol
2686 -Wvolatile-register-var @gol
2689 Note that some warning flags are not implied by @option{-Wall}. Some of
2690 them warn about constructions that users generally do not consider
2691 questionable, but which occasionally you might wish to check for;
2692 others warn about constructions that are necessary or hard to avoid in
2693 some cases, and there is no simple way to modify the code to suppress
2694 the warning. Some of them are enabled by @option{-Wextra} but many of
2695 them must be enabled individually.
2701 This enables some extra warning flags that are not enabled by
2702 @option{-Wall}. (This option used to be called @option{-W}. The older
2703 name is still supported, but the newer name is more descriptive.)
2705 @gccoptlist{-Wclobbered @gol
2707 -Wignored-qualifiers @gol
2708 -Wmissing-field-initializers @gol
2709 -Wmissing-parameter-type @r{(C only)} @gol
2710 -Wold-style-declaration @r{(C only)} @gol
2711 -Woverride-init @gol
2714 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2715 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2718 The option @option{-Wextra} also prints warning messages for the
2724 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2725 @samp{>}, or @samp{>=}.
2728 (C++ only) An enumerator and a non-enumerator both appear in a
2729 conditional expression.
2732 (C++ only) A non-static reference or non-static @samp{const} member
2733 appears in a class without constructors.
2736 (C++ only) Ambiguous virtual bases.
2739 (C++ only) Subscripting an array which has been declared @samp{register}.
2742 (C++ only) Taking the address of a variable which has been declared
2746 (C++ only) A base class is not initialized in a derived class' copy
2754 Inhibit warning messages about the use of @samp{#import}.
2756 @item -Wchar-subscripts
2757 @opindex Wchar-subscripts
2758 @opindex Wno-char-subscripts
2759 Warn if an array subscript has type @code{char}. This is a common cause
2760 of error, as programmers often forget that this type is signed on some
2762 This warning is enabled by @option{-Wall}.
2766 @opindex Wno-comment
2767 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2768 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2769 This warning is enabled by @option{-Wall}.
2774 @opindex ffreestanding
2775 @opindex fno-builtin
2776 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2777 the arguments supplied have types appropriate to the format string
2778 specified, and that the conversions specified in the format string make
2779 sense. This includes standard functions, and others specified by format
2780 attributes (@pxref{Function Attributes}), in the @code{printf},
2781 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2782 not in the C standard) families (or other target-specific families).
2783 Which functions are checked without format attributes having been
2784 specified depends on the standard version selected, and such checks of
2785 functions without the attribute specified are disabled by
2786 @option{-ffreestanding} or @option{-fno-builtin}.
2788 The formats are checked against the format features supported by GNU
2789 libc version 2.2. These include all ISO C90 and C99 features, as well
2790 as features from the Single Unix Specification and some BSD and GNU
2791 extensions. Other library implementations may not support all these
2792 features; GCC does not support warning about features that go beyond a
2793 particular library's limitations. However, if @option{-pedantic} is used
2794 with @option{-Wformat}, warnings will be given about format features not
2795 in the selected standard version (but not for @code{strfmon} formats,
2796 since those are not in any version of the C standard). @xref{C Dialect
2797 Options,,Options Controlling C Dialect}.
2799 Since @option{-Wformat} also checks for null format arguments for
2800 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2802 @option{-Wformat} is included in @option{-Wall}. For more control over some
2803 aspects of format checking, the options @option{-Wformat-y2k},
2804 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2805 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2806 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2809 @opindex Wformat-y2k
2810 @opindex Wno-format-y2k
2811 If @option{-Wformat} is specified, also warn about @code{strftime}
2812 formats which may yield only a two-digit year.
2814 @item -Wno-format-extra-args
2815 @opindex Wno-format-extra-args
2816 @opindex Wformat-extra-args
2817 If @option{-Wformat} is specified, do not warn about excess arguments to a
2818 @code{printf} or @code{scanf} format function. The C standard specifies
2819 that such arguments are ignored.
2821 Where the unused arguments lie between used arguments that are
2822 specified with @samp{$} operand number specifications, normally
2823 warnings are still given, since the implementation could not know what
2824 type to pass to @code{va_arg} to skip the unused arguments. However,
2825 in the case of @code{scanf} formats, this option will suppress the
2826 warning if the unused arguments are all pointers, since the Single
2827 Unix Specification says that such unused arguments are allowed.
2829 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2830 @opindex Wno-format-zero-length
2831 @opindex Wformat-zero-length
2832 If @option{-Wformat} is specified, do not warn about zero-length formats.
2833 The C standard specifies that zero-length formats are allowed.
2835 @item -Wformat-nonliteral
2836 @opindex Wformat-nonliteral
2837 @opindex Wno-format-nonliteral
2838 If @option{-Wformat} is specified, also warn if the format string is not a
2839 string literal and so cannot be checked, unless the format function
2840 takes its format arguments as a @code{va_list}.
2842 @item -Wformat-security
2843 @opindex Wformat-security
2844 @opindex Wno-format-security
2845 If @option{-Wformat} is specified, also warn about uses of format
2846 functions that represent possible security problems. At present, this
2847 warns about calls to @code{printf} and @code{scanf} functions where the
2848 format string is not a string literal and there are no format arguments,
2849 as in @code{printf (foo);}. This may be a security hole if the format
2850 string came from untrusted input and contains @samp{%n}. (This is
2851 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2852 in future warnings may be added to @option{-Wformat-security} that are not
2853 included in @option{-Wformat-nonliteral}.)
2857 @opindex Wno-format=2
2858 Enable @option{-Wformat} plus format checks not included in
2859 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2860 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2862 @item -Wnonnull @r{(C and Objective-C only)}
2864 @opindex Wno-nonnull
2865 Warn about passing a null pointer for arguments marked as
2866 requiring a non-null value by the @code{nonnull} function attribute.
2868 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2869 can be disabled with the @option{-Wno-nonnull} option.
2871 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2873 @opindex Wno-init-self
2874 Warn about uninitialized variables which are initialized with themselves.
2875 Note this option can only be used with the @option{-Wuninitialized} option,
2876 which in turn only works with @option{-O1} and above.
2878 For example, GCC will warn about @code{i} being uninitialized in the
2879 following snippet only when @option{-Winit-self} has been specified:
2890 @item -Wimplicit-int @r{(C and Objective-C only)}
2891 @opindex Wimplicit-int
2892 @opindex Wno-implicit-int
2893 Warn when a declaration does not specify a type.
2894 This warning is enabled by @option{-Wall}.
2896 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2897 @opindex Wimplicit-function-declaration
2898 @opindex Wno-implicit-function-declaration
2899 Give a warning whenever a function is used before being declared. In
2900 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2901 enabled by default and it is made into an error by
2902 @option{-pedantic-errors}. This warning is also enabled by
2907 @opindex Wno-implicit
2908 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2909 This warning is enabled by @option{-Wall}.
2911 @item -Wignored-qualifiers @r{(C and C++ only)}
2912 @opindex Wignored-qualifiers
2913 @opindex Wno-ignored-qualifiers
2914 Warn if the return type of a function has a type qualifier
2915 such as @code{const}. For ISO C such a type qualifier has no effect,
2916 since the value returned by a function is not an lvalue.
2917 For C++, the warning is only emitted for scalar types or @code{void}.
2918 ISO C prohibits qualified @code{void} return types on function
2919 definitions, so such return types always receive a warning
2920 even without this option.
2922 This warning is also enabled by @option{-Wextra}.
2927 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2928 function with external linkage, returning int, taking either zero
2929 arguments, two, or three arguments of appropriate types.
2930 This warning is enabled by @option{-Wall}.
2932 @item -Wmissing-braces
2933 @opindex Wmissing-braces
2934 @opindex Wno-missing-braces
2935 Warn if an aggregate or union initializer is not fully bracketed. In
2936 the following example, the initializer for @samp{a} is not fully
2937 bracketed, but that for @samp{b} is fully bracketed.
2940 int a[2][2] = @{ 0, 1, 2, 3 @};
2941 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2944 This warning is enabled by @option{-Wall}.
2946 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2947 @opindex Wmissing-include-dirs
2948 @opindex Wno-missing-include-dirs
2949 Warn if a user-supplied include directory does not exist.
2952 @opindex Wparentheses
2953 @opindex Wno-parentheses
2954 Warn if parentheses are omitted in certain contexts, such
2955 as when there is an assignment in a context where a truth value
2956 is expected, or when operators are nested whose precedence people
2957 often get confused about.
2959 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2960 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2961 interpretation from that of ordinary mathematical notation.
2963 Also warn about constructions where there may be confusion to which
2964 @code{if} statement an @code{else} branch belongs. Here is an example of
2979 In C/C++, every @code{else} branch belongs to the innermost possible
2980 @code{if} statement, which in this example is @code{if (b)}. This is
2981 often not what the programmer expected, as illustrated in the above
2982 example by indentation the programmer chose. When there is the
2983 potential for this confusion, GCC will issue a warning when this flag
2984 is specified. To eliminate the warning, add explicit braces around
2985 the innermost @code{if} statement so there is no way the @code{else}
2986 could belong to the enclosing @code{if}. The resulting code would
3003 This warning is enabled by @option{-Wall}.
3005 @item -Wsequence-point
3006 @opindex Wsequence-point
3007 @opindex Wno-sequence-point
3008 Warn about code that may have undefined semantics because of violations
3009 of sequence point rules in the C and C++ standards.
3011 The C and C++ standards defines the order in which expressions in a C/C++
3012 program are evaluated in terms of @dfn{sequence points}, which represent
3013 a partial ordering between the execution of parts of the program: those
3014 executed before the sequence point, and those executed after it. These
3015 occur after the evaluation of a full expression (one which is not part
3016 of a larger expression), after the evaluation of the first operand of a
3017 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3018 function is called (but after the evaluation of its arguments and the
3019 expression denoting the called function), and in certain other places.
3020 Other than as expressed by the sequence point rules, the order of
3021 evaluation of subexpressions of an expression is not specified. All
3022 these rules describe only a partial order rather than a total order,
3023 since, for example, if two functions are called within one expression
3024 with no sequence point between them, the order in which the functions
3025 are called is not specified. However, the standards committee have
3026 ruled that function calls do not overlap.
3028 It is not specified when between sequence points modifications to the
3029 values of objects take effect. Programs whose behavior depends on this
3030 have undefined behavior; the C and C++ standards specify that ``Between
3031 the previous and next sequence point an object shall have its stored
3032 value modified at most once by the evaluation of an expression.
3033 Furthermore, the prior value shall be read only to determine the value
3034 to be stored.''. If a program breaks these rules, the results on any
3035 particular implementation are entirely unpredictable.
3037 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3038 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3039 diagnosed by this option, and it may give an occasional false positive
3040 result, but in general it has been found fairly effective at detecting
3041 this sort of problem in programs.
3043 The standard is worded confusingly, therefore there is some debate
3044 over the precise meaning of the sequence point rules in subtle cases.
3045 Links to discussions of the problem, including proposed formal
3046 definitions, may be found on the GCC readings page, at
3047 @w{@uref{http://gcc.gnu.org/readings.html}}.
3049 This warning is enabled by @option{-Wall} for C and C++.
3052 @opindex Wreturn-type
3053 @opindex Wno-return-type
3054 Warn whenever a function is defined with a return-type that defaults
3055 to @code{int}. Also warn about any @code{return} statement with no
3056 return-value in a function whose return-type is not @code{void}
3057 (falling off the end of the function body is considered returning
3058 without a value), and about a @code{return} statement with a
3059 expression in a function whose return-type is @code{void}.
3061 For C++, a function without return type always produces a diagnostic
3062 message, even when @option{-Wno-return-type} is specified. The only
3063 exceptions are @samp{main} and functions defined in system headers.
3065 This warning is enabled by @option{-Wall}.
3070 Warn whenever a @code{switch} statement has an index of enumerated type
3071 and lacks a @code{case} for one or more of the named codes of that
3072 enumeration. (The presence of a @code{default} label prevents this
3073 warning.) @code{case} labels outside the enumeration range also
3074 provoke warnings when this option is used.
3075 This warning is enabled by @option{-Wall}.
3077 @item -Wswitch-default
3078 @opindex Wswitch-default
3079 @opindex Wno-switch-default
3080 Warn whenever a @code{switch} statement does not have a @code{default}
3084 @opindex Wswitch-enum
3085 @opindex Wno-switch-enum
3086 Warn whenever a @code{switch} statement has an index of enumerated type
3087 and lacks a @code{case} for one or more of the named codes of that
3088 enumeration. @code{case} labels outside the enumeration range also
3089 provoke warnings when this option is used.
3093 @opindex Wno-trigraphs
3094 Warn if any trigraphs are encountered that might change the meaning of
3095 the program (trigraphs within comments are not warned about).
3096 This warning is enabled by @option{-Wall}.
3098 @item -Wunused-function
3099 @opindex Wunused-function
3100 @opindex Wno-unused-function
3101 Warn whenever a static function is declared but not defined or a
3102 non-inline static function is unused.
3103 This warning is enabled by @option{-Wall}.
3105 @item -Wunused-label
3106 @opindex Wunused-label
3107 @opindex Wno-unused-label
3108 Warn whenever a label is declared but not used.
3109 This warning is enabled by @option{-Wall}.
3111 To suppress this warning use the @samp{unused} attribute
3112 (@pxref{Variable Attributes}).
3114 @item -Wunused-parameter
3115 @opindex Wunused-parameter
3116 @opindex Wno-unused-parameter
3117 Warn whenever a function parameter is unused aside from its declaration.
3119 To suppress this warning use the @samp{unused} attribute
3120 (@pxref{Variable Attributes}).
3122 @item -Wunused-variable
3123 @opindex Wunused-variable
3124 @opindex Wno-unused-variable
3125 Warn whenever a local variable or non-constant static variable is unused
3126 aside from its declaration.
3127 This warning is enabled by @option{-Wall}.
3129 To suppress this warning use the @samp{unused} attribute
3130 (@pxref{Variable Attributes}).
3132 @item -Wunused-value
3133 @opindex Wunused-value
3134 @opindex Wno-unused-value
3135 Warn whenever a statement computes a result that is explicitly not
3136 used. To suppress this warning cast the unused expression to
3137 @samp{void}. This includes an expression-statement or the left-hand
3138 side of a comma expression that contains no side effects. For example,
3139 an expression such as @samp{x[i,j]} will cause a warning, while
3140 @samp{x[(void)i,j]} will not.
3142 This warning is enabled by @option{-Wall}.
3147 All the above @option{-Wunused} options combined.
3149 In order to get a warning about an unused function parameter, you must
3150 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3151 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3153 @item -Wuninitialized
3154 @opindex Wuninitialized
3155 @opindex Wno-uninitialized
3156 Warn if an automatic variable is used without first being initialized or
3157 if a variable may be clobbered by a @code{setjmp} call.
3159 These warnings are possible only in optimizing compilation,
3160 because they require data flow information that is computed only
3161 when optimizing. If you do not specify @option{-O}, you will not get
3162 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
3163 requiring @option{-O}.
3165 If you want to warn about code which uses the uninitialized value of the
3166 variable in its own initializer, use the @option{-Winit-self} option.
3168 These warnings occur for individual uninitialized or clobbered
3169 elements of structure, union or array variables as well as for
3170 variables which are uninitialized or clobbered as a whole. They do
3171 not occur for variables or elements declared @code{volatile}. Because
3172 these warnings depend on optimization, the exact variables or elements
3173 for which there are warnings will depend on the precise optimization
3174 options and version of GCC used.
3176 Note that there may be no warning about a variable that is used only
3177 to compute a value that itself is never used, because such
3178 computations may be deleted by data flow analysis before the warnings
3181 These warnings are made optional because GCC is not smart
3182 enough to see all the reasons why the code might be correct
3183 despite appearing to have an error. Here is one example of how
3204 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3205 always initialized, but GCC doesn't know this. Here is
3206 another common case:
3211 if (change_y) save_y = y, y = new_y;
3213 if (change_y) y = save_y;
3218 This has no bug because @code{save_y} is used only if it is set.
3220 @cindex @code{longjmp} warnings
3221 This option also warns when a non-volatile automatic variable might be
3222 changed by a call to @code{longjmp}. These warnings as well are possible
3223 only in optimizing compilation.
3225 The compiler sees only the calls to @code{setjmp}. It cannot know
3226 where @code{longjmp} will be called; in fact, a signal handler could
3227 call it at any point in the code. As a result, you may get a warning
3228 even when there is in fact no problem because @code{longjmp} cannot
3229 in fact be called at the place which would cause a problem.
3231 Some spurious warnings can be avoided if you declare all the functions
3232 you use that never return as @code{noreturn}. @xref{Function
3235 This warning is enabled by @option{-Wall} or @option{-Wextra} in
3236 optimizing compilations (@option{-O1} and above).
3238 @item -Wunknown-pragmas
3239 @opindex Wunknown-pragmas
3240 @opindex Wno-unknown-pragmas
3241 @cindex warning for unknown pragmas
3242 @cindex unknown pragmas, warning
3243 @cindex pragmas, warning of unknown
3244 Warn when a #pragma directive is encountered which is not understood by
3245 GCC@. If this command line option is used, warnings will even be issued
3246 for unknown pragmas in system header files. This is not the case if
3247 the warnings were only enabled by the @option{-Wall} command line option.
3250 @opindex Wno-pragmas
3252 Do not warn about misuses of pragmas, such as incorrect parameters,
3253 invalid syntax, or conflicts between pragmas. See also
3254 @samp{-Wunknown-pragmas}.
3256 @item -Wstrict-aliasing
3257 @opindex Wstrict-aliasing
3258 @opindex Wno-strict-aliasing
3259 This option is only active when @option{-fstrict-aliasing} is active.
3260 It warns about code which might break the strict aliasing rules that the
3261 compiler is using for optimization. The warning does not catch all
3262 cases, but does attempt to catch the more common pitfalls. It is
3263 included in @option{-Wall}.
3264 It is equivalent to @option{-Wstrict-aliasing=3}
3266 @item -Wstrict-aliasing=n
3267 @opindex Wstrict-aliasing=n
3268 @opindex Wno-strict-aliasing=n
3269 This option is only active when @option{-fstrict-aliasing} is active.
3270 It warns about code which might break the strict aliasing rules that the
3271 compiler is using for optimization.
3272 Higher levels correspond to higher accuracy (fewer false positives).
3273 Higher levels also correspond to more effort, similar to the way -O works.
3274 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3277 Level 1: Most aggressive, quick, least accurate.
3278 Possibly useful when higher levels
3279 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3280 false negatives. However, it has many false positives.
3281 Warns for all pointer conversions between possibly incompatible types,
3282 even if never dereferenced. Runs in the frontend only.
3284 Level 2: Aggressive, quick, not too precise.
3285 May still have many false positives (not as many as level 1 though),
3286 and few false negatives (but possibly more than level 1).
3287 Unlike level 1, it only warns when an address is taken. Warns about
3288 incomplete types. Runs in the frontend only.
3290 Level 3 (default for @option{-Wstrict-aliasing}):
3291 Should have very few false positives and few false
3292 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3293 Takes care of the common punn+dereference pattern in the frontend:
3294 @code{*(int*)&some_float}.
3295 If optimization is enabled, it also runs in the backend, where it deals
3296 with multiple statement cases using flow-sensitive points-to information.
3297 Only warns when the converted pointer is dereferenced.
3298 Does not warn about incomplete types.
3300 @item -Wstrict-overflow
3301 @itemx -Wstrict-overflow=@var{n}
3302 @opindex Wstrict-overflow
3303 @opindex Wno-strict-overflow
3304 This option is only active when @option{-fstrict-overflow} is active.
3305 It warns about cases where the compiler optimizes based on the
3306 assumption that signed overflow does not occur. Note that it does not
3307 warn about all cases where the code might overflow: it only warns
3308 about cases where the compiler implements some optimization. Thus
3309 this warning depends on the optimization level.
3311 An optimization which assumes that signed overflow does not occur is
3312 perfectly safe if the values of the variables involved are such that
3313 overflow never does, in fact, occur. Therefore this warning can
3314 easily give a false positive: a warning about code which is not
3315 actually a problem. To help focus on important issues, several
3316 warning levels are defined. No warnings are issued for the use of
3317 undefined signed overflow when estimating how many iterations a loop
3318 will require, in particular when determining whether a loop will be
3322 @item -Wstrict-overflow=1
3323 Warn about cases which are both questionable and easy to avoid. For
3324 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3325 compiler will simplify this to @code{1}. This level of
3326 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3327 are not, and must be explicitly requested.
3329 @item -Wstrict-overflow=2
3330 Also warn about other cases where a comparison is simplified to a
3331 constant. For example: @code{abs (x) >= 0}. This can only be
3332 simplified when @option{-fstrict-overflow} is in effect, because
3333 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3334 zero. @option{-Wstrict-overflow} (with no level) is the same as
3335 @option{-Wstrict-overflow=2}.
3337 @item -Wstrict-overflow=3
3338 Also warn about other cases where a comparison is simplified. For
3339 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3341 @item -Wstrict-overflow=4
3342 Also warn about other simplifications not covered by the above cases.
3343 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3345 @item -Wstrict-overflow=5
3346 Also warn about cases where the compiler reduces the magnitude of a
3347 constant involved in a comparison. For example: @code{x + 2 > y} will
3348 be simplified to @code{x + 1 >= y}. This is reported only at the
3349 highest warning level because this simplification applies to many
3350 comparisons, so this warning level will give a very large number of
3354 @item -Warray-bounds
3355 @opindex Wno-array-bounds
3356 @opindex Warray-bounds
3357 This option is only active when @option{-ftree-vrp} is active
3358 (default for -O2 and above). It warns about subscripts to arrays
3359 that are always out of bounds. This warning is enabled by @option{-Wall}.
3361 @item -Wno-div-by-zero
3362 @opindex Wno-div-by-zero
3363 @opindex Wdiv-by-zero
3364 Do not warn about compile-time integer division by zero. Floating point
3365 division by zero is not warned about, as it can be a legitimate way of
3366 obtaining infinities and NaNs.
3368 @item -Wsystem-headers
3369 @opindex Wsystem-headers
3370 @opindex Wno-system-headers
3371 @cindex warnings from system headers
3372 @cindex system headers, warnings from
3373 Print warning messages for constructs found in system header files.
3374 Warnings from system headers are normally suppressed, on the assumption
3375 that they usually do not indicate real problems and would only make the
3376 compiler output harder to read. Using this command line option tells
3377 GCC to emit warnings from system headers as if they occurred in user
3378 code. However, note that using @option{-Wall} in conjunction with this
3379 option will @emph{not} warn about unknown pragmas in system
3380 headers---for that, @option{-Wunknown-pragmas} must also be used.
3383 @opindex Wfloat-equal
3384 @opindex Wno-float-equal
3385 Warn if floating point values are used in equality comparisons.
3387 The idea behind this is that sometimes it is convenient (for the
3388 programmer) to consider floating-point values as approximations to
3389 infinitely precise real numbers. If you are doing this, then you need
3390 to compute (by analyzing the code, or in some other way) the maximum or
3391 likely maximum error that the computation introduces, and allow for it
3392 when performing comparisons (and when producing output, but that's a
3393 different problem). In particular, instead of testing for equality, you
3394 would check to see whether the two values have ranges that overlap; and
3395 this is done with the relational operators, so equality comparisons are
3398 @item -Wtraditional @r{(C and Objective-C only)}
3399 @opindex Wtraditional
3400 @opindex Wno-traditional
3401 Warn about certain constructs that behave differently in traditional and
3402 ISO C@. Also warn about ISO C constructs that have no traditional C
3403 equivalent, and/or problematic constructs which should be avoided.
3407 Macro parameters that appear within string literals in the macro body.
3408 In traditional C macro replacement takes place within string literals,
3409 but does not in ISO C@.
3412 In traditional C, some preprocessor directives did not exist.
3413 Traditional preprocessors would only consider a line to be a directive
3414 if the @samp{#} appeared in column 1 on the line. Therefore
3415 @option{-Wtraditional} warns about directives that traditional C
3416 understands but would ignore because the @samp{#} does not appear as the
3417 first character on the line. It also suggests you hide directives like
3418 @samp{#pragma} not understood by traditional C by indenting them. Some
3419 traditional implementations would not recognize @samp{#elif}, so it
3420 suggests avoiding it altogether.
3423 A function-like macro that appears without arguments.
3426 The unary plus operator.
3429 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3430 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3431 constants.) Note, these suffixes appear in macros defined in the system
3432 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3433 Use of these macros in user code might normally lead to spurious
3434 warnings, however GCC's integrated preprocessor has enough context to
3435 avoid warning in these cases.
3438 A function declared external in one block and then used after the end of
3442 A @code{switch} statement has an operand of type @code{long}.
3445 A non-@code{static} function declaration follows a @code{static} one.
3446 This construct is not accepted by some traditional C compilers.
3449 The ISO type of an integer constant has a different width or
3450 signedness from its traditional type. This warning is only issued if
3451 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3452 typically represent bit patterns, are not warned about.
3455 Usage of ISO string concatenation is detected.
3458 Initialization of automatic aggregates.
3461 Identifier conflicts with labels. Traditional C lacks a separate
3462 namespace for labels.
3465 Initialization of unions. If the initializer is zero, the warning is
3466 omitted. This is done under the assumption that the zero initializer in
3467 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3468 initializer warnings and relies on default initialization to zero in the
3472 Conversions by prototypes between fixed/floating point values and vice
3473 versa. The absence of these prototypes when compiling with traditional
3474 C would cause serious problems. This is a subset of the possible
3475 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3478 Use of ISO C style function definitions. This warning intentionally is
3479 @emph{not} issued for prototype declarations or variadic functions
3480 because these ISO C features will appear in your code when using
3481 libiberty's traditional C compatibility macros, @code{PARAMS} and
3482 @code{VPARAMS}. This warning is also bypassed for nested functions
3483 because that feature is already a GCC extension and thus not relevant to
3484 traditional C compatibility.
3487 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3488 @opindex Wtraditional-conversion
3489 @opindex Wno-traditional-conversion
3490 Warn if a prototype causes a type conversion that is different from what
3491 would happen to the same argument in the absence of a prototype. This
3492 includes conversions of fixed point to floating and vice versa, and
3493 conversions changing the width or signedness of a fixed point argument
3494 except when the same as the default promotion.
3496 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3497 @opindex Wdeclaration-after-statement
3498 @opindex Wno-declaration-after-statement
3499 Warn when a declaration is found after a statement in a block. This
3500 construct, known from C++, was introduced with ISO C99 and is by default
3501 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3502 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3507 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3509 @item -Wno-endif-labels
3510 @opindex Wno-endif-labels
3511 @opindex Wendif-labels
3512 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3517 Warn whenever a local variable shadows another local variable, parameter or
3518 global variable or whenever a built-in function is shadowed.
3520 @item -Wlarger-than=@var{len}
3521 @opindex Wlarger-than=@var{len}
3522 @opindex Wlarger-than-@var{len}
3523 Warn whenever an object of larger than @var{len} bytes is defined.
3525 @item -Wframe-larger-than=@var{len}
3526 @opindex Wframe-larger-than
3527 Warn whenever the size of a function frame is larger than @var{len} bytes.
3529 @item -Wunsafe-loop-optimizations
3530 @opindex Wunsafe-loop-optimizations
3531 @opindex Wno-unsafe-loop-optimizations
3532 Warn if the loop cannot be optimized because the compiler could not
3533 assume anything on the bounds of the loop indices. With
3534 @option{-funsafe-loop-optimizations} warn if the compiler made
3537 @item -Wpointer-arith
3538 @opindex Wpointer-arith
3539 @opindex Wno-pointer-arith
3540 Warn about anything that depends on the ``size of'' a function type or
3541 of @code{void}. GNU C assigns these types a size of 1, for
3542 convenience in calculations with @code{void *} pointers and pointers
3543 to functions. In C++, warn also when an arithmetic operation involves
3544 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3547 @opindex Wtype-limits
3548 @opindex Wno-type-limits
3549 Warn if a comparison is always true or always false due to the limited
3550 range of the data type, but do not warn for constant expressions. For
3551 example, warn if an unsigned variable is compared against zero with
3552 @samp{<} or @samp{>=}. This warning is also enabled by
3555 @item -Wbad-function-cast @r{(C and Objective-C only)}
3556 @opindex Wbad-function-cast
3557 @opindex Wno-bad-function-cast
3558 Warn whenever a function call is cast to a non-matching type.
3559 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3561 @item -Wc++-compat @r{(C and Objective-C only)}
3562 Warn about ISO C constructs that are outside of the common subset of
3563 ISO C and ISO C++, e.g.@: request for implicit conversion from
3564 @code{void *} to a pointer to non-@code{void} type.
3566 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3567 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3568 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3569 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3573 @opindex Wno-cast-qual
3574 Warn whenever a pointer is cast so as to remove a type qualifier from
3575 the target type. For example, warn if a @code{const char *} is cast
3576 to an ordinary @code{char *}.
3579 @opindex Wcast-align
3580 @opindex Wno-cast-align
3581 Warn whenever a pointer is cast such that the required alignment of the
3582 target is increased. For example, warn if a @code{char *} is cast to
3583 an @code{int *} on machines where integers can only be accessed at
3584 two- or four-byte boundaries.
3586 @item -Wwrite-strings
3587 @opindex Wwrite-strings
3588 @opindex Wno-write-strings
3589 When compiling C, give string constants the type @code{const
3590 char[@var{length}]} so that
3591 copying the address of one into a non-@code{const} @code{char *}
3592 pointer will get a warning; when compiling C++, warn about the
3593 deprecated conversion from string literals to @code{char *}. This
3594 warning, by default, is enabled for C++ programs.
3595 These warnings will help you find at
3596 compile time code that can try to write into a string constant, but
3597 only if you have been very careful about using @code{const} in
3598 declarations and prototypes. Otherwise, it will just be a nuisance;
3599 this is why we did not make @option{-Wall} request these warnings.
3603 @opindex Wno-clobbered
3604 Warn for variables that might be changed by @samp{longjmp} or
3605 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3608 @opindex Wconversion
3609 @opindex Wno-conversion
3610 Warn for implicit conversions that may alter a value. This includes
3611 conversions between real and integer, like @code{abs (x)} when
3612 @code{x} is @code{double}; conversions between signed and unsigned,
3613 like @code{unsigned ui = -1}; and conversions to smaller types, like
3614 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3615 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3616 changed by the conversion like in @code{abs (2.0)}. Warnings about
3617 conversions between signed and unsigned integers can be disabled by
3618 using @option{-Wno-sign-conversion}.
3620 For C++, also warn for conversions between @code{NULL} and non-pointer
3621 types; confusing overload resolution for user-defined conversions; and
3622 conversions that will never use a type conversion operator:
3623 conversions to @code{void}, the same type, a base class or a reference
3624 to them. Warnings about conversions between signed and unsigned
3625 integers are disabled by default in C++ unless
3626 @option{-Wsign-conversion} is explicitly enabled.
3629 @opindex Wempty-body
3630 @opindex Wno-empty-body
3631 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3632 while} statement. Additionally, in C++, warn when an empty body occurs
3633 in a @samp{while} or @samp{for} statement with no whitespacing before
3634 the semicolon. This warning is also enabled by @option{-Wextra}.
3636 @item -Wsign-compare
3637 @opindex Wsign-compare
3638 @opindex Wno-sign-compare
3639 @cindex warning for comparison of signed and unsigned values
3640 @cindex comparison of signed and unsigned values, warning
3641 @cindex signed and unsigned values, comparison warning
3642 Warn when a comparison between signed and unsigned values could produce
3643 an incorrect result when the signed value is converted to unsigned.
3644 This warning is also enabled by @option{-Wextra}; to get the other warnings
3645 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3647 @item -Wsign-conversion
3648 @opindex Wsign-conversion
3649 @opindex Wno-sign-conversion
3650 Warn for implicit conversions that may change the sign of an integer
3651 value, like assigning a signed integer expression to an unsigned
3652 integer variable. An explicit cast silences the warning. In C, this
3653 option is enabled also by @option{-Wconversion}.
3657 @opindex Wno-address
3658 Warn about suspicious uses of memory addresses. These include using
3659 the address of a function in a conditional expression, such as
3660 @code{void func(void); if (func)}, and comparisons against the memory
3661 address of a string literal, such as @code{if (x == "abc")}. Such
3662 uses typically indicate a programmer error: the address of a function
3663 always evaluates to true, so their use in a conditional usually
3664 indicate that the programmer forgot the parentheses in a function
3665 call; and comparisons against string literals result in unspecified
3666 behavior and are not portable in C, so they usually indicate that the
3667 programmer intended to use @code{strcmp}. This warning is enabled by
3671 @opindex Wlogical-op
3672 @opindex Wno-logical-op
3673 Warn about suspicious uses of logical operators in expressions.
3674 This includes using logical operators in contexts where a
3675 bit-wise operator is likely to be expected.
3677 @item -Waggregate-return
3678 @opindex Waggregate-return
3679 @opindex Wno-aggregate-return
3680 Warn if any functions that return structures or unions are defined or
3681 called. (In languages where you can return an array, this also elicits
3684 @item -Wno-attributes
3685 @opindex Wno-attributes
3686 @opindex Wattributes
3687 Do not warn if an unexpected @code{__attribute__} is used, such as
3688 unrecognized attributes, function attributes applied to variables,
3689 etc. This will not stop errors for incorrect use of supported
3692 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3693 @opindex Wstrict-prototypes
3694 @opindex Wno-strict-prototypes
3695 Warn if a function is declared or defined without specifying the
3696 argument types. (An old-style function definition is permitted without
3697 a warning if preceded by a declaration which specifies the argument
3700 @item -Wold-style-declaration @r{(C and Objective-C only)}
3701 @opindex Wold-style-declaration
3702 @opindex Wno-old-style-declaration
3703 Warn for obsolescent usages, according to the C Standard, in a
3704 declaration. For example, warn if storage-class specifiers like
3705 @code{static} are not the first things in a declaration. This warning
3706 is also enabled by @option{-Wextra}.
3708 @item -Wold-style-definition @r{(C and Objective-C only)}
3709 @opindex Wold-style-definition
3710 @opindex Wno-old-style-definition
3711 Warn if an old-style function definition is used. A warning is given
3712 even if there is a previous prototype.
3714 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3715 @opindex Wmissing-parameter-type
3716 @opindex Wno-missing-parameter-type
3717 A function parameter is declared without a type specifier in K&R-style
3724 This warning is also enabled by @option{-Wextra}.
3726 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3727 @opindex Wmissing-prototypes
3728 @opindex Wno-missing-prototypes
3729 Warn if a global function is defined without a previous prototype
3730 declaration. This warning is issued even if the definition itself
3731 provides a prototype. The aim is to detect global functions that fail
3732 to be declared in header files.
3734 @item -Wmissing-declarations
3735 @opindex Wmissing-declarations
3736 @opindex Wno-missing-declarations
3737 Warn if a global function is defined without a previous declaration.
3738 Do so even if the definition itself provides a prototype.
3739 Use this option to detect global functions that are not declared in
3740 header files. In C++, no warnings are issued for function templates,
3741 or for inline functions, or for functions in anonymous namespaces.
3743 @item -Wmissing-field-initializers
3744 @opindex Wmissing-field-initializers
3745 @opindex Wno-missing-field-initializers
3749 Warn if a structure's initializer has some fields missing. For
3750 example, the following code would cause such a warning, because
3751 @code{x.h} is implicitly zero:
3754 struct s @{ int f, g, h; @};
3755 struct s x = @{ 3, 4 @};
3758 This option does not warn about designated initializers, so the following
3759 modification would not trigger a warning:
3762 struct s @{ int f, g, h; @};
3763 struct s x = @{ .f = 3, .g = 4 @};
3766 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3767 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3769 @item -Wmissing-noreturn
3770 @opindex Wmissing-noreturn
3771 @opindex Wno-missing-noreturn
3772 Warn about functions which might be candidates for attribute @code{noreturn}.
3773 Note these are only possible candidates, not absolute ones. Care should
3774 be taken to manually verify functions actually do not ever return before
3775 adding the @code{noreturn} attribute, otherwise subtle code generation
3776 bugs could be introduced. You will not get a warning for @code{main} in
3777 hosted C environments.
3779 @item -Wmissing-format-attribute
3780 @opindex Wmissing-format-attribute
3781 @opindex Wno-missing-format-attribute
3784 Warn about function pointers which might be candidates for @code{format}
3785 attributes. Note these are only possible candidates, not absolute ones.
3786 GCC will guess that function pointers with @code{format} attributes that
3787 are used in assignment, initialization, parameter passing or return
3788 statements should have a corresponding @code{format} attribute in the
3789 resulting type. I.e.@: the left-hand side of the assignment or
3790 initialization, the type of the parameter variable, or the return type
3791 of the containing function respectively should also have a @code{format}
3792 attribute to avoid the warning.
3794 GCC will also warn about function definitions which might be
3795 candidates for @code{format} attributes. Again, these are only
3796 possible candidates. GCC will guess that @code{format} attributes
3797 might be appropriate for any function that calls a function like
3798 @code{vprintf} or @code{vscanf}, but this might not always be the
3799 case, and some functions for which @code{format} attributes are
3800 appropriate may not be detected.
3802 @item -Wno-multichar
3803 @opindex Wno-multichar
3805 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3806 Usually they indicate a typo in the user's code, as they have
3807 implementation-defined values, and should not be used in portable code.
3809 @item -Wnormalized=<none|id|nfc|nfkc>
3810 @opindex Wnormalized=
3813 @cindex character set, input normalization
3814 In ISO C and ISO C++, two identifiers are different if they are
3815 different sequences of characters. However, sometimes when characters
3816 outside the basic ASCII character set are used, you can have two
3817 different character sequences that look the same. To avoid confusion,
3818 the ISO 10646 standard sets out some @dfn{normalization rules} which
3819 when applied ensure that two sequences that look the same are turned into
3820 the same sequence. GCC can warn you if you are using identifiers which
3821 have not been normalized; this option controls that warning.
3823 There are four levels of warning that GCC supports. The default is
3824 @option{-Wnormalized=nfc}, which warns about any identifier which is
3825 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3826 recommended form for most uses.
3828 Unfortunately, there are some characters which ISO C and ISO C++ allow
3829 in identifiers that when turned into NFC aren't allowable as
3830 identifiers. That is, there's no way to use these symbols in portable
3831 ISO C or C++ and have all your identifiers in NFC@.
3832 @option{-Wnormalized=id} suppresses the warning for these characters.
3833 It is hoped that future versions of the standards involved will correct
3834 this, which is why this option is not the default.
3836 You can switch the warning off for all characters by writing
3837 @option{-Wnormalized=none}. You would only want to do this if you
3838 were using some other normalization scheme (like ``D''), because
3839 otherwise you can easily create bugs that are literally impossible to see.
3841 Some characters in ISO 10646 have distinct meanings but look identical
3842 in some fonts or display methodologies, especially once formatting has
3843 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3844 LETTER N'', will display just like a regular @code{n} which has been
3845 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3846 normalization scheme to convert all these into a standard form as
3847 well, and GCC will warn if your code is not in NFKC if you use
3848 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3849 about every identifier that contains the letter O because it might be
3850 confused with the digit 0, and so is not the default, but may be
3851 useful as a local coding convention if the programming environment is
3852 unable to be fixed to display these characters distinctly.
3854 @item -Wno-deprecated-declarations
3855 @opindex Wno-deprecated-declarations
3856 @opindex Wdeprecated-declarations
3857 Do not warn about uses of functions (@pxref{Function Attributes}),
3858 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3859 Attributes}) marked as deprecated by using the @code{deprecated}
3863 @opindex Wno-overflow
3865 Do not warn about compile-time overflow in constant expressions.
3867 @item -Woverride-init @r{(C and Objective-C only)}
3868 @opindex Woverride-init
3869 @opindex Wno-override-init
3873 Warn if an initialized field without side effects is overridden when
3874 using designated initializers (@pxref{Designated Inits, , Designated
3877 This warning is included in @option{-Wextra}. To get other
3878 @option{-Wextra} warnings without this one, use @samp{-Wextra
3879 -Wno-override-init}.
3884 Warn if a structure is given the packed attribute, but the packed
3885 attribute has no effect on the layout or size of the structure.
3886 Such structures may be mis-aligned for little benefit. For
3887 instance, in this code, the variable @code{f.x} in @code{struct bar}
3888 will be misaligned even though @code{struct bar} does not itself
3889 have the packed attribute:
3896 @} __attribute__((packed));
3907 Warn if padding is included in a structure, either to align an element
3908 of the structure or to align the whole structure. Sometimes when this
3909 happens it is possible to rearrange the fields of the structure to
3910 reduce the padding and so make the structure smaller.
3912 @item -Wredundant-decls
3913 @opindex Wredundant-decls
3914 @opindex Wno-redundant-decls
3915 Warn if anything is declared more than once in the same scope, even in
3916 cases where multiple declaration is valid and changes nothing.
3918 @item -Wnested-externs @r{(C and Objective-C only)}
3919 @opindex Wnested-externs
3920 @opindex Wno-nested-externs
3921 Warn if an @code{extern} declaration is encountered within a function.
3923 @item -Wunreachable-code
3924 @opindex Wunreachable-code
3925 @opindex Wno-unreachable-code
3926 Warn if the compiler detects that code will never be executed.
3928 This option is intended to warn when the compiler detects that at
3929 least a whole line of source code will never be executed, because
3930 some condition is never satisfied or because it is after a
3931 procedure that never returns.
3933 It is possible for this option to produce a warning even though there
3934 are circumstances under which part of the affected line can be executed,
3935 so care should be taken when removing apparently-unreachable code.
3937 For instance, when a function is inlined, a warning may mean that the
3938 line is unreachable in only one inlined copy of the function.
3940 This option is not made part of @option{-Wall} because in a debugging
3941 version of a program there is often substantial code which checks
3942 correct functioning of the program and is, hopefully, unreachable
3943 because the program does work. Another common use of unreachable
3944 code is to provide behavior which is selectable at compile-time.
3949 Warn if a function can not be inlined and it was declared as inline.
3950 Even with this option, the compiler will not warn about failures to
3951 inline functions declared in system headers.
3953 The compiler uses a variety of heuristics to determine whether or not
3954 to inline a function. For example, the compiler takes into account
3955 the size of the function being inlined and the amount of inlining
3956 that has already been done in the current function. Therefore,
3957 seemingly insignificant changes in the source program can cause the
3958 warnings produced by @option{-Winline} to appear or disappear.
3960 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
3961 @opindex Wno-invalid-offsetof
3962 @opindex Winvalid-offsetof
3963 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3964 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3965 to a non-POD type is undefined. In existing C++ implementations,
3966 however, @samp{offsetof} typically gives meaningful results even when
3967 applied to certain kinds of non-POD types. (Such as a simple
3968 @samp{struct} that fails to be a POD type only by virtue of having a
3969 constructor.) This flag is for users who are aware that they are
3970 writing nonportable code and who have deliberately chosen to ignore the
3973 The restrictions on @samp{offsetof} may be relaxed in a future version
3974 of the C++ standard.
3976 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
3977 @opindex Wno-int-to-pointer-cast
3978 @opindex Wint-to-pointer-cast
3979 Suppress warnings from casts to pointer type of an integer of a
3982 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
3983 @opindex Wno-pointer-to-int-cast
3984 @opindex Wpointer-to-int-cast
3985 Suppress warnings from casts from a pointer to an integer type of a
3989 @opindex Winvalid-pch
3990 @opindex Wno-invalid-pch
3991 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3992 the search path but can't be used.
3996 @opindex Wno-long-long
3997 Warn if @samp{long long} type is used. This is default. To inhibit
3998 the warning messages, use @option{-Wno-long-long}. Flags
3999 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4000 only when @option{-pedantic} flag is used.
4002 @item -Wvariadic-macros
4003 @opindex Wvariadic-macros
4004 @opindex Wno-variadic-macros
4005 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4006 alternate syntax when in pedantic ISO C99 mode. This is default.
4007 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4012 Warn if variable length array is used in the code.
4013 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4014 the variable length array.
4016 @item -Wvolatile-register-var
4017 @opindex Wvolatile-register-var
4018 @opindex Wno-volatile-register-var
4019 Warn if a register variable is declared volatile. The volatile
4020 modifier does not inhibit all optimizations that may eliminate reads
4021 and/or writes to register variables. This warning is enabled by
4024 @item -Wdisabled-optimization
4025 @opindex Wdisabled-optimization
4026 @opindex Wno-disabled-optimization
4027 Warn if a requested optimization pass is disabled. This warning does
4028 not generally indicate that there is anything wrong with your code; it
4029 merely indicates that GCC's optimizers were unable to handle the code
4030 effectively. Often, the problem is that your code is too big or too
4031 complex; GCC will refuse to optimize programs when the optimization
4032 itself is likely to take inordinate amounts of time.
4034 @item -Wpointer-sign @r{(C and Objective-C only)}
4035 @opindex Wpointer-sign
4036 @opindex Wno-pointer-sign
4037 Warn for pointer argument passing or assignment with different signedness.
4038 This option is only supported for C and Objective-C@. It is implied by
4039 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4040 @option{-Wno-pointer-sign}.
4042 @item -Wstack-protector
4043 @opindex Wstack-protector
4044 @opindex Wno-stack-protector
4045 This option is only active when @option{-fstack-protector} is active. It
4046 warns about functions that will not be protected against stack smashing.
4048 @item -Woverlength-strings
4049 @opindex Woverlength-strings
4050 @opindex Wno-overlength-strings
4051 Warn about string constants which are longer than the ``minimum
4052 maximum'' length specified in the C standard. Modern compilers
4053 generally allow string constants which are much longer than the
4054 standard's minimum limit, but very portable programs should avoid
4055 using longer strings.
4057 The limit applies @emph{after} string constant concatenation, and does
4058 not count the trailing NUL@. In C89, the limit was 509 characters; in
4059 C99, it was raised to 4095. C++98 does not specify a normative
4060 minimum maximum, so we do not diagnose overlength strings in C++@.
4062 This option is implied by @option{-pedantic}, and can be disabled with
4063 @option{-Wno-overlength-strings}.
4066 @node Debugging Options
4067 @section Options for Debugging Your Program or GCC
4068 @cindex options, debugging
4069 @cindex debugging information options
4071 GCC has various special options that are used for debugging
4072 either your program or GCC:
4077 Produce debugging information in the operating system's native format
4078 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4081 On most systems that use stabs format, @option{-g} enables use of extra
4082 debugging information that only GDB can use; this extra information
4083 makes debugging work better in GDB but will probably make other debuggers
4085 refuse to read the program. If you want to control for certain whether
4086 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4087 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4089 GCC allows you to use @option{-g} with
4090 @option{-O}. The shortcuts taken by optimized code may occasionally
4091 produce surprising results: some variables you declared may not exist
4092 at all; flow of control may briefly move where you did not expect it;
4093 some statements may not be executed because they compute constant
4094 results or their values were already at hand; some statements may
4095 execute in different places because they were moved out of loops.
4097 Nevertheless it proves possible to debug optimized output. This makes
4098 it reasonable to use the optimizer for programs that might have bugs.
4100 The following options are useful when GCC is generated with the
4101 capability for more than one debugging format.
4105 Produce debugging information for use by GDB@. This means to use the
4106 most expressive format available (DWARF 2, stabs, or the native format
4107 if neither of those are supported), including GDB extensions if at all
4112 Produce debugging information in stabs format (if that is supported),
4113 without GDB extensions. This is the format used by DBX on most BSD
4114 systems. On MIPS, Alpha and System V Release 4 systems this option
4115 produces stabs debugging output which is not understood by DBX or SDB@.
4116 On System V Release 4 systems this option requires the GNU assembler.
4118 @item -feliminate-unused-debug-symbols
4119 @opindex feliminate-unused-debug-symbols
4120 Produce debugging information in stabs format (if that is supported),
4121 for only symbols that are actually used.
4123 @item -femit-class-debug-always
4124 Instead of emitting debugging information for a C++ class in only one
4125 object file, emit it in all object files using the class. This option
4126 should be used only with debuggers that are unable to handle the way GCC
4127 normally emits debugging information for classes because using this
4128 option will increase the size of debugging information by as much as a
4133 Produce debugging information in stabs format (if that is supported),
4134 using GNU extensions understood only by the GNU debugger (GDB)@. The
4135 use of these extensions is likely to make other debuggers crash or
4136 refuse to read the program.
4140 Produce debugging information in COFF format (if that is supported).
4141 This is the format used by SDB on most System V systems prior to
4146 Produce debugging information in XCOFF format (if that is supported).
4147 This is the format used by the DBX debugger on IBM RS/6000 systems.
4151 Produce debugging information in XCOFF format (if that is supported),
4152 using GNU extensions understood only by the GNU debugger (GDB)@. The
4153 use of these extensions is likely to make other debuggers crash or
4154 refuse to read the program, and may cause assemblers other than the GNU
4155 assembler (GAS) to fail with an error.
4159 Produce debugging information in DWARF version 2 format (if that is
4160 supported). This is the format used by DBX on IRIX 6. With this
4161 option, GCC uses features of DWARF version 3 when they are useful;
4162 version 3 is upward compatible with version 2, but may still cause
4163 problems for older debuggers.
4167 Produce debugging information in VMS debug format (if that is
4168 supported). This is the format used by DEBUG on VMS systems.
4171 @itemx -ggdb@var{level}
4172 @itemx -gstabs@var{level}
4173 @itemx -gcoff@var{level}
4174 @itemx -gxcoff@var{level}
4175 @itemx -gvms@var{level}
4176 Request debugging information and also use @var{level} to specify how
4177 much information. The default level is 2.
4179 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4182 Level 1 produces minimal information, enough for making backtraces in
4183 parts of the program that you don't plan to debug. This includes
4184 descriptions of functions and external variables, but no information
4185 about local variables and no line numbers.
4187 Level 3 includes extra information, such as all the macro definitions
4188 present in the program. Some debuggers support macro expansion when
4189 you use @option{-g3}.
4191 @option{-gdwarf-2} does not accept a concatenated debug level, because
4192 GCC used to support an option @option{-gdwarf} that meant to generate
4193 debug information in version 1 of the DWARF format (which is very
4194 different from version 2), and it would have been too confusing. That
4195 debug format is long obsolete, but the option cannot be changed now.
4196 Instead use an additional @option{-g@var{level}} option to change the
4197 debug level for DWARF2.
4199 @item -feliminate-dwarf2-dups
4200 @opindex feliminate-dwarf2-dups
4201 Compress DWARF2 debugging information by eliminating duplicated
4202 information about each symbol. This option only makes sense when
4203 generating DWARF2 debugging information with @option{-gdwarf-2}.
4205 @item -femit-struct-debug-baseonly
4206 Emit debug information for struct-like types
4207 only when the base name of the compilation source file
4208 matches the base name of file in which the struct was defined.
4210 This option substantially reduces the size of debugging information,
4211 but at significant potential loss in type information to the debugger.
4212 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4213 See @option{-femit-struct-debug-detailed} for more detailed control.
4215 This option works only with DWARF 2.
4217 @item -femit-struct-debug-reduced
4218 Emit debug information for struct-like types
4219 only when the base name of the compilation source file
4220 matches the base name of file in which the type was defined,
4221 unless the struct is a template or defined in a system header.
4223 This option significantly reduces the size of debugging information,
4224 with some potential loss in type information to the debugger.
4225 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4226 See @option{-femit-struct-debug-detailed} for more detailed control.
4228 This option works only with DWARF 2.
4230 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4231 Specify the struct-like types
4232 for which the compiler will generate debug information.
4233 The intent is to reduce duplicate struct debug information
4234 between different object files within the same program.
4236 This option is a detailed version of
4237 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4238 which will serve for most needs.
4240 A specification has the syntax
4241 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4243 The optional first word limits the specification to
4244 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4245 A struct type is used directly when it is the type of a variable, member.
4246 Indirect uses arise through pointers to structs.
4247 That is, when use of an incomplete struct would be legal, the use is indirect.
4249 @samp{struct one direct; struct two * indirect;}.
4251 The optional second word limits the specification to
4252 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4253 Generic structs are a bit complicated to explain.
4254 For C++, these are non-explicit specializations of template classes,
4255 or non-template classes within the above.
4256 Other programming languages have generics,
4257 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4259 The third word specifies the source files for those
4260 structs for which the compiler will emit debug information.
4261 The values @samp{none} and @samp{any} have the normal meaning.
4262 The value @samp{base} means that
4263 the base of name of the file in which the type declaration appears
4264 must match the base of the name of the main compilation file.
4265 In practice, this means that
4266 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4267 but types declared in other header will not.
4268 The value @samp{sys} means those types satisfying @samp{base}
4269 or declared in system or compiler headers.
4271 You may need to experiment to determine the best settings for your application.
4273 The default is @samp{-femit-struct-debug-detailed=all}.
4275 This option works only with DWARF 2.
4277 @item -fno-merge-debug-strings
4278 @opindex fmerge-debug-strings
4279 @opindex fno-merge-debug-strings
4280 Direct the linker to merge together strings which are identical in
4281 different object files. This is not supported by all assemblers or
4282 linker. This decreases the size of the debug information in the
4283 output file at the cost of increasing link processing time. This is
4286 @item -fdebug-prefix-map=@var{old}=@var{new}
4287 @opindex fdebug-prefix-map
4288 When compiling files in directory @file{@var{old}}, record debugging
4289 information describing them as in @file{@var{new}} instead.
4291 @cindex @command{prof}
4294 Generate extra code to write profile information suitable for the
4295 analysis program @command{prof}. You must use this option when compiling
4296 the source files you want data about, and you must also use it when
4299 @cindex @command{gprof}
4302 Generate extra code to write profile information suitable for the
4303 analysis program @command{gprof}. You must use this option when compiling
4304 the source files you want data about, and you must also use it when
4309 Makes the compiler print out each function name as it is compiled, and
4310 print some statistics about each pass when it finishes.
4313 @opindex ftime-report
4314 Makes the compiler print some statistics about the time consumed by each
4315 pass when it finishes.
4318 @opindex fmem-report
4319 Makes the compiler print some statistics about permanent memory
4320 allocation when it finishes.
4322 @item -fpre-ipa-mem-report
4323 @opindex fpre-ipa-mem-report
4324 @item -fpost-ipa-mem-report
4325 @opindex fpost-ipa-mem-report
4326 Makes the compiler print some statistics about permanent memory
4327 allocation before or after interprocedural optimization.
4329 @item -fprofile-arcs
4330 @opindex fprofile-arcs
4331 Add code so that program flow @dfn{arcs} are instrumented. During
4332 execution the program records how many times each branch and call is
4333 executed and how many times it is taken or returns. When the compiled
4334 program exits it saves this data to a file called
4335 @file{@var{auxname}.gcda} for each source file. The data may be used for
4336 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4337 test coverage analysis (@option{-ftest-coverage}). Each object file's
4338 @var{auxname} is generated from the name of the output file, if
4339 explicitly specified and it is not the final executable, otherwise it is
4340 the basename of the source file. In both cases any suffix is removed
4341 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4342 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4343 @xref{Cross-profiling}.
4345 @cindex @command{gcov}
4349 This option is used to compile and link code instrumented for coverage
4350 analysis. The option is a synonym for @option{-fprofile-arcs}
4351 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4352 linking). See the documentation for those options for more details.
4357 Compile the source files with @option{-fprofile-arcs} plus optimization
4358 and code generation options. For test coverage analysis, use the
4359 additional @option{-ftest-coverage} option. You do not need to profile
4360 every source file in a program.
4363 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4364 (the latter implies the former).
4367 Run the program on a representative workload to generate the arc profile
4368 information. This may be repeated any number of times. You can run
4369 concurrent instances of your program, and provided that the file system
4370 supports locking, the data files will be correctly updated. Also
4371 @code{fork} calls are detected and correctly handled (double counting
4375 For profile-directed optimizations, compile the source files again with
4376 the same optimization and code generation options plus
4377 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4378 Control Optimization}).
4381 For test coverage analysis, use @command{gcov} to produce human readable
4382 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4383 @command{gcov} documentation for further information.
4387 With @option{-fprofile-arcs}, for each function of your program GCC
4388 creates a program flow graph, then finds a spanning tree for the graph.
4389 Only arcs that are not on the spanning tree have to be instrumented: the
4390 compiler adds code to count the number of times that these arcs are
4391 executed. When an arc is the only exit or only entrance to a block, the
4392 instrumentation code can be added to the block; otherwise, a new basic
4393 block must be created to hold the instrumentation code.
4396 @item -ftest-coverage
4397 @opindex ftest-coverage
4398 Produce a notes file that the @command{gcov} code-coverage utility
4399 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4400 show program coverage. Each source file's note file is called
4401 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4402 above for a description of @var{auxname} and instructions on how to
4403 generate test coverage data. Coverage data will match the source files
4404 more closely, if you do not optimize.
4406 @item -fdbg-cnt-list
4407 @opindex fdbg-cnt-list
4408 Print the name and the counter upperbound for all debug counters.
4410 @item -fdbg-cnt=@var{counter-value-list}
4412 Set the internal debug counter upperbound. @var{counter-value-list}
4413 is a comma-separated list of @var{name}:@var{value} pairs
4414 which sets the upperbound of each debug counter @var{name} to @var{value}.
4415 All debug counters have the initial upperbound of @var{UINT_MAX},
4416 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4417 e.g. With -fdbg-cnt=dce:10,tail_call:0
4418 dbg_cnt(dce) will return true only for first 10 invocations
4419 and dbg_cnt(tail_call) will return false always.
4421 @item -d@var{letters}
4422 @item -fdump-rtl-@var{pass}
4424 Says to make debugging dumps during compilation at times specified by
4425 @var{letters}. This is used for debugging the RTL-based passes of the
4426 compiler. The file names for most of the dumps are made by appending a
4427 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4428 from the name of the output file, if explicitly specified and it is not
4429 an executable, otherwise it is the basename of the source file. These
4430 switches may have different effects when @option{-E} is used for
4433 Most debug dumps can be enabled either passing a letter to the @option{-d}
4434 option, or with a long @option{-fdump-rtl} switch; here are the possible
4435 letters for use in @var{letters} and @var{pass}, and their meanings:
4440 Annotate the assembler output with miscellaneous debugging information.
4443 @itemx -fdump-rtl-bbro
4445 @opindex fdump-rtl-bbro
4446 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4449 @itemx -fdump-rtl-combine
4451 @opindex fdump-rtl-combine
4452 Dump after the RTL instruction combination pass, to the file
4453 @file{@var{file}.129r.combine}.
4456 @itemx -fdump-rtl-ce1
4457 @itemx -fdump-rtl-ce2
4459 @opindex fdump-rtl-ce1
4460 @opindex fdump-rtl-ce2
4461 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4462 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4463 and @option{-fdump-rtl-ce2} enable dumping after the second if
4464 conversion, to the file @file{@var{file}.130r.ce2}.
4467 @itemx -fdump-rtl-btl
4468 @itemx -fdump-rtl-dbr
4470 @opindex fdump-rtl-btl
4471 @opindex fdump-rtl-dbr
4472 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4473 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4474 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4475 scheduling, to @file{@var{file}.36.dbr}.
4479 Dump all macro definitions, at the end of preprocessing, in addition to
4483 @itemx -fdump-rtl-ce3
4485 @opindex fdump-rtl-ce3
4486 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4489 @itemx -fdump-rtl-cfg
4490 @itemx -fdump-rtl-life
4492 @opindex fdump-rtl-cfg
4493 @opindex fdump-rtl-life
4494 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4495 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4496 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4497 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4500 @itemx -fdump-rtl-greg
4502 @opindex fdump-rtl-greg
4503 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4506 @itemx -fdump-rtl-gcse
4507 @itemx -fdump-rtl-bypass
4509 @opindex fdump-rtl-gcse
4510 @opindex fdump-rtl-bypass
4511 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4512 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4513 enable dumping after jump bypassing and control flow optimizations, to
4514 @file{@var{file}.115r.bypass}.
4517 @itemx -fdump-rtl-eh
4519 @opindex fdump-rtl-eh
4520 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4523 @itemx -fdump-rtl-sibling
4525 @opindex fdump-rtl-sibling
4526 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4529 @itemx -fdump-rtl-jump
4531 @opindex fdump-rtl-jump
4532 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4535 @itemx -fdump-rtl-stack
4537 @opindex fdump-rtl-stack
4538 Dump after conversion from GCC's "flat register file" registers to the
4539 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4542 @itemx -fdump-rtl-lreg
4544 @opindex fdump-rtl-lreg
4545 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4548 @itemx -fdump-rtl-loop2
4550 @opindex fdump-rtl-loop2
4551 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4552 loop optimization pass, to @file{@var{file}.119r.loop2},
4553 @file{@var{file}.120r.loop2_init},
4554 @file{@var{file}.121r.loop2_invariant}, and
4555 @file{@var{file}.125r.loop2_done}.
4558 @itemx -fdump-rtl-sms
4560 @opindex fdump-rtl-sms
4561 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4564 @itemx -fdump-rtl-mach
4566 @opindex fdump-rtl-mach
4567 Dump after performing the machine dependent reorganization pass, to
4568 @file{@var{file}.155r.mach} if that pass exists.
4571 @itemx -fdump-rtl-rnreg
4573 @opindex fdump-rtl-rnreg
4574 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4577 @itemx -fdump-rtl-regmove
4579 @opindex fdump-rtl-regmove
4580 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4583 @itemx -fdump-rtl-postreload
4585 @opindex fdump-rtl-postreload
4586 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4589 @itemx -fdump-rtl-expand
4591 @opindex fdump-rtl-expand
4592 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4595 @itemx -fdump-rtl-sched2
4597 @opindex fdump-rtl-sched2
4598 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4601 @itemx -fdump-rtl-cse
4603 @opindex fdump-rtl-cse
4604 Dump after CSE (including the jump optimization that sometimes follows
4605 CSE), to @file{@var{file}.113r.cse}.
4608 @itemx -fdump-rtl-sched1
4610 @opindex fdump-rtl-sched1
4611 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4614 @itemx -fdump-rtl-cse2
4616 @opindex fdump-rtl-cse2
4617 Dump after the second CSE pass (including the jump optimization that
4618 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4621 @itemx -fdump-rtl-tracer
4623 @opindex fdump-rtl-tracer
4624 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4627 @itemx -fdump-rtl-vpt
4628 @itemx -fdump-rtl-vartrack
4630 @opindex fdump-rtl-vpt
4631 @opindex fdump-rtl-vartrack
4632 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4633 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4634 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4635 to @file{@var{file}.154r.vartrack}.
4638 @itemx -fdump-rtl-flow2
4640 @opindex fdump-rtl-flow2
4641 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4644 @itemx -fdump-rtl-peephole2
4646 @opindex fdump-rtl-peephole2
4647 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4650 @itemx -fdump-rtl-web
4652 @opindex fdump-rtl-web
4653 Dump after live range splitting, to @file{@var{file}.126r.web}.
4656 @itemx -fdump-rtl-all
4658 @opindex fdump-rtl-all
4659 Produce all the dumps listed above.
4663 Produce a core dump whenever an error occurs.
4667 Print statistics on memory usage, at the end of the run, to
4672 Annotate the assembler output with a comment indicating which
4673 pattern and alternative was used. The length of each instruction is
4678 Dump the RTL in the assembler output as a comment before each instruction.
4679 Also turns on @option{-dp} annotation.
4683 For each of the other indicated dump files (either with @option{-d} or
4684 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4685 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4689 Just generate RTL for a function instead of compiling it. Usually used
4690 with @samp{r} (@option{-fdump-rtl-expand}).
4694 Dump debugging information during parsing, to standard error.
4698 @opindex fdump-noaddr
4699 When doing debugging dumps (see @option{-d} option above), suppress
4700 address output. This makes it more feasible to use diff on debugging
4701 dumps for compiler invocations with different compiler binaries and/or
4702 different text / bss / data / heap / stack / dso start locations.
4704 @item -fdump-unnumbered
4705 @opindex fdump-unnumbered
4706 When doing debugging dumps (see @option{-d} option above), suppress instruction
4707 numbers and address output. This makes it more feasible to
4708 use diff on debugging dumps for compiler invocations with different
4709 options, in particular with and without @option{-g}.
4711 @item -fdump-translation-unit @r{(C++ only)}
4712 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4713 @opindex fdump-translation-unit
4714 Dump a representation of the tree structure for the entire translation
4715 unit to a file. The file name is made by appending @file{.tu} to the
4716 source file name. If the @samp{-@var{options}} form is used, @var{options}
4717 controls the details of the dump as described for the
4718 @option{-fdump-tree} options.
4720 @item -fdump-class-hierarchy @r{(C++ only)}
4721 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4722 @opindex fdump-class-hierarchy
4723 Dump a representation of each class's hierarchy and virtual function
4724 table layout to a file. The file name is made by appending @file{.class}
4725 to the source file name. If the @samp{-@var{options}} form is used,
4726 @var{options} controls the details of the dump as described for the
4727 @option{-fdump-tree} options.
4729 @item -fdump-ipa-@var{switch}
4731 Control the dumping at various stages of inter-procedural analysis
4732 language tree to a file. The file name is generated by appending a switch
4733 specific suffix to the source file name. The following dumps are possible:
4737 Enables all inter-procedural analysis dumps.
4740 Dumps information about call-graph optimization, unused function removal,
4741 and inlining decisions.
4744 Dump after function inlining.
4748 @item -fdump-tree-@var{switch}
4749 @itemx -fdump-tree-@var{switch}-@var{options}
4751 Control the dumping at various stages of processing the intermediate
4752 language tree to a file. The file name is generated by appending a switch
4753 specific suffix to the source file name. If the @samp{-@var{options}}
4754 form is used, @var{options} is a list of @samp{-} separated options that
4755 control the details of the dump. Not all options are applicable to all
4756 dumps, those which are not meaningful will be ignored. The following
4757 options are available
4761 Print the address of each node. Usually this is not meaningful as it
4762 changes according to the environment and source file. Its primary use
4763 is for tying up a dump file with a debug environment.
4765 Inhibit dumping of members of a scope or body of a function merely
4766 because that scope has been reached. Only dump such items when they
4767 are directly reachable by some other path. When dumping pretty-printed
4768 trees, this option inhibits dumping the bodies of control structures.
4770 Print a raw representation of the tree. By default, trees are
4771 pretty-printed into a C-like representation.
4773 Enable more detailed dumps (not honored by every dump option).
4775 Enable dumping various statistics about the pass (not honored by every dump
4778 Enable showing basic block boundaries (disabled in raw dumps).
4780 Enable showing virtual operands for every statement.
4782 Enable showing line numbers for statements.
4784 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4786 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4789 The following tree dumps are possible:
4793 Dump before any tree based optimization, to @file{@var{file}.original}.
4796 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4799 @opindex fdump-tree-gimple
4800 Dump each function before and after the gimplification pass to a file. The
4801 file name is made by appending @file{.gimple} to the source file name.
4804 @opindex fdump-tree-cfg
4805 Dump the control flow graph of each function to a file. The file name is
4806 made by appending @file{.cfg} to the source file name.
4809 @opindex fdump-tree-vcg
4810 Dump the control flow graph of each function to a file in VCG format. The
4811 file name is made by appending @file{.vcg} to the source file name. Note
4812 that if the file contains more than one function, the generated file cannot
4813 be used directly by VCG@. You will need to cut and paste each function's
4814 graph into its own separate file first.
4817 @opindex fdump-tree-ch
4818 Dump each function after copying loop headers. The file name is made by
4819 appending @file{.ch} to the source file name.
4822 @opindex fdump-tree-ssa
4823 Dump SSA related information to a file. The file name is made by appending
4824 @file{.ssa} to the source file name.
4827 @opindex fdump-tree-salias
4828 Dump structure aliasing variable information to a file. This file name
4829 is made by appending @file{.salias} to the source file name.
4832 @opindex fdump-tree-alias
4833 Dump aliasing information for each function. The file name is made by
4834 appending @file{.alias} to the source file name.
4837 @opindex fdump-tree-ccp
4838 Dump each function after CCP@. The file name is made by appending
4839 @file{.ccp} to the source file name.
4842 @opindex fdump-tree-storeccp
4843 Dump each function after STORE-CCP@. The file name is made by appending
4844 @file{.storeccp} to the source file name.
4847 @opindex fdump-tree-pre
4848 Dump trees after partial redundancy elimination. The file name is made
4849 by appending @file{.pre} to the source file name.
4852 @opindex fdump-tree-fre
4853 Dump trees after full redundancy elimination. The file name is made
4854 by appending @file{.fre} to the source file name.
4857 @opindex fdump-tree-copyprop
4858 Dump trees after copy propagation. The file name is made
4859 by appending @file{.copyprop} to the source file name.
4861 @item store_copyprop
4862 @opindex fdump-tree-store_copyprop
4863 Dump trees after store copy-propagation. The file name is made
4864 by appending @file{.store_copyprop} to the source file name.
4867 @opindex fdump-tree-dce
4868 Dump each function after dead code elimination. The file name is made by
4869 appending @file{.dce} to the source file name.
4872 @opindex fdump-tree-mudflap
4873 Dump each function after adding mudflap instrumentation. The file name is
4874 made by appending @file{.mudflap} to the source file name.
4877 @opindex fdump-tree-sra
4878 Dump each function after performing scalar replacement of aggregates. The
4879 file name is made by appending @file{.sra} to the source file name.
4882 @opindex fdump-tree-sink
4883 Dump each function after performing code sinking. The file name is made
4884 by appending @file{.sink} to the source file name.
4887 @opindex fdump-tree-dom
4888 Dump each function after applying dominator tree optimizations. The file
4889 name is made by appending @file{.dom} to the source file name.
4892 @opindex fdump-tree-dse
4893 Dump each function after applying dead store elimination. The file
4894 name is made by appending @file{.dse} to the source file name.
4897 @opindex fdump-tree-phiopt
4898 Dump each function after optimizing PHI nodes into straightline code. The file
4899 name is made by appending @file{.phiopt} to the source file name.
4902 @opindex fdump-tree-forwprop
4903 Dump each function after forward propagating single use variables. The file
4904 name is made by appending @file{.forwprop} to the source file name.
4907 @opindex fdump-tree-copyrename
4908 Dump each function after applying the copy rename optimization. The file
4909 name is made by appending @file{.copyrename} to the source file name.
4912 @opindex fdump-tree-nrv
4913 Dump each function after applying the named return value optimization on
4914 generic trees. The file name is made by appending @file{.nrv} to the source
4918 @opindex fdump-tree-vect
4919 Dump each function after applying vectorization of loops. The file name is
4920 made by appending @file{.vect} to the source file name.
4923 @opindex fdump-tree-vrp
4924 Dump each function after Value Range Propagation (VRP). The file name
4925 is made by appending @file{.vrp} to the source file name.
4928 @opindex fdump-tree-all
4929 Enable all the available tree dumps with the flags provided in this option.
4932 @item -ftree-vectorizer-verbose=@var{n}
4933 @opindex ftree-vectorizer-verbose
4934 This option controls the amount of debugging output the vectorizer prints.
4935 This information is written to standard error, unless
4936 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4937 in which case it is output to the usual dump listing file, @file{.vect}.
4938 For @var{n}=0 no diagnostic information is reported.
4939 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4940 and the total number of loops that got vectorized.
4941 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4942 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
4943 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4944 level that @option{-fdump-tree-vect-stats} uses.
4945 Higher verbosity levels mean either more information dumped for each
4946 reported loop, or same amount of information reported for more loops:
4947 If @var{n}=3, alignment related information is added to the reports.
4948 If @var{n}=4, data-references related information (e.g.@: memory dependences,
4949 memory access-patterns) is added to the reports.
4950 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4951 that did not pass the first analysis phase (i.e., may not be countable, or
4952 may have complicated control-flow).
4953 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4954 For @var{n}=7, all the information the vectorizer generates during its
4955 analysis and transformation is reported. This is the same verbosity level
4956 that @option{-fdump-tree-vect-details} uses.
4958 @item -frandom-seed=@var{string}
4959 @opindex frandom-string
4960 This option provides a seed that GCC uses when it would otherwise use
4961 random numbers. It is used to generate certain symbol names
4962 that have to be different in every compiled file. It is also used to
4963 place unique stamps in coverage data files and the object files that
4964 produce them. You can use the @option{-frandom-seed} option to produce
4965 reproducibly identical object files.
4967 The @var{string} should be different for every file you compile.
4969 @item -fsched-verbose=@var{n}
4970 @opindex fsched-verbose
4971 On targets that use instruction scheduling, this option controls the
4972 amount of debugging output the scheduler prints. This information is
4973 written to standard error, unless @option{-dS} or @option{-dR} is
4974 specified, in which case it is output to the usual dump
4975 listing file, @file{.sched} or @file{.sched2} respectively. However
4976 for @var{n} greater than nine, the output is always printed to standard
4979 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4980 same information as @option{-dRS}. For @var{n} greater than one, it
4981 also output basic block probabilities, detailed ready list information
4982 and unit/insn info. For @var{n} greater than two, it includes RTL
4983 at abort point, control-flow and regions info. And for @var{n} over
4984 four, @option{-fsched-verbose} also includes dependence info.
4988 Store the usual ``temporary'' intermediate files permanently; place them
4989 in the current directory and name them based on the source file. Thus,
4990 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4991 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4992 preprocessed @file{foo.i} output file even though the compiler now
4993 normally uses an integrated preprocessor.
4995 When used in combination with the @option{-x} command line option,
4996 @option{-save-temps} is sensible enough to avoid over writing an
4997 input source file with the same extension as an intermediate file.
4998 The corresponding intermediate file may be obtained by renaming the
4999 source file before using @option{-save-temps}.
5003 Report the CPU time taken by each subprocess in the compilation
5004 sequence. For C source files, this is the compiler proper and assembler
5005 (plus the linker if linking is done). The output looks like this:
5012 The first number on each line is the ``user time'', that is time spent
5013 executing the program itself. The second number is ``system time'',
5014 time spent executing operating system routines on behalf of the program.
5015 Both numbers are in seconds.
5017 @item -fvar-tracking
5018 @opindex fvar-tracking
5019 Run variable tracking pass. It computes where variables are stored at each
5020 position in code. Better debugging information is then generated
5021 (if the debugging information format supports this information).
5023 It is enabled by default when compiling with optimization (@option{-Os},
5024 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5025 the debug info format supports it.
5027 @item -print-file-name=@var{library}
5028 @opindex print-file-name
5029 Print the full absolute name of the library file @var{library} that
5030 would be used when linking---and don't do anything else. With this
5031 option, GCC does not compile or link anything; it just prints the
5034 @item -print-multi-directory
5035 @opindex print-multi-directory
5036 Print the directory name corresponding to the multilib selected by any
5037 other switches present in the command line. This directory is supposed
5038 to exist in @env{GCC_EXEC_PREFIX}.
5040 @item -print-multi-lib
5041 @opindex print-multi-lib
5042 Print the mapping from multilib directory names to compiler switches
5043 that enable them. The directory name is separated from the switches by
5044 @samp{;}, and each switch starts with an @samp{@@} instead of the
5045 @samp{-}, without spaces between multiple switches. This is supposed to
5046 ease shell-processing.
5048 @item -print-prog-name=@var{program}
5049 @opindex print-prog-name
5050 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5052 @item -print-libgcc-file-name
5053 @opindex print-libgcc-file-name
5054 Same as @option{-print-file-name=libgcc.a}.
5056 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5057 but you do want to link with @file{libgcc.a}. You can do
5060 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5063 @item -print-search-dirs
5064 @opindex print-search-dirs
5065 Print the name of the configured installation directory and a list of
5066 program and library directories @command{gcc} will search---and don't do anything else.
5068 This is useful when @command{gcc} prints the error message
5069 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5070 To resolve this you either need to put @file{cpp0} and the other compiler
5071 components where @command{gcc} expects to find them, or you can set the environment
5072 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5073 Don't forget the trailing @samp{/}.
5074 @xref{Environment Variables}.
5076 @item -print-sysroot-headers-suffix
5077 @opindex print-sysroot-headers-suffix
5078 Print the suffix added to the target sysroot when searching for
5079 headers, or give an error if the compiler is not configured with such
5080 a suffix---and don't do anything else.
5083 @opindex dumpmachine
5084 Print the compiler's target machine (for example,
5085 @samp{i686-pc-linux-gnu})---and don't do anything else.
5088 @opindex dumpversion
5089 Print the compiler version (for example, @samp{3.0})---and don't do
5094 Print the compiler's built-in specs---and don't do anything else. (This
5095 is used when GCC itself is being built.) @xref{Spec Files}.
5097 @item -feliminate-unused-debug-types
5098 @opindex feliminate-unused-debug-types
5099 Normally, when producing DWARF2 output, GCC will emit debugging
5100 information for all types declared in a compilation
5101 unit, regardless of whether or not they are actually used
5102 in that compilation unit. Sometimes this is useful, such as
5103 if, in the debugger, you want to cast a value to a type that is
5104 not actually used in your program (but is declared). More often,
5105 however, this results in a significant amount of wasted space.
5106 With this option, GCC will avoid producing debug symbol output
5107 for types that are nowhere used in the source file being compiled.
5110 @node Optimize Options
5111 @section Options That Control Optimization
5112 @cindex optimize options
5113 @cindex options, optimization
5115 These options control various sorts of optimizations.
5117 Without any optimization option, the compiler's goal is to reduce the
5118 cost of compilation and to make debugging produce the expected
5119 results. Statements are independent: if you stop the program with a
5120 breakpoint between statements, you can then assign a new value to any
5121 variable or change the program counter to any other statement in the
5122 function and get exactly the results you would expect from the source
5125 Turning on optimization flags makes the compiler attempt to improve
5126 the performance and/or code size at the expense of compilation time
5127 and possibly the ability to debug the program.
5129 The compiler performs optimization based on the knowledge it has of
5130 the program. Optimization levels @option{-O} and above, in
5131 particular, enable @emph{unit-at-a-time} mode, which allows the
5132 compiler to consider information gained from later functions in
5133 the file when compiling a function. Compiling multiple files at
5134 once to a single output file in @emph{unit-at-a-time} mode allows
5135 the compiler to use information gained from all of the files when
5136 compiling each of them.
5138 Not all optimizations are controlled directly by a flag. Only
5139 optimizations that have a flag are listed.
5146 Optimize. Optimizing compilation takes somewhat more time, and a lot
5147 more memory for a large function.
5149 With @option{-O}, the compiler tries to reduce code size and execution
5150 time, without performing any optimizations that take a great deal of
5153 @option{-O} turns on the following optimization flags:
5156 -fcprop-registers @gol
5159 -fdelayed-branch @gol
5161 -fguess-branch-probability @gol
5162 -fif-conversion2 @gol
5163 -fif-conversion @gol
5164 -finline-small-functions @gol
5165 -fipa-pure-const @gol
5166 -fipa-reference @gol
5168 -fsplit-wide-types @gol
5171 -ftree-copyrename @gol
5173 -ftree-dominator-opts @gol
5180 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5181 where doing so does not interfere with debugging.
5185 Optimize even more. GCC performs nearly all supported optimizations
5186 that do not involve a space-speed tradeoff. The compiler does not
5187 perform loop unrolling or function inlining when you specify @option{-O2}.
5188 As compared to @option{-O}, this option increases both compilation time
5189 and the performance of the generated code.
5191 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5192 also turns on the following optimization flags:
5193 @gccoptlist{-fthread-jumps @gol
5194 -falign-functions -falign-jumps @gol
5195 -falign-loops -falign-labels @gol
5198 -fcse-follow-jumps -fcse-skip-blocks @gol
5199 -fdelete-null-pointer-checks @gol
5200 -fexpensive-optimizations @gol
5201 -fgcse -fgcse-lm @gol
5202 -foptimize-sibling-calls @gol
5205 -freorder-blocks -freorder-functions @gol
5206 -frerun-cse-after-loop @gol
5207 -fsched-interblock -fsched-spec @gol
5208 -fschedule-insns -fschedule-insns2 @gol
5209 -fstrict-aliasing -fstrict-overflow @gol
5213 Please note the warning under @option{-fgcse} about
5214 invoking @option{-O2} on programs that use computed gotos.
5218 Optimize yet more. @option{-O3} turns on all optimizations specified by
5219 @option{-O2} and also turns on the @option{-finline-functions},
5220 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5221 @option{-fgcse-after-reload} options.
5225 Reduce compilation time and make debugging produce the expected
5226 results. This is the default.
5230 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5231 do not typically increase code size. It also performs further
5232 optimizations designed to reduce code size.
5234 @option{-Os} disables the following optimization flags:
5235 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5236 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5237 -fprefetch-loop-arrays -ftree-vect-loop-version}
5239 If you use multiple @option{-O} options, with or without level numbers,
5240 the last such option is the one that is effective.
5243 Options of the form @option{-f@var{flag}} specify machine-independent
5244 flags. Most flags have both positive and negative forms; the negative
5245 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5246 below, only one of the forms is listed---the one you typically will
5247 use. You can figure out the other form by either removing @samp{no-}
5250 The following options control specific optimizations. They are either
5251 activated by @option{-O} options or are related to ones that are. You
5252 can use the following flags in the rare cases when ``fine-tuning'' of
5253 optimizations to be performed is desired.
5256 @item -fno-default-inline
5257 @opindex fno-default-inline
5258 Do not make member functions inline by default merely because they are
5259 defined inside the class scope (C++ only). Otherwise, when you specify
5260 @w{@option{-O}}, member functions defined inside class scope are compiled
5261 inline by default; i.e., you don't need to add @samp{inline} in front of
5262 the member function name.
5264 @item -fno-defer-pop
5265 @opindex fno-defer-pop
5266 Always pop the arguments to each function call as soon as that function
5267 returns. For machines which must pop arguments after a function call,
5268 the compiler normally lets arguments accumulate on the stack for several
5269 function calls and pops them all at once.
5271 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5273 @item -fforward-propagate
5274 @opindex fforward-propagate
5275 Perform a forward propagation pass on RTL@. The pass tries to combine two
5276 instructions and checks if the result can be simplified. If loop unrolling
5277 is active, two passes are performed and the second is scheduled after
5280 This option is enabled by default at optimization levels @option{-O2},
5281 @option{-O3}, @option{-Os}.
5283 @item -fomit-frame-pointer
5284 @opindex fomit-frame-pointer
5285 Don't keep the frame pointer in a register for functions that
5286 don't need one. This avoids the instructions to save, set up and
5287 restore frame pointers; it also makes an extra register available
5288 in many functions. @strong{It also makes debugging impossible on
5291 On some machines, such as the VAX, this flag has no effect, because
5292 the standard calling sequence automatically handles the frame pointer
5293 and nothing is saved by pretending it doesn't exist. The
5294 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5295 whether a target machine supports this flag. @xref{Registers,,Register
5296 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5298 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5300 @item -foptimize-sibling-calls
5301 @opindex foptimize-sibling-calls
5302 Optimize sibling and tail recursive calls.
5304 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5308 Don't pay attention to the @code{inline} keyword. Normally this option
5309 is used to keep the compiler from expanding any functions inline.
5310 Note that if you are not optimizing, no functions can be expanded inline.
5312 @item -finline-small-functions
5313 @opindex finline-small-functions
5314 Integrate functions into their callers when their body is smaller than expected
5315 function call code (so overall size of program gets smaller). The compiler
5316 heuristically decides which functions are simple enough to be worth integrating
5319 Enabled at level @option{-O2}.
5321 @item -finline-functions
5322 @opindex finline-functions
5323 Integrate all simple functions into their callers. The compiler
5324 heuristically decides which functions are simple enough to be worth
5325 integrating in this way.
5327 If all calls to a given function are integrated, and the function is
5328 declared @code{static}, then the function is normally not output as
5329 assembler code in its own right.
5331 Enabled at level @option{-O3}.
5333 @item -finline-functions-called-once
5334 @opindex finline-functions-called-once
5335 Consider all @code{static} functions called once for inlining into their
5336 caller even if they are not marked @code{inline}. If a call to a given
5337 function is integrated, then the function is not output as assembler code
5340 Enabled if @option{-funit-at-a-time} is enabled.
5342 @item -fearly-inlining
5343 @opindex fearly-inlining
5344 Inline functions marked by @code{always_inline} and functions whose body seems
5345 smaller than the function call overhead early before doing
5346 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5347 makes profiling significantly cheaper and usually inlining faster on programs
5348 having large chains of nested wrapper functions.
5352 @item -finline-limit=@var{n}
5353 @opindex finline-limit
5354 By default, GCC limits the size of functions that can be inlined. This flag
5355 allows coarse control of this limit. @var{n} is the size of functions that
5356 can be inlined in number of pseudo instructions.
5358 Inlining is actually controlled by a number of parameters, which may be
5359 specified individually by using @option{--param @var{name}=@var{value}}.
5360 The @option{-finline-limit=@var{n}} option sets some of these parameters
5364 @item max-inline-insns-single
5365 is set to @var{n}/2.
5366 @item max-inline-insns-auto
5367 is set to @var{n}/2.
5370 See below for a documentation of the individual
5371 parameters controlling inlining and for the defaults of these parameters.
5373 @emph{Note:} there may be no value to @option{-finline-limit} that results
5374 in default behavior.
5376 @emph{Note:} pseudo instruction represents, in this particular context, an
5377 abstract measurement of function's size. In no way does it represent a count
5378 of assembly instructions and as such its exact meaning might change from one
5379 release to an another.
5381 @item -fkeep-inline-functions
5382 @opindex fkeep-inline-functions
5383 In C, emit @code{static} functions that are declared @code{inline}
5384 into the object file, even if the function has been inlined into all
5385 of its callers. This switch does not affect functions using the
5386 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5387 inline functions into the object file.
5389 @item -fkeep-static-consts
5390 @opindex fkeep-static-consts
5391 Emit variables declared @code{static const} when optimization isn't turned
5392 on, even if the variables aren't referenced.
5394 GCC enables this option by default. If you want to force the compiler to
5395 check if the variable was referenced, regardless of whether or not
5396 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5398 @item -fmerge-constants
5399 @opindex fmerge-constants
5400 Attempt to merge identical constants (string constants and floating point
5401 constants) across compilation units.
5403 This option is the default for optimized compilation if the assembler and
5404 linker support it. Use @option{-fno-merge-constants} to inhibit this
5407 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5409 @item -fmerge-all-constants
5410 @opindex fmerge-all-constants
5411 Attempt to merge identical constants and identical variables.
5413 This option implies @option{-fmerge-constants}. In addition to
5414 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5415 arrays or initialized constant variables with integral or floating point
5416 types. Languages like C or C++ require each non-automatic variable to
5417 have distinct location, so using this option will result in non-conforming
5420 @item -fmodulo-sched
5421 @opindex fmodulo-sched
5422 Perform swing modulo scheduling immediately before the first scheduling
5423 pass. This pass looks at innermost loops and reorders their
5424 instructions by overlapping different iterations.
5426 @item -fmodulo-sched-allow-regmoves
5427 @opindex fmodulo-sched-allow-regmoves
5428 Perform more aggressive SMS based modulo scheduling with register moves
5429 allowed. By setting this flag certain anti-dependences edges will be
5430 deleted which will trigger the generation of reg-moves based on the
5431 life-range analysis. This option is effective only with
5432 @option{-fmodulo-sched} enabled.
5434 @item -fno-branch-count-reg
5435 @opindex fno-branch-count-reg
5436 Do not use ``decrement and branch'' instructions on a count register,
5437 but instead generate a sequence of instructions that decrement a
5438 register, compare it against zero, then branch based upon the result.
5439 This option is only meaningful on architectures that support such
5440 instructions, which include x86, PowerPC, IA-64 and S/390.
5442 The default is @option{-fbranch-count-reg}.
5444 @item -fno-function-cse
5445 @opindex fno-function-cse
5446 Do not put function addresses in registers; make each instruction that
5447 calls a constant function contain the function's address explicitly.
5449 This option results in less efficient code, but some strange hacks
5450 that alter the assembler output may be confused by the optimizations
5451 performed when this option is not used.
5453 The default is @option{-ffunction-cse}
5455 @item -fno-zero-initialized-in-bss
5456 @opindex fno-zero-initialized-in-bss
5457 If the target supports a BSS section, GCC by default puts variables that
5458 are initialized to zero into BSS@. This can save space in the resulting
5461 This option turns off this behavior because some programs explicitly
5462 rely on variables going to the data section. E.g., so that the
5463 resulting executable can find the beginning of that section and/or make
5464 assumptions based on that.
5466 The default is @option{-fzero-initialized-in-bss}.
5468 @item -fmudflap -fmudflapth -fmudflapir
5472 @cindex bounds checking
5474 For front-ends that support it (C and C++), instrument all risky
5475 pointer/array dereferencing operations, some standard library
5476 string/heap functions, and some other associated constructs with
5477 range/validity tests. Modules so instrumented should be immune to
5478 buffer overflows, invalid heap use, and some other classes of C/C++
5479 programming errors. The instrumentation relies on a separate runtime
5480 library (@file{libmudflap}), which will be linked into a program if
5481 @option{-fmudflap} is given at link time. Run-time behavior of the
5482 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5483 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5486 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5487 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5488 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5489 instrumentation should ignore pointer reads. This produces less
5490 instrumentation (and therefore faster execution) and still provides
5491 some protection against outright memory corrupting writes, but allows
5492 erroneously read data to propagate within a program.
5494 @item -fthread-jumps
5495 @opindex fthread-jumps
5496 Perform optimizations where we check to see if a jump branches to a
5497 location where another comparison subsumed by the first is found. If
5498 so, the first branch is redirected to either the destination of the
5499 second branch or a point immediately following it, depending on whether
5500 the condition is known to be true or false.
5502 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5504 @item -fsplit-wide-types
5505 @opindex fsplit-wide-types
5506 When using a type that occupies multiple registers, such as @code{long
5507 long} on a 32-bit system, split the registers apart and allocate them
5508 independently. This normally generates better code for those types,
5509 but may make debugging more difficult.
5511 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5514 @item -fcse-follow-jumps
5515 @opindex fcse-follow-jumps
5516 In common subexpression elimination (CSE), scan through jump instructions
5517 when the target of the jump is not reached by any other path. For
5518 example, when CSE encounters an @code{if} statement with an
5519 @code{else} clause, CSE will follow the jump when the condition
5522 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5524 @item -fcse-skip-blocks
5525 @opindex fcse-skip-blocks
5526 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5527 follow jumps which conditionally skip over blocks. When CSE
5528 encounters a simple @code{if} statement with no else clause,
5529 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5530 body of the @code{if}.
5532 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5534 @item -frerun-cse-after-loop
5535 @opindex frerun-cse-after-loop
5536 Re-run common subexpression elimination after loop optimizations has been
5539 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5543 Perform a global common subexpression elimination pass.
5544 This pass also performs global constant and copy propagation.
5546 @emph{Note:} When compiling a program using computed gotos, a GCC
5547 extension, you may get better runtime performance if you disable
5548 the global common subexpression elimination pass by adding
5549 @option{-fno-gcse} to the command line.
5551 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5555 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5556 attempt to move loads which are only killed by stores into themselves. This
5557 allows a loop containing a load/store sequence to be changed to a load outside
5558 the loop, and a copy/store within the loop.
5560 Enabled by default when gcse is enabled.
5564 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5565 global common subexpression elimination. This pass will attempt to move
5566 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5567 loops containing a load/store sequence can be changed to a load before
5568 the loop and a store after the loop.
5570 Not enabled at any optimization level.
5574 When @option{-fgcse-las} is enabled, the global common subexpression
5575 elimination pass eliminates redundant loads that come after stores to the
5576 same memory location (both partial and full redundancies).
5578 Not enabled at any optimization level.
5580 @item -fgcse-after-reload
5581 @opindex fgcse-after-reload
5582 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5583 pass is performed after reload. The purpose of this pass is to cleanup
5586 @item -funsafe-loop-optimizations
5587 @opindex funsafe-loop-optimizations
5588 If given, the loop optimizer will assume that loop indices do not
5589 overflow, and that the loops with nontrivial exit condition are not
5590 infinite. This enables a wider range of loop optimizations even if
5591 the loop optimizer itself cannot prove that these assumptions are valid.
5592 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5593 if it finds this kind of loop.
5595 @item -fcrossjumping
5596 @opindex fcrossjumping
5597 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5598 resulting code may or may not perform better than without cross-jumping.
5600 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5602 @item -fauto-inc-dec
5603 @opindex fauto-inc-dec
5604 Combine increments or decrements of addresses with memory accesses.
5605 This pass is always skipped on architectures that do not have
5606 instructions to support this. Enabled by default at @option{-O} and
5607 higher on architectures that support this.
5611 Perform dead code elimination (DCE) on RTL@.
5612 Enabled by default at @option{-O} and higher.
5616 Perform dead store elimination (DSE) on RTL@.
5617 Enabled by default at @option{-O} and higher.
5619 @item -fif-conversion
5620 @opindex fif-conversion
5621 Attempt to transform conditional jumps into branch-less equivalents. This
5622 include use of conditional moves, min, max, set flags and abs instructions, and
5623 some tricks doable by standard arithmetics. The use of conditional execution
5624 on chips where it is available is controlled by @code{if-conversion2}.
5626 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5628 @item -fif-conversion2
5629 @opindex fif-conversion2
5630 Use conditional execution (where available) to transform conditional jumps into
5631 branch-less equivalents.
5633 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5635 @item -fdelete-null-pointer-checks
5636 @opindex fdelete-null-pointer-checks
5637 Use global dataflow analysis to identify and eliminate useless checks
5638 for null pointers. The compiler assumes that dereferencing a null
5639 pointer would have halted the program. If a pointer is checked after
5640 it has already been dereferenced, it cannot be null.
5642 In some environments, this assumption is not true, and programs can
5643 safely dereference null pointers. Use
5644 @option{-fno-delete-null-pointer-checks} to disable this optimization
5645 for programs which depend on that behavior.
5647 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5649 @item -fexpensive-optimizations
5650 @opindex fexpensive-optimizations
5651 Perform a number of minor optimizations that are relatively expensive.
5653 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5655 @item -foptimize-register-move
5657 @opindex foptimize-register-move
5659 Attempt to reassign register numbers in move instructions and as
5660 operands of other simple instructions in order to maximize the amount of
5661 register tying. This is especially helpful on machines with two-operand
5664 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5667 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5669 @item -fdelayed-branch
5670 @opindex fdelayed-branch
5671 If supported for the target machine, attempt to reorder instructions
5672 to exploit instruction slots available after delayed branch
5675 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5677 @item -fschedule-insns
5678 @opindex fschedule-insns
5679 If supported for the target machine, attempt to reorder instructions to
5680 eliminate execution stalls due to required data being unavailable. This
5681 helps machines that have slow floating point or memory load instructions
5682 by allowing other instructions to be issued until the result of the load
5683 or floating point instruction is required.
5685 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5687 @item -fschedule-insns2
5688 @opindex fschedule-insns2
5689 Similar to @option{-fschedule-insns}, but requests an additional pass of
5690 instruction scheduling after register allocation has been done. This is
5691 especially useful on machines with a relatively small number of
5692 registers and where memory load instructions take more than one cycle.
5694 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5696 @item -fno-sched-interblock
5697 @opindex fno-sched-interblock
5698 Don't schedule instructions across basic blocks. This is normally
5699 enabled by default when scheduling before register allocation, i.e.@:
5700 with @option{-fschedule-insns} or at @option{-O2} or higher.
5702 @item -fno-sched-spec
5703 @opindex fno-sched-spec
5704 Don't allow speculative motion of non-load instructions. This is normally
5705 enabled by default when scheduling before register allocation, i.e.@:
5706 with @option{-fschedule-insns} or at @option{-O2} or higher.
5708 @item -fsched-spec-load
5709 @opindex fsched-spec-load
5710 Allow speculative motion of some load instructions. This only makes
5711 sense when scheduling before register allocation, i.e.@: with
5712 @option{-fschedule-insns} or at @option{-O2} or higher.
5714 @item -fsched-spec-load-dangerous
5715 @opindex fsched-spec-load-dangerous
5716 Allow speculative motion of more load instructions. This only makes
5717 sense when scheduling before register allocation, i.e.@: with
5718 @option{-fschedule-insns} or at @option{-O2} or higher.
5720 @item -fsched-stalled-insns
5721 @itemx -fsched-stalled-insns=@var{n}
5722 @opindex fsched-stalled-insns
5723 Define how many insns (if any) can be moved prematurely from the queue
5724 of stalled insns into the ready list, during the second scheduling pass.
5725 @option{-fno-sched-stalled-insns} means that no insns will be moved
5726 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5727 on how many queued insns can be moved prematurely.
5728 @option{-fsched-stalled-insns} without a value is equivalent to
5729 @option{-fsched-stalled-insns=1}.
5731 @item -fsched-stalled-insns-dep
5732 @itemx -fsched-stalled-insns-dep=@var{n}
5733 @opindex fsched-stalled-insns-dep
5734 Define how many insn groups (cycles) will be examined for a dependency
5735 on a stalled insn that is candidate for premature removal from the queue
5736 of stalled insns. This has an effect only during the second scheduling pass,
5737 and only if @option{-fsched-stalled-insns} is used.
5738 @option{-fno-sched-stalled-insns-dep} is equivalent to
5739 @option{-fsched-stalled-insns-dep=0}.
5740 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5741 @option{-fsched-stalled-insns-dep=1}.
5743 @item -fsched2-use-superblocks
5744 @opindex fsched2-use-superblocks
5745 When scheduling after register allocation, do use superblock scheduling
5746 algorithm. Superblock scheduling allows motion across basic block boundaries
5747 resulting on faster schedules. This option is experimental, as not all machine
5748 descriptions used by GCC model the CPU closely enough to avoid unreliable
5749 results from the algorithm.
5751 This only makes sense when scheduling after register allocation, i.e.@: with
5752 @option{-fschedule-insns2} or at @option{-O2} or higher.
5754 @item -fsched2-use-traces
5755 @opindex fsched2-use-traces
5756 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5757 allocation and additionally perform code duplication in order to increase the
5758 size of superblocks using tracer pass. See @option{-ftracer} for details on
5761 This mode should produce faster but significantly longer programs. Also
5762 without @option{-fbranch-probabilities} the traces constructed may not
5763 match the reality and hurt the performance. This only makes
5764 sense when scheduling after register allocation, i.e.@: with
5765 @option{-fschedule-insns2} or at @option{-O2} or higher.
5769 Eliminate redundant sign extension instructions and move the non-redundant
5770 ones to optimal placement using lazy code motion (LCM).
5772 @item -freschedule-modulo-scheduled-loops
5773 @opindex freschedule-modulo-scheduled-loops
5774 The modulo scheduling comes before the traditional scheduling, if a loop
5775 was modulo scheduled we may want to prevent the later scheduling passes
5776 from changing its schedule, we use this option to control that.
5778 @item -fcaller-saves
5779 @opindex fcaller-saves
5780 Enable values to be allocated in registers that will be clobbered by
5781 function calls, by emitting extra instructions to save and restore the
5782 registers around such calls. Such allocation is done only when it
5783 seems to result in better code than would otherwise be produced.
5785 This option is always enabled by default on certain machines, usually
5786 those which have no call-preserved registers to use instead.
5788 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5790 @item -ftree-reassoc
5791 @opindex ftree-reassoc
5792 Perform reassociation on trees. This flag is enabled by default
5793 at @option{-O} and higher.
5797 Perform partial redundancy elimination (PRE) on trees. This flag is
5798 enabled by default at @option{-O2} and @option{-O3}.
5802 Perform full redundancy elimination (FRE) on trees. The difference
5803 between FRE and PRE is that FRE only considers expressions
5804 that are computed on all paths leading to the redundant computation.
5805 This analysis is faster than PRE, though it exposes fewer redundancies.
5806 This flag is enabled by default at @option{-O} and higher.
5808 @item -ftree-copy-prop
5809 @opindex ftree-copy-prop
5810 Perform copy propagation on trees. This pass eliminates unnecessary
5811 copy operations. This flag is enabled by default at @option{-O} and
5815 @opindex ftree-salias
5816 Perform structural alias analysis on trees. This flag
5817 is enabled by default at @option{-O} and higher.
5819 @item -fipa-pure-const
5820 @opindex fipa-pure-const
5821 Discover which functions are pure or constant.
5822 Enabled by default at @option{-O} and higher.
5824 @item -fipa-reference
5825 @opindex fipa-reference
5826 Discover which static variables do not escape cannot escape the
5828 Enabled by default at @option{-O} and higher.
5830 @item -fipa-struct-reorg
5831 @opindex fipa-struct-reorg
5832 Perform structure reorganization optimization, that change C-like structures
5833 layout in order to better utilize spatial locality. This transformation is
5834 affective for programs containing arrays of structures. Available in two
5835 compilation modes: profile-based (enabled with @option{-fprofile-generate})
5836 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
5837 to provide the safety of this transformation. It works only in whole program
5838 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
5839 enabled. Structures considered @samp{cold} by this transformation are not
5840 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
5842 With this flag, the program debug info reflects a new structure layout.
5846 Perform interprocedural pointer analysis.
5850 Perform interprocedural constant propagation.
5851 This optimization analyzes the program to determine when values passed
5852 to functions are constants and then optimizes accordingly.
5853 This optimization can substantially increase performance
5854 if the application has constants passed to functions, but
5855 because this optimization can create multiple copies of functions,
5856 it may significantly increase code size.
5858 @item -fipa-matrix-reorg
5859 @opindex fipa-matrix-reorg
5860 Perform matrix flattening and transposing.
5861 Matrix flattening tries to replace a m-dimensional matrix
5862 with its equivalent n-dimensional matrix, where n < m.
5863 This reduces the level of indirection needed for accessing the elements
5864 of the matrix. The second optimization is matrix transposing that
5865 attemps to change the order of the matrix's dimensions in order to
5866 improve cache locality.
5867 Both optimizations need fwhole-program flag.
5868 Transposing is enabled only if profiling information is avaliable.
5873 Perform forward store motion on trees. This flag is
5874 enabled by default at @option{-O} and higher.
5878 Perform sparse conditional constant propagation (CCP) on trees. This
5879 pass only operates on local scalar variables and is enabled by default
5880 at @option{-O} and higher.
5882 @item -ftree-store-ccp
5883 @opindex ftree-store-ccp
5884 Perform sparse conditional constant propagation (CCP) on trees. This
5885 pass operates on both local scalar variables and memory stores and
5886 loads (global variables, structures, arrays, etc). This flag is
5887 enabled by default at @option{-O2} and higher.
5891 Perform dead code elimination (DCE) on trees. This flag is enabled by
5892 default at @option{-O} and higher.
5894 @item -ftree-dominator-opts
5895 @opindex ftree-dominator-opts
5896 Perform a variety of simple scalar cleanups (constant/copy
5897 propagation, redundancy elimination, range propagation and expression
5898 simplification) based on a dominator tree traversal. This also
5899 performs jump threading (to reduce jumps to jumps). This flag is
5900 enabled by default at @option{-O} and higher.
5904 Perform dead store elimination (DSE) on trees. A dead store is a store into
5905 a memory location which will later be overwritten by another store without
5906 any intervening loads. In this case the earlier store can be deleted. This
5907 flag is enabled by default at @option{-O} and higher.
5911 Perform loop header copying on trees. This is beneficial since it increases
5912 effectiveness of code motion optimizations. It also saves one jump. This flag
5913 is enabled by default at @option{-O} and higher. It is not enabled
5914 for @option{-Os}, since it usually increases code size.
5916 @item -ftree-loop-optimize
5917 @opindex ftree-loop-optimize
5918 Perform loop optimizations on trees. This flag is enabled by default
5919 at @option{-O} and higher.
5921 @item -ftree-loop-linear
5922 @opindex ftree-loop-linear
5923 Perform linear loop transformations on tree. This flag can improve cache
5924 performance and allow further loop optimizations to take place.
5926 @item -fcheck-data-deps
5927 @opindex fcheck-data-deps
5928 Compare the results of several data dependence analyzers. This option
5929 is used for debugging the data dependence analyzers.
5931 @item -ftree-loop-im
5932 @opindex ftree-loop-im
5933 Perform loop invariant motion on trees. This pass moves only invariants that
5934 would be hard to handle at RTL level (function calls, operations that expand to
5935 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5936 operands of conditions that are invariant out of the loop, so that we can use
5937 just trivial invariantness analysis in loop unswitching. The pass also includes
5940 @item -ftree-loop-ivcanon
5941 @opindex ftree-loop-ivcanon
5942 Create a canonical counter for number of iterations in the loop for that
5943 determining number of iterations requires complicated analysis. Later
5944 optimizations then may determine the number easily. Useful especially
5945 in connection with unrolling.
5949 Perform induction variable optimizations (strength reduction, induction
5950 variable merging and induction variable elimination) on trees.
5952 @item -ftree-parallelize-loops=n
5953 @opindex ftree-parallelize-loops
5954 Parallelize loops, i.e., split their iteration space to run in n threads.
5955 This is only possible for loops whose iterations are independent
5956 and can be arbitrarily reordered. The optimization is only
5957 profitable on multiprocessor machines, for loops that are CPU-intensive,
5958 rather than constrained e.g.@: by memory bandwidth. This option
5959 implies @option{-pthread}, and thus is only supported on targets
5960 that have support for @option{-pthread}.
5964 Perform scalar replacement of aggregates. This pass replaces structure
5965 references with scalars to prevent committing structures to memory too
5966 early. This flag is enabled by default at @option{-O} and higher.
5968 @item -ftree-copyrename
5969 @opindex ftree-copyrename
5970 Perform copy renaming on trees. This pass attempts to rename compiler
5971 temporaries to other variables at copy locations, usually resulting in
5972 variable names which more closely resemble the original variables. This flag
5973 is enabled by default at @option{-O} and higher.
5977 Perform temporary expression replacement during the SSA->normal phase. Single
5978 use/single def temporaries are replaced at their use location with their
5979 defining expression. This results in non-GIMPLE code, but gives the expanders
5980 much more complex trees to work on resulting in better RTL generation. This is
5981 enabled by default at @option{-O} and higher.
5983 @item -ftree-vectorize
5984 @opindex ftree-vectorize
5985 Perform loop vectorization on trees.
5987 @item -ftree-vect-loop-version
5988 @opindex ftree-vect-loop-version
5989 Perform loop versioning when doing loop vectorization on trees. When a loop
5990 appears to be vectorizable except that data alignment or data dependence cannot
5991 be determined at compile time then vectorized and non-vectorized versions of
5992 the loop are generated along with runtime checks for alignment or dependence
5993 to control which version is executed. This option is enabled by default
5994 except at level @option{-Os} where it is disabled.
5996 @item -fvect-cost-model
5997 @opindex fvect-cost-model
5998 Enable cost model for vectorization.
6002 Perform Value Range Propagation on trees. This is similar to the
6003 constant propagation pass, but instead of values, ranges of values are
6004 propagated. This allows the optimizers to remove unnecessary range
6005 checks like array bound checks and null pointer checks. This is
6006 enabled by default at @option{-O2} and higher. Null pointer check
6007 elimination is only done if @option{-fdelete-null-pointer-checks} is
6012 Perform tail duplication to enlarge superblock size. This transformation
6013 simplifies the control flow of the function allowing other optimizations to do
6016 @item -funroll-loops
6017 @opindex funroll-loops
6018 Unroll loops whose number of iterations can be determined at compile
6019 time or upon entry to the loop. @option{-funroll-loops} implies
6020 @option{-frerun-cse-after-loop}. This option makes code larger,
6021 and may or may not make it run faster.
6023 @item -funroll-all-loops
6024 @opindex funroll-all-loops
6025 Unroll all loops, even if their number of iterations is uncertain when
6026 the loop is entered. This usually makes programs run more slowly.
6027 @option{-funroll-all-loops} implies the same options as
6028 @option{-funroll-loops},
6030 @item -fsplit-ivs-in-unroller
6031 @opindex fsplit-ivs-in-unroller
6032 Enables expressing of values of induction variables in later iterations
6033 of the unrolled loop using the value in the first iteration. This breaks
6034 long dependency chains, thus improving efficiency of the scheduling passes.
6036 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6037 same effect. However in cases the loop body is more complicated than
6038 a single basic block, this is not reliable. It also does not work at all
6039 on some of the architectures due to restrictions in the CSE pass.
6041 This optimization is enabled by default.
6043 @item -fvariable-expansion-in-unroller
6044 @opindex fvariable-expansion-in-unroller
6045 With this option, the compiler will create multiple copies of some
6046 local variables when unrolling a loop which can result in superior code.
6048 @item -fpredictive-commoning
6049 @opindex fpredictive-commoning
6050 Perform predictive commoning optimization, i.e., reusing computations
6051 (especially memory loads and stores) performed in previous
6052 iterations of loops.
6054 This option is enabled at level @option{-O3}.
6056 @item -fprefetch-loop-arrays
6057 @opindex fprefetch-loop-arrays
6058 If supported by the target machine, generate instructions to prefetch
6059 memory to improve the performance of loops that access large arrays.
6061 This option may generate better or worse code; results are highly
6062 dependent on the structure of loops within the source code.
6064 Disabled at level @option{-Os}.
6067 @itemx -fno-peephole2
6068 @opindex fno-peephole
6069 @opindex fno-peephole2
6070 Disable any machine-specific peephole optimizations. The difference
6071 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6072 are implemented in the compiler; some targets use one, some use the
6073 other, a few use both.
6075 @option{-fpeephole} is enabled by default.
6076 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6078 @item -fno-guess-branch-probability
6079 @opindex fno-guess-branch-probability
6080 Do not guess branch probabilities using heuristics.
6082 GCC will use heuristics to guess branch probabilities if they are
6083 not provided by profiling feedback (@option{-fprofile-arcs}). These
6084 heuristics are based on the control flow graph. If some branch probabilities
6085 are specified by @samp{__builtin_expect}, then the heuristics will be
6086 used to guess branch probabilities for the rest of the control flow graph,
6087 taking the @samp{__builtin_expect} info into account. The interactions
6088 between the heuristics and @samp{__builtin_expect} can be complex, and in
6089 some cases, it may be useful to disable the heuristics so that the effects
6090 of @samp{__builtin_expect} are easier to understand.
6092 The default is @option{-fguess-branch-probability} at levels
6093 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6095 @item -freorder-blocks
6096 @opindex freorder-blocks
6097 Reorder basic blocks in the compiled function in order to reduce number of
6098 taken branches and improve code locality.
6100 Enabled at levels @option{-O2}, @option{-O3}.
6102 @item -freorder-blocks-and-partition
6103 @opindex freorder-blocks-and-partition
6104 In addition to reordering basic blocks in the compiled function, in order
6105 to reduce number of taken branches, partitions hot and cold basic blocks
6106 into separate sections of the assembly and .o files, to improve
6107 paging and cache locality performance.
6109 This optimization is automatically turned off in the presence of
6110 exception handling, for linkonce sections, for functions with a user-defined
6111 section attribute and on any architecture that does not support named
6114 @item -freorder-functions
6115 @opindex freorder-functions
6116 Reorder functions in the object file in order to
6117 improve code locality. This is implemented by using special
6118 subsections @code{.text.hot} for most frequently executed functions and
6119 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6120 the linker so object file format must support named sections and linker must
6121 place them in a reasonable way.
6123 Also profile feedback must be available in to make this option effective. See
6124 @option{-fprofile-arcs} for details.
6126 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6128 @item -fstrict-aliasing
6129 @opindex fstrict-aliasing
6130 Allows the compiler to assume the strictest aliasing rules applicable to
6131 the language being compiled. For C (and C++), this activates
6132 optimizations based on the type of expressions. In particular, an
6133 object of one type is assumed never to reside at the same address as an
6134 object of a different type, unless the types are almost the same. For
6135 example, an @code{unsigned int} can alias an @code{int}, but not a
6136 @code{void*} or a @code{double}. A character type may alias any other
6139 Pay special attention to code like this:
6152 The practice of reading from a different union member than the one most
6153 recently written to (called ``type-punning'') is common. Even with
6154 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6155 is accessed through the union type. So, the code above will work as
6156 expected. However, this code might not:
6167 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6169 @item -fstrict-overflow
6170 @opindex fstrict-overflow
6171 Allow the compiler to assume strict signed overflow rules, depending
6172 on the language being compiled. For C (and C++) this means that
6173 overflow when doing arithmetic with signed numbers is undefined, which
6174 means that the compiler may assume that it will not happen. This
6175 permits various optimizations. For example, the compiler will assume
6176 that an expression like @code{i + 10 > i} will always be true for
6177 signed @code{i}. This assumption is only valid if signed overflow is
6178 undefined, as the expression is false if @code{i + 10} overflows when
6179 using twos complement arithmetic. When this option is in effect any
6180 attempt to determine whether an operation on signed numbers will
6181 overflow must be written carefully to not actually involve overflow.
6183 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6184 that signed overflow is fully defined: it wraps. When
6185 @option{-fwrapv} is used, there is no difference between
6186 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
6187 @option{-fwrapv} certain types of overflow are permitted. For
6188 example, if the compiler gets an overflow when doing arithmetic on
6189 constants, the overflowed value can still be used with
6190 @option{-fwrapv}, but not otherwise.
6192 The @option{-fstrict-overflow} option is enabled at levels
6193 @option{-O2}, @option{-O3}, @option{-Os}.
6195 @item -falign-functions
6196 @itemx -falign-functions=@var{n}
6197 @opindex falign-functions
6198 Align the start of functions to the next power-of-two greater than
6199 @var{n}, skipping up to @var{n} bytes. For instance,
6200 @option{-falign-functions=32} aligns functions to the next 32-byte
6201 boundary, but @option{-falign-functions=24} would align to the next
6202 32-byte boundary only if this can be done by skipping 23 bytes or less.
6204 @option{-fno-align-functions} and @option{-falign-functions=1} are
6205 equivalent and mean that functions will not be aligned.
6207 Some assemblers only support this flag when @var{n} is a power of two;
6208 in that case, it is rounded up.
6210 If @var{n} is not specified or is zero, use a machine-dependent default.
6212 Enabled at levels @option{-O2}, @option{-O3}.
6214 @item -falign-labels
6215 @itemx -falign-labels=@var{n}
6216 @opindex falign-labels
6217 Align all branch targets to a power-of-two boundary, skipping up to
6218 @var{n} bytes like @option{-falign-functions}. This option can easily
6219 make code slower, because it must insert dummy operations for when the
6220 branch target is reached in the usual flow of the code.
6222 @option{-fno-align-labels} and @option{-falign-labels=1} are
6223 equivalent and mean that labels will not be aligned.
6225 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6226 are greater than this value, then their values are used instead.
6228 If @var{n} is not specified or is zero, use a machine-dependent default
6229 which is very likely to be @samp{1}, meaning no alignment.
6231 Enabled at levels @option{-O2}, @option{-O3}.
6234 @itemx -falign-loops=@var{n}
6235 @opindex falign-loops
6236 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6237 like @option{-falign-functions}. The hope is that the loop will be
6238 executed many times, which will make up for any execution of the dummy
6241 @option{-fno-align-loops} and @option{-falign-loops=1} are
6242 equivalent and mean that loops will not be aligned.
6244 If @var{n} is not specified or is zero, use a machine-dependent default.
6246 Enabled at levels @option{-O2}, @option{-O3}.
6249 @itemx -falign-jumps=@var{n}
6250 @opindex falign-jumps
6251 Align branch targets to a power-of-two boundary, for branch targets
6252 where the targets can only be reached by jumping, skipping up to @var{n}
6253 bytes like @option{-falign-functions}. In this case, no dummy operations
6256 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6257 equivalent and mean that loops will not be aligned.
6259 If @var{n} is not specified or is zero, use a machine-dependent default.
6261 Enabled at levels @option{-O2}, @option{-O3}.
6263 @item -funit-at-a-time
6264 @opindex funit-at-a-time
6265 Parse the whole compilation unit before starting to produce code.
6266 This allows some extra optimizations to take place but consumes
6267 more memory (in general). There are some compatibility issues
6268 with @emph{unit-at-a-time} mode:
6271 enabling @emph{unit-at-a-time} mode may change the order
6272 in which functions, variables, and top-level @code{asm} statements
6273 are emitted, and will likely break code relying on some particular
6274 ordering. The majority of such top-level @code{asm} statements,
6275 though, can be replaced by @code{section} attributes. The
6276 @option{fno-toplevel-reorder} option may be used to keep the ordering
6277 used in the input file, at the cost of some optimizations.
6280 @emph{unit-at-a-time} mode removes unreferenced static variables
6281 and functions. This may result in undefined references
6282 when an @code{asm} statement refers directly to variables or functions
6283 that are otherwise unused. In that case either the variable/function
6284 shall be listed as an operand of the @code{asm} statement operand or,
6285 in the case of top-level @code{asm} statements the attribute @code{used}
6286 shall be used on the declaration.
6289 Static functions now can use non-standard passing conventions that
6290 may break @code{asm} statements calling functions directly. Again,
6291 attribute @code{used} will prevent this behavior.
6294 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6295 but this scheme may not be supported by future releases of GCC@.
6297 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6299 @item -fno-toplevel-reorder
6300 @opindex fno-toplevel-reorder
6301 Do not reorder top-level functions, variables, and @code{asm}
6302 statements. Output them in the same order that they appear in the
6303 input file. When this option is used, unreferenced static variables
6304 will not be removed. This option is intended to support existing code
6305 which relies on a particular ordering. For new code, it is better to
6310 Constructs webs as commonly used for register allocation purposes and assign
6311 each web individual pseudo register. This allows the register allocation pass
6312 to operate on pseudos directly, but also strengthens several other optimization
6313 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6314 however, make debugging impossible, since variables will no longer stay in a
6317 Enabled by default with @option{-funroll-loops}.
6319 @item -fwhole-program
6320 @opindex fwhole-program
6321 Assume that the current compilation unit represents whole program being
6322 compiled. All public functions and variables with the exception of @code{main}
6323 and those merged by attribute @code{externally_visible} become static functions
6324 and in a affect gets more aggressively optimized by interprocedural optimizers.
6325 While this option is equivalent to proper use of @code{static} keyword for
6326 programs consisting of single file, in combination with option
6327 @option{--combine} this flag can be used to compile most of smaller scale C
6328 programs since the functions and variables become local for the whole combined
6329 compilation unit, not for the single source file itself.
6331 This option is not supported for Fortran programs.
6333 @item -fcprop-registers
6334 @opindex fcprop-registers
6335 After register allocation and post-register allocation instruction splitting,
6336 we perform a copy-propagation pass to try to reduce scheduling dependencies
6337 and occasionally eliminate the copy.
6339 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6341 @item -fprofile-generate
6342 @opindex fprofile-generate
6344 Enable options usually used for instrumenting application to produce
6345 profile useful for later recompilation with profile feedback based
6346 optimization. You must use @option{-fprofile-generate} both when
6347 compiling and when linking your program.
6349 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6352 @opindex fprofile-use
6353 Enable profile feedback directed optimizations, and optimizations
6354 generally profitable only with profile feedback available.
6356 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6357 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6359 By default, GCC emits an error message if the feedback profiles do not
6360 match the source code. This error can be turned into a warning by using
6361 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6365 The following options control compiler behavior regarding floating
6366 point arithmetic. These options trade off between speed and
6367 correctness. All must be specifically enabled.
6371 @opindex ffloat-store
6372 Do not store floating point variables in registers, and inhibit other
6373 options that might change whether a floating point value is taken from a
6376 @cindex floating point precision
6377 This option prevents undesirable excess precision on machines such as
6378 the 68000 where the floating registers (of the 68881) keep more
6379 precision than a @code{double} is supposed to have. Similarly for the
6380 x86 architecture. For most programs, the excess precision does only
6381 good, but a few programs rely on the precise definition of IEEE floating
6382 point. Use @option{-ffloat-store} for such programs, after modifying
6383 them to store all pertinent intermediate computations into variables.
6387 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6388 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6389 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6391 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6393 This option is not turned on by any @option{-O} option since
6394 it can result in incorrect output for programs which depend on
6395 an exact implementation of IEEE or ISO rules/specifications for
6396 math functions. It may, however, yield faster code for programs
6397 that do not require the guarantees of these specifications.
6399 @item -fno-math-errno
6400 @opindex fno-math-errno
6401 Do not set ERRNO after calling math functions that are executed
6402 with a single instruction, e.g., sqrt. A program that relies on
6403 IEEE exceptions for math error handling may want to use this flag
6404 for speed while maintaining IEEE arithmetic compatibility.
6406 This option is not turned on by any @option{-O} option since
6407 it can result in incorrect output for programs which depend on
6408 an exact implementation of IEEE or ISO rules/specifications for
6409 math functions. It may, however, yield faster code for programs
6410 that do not require the guarantees of these specifications.
6412 The default is @option{-fmath-errno}.
6414 On Darwin systems, the math library never sets @code{errno}. There is
6415 therefore no reason for the compiler to consider the possibility that
6416 it might, and @option{-fno-math-errno} is the default.
6418 @item -funsafe-math-optimizations
6419 @opindex funsafe-math-optimizations
6421 Allow optimizations for floating-point arithmetic that (a) assume
6422 that arguments and results are valid and (b) may violate IEEE or
6423 ANSI standards. When used at link-time, it may include libraries
6424 or startup files that change the default FPU control word or other
6425 similar optimizations.
6427 This option is not turned on by any @option{-O} option since
6428 it can result in incorrect output for programs which depend on
6429 an exact implementation of IEEE or ISO rules/specifications for
6430 math functions. It may, however, yield faster code for programs
6431 that do not require the guarantees of these specifications.
6432 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6433 @option{-fassociative-math} and @option{-freciprocal-math}.
6435 The default is @option{-fno-unsafe-math-optimizations}.
6437 @item -fassociative-math
6438 @opindex fassociative-math
6440 Allow re-association of operands in series of floating-point operations.
6441 This violates the ISO C and C++ language standard by possibly changing
6442 computation result. NOTE: re-ordering may change the sign of zero as
6443 well as ignore NaNs and inhibit or create underflow or overflow (and
6444 thus cannot be used on a code which relies on rounding behavior like
6445 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6446 and thus may not be used when ordered comparisons are required.
6447 This option requires that both @option{-fno-signed-zeros} and
6448 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6449 much sense with @option{-frounding-math}.
6451 The default is @option{-fno-associative-math}.
6453 @item -freciprocal-math
6454 @opindex freciprocal-math
6456 Allow the reciprocal of a value to be used instead of dividing by
6457 the value if this enables optimizations. For example @code{x / y}
6458 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6459 is subject to common subexpression elimination. Note that this loses
6460 precision and increases the number of flops operating on the value.
6462 The default is @option{-fno-reciprocal-math}.
6464 @item -ffinite-math-only
6465 @opindex ffinite-math-only
6466 Allow optimizations for floating-point arithmetic that assume
6467 that arguments and results are not NaNs or +-Infs.
6469 This option is not turned on by any @option{-O} option since
6470 it can result in incorrect output for programs which depend on
6471 an exact implementation of IEEE or ISO rules/specifications for
6472 math functions. It may, however, yield faster code for programs
6473 that do not require the guarantees of these specifications.
6475 The default is @option{-fno-finite-math-only}.
6477 @item -fno-signed-zeros
6478 @opindex fno-signed-zeros
6479 Allow optimizations for floating point arithmetic that ignore the
6480 signedness of zero. IEEE arithmetic specifies the behavior of
6481 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6482 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6483 This option implies that the sign of a zero result isn't significant.
6485 The default is @option{-fsigned-zeros}.
6487 @item -fno-trapping-math
6488 @opindex fno-trapping-math
6489 Compile code assuming that floating-point operations cannot generate
6490 user-visible traps. These traps include division by zero, overflow,
6491 underflow, inexact result and invalid operation. This option requires
6492 that @option{-fno-signaling-nans} be in effect. Setting this option may
6493 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6495 This option should never be turned on by any @option{-O} option since
6496 it can result in incorrect output for programs which depend on
6497 an exact implementation of IEEE or ISO rules/specifications for
6500 The default is @option{-ftrapping-math}.
6502 @item -frounding-math
6503 @opindex frounding-math
6504 Disable transformations and optimizations that assume default floating
6505 point rounding behavior. This is round-to-zero for all floating point
6506 to integer conversions, and round-to-nearest for all other arithmetic
6507 truncations. This option should be specified for programs that change
6508 the FP rounding mode dynamically, or that may be executed with a
6509 non-default rounding mode. This option disables constant folding of
6510 floating point expressions at compile-time (which may be affected by
6511 rounding mode) and arithmetic transformations that are unsafe in the
6512 presence of sign-dependent rounding modes.
6514 The default is @option{-fno-rounding-math}.
6516 This option is experimental and does not currently guarantee to
6517 disable all GCC optimizations that are affected by rounding mode.
6518 Future versions of GCC may provide finer control of this setting
6519 using C99's @code{FENV_ACCESS} pragma. This command line option
6520 will be used to specify the default state for @code{FENV_ACCESS}.
6522 @item -frtl-abstract-sequences
6523 @opindex frtl-abstract-sequences
6524 It is a size optimization method. This option is to find identical
6525 sequences of code, which can be turned into pseudo-procedures and
6526 then replace all occurrences with calls to the newly created
6527 subroutine. It is kind of an opposite of @option{-finline-functions}.
6528 This optimization runs at RTL level.
6530 @item -fsignaling-nans
6531 @opindex fsignaling-nans
6532 Compile code assuming that IEEE signaling NaNs may generate user-visible
6533 traps during floating-point operations. Setting this option disables
6534 optimizations that may change the number of exceptions visible with
6535 signaling NaNs. This option implies @option{-ftrapping-math}.
6537 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6540 The default is @option{-fno-signaling-nans}.
6542 This option is experimental and does not currently guarantee to
6543 disable all GCC optimizations that affect signaling NaN behavior.
6545 @item -fsingle-precision-constant
6546 @opindex fsingle-precision-constant
6547 Treat floating point constant as single precision constant instead of
6548 implicitly converting it to double precision constant.
6550 @item -fcx-limited-range
6551 @opindex fcx-limited-range
6552 When enabled, this option states that a range reduction step is not
6553 needed when performing complex division. Also, there is no checking
6554 whether the result of a complex multiplication or division is @code{NaN
6555 + I*NaN}, with an attempt to rescue the situation in that case. The
6556 default is @option{-fno-cx-limited-range}, but is enabled by
6557 @option{-ffast-math}.
6559 This option controls the default setting of the ISO C99
6560 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6563 @item -fcx-fortran-rules
6564 @opindex fcx-fortran-rules
6565 Complex multiplication and division follow Fortran rules. Range
6566 reduction is done as part of complex division, but there is no checking
6567 whether the result of a complex multiplication or division is @code{NaN
6568 + I*NaN}, with an attempt to rescue the situation in that case.
6570 The default is @option{-fno-cx-fortran-rules}.
6574 The following options control optimizations that may improve
6575 performance, but are not enabled by any @option{-O} options. This
6576 section includes experimental options that may produce broken code.
6579 @item -fbranch-probabilities
6580 @opindex fbranch-probabilities
6581 After running a program compiled with @option{-fprofile-arcs}
6582 (@pxref{Debugging Options,, Options for Debugging Your Program or
6583 @command{gcc}}), you can compile it a second time using
6584 @option{-fbranch-probabilities}, to improve optimizations based on
6585 the number of times each branch was taken. When the program
6586 compiled with @option{-fprofile-arcs} exits it saves arc execution
6587 counts to a file called @file{@var{sourcename}.gcda} for each source
6588 file. The information in this data file is very dependent on the
6589 structure of the generated code, so you must use the same source code
6590 and the same optimization options for both compilations.
6592 With @option{-fbranch-probabilities}, GCC puts a
6593 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6594 These can be used to improve optimization. Currently, they are only
6595 used in one place: in @file{reorg.c}, instead of guessing which path a
6596 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6597 exactly determine which path is taken more often.
6599 @item -fprofile-values
6600 @opindex fprofile-values
6601 If combined with @option{-fprofile-arcs}, it adds code so that some
6602 data about values of expressions in the program is gathered.
6604 With @option{-fbranch-probabilities}, it reads back the data gathered
6605 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6606 notes to instructions for their later usage in optimizations.
6608 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6612 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6613 a code to gather information about values of expressions.
6615 With @option{-fbranch-probabilities}, it reads back the data gathered
6616 and actually performs the optimizations based on them.
6617 Currently the optimizations include specialization of division operation
6618 using the knowledge about the value of the denominator.
6620 @item -frename-registers
6621 @opindex frename-registers
6622 Attempt to avoid false dependencies in scheduled code by making use
6623 of registers left over after register allocation. This optimization
6624 will most benefit processors with lots of registers. Depending on the
6625 debug information format adopted by the target, however, it can
6626 make debugging impossible, since variables will no longer stay in
6627 a ``home register''.
6629 Enabled by default with @option{-funroll-loops}.
6633 Perform tail duplication to enlarge superblock size. This transformation
6634 simplifies the control flow of the function allowing other optimizations to do
6637 Enabled with @option{-fprofile-use}.
6639 @item -funroll-loops
6640 @opindex funroll-loops
6641 Unroll loops whose number of iterations can be determined at compile time or
6642 upon entry to the loop. @option{-funroll-loops} implies
6643 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6644 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6645 small constant number of iterations). This option makes code larger, and may
6646 or may not make it run faster.
6648 Enabled with @option{-fprofile-use}.
6650 @item -funroll-all-loops
6651 @opindex funroll-all-loops
6652 Unroll all loops, even if their number of iterations is uncertain when
6653 the loop is entered. This usually makes programs run more slowly.
6654 @option{-funroll-all-loops} implies the same options as
6655 @option{-funroll-loops}.
6658 @opindex fpeel-loops
6659 Peels the loops for that there is enough information that they do not
6660 roll much (from profile feedback). It also turns on complete loop peeling
6661 (i.e.@: complete removal of loops with small constant number of iterations).
6663 Enabled with @option{-fprofile-use}.
6665 @item -fmove-loop-invariants
6666 @opindex fmove-loop-invariants
6667 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6668 at level @option{-O1}
6670 @item -funswitch-loops
6671 @opindex funswitch-loops
6672 Move branches with loop invariant conditions out of the loop, with duplicates
6673 of the loop on both branches (modified according to result of the condition).
6675 @item -ffunction-sections
6676 @itemx -fdata-sections
6677 @opindex ffunction-sections
6678 @opindex fdata-sections
6679 Place each function or data item into its own section in the output
6680 file if the target supports arbitrary sections. The name of the
6681 function or the name of the data item determines the section's name
6684 Use these options on systems where the linker can perform optimizations
6685 to improve locality of reference in the instruction space. Most systems
6686 using the ELF object format and SPARC processors running Solaris 2 have
6687 linkers with such optimizations. AIX may have these optimizations in
6690 Only use these options when there are significant benefits from doing
6691 so. When you specify these options, the assembler and linker will
6692 create larger object and executable files and will also be slower.
6693 You will not be able to use @code{gprof} on all systems if you
6694 specify this option and you may have problems with debugging if
6695 you specify both this option and @option{-g}.
6697 @item -fbranch-target-load-optimize
6698 @opindex fbranch-target-load-optimize
6699 Perform branch target register load optimization before prologue / epilogue
6701 The use of target registers can typically be exposed only during reload,
6702 thus hoisting loads out of loops and doing inter-block scheduling needs
6703 a separate optimization pass.
6705 @item -fbranch-target-load-optimize2
6706 @opindex fbranch-target-load-optimize2
6707 Perform branch target register load optimization after prologue / epilogue
6710 @item -fbtr-bb-exclusive
6711 @opindex fbtr-bb-exclusive
6712 When performing branch target register load optimization, don't reuse
6713 branch target registers in within any basic block.
6715 @item -fstack-protector
6716 @opindex fstack-protector
6717 Emit extra code to check for buffer overflows, such as stack smashing
6718 attacks. This is done by adding a guard variable to functions with
6719 vulnerable objects. This includes functions that call alloca, and
6720 functions with buffers larger than 8 bytes. The guards are initialized
6721 when a function is entered and then checked when the function exits.
6722 If a guard check fails, an error message is printed and the program exits.
6724 @item -fstack-protector-all
6725 @opindex fstack-protector-all
6726 Like @option{-fstack-protector} except that all functions are protected.
6728 @item -fsection-anchors
6729 @opindex fsection-anchors
6730 Try to reduce the number of symbolic address calculations by using
6731 shared ``anchor'' symbols to address nearby objects. This transformation
6732 can help to reduce the number of GOT entries and GOT accesses on some
6735 For example, the implementation of the following function @code{foo}:
6739 int foo (void) @{ return a + b + c; @}
6742 would usually calculate the addresses of all three variables, but if you
6743 compile it with @option{-fsection-anchors}, it will access the variables
6744 from a common anchor point instead. The effect is similar to the
6745 following pseudocode (which isn't valid C):
6750 register int *xr = &x;
6751 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6755 Not all targets support this option.
6757 @item --param @var{name}=@var{value}
6759 In some places, GCC uses various constants to control the amount of
6760 optimization that is done. For example, GCC will not inline functions
6761 that contain more that a certain number of instructions. You can
6762 control some of these constants on the command-line using the
6763 @option{--param} option.
6765 The names of specific parameters, and the meaning of the values, are
6766 tied to the internals of the compiler, and are subject to change
6767 without notice in future releases.
6769 In each case, the @var{value} is an integer. The allowable choices for
6770 @var{name} are given in the following table:
6773 @item salias-max-implicit-fields
6774 The maximum number of fields in a variable without direct
6775 structure accesses for which structure aliasing will consider trying
6776 to track each field. The default is 5
6778 @item salias-max-array-elements
6779 The maximum number of elements an array can have and its elements
6780 still be tracked individually by structure aliasing. The default is 4
6782 @item sra-max-structure-size
6783 The maximum structure size, in bytes, at which the scalar replacement
6784 of aggregates (SRA) optimization will perform block copies. The
6785 default value, 0, implies that GCC will select the most appropriate
6788 @item sra-field-structure-ratio
6789 The threshold ratio (as a percentage) between instantiated fields and
6790 the complete structure size. We say that if the ratio of the number
6791 of bytes in instantiated fields to the number of bytes in the complete
6792 structure exceeds this parameter, then block copies are not used. The
6795 @item struct-reorg-cold-struct-ratio
6796 The threshold ratio (as a percentage) between a structure frequency
6797 and the frequency of the hottest structure in the program. This parameter
6798 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
6799 We say that if the ratio of a structure frequency, calculated by profiling,
6800 to the hottest structure frequency in the program is less than this
6801 parameter, then structure reorganization is not applied to this structure.
6804 @item max-crossjump-edges
6805 The maximum number of incoming edges to consider for crossjumping.
6806 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6807 the number of edges incoming to each block. Increasing values mean
6808 more aggressive optimization, making the compile time increase with
6809 probably small improvement in executable size.
6811 @item min-crossjump-insns
6812 The minimum number of instructions which must be matched at the end
6813 of two blocks before crossjumping will be performed on them. This
6814 value is ignored in the case where all instructions in the block being
6815 crossjumped from are matched. The default value is 5.
6817 @item max-grow-copy-bb-insns
6818 The maximum code size expansion factor when copying basic blocks
6819 instead of jumping. The expansion is relative to a jump instruction.
6820 The default value is 8.
6822 @item max-goto-duplication-insns
6823 The maximum number of instructions to duplicate to a block that jumps
6824 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6825 passes, GCC factors computed gotos early in the compilation process,
6826 and unfactors them as late as possible. Only computed jumps at the
6827 end of a basic blocks with no more than max-goto-duplication-insns are
6828 unfactored. The default value is 8.
6830 @item max-delay-slot-insn-search
6831 The maximum number of instructions to consider when looking for an
6832 instruction to fill a delay slot. If more than this arbitrary number of
6833 instructions is searched, the time savings from filling the delay slot
6834 will be minimal so stop searching. Increasing values mean more
6835 aggressive optimization, making the compile time increase with probably
6836 small improvement in executable run time.
6838 @item max-delay-slot-live-search
6839 When trying to fill delay slots, the maximum number of instructions to
6840 consider when searching for a block with valid live register
6841 information. Increasing this arbitrarily chosen value means more
6842 aggressive optimization, increasing the compile time. This parameter
6843 should be removed when the delay slot code is rewritten to maintain the
6846 @item max-gcse-memory
6847 The approximate maximum amount of memory that will be allocated in
6848 order to perform the global common subexpression elimination
6849 optimization. If more memory than specified is required, the
6850 optimization will not be done.
6852 @item max-gcse-passes
6853 The maximum number of passes of GCSE to run. The default is 1.
6855 @item max-pending-list-length
6856 The maximum number of pending dependencies scheduling will allow
6857 before flushing the current state and starting over. Large functions
6858 with few branches or calls can create excessively large lists which
6859 needlessly consume memory and resources.
6861 @item max-inline-insns-single
6862 Several parameters control the tree inliner used in gcc.
6863 This number sets the maximum number of instructions (counted in GCC's
6864 internal representation) in a single function that the tree inliner
6865 will consider for inlining. This only affects functions declared
6866 inline and methods implemented in a class declaration (C++).
6867 The default value is 450.
6869 @item max-inline-insns-auto
6870 When you use @option{-finline-functions} (included in @option{-O3}),
6871 a lot of functions that would otherwise not be considered for inlining
6872 by the compiler will be investigated. To those functions, a different
6873 (more restrictive) limit compared to functions declared inline can
6875 The default value is 90.
6877 @item large-function-insns
6878 The limit specifying really large functions. For functions larger than this
6879 limit after inlining inlining is constrained by
6880 @option{--param large-function-growth}. This parameter is useful primarily
6881 to avoid extreme compilation time caused by non-linear algorithms used by the
6883 This parameter is ignored when @option{-funit-at-a-time} is not used.
6884 The default value is 2700.
6886 @item large-function-growth
6887 Specifies maximal growth of large function caused by inlining in percents.
6888 This parameter is ignored when @option{-funit-at-a-time} is not used.
6889 The default value is 100 which limits large function growth to 2.0 times
6892 @item large-unit-insns
6893 The limit specifying large translation unit. Growth caused by inlining of
6894 units larger than this limit is limited by @option{--param inline-unit-growth}.
6895 For small units this might be too tight (consider unit consisting of function A
6896 that is inline and B that just calls A three time. If B is small relative to
6897 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6898 large units consisting of small inlineable functions however the overall unit
6899 growth limit is needed to avoid exponential explosion of code size. Thus for
6900 smaller units, the size is increased to @option{--param large-unit-insns}
6901 before applying @option{--param inline-unit-growth}. The default is 10000
6903 @item inline-unit-growth
6904 Specifies maximal overall growth of the compilation unit caused by inlining.
6905 This parameter is ignored when @option{-funit-at-a-time} is not used.
6906 The default value is 30 which limits unit growth to 1.3 times the original
6909 @item large-stack-frame
6910 The limit specifying large stack frames. While inlining the algorithm is trying
6911 to not grow past this limit too much. Default value is 256 bytes.
6913 @item large-stack-frame-growth
6914 Specifies maximal growth of large stack frames caused by inlining in percents.
6915 The default value is 1000 which limits large stack frame growth to 11 times
6918 @item max-inline-insns-recursive
6919 @itemx max-inline-insns-recursive-auto
6920 Specifies maximum number of instructions out-of-line copy of self recursive inline
6921 function can grow into by performing recursive inlining.
6923 For functions declared inline @option{--param max-inline-insns-recursive} is
6924 taken into account. For function not declared inline, recursive inlining
6925 happens only when @option{-finline-functions} (included in @option{-O3}) is
6926 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6927 default value is 450.
6929 @item max-inline-recursive-depth
6930 @itemx max-inline-recursive-depth-auto
6931 Specifies maximum recursion depth used by the recursive inlining.
6933 For functions declared inline @option{--param max-inline-recursive-depth} is
6934 taken into account. For function not declared inline, recursive inlining
6935 happens only when @option{-finline-functions} (included in @option{-O3}) is
6936 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6939 @item min-inline-recursive-probability
6940 Recursive inlining is profitable only for function having deep recursion
6941 in average and can hurt for function having little recursion depth by
6942 increasing the prologue size or complexity of function body to other
6945 When profile feedback is available (see @option{-fprofile-generate}) the actual
6946 recursion depth can be guessed from probability that function will recurse via
6947 given call expression. This parameter limits inlining only to call expression
6948 whose probability exceeds given threshold (in percents). The default value is
6951 @item inline-call-cost
6952 Specify cost of call instruction relative to simple arithmetics operations
6953 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6954 functions and at the same time increases size of leaf function that is believed to
6955 reduce function size by being inlined. In effect it increases amount of
6956 inlining for code having large abstraction penalty (many functions that just
6957 pass the arguments to other functions) and decrease inlining for code with low
6958 abstraction penalty. The default value is 12.
6960 @item min-vect-loop-bound
6961 The minimum number of iterations under which a loop will not get vectorized
6962 when @option{-ftree-vectorize} is used. The number of iterations after
6963 vectorization needs to be greater than the value specified by this option
6964 to allow vectorization. The default value is 0.
6966 @item max-unrolled-insns
6967 The maximum number of instructions that a loop should have if that loop
6968 is unrolled, and if the loop is unrolled, it determines how many times
6969 the loop code is unrolled.
6971 @item max-average-unrolled-insns
6972 The maximum number of instructions biased by probabilities of their execution
6973 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6974 it determines how many times the loop code is unrolled.
6976 @item max-unroll-times
6977 The maximum number of unrollings of a single loop.
6979 @item max-peeled-insns
6980 The maximum number of instructions that a loop should have if that loop
6981 is peeled, and if the loop is peeled, it determines how many times
6982 the loop code is peeled.
6984 @item max-peel-times
6985 The maximum number of peelings of a single loop.
6987 @item max-completely-peeled-insns
6988 The maximum number of insns of a completely peeled loop.
6990 @item max-completely-peel-times
6991 The maximum number of iterations of a loop to be suitable for complete peeling.
6993 @item max-unswitch-insns
6994 The maximum number of insns of an unswitched loop.
6996 @item max-unswitch-level
6997 The maximum number of branches unswitched in a single loop.
7000 The minimum cost of an expensive expression in the loop invariant motion.
7002 @item iv-consider-all-candidates-bound
7003 Bound on number of candidates for induction variables below that
7004 all candidates are considered for each use in induction variable
7005 optimizations. Only the most relevant candidates are considered
7006 if there are more candidates, to avoid quadratic time complexity.
7008 @item iv-max-considered-uses
7009 The induction variable optimizations give up on loops that contain more
7010 induction variable uses.
7012 @item iv-always-prune-cand-set-bound
7013 If number of candidates in the set is smaller than this value,
7014 we always try to remove unnecessary ivs from the set during its
7015 optimization when a new iv is added to the set.
7017 @item scev-max-expr-size
7018 Bound on size of expressions used in the scalar evolutions analyzer.
7019 Large expressions slow the analyzer.
7021 @item omega-max-vars
7022 The maximum number of variables in an Omega constraint system.
7023 The default value is 128.
7025 @item omega-max-geqs
7026 The maximum number of inequalities in an Omega constraint system.
7027 The default value is 256.
7030 The maximum number of equalities in an Omega constraint system.
7031 The default value is 128.
7033 @item omega-max-wild-cards
7034 The maximum number of wildcard variables that the Omega solver will
7035 be able to insert. The default value is 18.
7037 @item omega-hash-table-size
7038 The size of the hash table in the Omega solver. The default value is
7041 @item omega-max-keys
7042 The maximal number of keys used by the Omega solver. The default
7045 @item omega-eliminate-redundant-constraints
7046 When set to 1, use expensive methods to eliminate all redundant
7047 constraints. The default value is 0.
7049 @item vect-max-version-for-alignment-checks
7050 The maximum number of runtime checks that can be performed when
7051 doing loop versioning for alignment in the vectorizer. See option
7052 ftree-vect-loop-version for more information.
7054 @item vect-max-version-for-alias-checks
7055 The maximum number of runtime checks that can be performed when
7056 doing loop versioning for alias in the vectorizer. See option
7057 ftree-vect-loop-version for more information.
7059 @item max-iterations-to-track
7061 The maximum number of iterations of a loop the brute force algorithm
7062 for analysis of # of iterations of the loop tries to evaluate.
7064 @item hot-bb-count-fraction
7065 Select fraction of the maximal count of repetitions of basic block in program
7066 given basic block needs to have to be considered hot.
7068 @item hot-bb-frequency-fraction
7069 Select fraction of the maximal frequency of executions of basic block in
7070 function given basic block needs to have to be considered hot
7072 @item max-predicted-iterations
7073 The maximum number of loop iterations we predict statically. This is useful
7074 in cases where function contain single loop with known bound and other loop
7075 with unknown. We predict the known number of iterations correctly, while
7076 the unknown number of iterations average to roughly 10. This means that the
7077 loop without bounds would appear artificially cold relative to the other one.
7079 @item align-threshold
7081 Select fraction of the maximal frequency of executions of basic block in
7082 function given basic block will get aligned.
7084 @item align-loop-iterations
7086 A loop expected to iterate at lest the selected number of iterations will get
7089 @item tracer-dynamic-coverage
7090 @itemx tracer-dynamic-coverage-feedback
7092 This value is used to limit superblock formation once the given percentage of
7093 executed instructions is covered. This limits unnecessary code size
7096 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7097 feedback is available. The real profiles (as opposed to statically estimated
7098 ones) are much less balanced allowing the threshold to be larger value.
7100 @item tracer-max-code-growth
7101 Stop tail duplication once code growth has reached given percentage. This is
7102 rather hokey argument, as most of the duplicates will be eliminated later in
7103 cross jumping, so it may be set to much higher values than is the desired code
7106 @item tracer-min-branch-ratio
7108 Stop reverse growth when the reverse probability of best edge is less than this
7109 threshold (in percent).
7111 @item tracer-min-branch-ratio
7112 @itemx tracer-min-branch-ratio-feedback
7114 Stop forward growth if the best edge do have probability lower than this
7117 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7118 compilation for profile feedback and one for compilation without. The value
7119 for compilation with profile feedback needs to be more conservative (higher) in
7120 order to make tracer effective.
7122 @item max-cse-path-length
7124 Maximum number of basic blocks on path that cse considers. The default is 10.
7127 The maximum instructions CSE process before flushing. The default is 1000.
7129 @item max-aliased-vops
7131 Maximum number of virtual operands per function allowed to represent
7132 aliases before triggering the alias partitioning heuristic. Alias
7133 partitioning reduces compile times and memory consumption needed for
7134 aliasing at the expense of precision loss in alias information. The
7135 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7138 Notice that if a function contains more memory statements than the
7139 value of this parameter, it is not really possible to achieve this
7140 reduction. In this case, the compiler will use the number of memory
7141 statements as the value for @option{max-aliased-vops}.
7143 @item avg-aliased-vops
7145 Average number of virtual operands per statement allowed to represent
7146 aliases before triggering the alias partitioning heuristic. This
7147 works in conjunction with @option{max-aliased-vops}. If a function
7148 contains more than @option{max-aliased-vops} virtual operators, then
7149 memory symbols will be grouped into memory partitions until either the
7150 total number of virtual operators is below @option{max-aliased-vops}
7151 or the average number of virtual operators per memory statement is
7152 below @option{avg-aliased-vops}. The default value for this parameter
7153 is 1 for -O1 and -O2, and 3 for -O3.
7155 @item ggc-min-expand
7157 GCC uses a garbage collector to manage its own memory allocation. This
7158 parameter specifies the minimum percentage by which the garbage
7159 collector's heap should be allowed to expand between collections.
7160 Tuning this may improve compilation speed; it has no effect on code
7163 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7164 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7165 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7166 GCC is not able to calculate RAM on a particular platform, the lower
7167 bound of 30% is used. Setting this parameter and
7168 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7169 every opportunity. This is extremely slow, but can be useful for
7172 @item ggc-min-heapsize
7174 Minimum size of the garbage collector's heap before it begins bothering
7175 to collect garbage. The first collection occurs after the heap expands
7176 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7177 tuning this may improve compilation speed, and has no effect on code
7180 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7181 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7182 with a lower bound of 4096 (four megabytes) and an upper bound of
7183 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7184 particular platform, the lower bound is used. Setting this parameter
7185 very large effectively disables garbage collection. Setting this
7186 parameter and @option{ggc-min-expand} to zero causes a full collection
7187 to occur at every opportunity.
7189 @item max-reload-search-insns
7190 The maximum number of instruction reload should look backward for equivalent
7191 register. Increasing values mean more aggressive optimization, making the
7192 compile time increase with probably slightly better performance. The default
7195 @item max-cselib-memory-locations
7196 The maximum number of memory locations cselib should take into account.
7197 Increasing values mean more aggressive optimization, making the compile time
7198 increase with probably slightly better performance. The default value is 500.
7200 @item max-flow-memory-locations
7201 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
7202 The default value is 100.
7204 @item reorder-blocks-duplicate
7205 @itemx reorder-blocks-duplicate-feedback
7207 Used by basic block reordering pass to decide whether to use unconditional
7208 branch or duplicate the code on its destination. Code is duplicated when its
7209 estimated size is smaller than this value multiplied by the estimated size of
7210 unconditional jump in the hot spots of the program.
7212 The @option{reorder-block-duplicate-feedback} is used only when profile
7213 feedback is available and may be set to higher values than
7214 @option{reorder-block-duplicate} since information about the hot spots is more
7217 @item max-sched-ready-insns
7218 The maximum number of instructions ready to be issued the scheduler should
7219 consider at any given time during the first scheduling pass. Increasing
7220 values mean more thorough searches, making the compilation time increase
7221 with probably little benefit. The default value is 100.
7223 @item max-sched-region-blocks
7224 The maximum number of blocks in a region to be considered for
7225 interblock scheduling. The default value is 10.
7227 @item max-sched-region-insns
7228 The maximum number of insns in a region to be considered for
7229 interblock scheduling. The default value is 100.
7232 The minimum probability (in percents) of reaching a source block
7233 for interblock speculative scheduling. The default value is 40.
7235 @item max-sched-extend-regions-iters
7236 The maximum number of iterations through CFG to extend regions.
7237 0 - disable region extension,
7238 N - do at most N iterations.
7239 The default value is 0.
7241 @item max-sched-insn-conflict-delay
7242 The maximum conflict delay for an insn to be considered for speculative motion.
7243 The default value is 3.
7245 @item sched-spec-prob-cutoff
7246 The minimal probability of speculation success (in percents), so that
7247 speculative insn will be scheduled.
7248 The default value is 40.
7250 @item max-last-value-rtl
7252 The maximum size measured as number of RTLs that can be recorded in an expression
7253 in combiner for a pseudo register as last known value of that register. The default
7256 @item integer-share-limit
7257 Small integer constants can use a shared data structure, reducing the
7258 compiler's memory usage and increasing its speed. This sets the maximum
7259 value of a shared integer constant. The default value is 256.
7261 @item min-virtual-mappings
7262 Specifies the minimum number of virtual mappings in the incremental
7263 SSA updater that should be registered to trigger the virtual mappings
7264 heuristic defined by virtual-mappings-ratio. The default value is
7267 @item virtual-mappings-ratio
7268 If the number of virtual mappings is virtual-mappings-ratio bigger
7269 than the number of virtual symbols to be updated, then the incremental
7270 SSA updater switches to a full update for those symbols. The default
7273 @item ssp-buffer-size
7274 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7275 protection when @option{-fstack-protection} is used.
7277 @item max-jump-thread-duplication-stmts
7278 Maximum number of statements allowed in a block that needs to be
7279 duplicated when threading jumps.
7281 @item max-fields-for-field-sensitive
7282 Maximum number of fields in a structure we will treat in
7283 a field sensitive manner during pointer analysis.
7285 @item prefetch-latency
7286 Estimate on average number of instructions that are executed before
7287 prefetch finishes. The distance we prefetch ahead is proportional
7288 to this constant. Increasing this number may also lead to less
7289 streams being prefetched (see @option{simultaneous-prefetches}).
7291 @item simultaneous-prefetches
7292 Maximum number of prefetches that can run at the same time.
7294 @item l1-cache-line-size
7295 The size of cache line in L1 cache, in bytes.
7298 The size of L1 cache, in kilobytes.
7301 The size of L2 cache, in kilobytes.
7303 @item use-canonical-types
7304 Whether the compiler should use the ``canonical'' type system. By
7305 default, this should always be 1, which uses a more efficient internal
7306 mechanism for comparing types in C++ and Objective-C++. However, if
7307 bugs in the canonical type system are causing compilation failures,
7308 set this value to 0 to disable canonical types.
7310 @item max-partial-antic-length
7311 Maximum length of the partial antic set computed during the tree
7312 partial redundancy elimination optimization (@option{-ftree-pre}) when
7313 optimizing at @option{-O3} and above. For some sorts of source code
7314 the enhanced partial redundancy elimination optimization can run away,
7315 consuming all of the memory available on the host machine. This
7316 parameter sets a limit on the length of the sets that are computed,
7317 which prevents the runaway behaviour. Setting a value of 0 for
7318 this paramter will allow an unlimited set length.
7320 @item sccvn-max-scc-size
7321 Maximum size of a strongly connected component (SCC) during SCCVN
7322 processing. If this limit is hit, SCCVN processing for the whole
7323 function will not be done and optimizations depending on it will
7324 be disabled. The default maximum SCC size is 10000.
7329 @node Preprocessor Options
7330 @section Options Controlling the Preprocessor
7331 @cindex preprocessor options
7332 @cindex options, preprocessor
7334 These options control the C preprocessor, which is run on each C source
7335 file before actual compilation.
7337 If you use the @option{-E} option, nothing is done except preprocessing.
7338 Some of these options make sense only together with @option{-E} because
7339 they cause the preprocessor output to be unsuitable for actual
7344 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7345 and pass @var{option} directly through to the preprocessor. If
7346 @var{option} contains commas, it is split into multiple options at the
7347 commas. However, many options are modified, translated or interpreted
7348 by the compiler driver before being passed to the preprocessor, and
7349 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7350 interface is undocumented and subject to change, so whenever possible
7351 you should avoid using @option{-Wp} and let the driver handle the
7354 @item -Xpreprocessor @var{option}
7355 @opindex preprocessor
7356 Pass @var{option} as an option to the preprocessor. You can use this to
7357 supply system-specific preprocessor options which GCC does not know how to
7360 If you want to pass an option that takes an argument, you must use
7361 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7364 @include cppopts.texi
7366 @node Assembler Options
7367 @section Passing Options to the Assembler
7369 @c prevent bad page break with this line
7370 You can pass options to the assembler.
7373 @item -Wa,@var{option}
7375 Pass @var{option} as an option to the assembler. If @var{option}
7376 contains commas, it is split into multiple options at the commas.
7378 @item -Xassembler @var{option}
7380 Pass @var{option} as an option to the assembler. You can use this to
7381 supply system-specific assembler options which GCC does not know how to
7384 If you want to pass an option that takes an argument, you must use
7385 @option{-Xassembler} twice, once for the option and once for the argument.
7390 @section Options for Linking
7391 @cindex link options
7392 @cindex options, linking
7394 These options come into play when the compiler links object files into
7395 an executable output file. They are meaningless if the compiler is
7396 not doing a link step.
7400 @item @var{object-file-name}
7401 A file name that does not end in a special recognized suffix is
7402 considered to name an object file or library. (Object files are
7403 distinguished from libraries by the linker according to the file
7404 contents.) If linking is done, these object files are used as input
7413 If any of these options is used, then the linker is not run, and
7414 object file names should not be used as arguments. @xref{Overall
7418 @item -l@var{library}
7419 @itemx -l @var{library}
7421 Search the library named @var{library} when linking. (The second
7422 alternative with the library as a separate argument is only for
7423 POSIX compliance and is not recommended.)
7425 It makes a difference where in the command you write this option; the
7426 linker searches and processes libraries and object files in the order they
7427 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7428 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7429 to functions in @samp{z}, those functions may not be loaded.
7431 The linker searches a standard list of directories for the library,
7432 which is actually a file named @file{lib@var{library}.a}. The linker
7433 then uses this file as if it had been specified precisely by name.
7435 The directories searched include several standard system directories
7436 plus any that you specify with @option{-L}.
7438 Normally the files found this way are library files---archive files
7439 whose members are object files. The linker handles an archive file by
7440 scanning through it for members which define symbols that have so far
7441 been referenced but not defined. But if the file that is found is an
7442 ordinary object file, it is linked in the usual fashion. The only
7443 difference between using an @option{-l} option and specifying a file name
7444 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7445 and searches several directories.
7449 You need this special case of the @option{-l} option in order to
7450 link an Objective-C or Objective-C++ program.
7453 @opindex nostartfiles
7454 Do not use the standard system startup files when linking.
7455 The standard system libraries are used normally, unless @option{-nostdlib}
7456 or @option{-nodefaultlibs} is used.
7458 @item -nodefaultlibs
7459 @opindex nodefaultlibs
7460 Do not use the standard system libraries when linking.
7461 Only the libraries you specify will be passed to the linker.
7462 The standard startup files are used normally, unless @option{-nostartfiles}
7463 is used. The compiler may generate calls to @code{memcmp},
7464 @code{memset}, @code{memcpy} and @code{memmove}.
7465 These entries are usually resolved by entries in
7466 libc. These entry points should be supplied through some other
7467 mechanism when this option is specified.
7471 Do not use the standard system startup files or libraries when linking.
7472 No startup files and only the libraries you specify will be passed to
7473 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7474 @code{memcpy} and @code{memmove}.
7475 These entries are usually resolved by entries in
7476 libc. These entry points should be supplied through some other
7477 mechanism when this option is specified.
7479 @cindex @option{-lgcc}, use with @option{-nostdlib}
7480 @cindex @option{-nostdlib} and unresolved references
7481 @cindex unresolved references and @option{-nostdlib}
7482 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7483 @cindex @option{-nodefaultlibs} and unresolved references
7484 @cindex unresolved references and @option{-nodefaultlibs}
7485 One of the standard libraries bypassed by @option{-nostdlib} and
7486 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7487 that GCC uses to overcome shortcomings of particular machines, or special
7488 needs for some languages.
7489 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7490 Collection (GCC) Internals},
7491 for more discussion of @file{libgcc.a}.)
7492 In most cases, you need @file{libgcc.a} even when you want to avoid
7493 other standard libraries. In other words, when you specify @option{-nostdlib}
7494 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7495 This ensures that you have no unresolved references to internal GCC
7496 library subroutines. (For example, @samp{__main}, used to ensure C++
7497 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7498 GNU Compiler Collection (GCC) Internals}.)
7502 Produce a position independent executable on targets which support it.
7503 For predictable results, you must also specify the same set of options
7504 that were used to generate code (@option{-fpie}, @option{-fPIE},
7505 or model suboptions) when you specify this option.
7509 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7510 that support it. This instructs the linker to add all symbols, not
7511 only used ones, to the dynamic symbol table. This option is needed
7512 for some uses of @code{dlopen} or to allow obtaining backtraces
7513 from within a program.
7517 Remove all symbol table and relocation information from the executable.
7521 On systems that support dynamic linking, this prevents linking with the shared
7522 libraries. On other systems, this option has no effect.
7526 Produce a shared object which can then be linked with other objects to
7527 form an executable. Not all systems support this option. For predictable
7528 results, you must also specify the same set of options that were used to
7529 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7530 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7531 needs to build supplementary stub code for constructors to work. On
7532 multi-libbed systems, @samp{gcc -shared} must select the correct support
7533 libraries to link against. Failing to supply the correct flags may lead
7534 to subtle defects. Supplying them in cases where they are not necessary
7537 @item -shared-libgcc
7538 @itemx -static-libgcc
7539 @opindex shared-libgcc
7540 @opindex static-libgcc
7541 On systems that provide @file{libgcc} as a shared library, these options
7542 force the use of either the shared or static version respectively.
7543 If no shared version of @file{libgcc} was built when the compiler was
7544 configured, these options have no effect.
7546 There are several situations in which an application should use the
7547 shared @file{libgcc} instead of the static version. The most common
7548 of these is when the application wishes to throw and catch exceptions
7549 across different shared libraries. In that case, each of the libraries
7550 as well as the application itself should use the shared @file{libgcc}.
7552 Therefore, the G++ and GCJ drivers automatically add
7553 @option{-shared-libgcc} whenever you build a shared library or a main
7554 executable, because C++ and Java programs typically use exceptions, so
7555 this is the right thing to do.
7557 If, instead, you use the GCC driver to create shared libraries, you may
7558 find that they will not always be linked with the shared @file{libgcc}.
7559 If GCC finds, at its configuration time, that you have a non-GNU linker
7560 or a GNU linker that does not support option @option{--eh-frame-hdr},
7561 it will link the shared version of @file{libgcc} into shared libraries
7562 by default. Otherwise, it will take advantage of the linker and optimize
7563 away the linking with the shared version of @file{libgcc}, linking with
7564 the static version of libgcc by default. This allows exceptions to
7565 propagate through such shared libraries, without incurring relocation
7566 costs at library load time.
7568 However, if a library or main executable is supposed to throw or catch
7569 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7570 for the languages used in the program, or using the option
7571 @option{-shared-libgcc}, such that it is linked with the shared
7576 Bind references to global symbols when building a shared object. Warn
7577 about any unresolved references (unless overridden by the link editor
7578 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7581 @item -Xlinker @var{option}
7583 Pass @var{option} as an option to the linker. You can use this to
7584 supply system-specific linker options which GCC does not know how to
7587 If you want to pass an option that takes an argument, you must use
7588 @option{-Xlinker} twice, once for the option and once for the argument.
7589 For example, to pass @option{-assert definitions}, you must write
7590 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7591 @option{-Xlinker "-assert definitions"}, because this passes the entire
7592 string as a single argument, which is not what the linker expects.
7594 @item -Wl,@var{option}
7596 Pass @var{option} as an option to the linker. If @var{option} contains
7597 commas, it is split into multiple options at the commas.
7599 @item -u @var{symbol}
7601 Pretend the symbol @var{symbol} is undefined, to force linking of
7602 library modules to define it. You can use @option{-u} multiple times with
7603 different symbols to force loading of additional library modules.
7606 @node Directory Options
7607 @section Options for Directory Search
7608 @cindex directory options
7609 @cindex options, directory search
7612 These options specify directories to search for header files, for
7613 libraries and for parts of the compiler:
7618 Add the directory @var{dir} to the head of the list of directories to be
7619 searched for header files. This can be used to override a system header
7620 file, substituting your own version, since these directories are
7621 searched before the system header file directories. However, you should
7622 not use this option to add directories that contain vendor-supplied
7623 system header files (use @option{-isystem} for that). If you use more than
7624 one @option{-I} option, the directories are scanned in left-to-right
7625 order; the standard system directories come after.
7627 If a standard system include directory, or a directory specified with
7628 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7629 option will be ignored. The directory will still be searched but as a
7630 system directory at its normal position in the system include chain.
7631 This is to ensure that GCC's procedure to fix buggy system headers and
7632 the ordering for the include_next directive are not inadvertently changed.
7633 If you really need to change the search order for system directories,
7634 use the @option{-nostdinc} and/or @option{-isystem} options.
7636 @item -iquote@var{dir}
7638 Add the directory @var{dir} to the head of the list of directories to
7639 be searched for header files only for the case of @samp{#include
7640 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7641 otherwise just like @option{-I}.
7645 Add directory @var{dir} to the list of directories to be searched
7648 @item -B@var{prefix}
7650 This option specifies where to find the executables, libraries,
7651 include files, and data files of the compiler itself.
7653 The compiler driver program runs one or more of the subprograms
7654 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7655 @var{prefix} as a prefix for each program it tries to run, both with and
7656 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7658 For each subprogram to be run, the compiler driver first tries the
7659 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7660 was not specified, the driver tries two standard prefixes, which are
7661 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7662 those results in a file name that is found, the unmodified program
7663 name is searched for using the directories specified in your
7664 @env{PATH} environment variable.
7666 The compiler will check to see if the path provided by the @option{-B}
7667 refers to a directory, and if necessary it will add a directory
7668 separator character at the end of the path.
7670 @option{-B} prefixes that effectively specify directory names also apply
7671 to libraries in the linker, because the compiler translates these
7672 options into @option{-L} options for the linker. They also apply to
7673 includes files in the preprocessor, because the compiler translates these
7674 options into @option{-isystem} options for the preprocessor. In this case,
7675 the compiler appends @samp{include} to the prefix.
7677 The run-time support file @file{libgcc.a} can also be searched for using
7678 the @option{-B} prefix, if needed. If it is not found there, the two
7679 standard prefixes above are tried, and that is all. The file is left
7680 out of the link if it is not found by those means.
7682 Another way to specify a prefix much like the @option{-B} prefix is to use
7683 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7686 As a special kludge, if the path provided by @option{-B} is
7687 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7688 9, then it will be replaced by @file{[dir/]include}. This is to help
7689 with boot-strapping the compiler.
7691 @item -specs=@var{file}
7693 Process @var{file} after the compiler reads in the standard @file{specs}
7694 file, in order to override the defaults that the @file{gcc} driver
7695 program uses when determining what switches to pass to @file{cc1},
7696 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7697 @option{-specs=@var{file}} can be specified on the command line, and they
7698 are processed in order, from left to right.
7700 @item --sysroot=@var{dir}
7702 Use @var{dir} as the logical root directory for headers and libraries.
7703 For example, if the compiler would normally search for headers in
7704 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7705 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7707 If you use both this option and the @option{-isysroot} option, then
7708 the @option{--sysroot} option will apply to libraries, but the
7709 @option{-isysroot} option will apply to header files.
7711 The GNU linker (beginning with version 2.16) has the necessary support
7712 for this option. If your linker does not support this option, the
7713 header file aspect of @option{--sysroot} will still work, but the
7714 library aspect will not.
7718 This option has been deprecated. Please use @option{-iquote} instead for
7719 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7720 Any directories you specify with @option{-I} options before the @option{-I-}
7721 option are searched only for the case of @samp{#include "@var{file}"};
7722 they are not searched for @samp{#include <@var{file}>}.
7724 If additional directories are specified with @option{-I} options after
7725 the @option{-I-}, these directories are searched for all @samp{#include}
7726 directives. (Ordinarily @emph{all} @option{-I} directories are used
7729 In addition, the @option{-I-} option inhibits the use of the current
7730 directory (where the current input file came from) as the first search
7731 directory for @samp{#include "@var{file}"}. There is no way to
7732 override this effect of @option{-I-}. With @option{-I.} you can specify
7733 searching the directory which was current when the compiler was
7734 invoked. That is not exactly the same as what the preprocessor does
7735 by default, but it is often satisfactory.
7737 @option{-I-} does not inhibit the use of the standard system directories
7738 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7745 @section Specifying subprocesses and the switches to pass to them
7748 @command{gcc} is a driver program. It performs its job by invoking a
7749 sequence of other programs to do the work of compiling, assembling and
7750 linking. GCC interprets its command-line parameters and uses these to
7751 deduce which programs it should invoke, and which command-line options
7752 it ought to place on their command lines. This behavior is controlled
7753 by @dfn{spec strings}. In most cases there is one spec string for each
7754 program that GCC can invoke, but a few programs have multiple spec
7755 strings to control their behavior. The spec strings built into GCC can
7756 be overridden by using the @option{-specs=} command-line switch to specify
7759 @dfn{Spec files} are plaintext files that are used to construct spec
7760 strings. They consist of a sequence of directives separated by blank
7761 lines. The type of directive is determined by the first non-whitespace
7762 character on the line and it can be one of the following:
7765 @item %@var{command}
7766 Issues a @var{command} to the spec file processor. The commands that can
7770 @item %include <@var{file}>
7772 Search for @var{file} and insert its text at the current point in the
7775 @item %include_noerr <@var{file}>
7776 @cindex %include_noerr
7777 Just like @samp{%include}, but do not generate an error message if the include
7778 file cannot be found.
7780 @item %rename @var{old_name} @var{new_name}
7782 Rename the spec string @var{old_name} to @var{new_name}.
7786 @item *[@var{spec_name}]:
7787 This tells the compiler to create, override or delete the named spec
7788 string. All lines after this directive up to the next directive or
7789 blank line are considered to be the text for the spec string. If this
7790 results in an empty string then the spec will be deleted. (Or, if the
7791 spec did not exist, then nothing will happened.) Otherwise, if the spec
7792 does not currently exist a new spec will be created. If the spec does
7793 exist then its contents will be overridden by the text of this
7794 directive, unless the first character of that text is the @samp{+}
7795 character, in which case the text will be appended to the spec.
7797 @item [@var{suffix}]:
7798 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7799 and up to the next directive or blank line are considered to make up the
7800 spec string for the indicated suffix. When the compiler encounters an
7801 input file with the named suffix, it will processes the spec string in
7802 order to work out how to compile that file. For example:
7809 This says that any input file whose name ends in @samp{.ZZ} should be
7810 passed to the program @samp{z-compile}, which should be invoked with the
7811 command-line switch @option{-input} and with the result of performing the
7812 @samp{%i} substitution. (See below.)
7814 As an alternative to providing a spec string, the text that follows a
7815 suffix directive can be one of the following:
7818 @item @@@var{language}
7819 This says that the suffix is an alias for a known @var{language}. This is
7820 similar to using the @option{-x} command-line switch to GCC to specify a
7821 language explicitly. For example:
7828 Says that .ZZ files are, in fact, C++ source files.
7831 This causes an error messages saying:
7834 @var{name} compiler not installed on this system.
7838 GCC already has an extensive list of suffixes built into it.
7839 This directive will add an entry to the end of the list of suffixes, but
7840 since the list is searched from the end backwards, it is effectively
7841 possible to override earlier entries using this technique.
7845 GCC has the following spec strings built into it. Spec files can
7846 override these strings or create their own. Note that individual
7847 targets can also add their own spec strings to this list.
7850 asm Options to pass to the assembler
7851 asm_final Options to pass to the assembler post-processor
7852 cpp Options to pass to the C preprocessor
7853 cc1 Options to pass to the C compiler
7854 cc1plus Options to pass to the C++ compiler
7855 endfile Object files to include at the end of the link
7856 link Options to pass to the linker
7857 lib Libraries to include on the command line to the linker
7858 libgcc Decides which GCC support library to pass to the linker
7859 linker Sets the name of the linker
7860 predefines Defines to be passed to the C preprocessor
7861 signed_char Defines to pass to CPP to say whether @code{char} is signed
7863 startfile Object files to include at the start of the link
7866 Here is a small example of a spec file:
7872 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7875 This example renames the spec called @samp{lib} to @samp{old_lib} and
7876 then overrides the previous definition of @samp{lib} with a new one.
7877 The new definition adds in some extra command-line options before
7878 including the text of the old definition.
7880 @dfn{Spec strings} are a list of command-line options to be passed to their
7881 corresponding program. In addition, the spec strings can contain
7882 @samp{%}-prefixed sequences to substitute variable text or to
7883 conditionally insert text into the command line. Using these constructs
7884 it is possible to generate quite complex command lines.
7886 Here is a table of all defined @samp{%}-sequences for spec
7887 strings. Note that spaces are not generated automatically around the
7888 results of expanding these sequences. Therefore you can concatenate them
7889 together or combine them with constant text in a single argument.
7893 Substitute one @samp{%} into the program name or argument.
7896 Substitute the name of the input file being processed.
7899 Substitute the basename of the input file being processed.
7900 This is the substring up to (and not including) the last period
7901 and not including the directory.
7904 This is the same as @samp{%b}, but include the file suffix (text after
7908 Marks the argument containing or following the @samp{%d} as a
7909 temporary file name, so that that file will be deleted if GCC exits
7910 successfully. Unlike @samp{%g}, this contributes no text to the
7913 @item %g@var{suffix}
7914 Substitute a file name that has suffix @var{suffix} and is chosen
7915 once per compilation, and mark the argument in the same way as
7916 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7917 name is now chosen in a way that is hard to predict even when previously
7918 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7919 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7920 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7921 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7922 was simply substituted with a file name chosen once per compilation,
7923 without regard to any appended suffix (which was therefore treated
7924 just like ordinary text), making such attacks more likely to succeed.
7926 @item %u@var{suffix}
7927 Like @samp{%g}, but generates a new temporary file name even if
7928 @samp{%u@var{suffix}} was already seen.
7930 @item %U@var{suffix}
7931 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7932 new one if there is no such last file name. In the absence of any
7933 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7934 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7935 would involve the generation of two distinct file names, one
7936 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7937 simply substituted with a file name chosen for the previous @samp{%u},
7938 without regard to any appended suffix.
7940 @item %j@var{suffix}
7941 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7942 writable, and if save-temps is off; otherwise, substitute the name
7943 of a temporary file, just like @samp{%u}. This temporary file is not
7944 meant for communication between processes, but rather as a junk
7947 @item %|@var{suffix}
7948 @itemx %m@var{suffix}
7949 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7950 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7951 all. These are the two most common ways to instruct a program that it
7952 should read from standard input or write to standard output. If you
7953 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7954 construct: see for example @file{f/lang-specs.h}.
7956 @item %.@var{SUFFIX}
7957 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7958 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7959 terminated by the next space or %.
7962 Marks the argument containing or following the @samp{%w} as the
7963 designated output file of this compilation. This puts the argument
7964 into the sequence of arguments that @samp{%o} will substitute later.
7967 Substitutes the names of all the output files, with spaces
7968 automatically placed around them. You should write spaces
7969 around the @samp{%o} as well or the results are undefined.
7970 @samp{%o} is for use in the specs for running the linker.
7971 Input files whose names have no recognized suffix are not compiled
7972 at all, but they are included among the output files, so they will
7976 Substitutes the suffix for object files. Note that this is
7977 handled specially when it immediately follows @samp{%g, %u, or %U},
7978 because of the need for those to form complete file names. The
7979 handling is such that @samp{%O} is treated exactly as if it had already
7980 been substituted, except that @samp{%g, %u, and %U} do not currently
7981 support additional @var{suffix} characters following @samp{%O} as they would
7982 following, for example, @samp{.o}.
7985 Substitutes the standard macro predefinitions for the
7986 current target machine. Use this when running @code{cpp}.
7989 Like @samp{%p}, but puts @samp{__} before and after the name of each
7990 predefined macro, except for macros that start with @samp{__} or with
7991 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7995 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7996 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7997 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7998 and @option{-imultilib} as necessary.
8001 Current argument is the name of a library or startup file of some sort.
8002 Search for that file in a standard list of directories and substitute
8003 the full name found.
8006 Print @var{str} as an error message. @var{str} is terminated by a newline.
8007 Use this when inconsistent options are detected.
8010 Substitute the contents of spec string @var{name} at this point.
8013 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8015 @item %x@{@var{option}@}
8016 Accumulate an option for @samp{%X}.
8019 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8023 Output the accumulated assembler options specified by @option{-Wa}.
8026 Output the accumulated preprocessor options specified by @option{-Wp}.
8029 Process the @code{asm} spec. This is used to compute the
8030 switches to be passed to the assembler.
8033 Process the @code{asm_final} spec. This is a spec string for
8034 passing switches to an assembler post-processor, if such a program is
8038 Process the @code{link} spec. This is the spec for computing the
8039 command line passed to the linker. Typically it will make use of the
8040 @samp{%L %G %S %D and %E} sequences.
8043 Dump out a @option{-L} option for each directory that GCC believes might
8044 contain startup files. If the target supports multilibs then the
8045 current multilib directory will be prepended to each of these paths.
8048 Process the @code{lib} spec. This is a spec string for deciding which
8049 libraries should be included on the command line to the linker.
8052 Process the @code{libgcc} spec. This is a spec string for deciding
8053 which GCC support library should be included on the command line to the linker.
8056 Process the @code{startfile} spec. This is a spec for deciding which
8057 object files should be the first ones passed to the linker. Typically
8058 this might be a file named @file{crt0.o}.
8061 Process the @code{endfile} spec. This is a spec string that specifies
8062 the last object files that will be passed to the linker.
8065 Process the @code{cpp} spec. This is used to construct the arguments
8066 to be passed to the C preprocessor.
8069 Process the @code{cc1} spec. This is used to construct the options to be
8070 passed to the actual C compiler (@samp{cc1}).
8073 Process the @code{cc1plus} spec. This is used to construct the options to be
8074 passed to the actual C++ compiler (@samp{cc1plus}).
8077 Substitute the variable part of a matched option. See below.
8078 Note that each comma in the substituted string is replaced by
8082 Remove all occurrences of @code{-S} from the command line. Note---this
8083 command is position dependent. @samp{%} commands in the spec string
8084 before this one will see @code{-S}, @samp{%} commands in the spec string
8085 after this one will not.
8087 @item %:@var{function}(@var{args})
8088 Call the named function @var{function}, passing it @var{args}.
8089 @var{args} is first processed as a nested spec string, then split
8090 into an argument vector in the usual fashion. The function returns
8091 a string which is processed as if it had appeared literally as part
8092 of the current spec.
8094 The following built-in spec functions are provided:
8098 The @code{getenv} spec function takes two arguments: an environment
8099 variable name and a string. If the environment variable is not
8100 defined, a fatal error is issued. Otherwise, the return value is the
8101 value of the environment variable concatenated with the string. For
8102 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8105 %:getenv(TOPDIR /include)
8108 expands to @file{/path/to/top/include}.
8110 @item @code{if-exists}
8111 The @code{if-exists} spec function takes one argument, an absolute
8112 pathname to a file. If the file exists, @code{if-exists} returns the
8113 pathname. Here is a small example of its usage:
8117 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8120 @item @code{if-exists-else}
8121 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8122 spec function, except that it takes two arguments. The first argument is
8123 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8124 returns the pathname. If it does not exist, it returns the second argument.
8125 This way, @code{if-exists-else} can be used to select one file or another,
8126 based on the existence of the first. Here is a small example of its usage:
8130 crt0%O%s %:if-exists(crti%O%s) \
8131 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8134 @item @code{replace-outfile}
8135 The @code{replace-outfile} spec function takes two arguments. It looks for the
8136 first argument in the outfiles array and replaces it with the second argument. Here
8137 is a small example of its usage:
8140 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8143 @item @code{print-asm-header}
8144 The @code{print-asm-header} function takes no arguments and simply
8145 prints a banner like:
8151 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8154 It is used to separate compiler options from assembler options
8155 in the @option{--target-help} output.
8159 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8160 If that switch was not specified, this substitutes nothing. Note that
8161 the leading dash is omitted when specifying this option, and it is
8162 automatically inserted if the substitution is performed. Thus the spec
8163 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8164 and would output the command line option @option{-foo}.
8166 @item %W@{@code{S}@}
8167 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8170 @item %@{@code{S}*@}
8171 Substitutes all the switches specified to GCC whose names start
8172 with @code{-S}, but which also take an argument. This is used for
8173 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8174 GCC considers @option{-o foo} as being
8175 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8176 text, including the space. Thus two arguments would be generated.
8178 @item %@{@code{S}*&@code{T}*@}
8179 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8180 (the order of @code{S} and @code{T} in the spec is not significant).
8181 There can be any number of ampersand-separated variables; for each the
8182 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8184 @item %@{@code{S}:@code{X}@}
8185 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8187 @item %@{!@code{S}:@code{X}@}
8188 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8190 @item %@{@code{S}*:@code{X}@}
8191 Substitutes @code{X} if one or more switches whose names start with
8192 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8193 once, no matter how many such switches appeared. However, if @code{%*}
8194 appears somewhere in @code{X}, then @code{X} will be substituted once
8195 for each matching switch, with the @code{%*} replaced by the part of
8196 that switch that matched the @code{*}.
8198 @item %@{.@code{S}:@code{X}@}
8199 Substitutes @code{X}, if processing a file with suffix @code{S}.
8201 @item %@{!.@code{S}:@code{X}@}
8202 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8204 @item %@{,@code{S}:@code{X}@}
8205 Substitutes @code{X}, if processing a file for language @code{S}.
8207 @item %@{!,@code{S}:@code{X}@}
8208 Substitutes @code{X}, if not processing a file for language @code{S}.
8210 @item %@{@code{S}|@code{P}:@code{X}@}
8211 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8212 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8213 @code{*} sequences as well, although they have a stronger binding than
8214 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8215 alternatives must be starred, and only the first matching alternative
8218 For example, a spec string like this:
8221 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8224 will output the following command-line options from the following input
8225 command-line options:
8230 -d fred.c -foo -baz -boggle
8231 -d jim.d -bar -baz -boggle
8234 @item %@{S:X; T:Y; :D@}
8236 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8237 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8238 be as many clauses as you need. This may be combined with @code{.},
8239 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8244 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8245 construct may contain other nested @samp{%} constructs or spaces, or
8246 even newlines. They are processed as usual, as described above.
8247 Trailing white space in @code{X} is ignored. White space may also
8248 appear anywhere on the left side of the colon in these constructs,
8249 except between @code{.} or @code{*} and the corresponding word.
8251 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8252 handled specifically in these constructs. If another value of
8253 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8254 @option{-W} switch is found later in the command line, the earlier
8255 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8256 just one letter, which passes all matching options.
8258 The character @samp{|} at the beginning of the predicate text is used to
8259 indicate that a command should be piped to the following command, but
8260 only if @option{-pipe} is specified.
8262 It is built into GCC which switches take arguments and which do not.
8263 (You might think it would be useful to generalize this to allow each
8264 compiler's spec to say which switches take arguments. But this cannot
8265 be done in a consistent fashion. GCC cannot even decide which input
8266 files have been specified without knowing which switches take arguments,
8267 and it must know which input files to compile in order to tell which
8270 GCC also knows implicitly that arguments starting in @option{-l} are to be
8271 treated as compiler output files, and passed to the linker in their
8272 proper position among the other output files.
8274 @c man begin OPTIONS
8276 @node Target Options
8277 @section Specifying Target Machine and Compiler Version
8278 @cindex target options
8279 @cindex cross compiling
8280 @cindex specifying machine version
8281 @cindex specifying compiler version and target machine
8282 @cindex compiler version, specifying
8283 @cindex target machine, specifying
8285 The usual way to run GCC is to run the executable called @file{gcc}, or
8286 @file{<machine>-gcc} when cross-compiling, or
8287 @file{<machine>-gcc-<version>} to run a version other than the one that
8288 was installed last. Sometimes this is inconvenient, so GCC provides
8289 options that will switch to another cross-compiler or version.
8292 @item -b @var{machine}
8294 The argument @var{machine} specifies the target machine for compilation.
8296 The value to use for @var{machine} is the same as was specified as the
8297 machine type when configuring GCC as a cross-compiler. For
8298 example, if a cross-compiler was configured with @samp{configure
8299 arm-elf}, meaning to compile for an arm processor with elf binaries,
8300 then you would specify @option{-b arm-elf} to run that cross compiler.
8301 Because there are other options beginning with @option{-b}, the
8302 configuration must contain a hyphen.
8304 @item -V @var{version}
8306 The argument @var{version} specifies which version of GCC to run.
8307 This is useful when multiple versions are installed. For example,
8308 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8311 The @option{-V} and @option{-b} options work by running the
8312 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8313 use them if you can just run that directly.
8315 @node Submodel Options
8316 @section Hardware Models and Configurations
8317 @cindex submodel options
8318 @cindex specifying hardware config
8319 @cindex hardware models and configurations, specifying
8320 @cindex machine dependent options
8322 Earlier we discussed the standard option @option{-b} which chooses among
8323 different installed compilers for completely different target
8324 machines, such as VAX vs.@: 68000 vs.@: 80386.
8326 In addition, each of these target machine types can have its own
8327 special options, starting with @samp{-m}, to choose among various
8328 hardware models or configurations---for example, 68010 vs 68020,
8329 floating coprocessor or none. A single installed version of the
8330 compiler can compile for any model or configuration, according to the
8333 Some configurations of the compiler also support additional special
8334 options, usually for compatibility with other compilers on the same
8337 @c This list is ordered alphanumerically by subsection name.
8338 @c It should be the same order and spelling as these options are listed
8339 @c in Machine Dependent Options
8345 * Blackfin Options::
8349 * DEC Alpha Options::
8350 * DEC Alpha/VMS Options::
8352 * GNU/Linux Options::
8355 * i386 and x86-64 Options::
8368 * RS/6000 and PowerPC Options::
8369 * S/390 and zSeries Options::
8374 * System V Options::
8379 * Xstormy16 Options::
8385 @subsection ARC Options
8388 These options are defined for ARC implementations:
8393 Compile code for little endian mode. This is the default.
8397 Compile code for big endian mode.
8400 @opindex mmangle-cpu
8401 Prepend the name of the cpu to all public symbol names.
8402 In multiple-processor systems, there are many ARC variants with different
8403 instruction and register set characteristics. This flag prevents code
8404 compiled for one cpu to be linked with code compiled for another.
8405 No facility exists for handling variants that are ``almost identical''.
8406 This is an all or nothing option.
8408 @item -mcpu=@var{cpu}
8410 Compile code for ARC variant @var{cpu}.
8411 Which variants are supported depend on the configuration.
8412 All variants support @option{-mcpu=base}, this is the default.
8414 @item -mtext=@var{text-section}
8415 @itemx -mdata=@var{data-section}
8416 @itemx -mrodata=@var{readonly-data-section}
8420 Put functions, data, and readonly data in @var{text-section},
8421 @var{data-section}, and @var{readonly-data-section} respectively
8422 by default. This can be overridden with the @code{section} attribute.
8423 @xref{Variable Attributes}.
8428 @subsection ARM Options
8431 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8435 @item -mabi=@var{name}
8437 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8438 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8441 @opindex mapcs-frame
8442 Generate a stack frame that is compliant with the ARM Procedure Call
8443 Standard for all functions, even if this is not strictly necessary for
8444 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8445 with this option will cause the stack frames not to be generated for
8446 leaf functions. The default is @option{-mno-apcs-frame}.
8450 This is a synonym for @option{-mapcs-frame}.
8453 @c not currently implemented
8454 @item -mapcs-stack-check
8455 @opindex mapcs-stack-check
8456 Generate code to check the amount of stack space available upon entry to
8457 every function (that actually uses some stack space). If there is
8458 insufficient space available then either the function
8459 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8460 called, depending upon the amount of stack space required. The run time
8461 system is required to provide these functions. The default is
8462 @option{-mno-apcs-stack-check}, since this produces smaller code.
8464 @c not currently implemented
8466 @opindex mapcs-float
8467 Pass floating point arguments using the float point registers. This is
8468 one of the variants of the APCS@. This option is recommended if the
8469 target hardware has a floating point unit or if a lot of floating point
8470 arithmetic is going to be performed by the code. The default is
8471 @option{-mno-apcs-float}, since integer only code is slightly increased in
8472 size if @option{-mapcs-float} is used.
8474 @c not currently implemented
8475 @item -mapcs-reentrant
8476 @opindex mapcs-reentrant
8477 Generate reentrant, position independent code. The default is
8478 @option{-mno-apcs-reentrant}.
8481 @item -mthumb-interwork
8482 @opindex mthumb-interwork
8483 Generate code which supports calling between the ARM and Thumb
8484 instruction sets. Without this option the two instruction sets cannot
8485 be reliably used inside one program. The default is
8486 @option{-mno-thumb-interwork}, since slightly larger code is generated
8487 when @option{-mthumb-interwork} is specified.
8489 @item -mno-sched-prolog
8490 @opindex mno-sched-prolog
8491 Prevent the reordering of instructions in the function prolog, or the
8492 merging of those instruction with the instructions in the function's
8493 body. This means that all functions will start with a recognizable set
8494 of instructions (or in fact one of a choice from a small set of
8495 different function prologues), and this information can be used to
8496 locate the start if functions inside an executable piece of code. The
8497 default is @option{-msched-prolog}.
8500 @opindex mhard-float
8501 Generate output containing floating point instructions. This is the
8505 @opindex msoft-float
8506 Generate output containing library calls for floating point.
8507 @strong{Warning:} the requisite libraries are not available for all ARM
8508 targets. Normally the facilities of the machine's usual C compiler are
8509 used, but this cannot be done directly in cross-compilation. You must make
8510 your own arrangements to provide suitable library functions for
8513 @option{-msoft-float} changes the calling convention in the output file;
8514 therefore, it is only useful if you compile @emph{all} of a program with
8515 this option. In particular, you need to compile @file{libgcc.a}, the
8516 library that comes with GCC, with @option{-msoft-float} in order for
8519 @item -mfloat-abi=@var{name}
8521 Specifies which ABI to use for floating point values. Permissible values
8522 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8524 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8525 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8526 of floating point instructions, but still uses the soft-float calling
8529 @item -mlittle-endian
8530 @opindex mlittle-endian
8531 Generate code for a processor running in little-endian mode. This is
8532 the default for all standard configurations.
8535 @opindex mbig-endian
8536 Generate code for a processor running in big-endian mode; the default is
8537 to compile code for a little-endian processor.
8539 @item -mwords-little-endian
8540 @opindex mwords-little-endian
8541 This option only applies when generating code for big-endian processors.
8542 Generate code for a little-endian word order but a big-endian byte
8543 order. That is, a byte order of the form @samp{32107654}. Note: this
8544 option should only be used if you require compatibility with code for
8545 big-endian ARM processors generated by versions of the compiler prior to
8548 @item -mcpu=@var{name}
8550 This specifies the name of the target ARM processor. GCC uses this name
8551 to determine what kind of instructions it can emit when generating
8552 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8553 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8554 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8555 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8556 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8557 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8558 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8559 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8560 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8561 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8562 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8563 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8564 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8565 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8566 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8567 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8569 @itemx -mtune=@var{name}
8571 This option is very similar to the @option{-mcpu=} option, except that
8572 instead of specifying the actual target processor type, and hence
8573 restricting which instructions can be used, it specifies that GCC should
8574 tune the performance of the code as if the target were of the type
8575 specified in this option, but still choosing the instructions that it
8576 will generate based on the cpu specified by a @option{-mcpu=} option.
8577 For some ARM implementations better performance can be obtained by using
8580 @item -march=@var{name}
8582 This specifies the name of the target ARM architecture. GCC uses this
8583 name to determine what kind of instructions it can emit when generating
8584 assembly code. This option can be used in conjunction with or instead
8585 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8586 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8587 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8588 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8589 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8591 @item -mfpu=@var{name}
8592 @itemx -mfpe=@var{number}
8593 @itemx -mfp=@var{number}
8597 This specifies what floating point hardware (or hardware emulation) is
8598 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8599 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8600 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8601 with older versions of GCC@.
8603 If @option{-msoft-float} is specified this specifies the format of
8604 floating point values.
8606 @item -mstructure-size-boundary=@var{n}
8607 @opindex mstructure-size-boundary
8608 The size of all structures and unions will be rounded up to a multiple
8609 of the number of bits set by this option. Permissible values are 8, 32
8610 and 64. The default value varies for different toolchains. For the COFF
8611 targeted toolchain the default value is 8. A value of 64 is only allowed
8612 if the underlying ABI supports it.
8614 Specifying the larger number can produce faster, more efficient code, but
8615 can also increase the size of the program. Different values are potentially
8616 incompatible. Code compiled with one value cannot necessarily expect to
8617 work with code or libraries compiled with another value, if they exchange
8618 information using structures or unions.
8620 @item -mabort-on-noreturn
8621 @opindex mabort-on-noreturn
8622 Generate a call to the function @code{abort} at the end of a
8623 @code{noreturn} function. It will be executed if the function tries to
8627 @itemx -mno-long-calls
8628 @opindex mlong-calls
8629 @opindex mno-long-calls
8630 Tells the compiler to perform function calls by first loading the
8631 address of the function into a register and then performing a subroutine
8632 call on this register. This switch is needed if the target function
8633 will lie outside of the 64 megabyte addressing range of the offset based
8634 version of subroutine call instruction.
8636 Even if this switch is enabled, not all function calls will be turned
8637 into long calls. The heuristic is that static functions, functions
8638 which have the @samp{short-call} attribute, functions that are inside
8639 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8640 definitions have already been compiled within the current compilation
8641 unit, will not be turned into long calls. The exception to this rule is
8642 that weak function definitions, functions with the @samp{long-call}
8643 attribute or the @samp{section} attribute, and functions that are within
8644 the scope of a @samp{#pragma long_calls} directive, will always be
8645 turned into long calls.
8647 This feature is not enabled by default. Specifying
8648 @option{-mno-long-calls} will restore the default behavior, as will
8649 placing the function calls within the scope of a @samp{#pragma
8650 long_calls_off} directive. Note these switches have no effect on how
8651 the compiler generates code to handle function calls via function
8654 @item -mnop-fun-dllimport
8655 @opindex mnop-fun-dllimport
8656 Disable support for the @code{dllimport} attribute.
8658 @item -msingle-pic-base
8659 @opindex msingle-pic-base
8660 Treat the register used for PIC addressing as read-only, rather than
8661 loading it in the prologue for each function. The run-time system is
8662 responsible for initializing this register with an appropriate value
8663 before execution begins.
8665 @item -mpic-register=@var{reg}
8666 @opindex mpic-register
8667 Specify the register to be used for PIC addressing. The default is R10
8668 unless stack-checking is enabled, when R9 is used.
8670 @item -mcirrus-fix-invalid-insns
8671 @opindex mcirrus-fix-invalid-insns
8672 @opindex mno-cirrus-fix-invalid-insns
8673 Insert NOPs into the instruction stream to in order to work around
8674 problems with invalid Maverick instruction combinations. This option
8675 is only valid if the @option{-mcpu=ep9312} option has been used to
8676 enable generation of instructions for the Cirrus Maverick floating
8677 point co-processor. This option is not enabled by default, since the
8678 problem is only present in older Maverick implementations. The default
8679 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8682 @item -mpoke-function-name
8683 @opindex mpoke-function-name
8684 Write the name of each function into the text section, directly
8685 preceding the function prologue. The generated code is similar to this:
8689 .ascii "arm_poke_function_name", 0
8692 .word 0xff000000 + (t1 - t0)
8693 arm_poke_function_name
8695 stmfd sp!, @{fp, ip, lr, pc@}
8699 When performing a stack backtrace, code can inspect the value of
8700 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8701 location @code{pc - 12} and the top 8 bits are set, then we know that
8702 there is a function name embedded immediately preceding this location
8703 and has length @code{((pc[-3]) & 0xff000000)}.
8707 Generate code for the Thumb instruction set. The default is to
8708 use the 32-bit ARM instruction set.
8709 This option automatically enables either 16-bit Thumb-1 or
8710 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8711 and @option{-march=@var{name}} options.
8714 @opindex mtpcs-frame
8715 Generate a stack frame that is compliant with the Thumb Procedure Call
8716 Standard for all non-leaf functions. (A leaf function is one that does
8717 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8719 @item -mtpcs-leaf-frame
8720 @opindex mtpcs-leaf-frame
8721 Generate a stack frame that is compliant with the Thumb Procedure Call
8722 Standard for all leaf functions. (A leaf function is one that does
8723 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8725 @item -mcallee-super-interworking
8726 @opindex mcallee-super-interworking
8727 Gives all externally visible functions in the file being compiled an ARM
8728 instruction set header which switches to Thumb mode before executing the
8729 rest of the function. This allows these functions to be called from
8730 non-interworking code.
8732 @item -mcaller-super-interworking
8733 @opindex mcaller-super-interworking
8734 Allows calls via function pointers (including virtual functions) to
8735 execute correctly regardless of whether the target code has been
8736 compiled for interworking or not. There is a small overhead in the cost
8737 of executing a function pointer if this option is enabled.
8739 @item -mtp=@var{name}
8741 Specify the access model for the thread local storage pointer. The valid
8742 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8743 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8744 (supported in the arm6k architecture), and @option{auto}, which uses the
8745 best available method for the selected processor. The default setting is
8751 @subsection AVR Options
8754 These options are defined for AVR implementations:
8757 @item -mmcu=@var{mcu}
8759 Specify ATMEL AVR instruction set or MCU type.
8761 Instruction set avr1 is for the minimal AVR core, not supported by the C
8762 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8763 attiny11, attiny12, attiny15, attiny28).
8765 Instruction set avr2 (default) is for the classic AVR core with up to
8766 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8767 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8768 at90c8534, at90s8535).
8770 Instruction set avr3 is for the classic AVR core with up to 128K program
8771 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8773 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8774 memory space (MCU types: atmega8, atmega83, atmega85).
8776 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8777 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8778 atmega64, atmega128, at43usb355, at94k).
8782 Output instruction sizes to the asm file.
8784 @item -minit-stack=@var{N}
8785 @opindex minit-stack
8786 Specify the initial stack address, which may be a symbol or numeric value,
8787 @samp{__stack} is the default.
8789 @item -mno-interrupts
8790 @opindex mno-interrupts
8791 Generated code is not compatible with hardware interrupts.
8792 Code size will be smaller.
8794 @item -mcall-prologues
8795 @opindex mcall-prologues
8796 Functions prologues/epilogues expanded as call to appropriate
8797 subroutines. Code size will be smaller.
8799 @item -mno-tablejump
8800 @opindex mno-tablejump
8801 Do not generate tablejump insns which sometimes increase code size.
8804 @opindex mtiny-stack
8805 Change only the low 8 bits of the stack pointer.
8809 Assume int to be 8 bit integer. This affects the sizes of all types: A
8810 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8811 and long long will be 4 bytes. Please note that this option does not
8812 comply to the C standards, but it will provide you with smaller code
8816 @node Blackfin Options
8817 @subsection Blackfin Options
8818 @cindex Blackfin Options
8821 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8823 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8824 can be one of @samp{bf522}, @samp{bf523}, @samp{bf524},
8825 @samp{bf525}, @samp{bf526}, @samp{bf527},
8826 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8827 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8828 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
8830 The optional @var{sirevision} specifies the silicon revision of the target
8831 Blackfin processor. Any workarounds available for the targeted silicon revision
8832 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8833 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8834 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8835 hexadecimal digits representing the major and minor numbers in the silicon
8836 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8837 is not defined. If @var{sirevision} is @samp{any}, the
8838 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8839 If this optional @var{sirevision} is not used, GCC assumes the latest known
8840 silicon revision of the targeted Blackfin processor.
8842 Support for @samp{bf561} is incomplete. For @samp{bf561},
8843 Only the processor macro is defined.
8844 Without this option, @samp{bf532} is used as the processor by default.
8845 The corresponding predefined processor macros for @var{cpu} is to
8846 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8847 provided by libgloss to be linked in if @option{-msim} is not given.
8851 Specifies that the program will be run on the simulator. This causes
8852 the simulator BSP provided by libgloss to be linked in. This option
8853 has effect only for @samp{bfin-elf} toolchain.
8854 Certain other options, such as @option{-mid-shared-library} and
8855 @option{-mfdpic}, imply @option{-msim}.
8857 @item -momit-leaf-frame-pointer
8858 @opindex momit-leaf-frame-pointer
8859 Don't keep the frame pointer in a register for leaf functions. This
8860 avoids the instructions to save, set up and restore frame pointers and
8861 makes an extra register available in leaf functions. The option
8862 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8863 which might make debugging harder.
8865 @item -mspecld-anomaly
8866 @opindex mspecld-anomaly
8867 When enabled, the compiler will ensure that the generated code does not
8868 contain speculative loads after jump instructions. If this option is used,
8869 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8871 @item -mno-specld-anomaly
8872 @opindex mno-specld-anomaly
8873 Don't generate extra code to prevent speculative loads from occurring.
8875 @item -mcsync-anomaly
8876 @opindex mcsync-anomaly
8877 When enabled, the compiler will ensure that the generated code does not
8878 contain CSYNC or SSYNC instructions too soon after conditional branches.
8879 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8881 @item -mno-csync-anomaly
8882 @opindex mno-csync-anomaly
8883 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8884 occurring too soon after a conditional branch.
8888 When enabled, the compiler is free to take advantage of the knowledge that
8889 the entire program fits into the low 64k of memory.
8892 @opindex mno-low-64k
8893 Assume that the program is arbitrarily large. This is the default.
8895 @item -mstack-check-l1
8896 @opindex mstack-check-l1
8897 Do stack checking using information placed into L1 scratchpad memory by the
8900 @item -mid-shared-library
8901 @opindex mid-shared-library
8902 Generate code that supports shared libraries via the library ID method.
8903 This allows for execute in place and shared libraries in an environment
8904 without virtual memory management. This option implies @option{-fPIC}.
8905 With a @samp{bfin-elf} target, this option implies @option{-msim}.
8907 @item -mno-id-shared-library
8908 @opindex mno-id-shared-library
8909 Generate code that doesn't assume ID based shared libraries are being used.
8910 This is the default.
8912 @item -mleaf-id-shared-library
8913 @opindex mleaf-id-shared-library
8914 Generate code that supports shared libraries via the library ID method,
8915 but assumes that this library or executable won't link against any other
8916 ID shared libraries. That allows the compiler to use faster code for jumps
8919 @item -mno-leaf-id-shared-library
8920 @opindex mno-leaf-id-shared-library
8921 Do not assume that the code being compiled won't link against any ID shared
8922 libraries. Slower code will be generated for jump and call insns.
8924 @item -mshared-library-id=n
8925 @opindex mshared-library-id
8926 Specified the identification number of the ID based shared library being
8927 compiled. Specifying a value of 0 will generate more compact code, specifying
8928 other values will force the allocation of that number to the current
8929 library but is no more space or time efficient than omitting this option.
8933 Generate code that allows the data segment to be located in a different
8934 area of memory from the text segment. This allows for execute in place in
8935 an environment without virtual memory management by eliminating relocations
8936 against the text section.
8939 @opindex mno-sep-data
8940 Generate code that assumes that the data segment follows the text segment.
8941 This is the default.
8944 @itemx -mno-long-calls
8945 @opindex mlong-calls
8946 @opindex mno-long-calls
8947 Tells the compiler to perform function calls by first loading the
8948 address of the function into a register and then performing a subroutine
8949 call on this register. This switch is needed if the target function
8950 will lie outside of the 24 bit addressing range of the offset based
8951 version of subroutine call instruction.
8953 This feature is not enabled by default. Specifying
8954 @option{-mno-long-calls} will restore the default behavior. Note these
8955 switches have no effect on how the compiler generates code to handle
8956 function calls via function pointers.
8960 Link with the fast floating-point library. This library relaxes some of
8961 the IEEE floating-point standard's rules for checking inputs against
8962 Not-a-Number (NAN), in the interest of performance.
8965 @opindex minline-plt
8966 Enable inlining of PLT entries in function calls to functions that are
8967 not known to bind locally. It has no effect without @option{-mfdpic}.
8971 @subsection CRIS Options
8972 @cindex CRIS Options
8974 These options are defined specifically for the CRIS ports.
8977 @item -march=@var{architecture-type}
8978 @itemx -mcpu=@var{architecture-type}
8981 Generate code for the specified architecture. The choices for
8982 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8983 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8984 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8987 @item -mtune=@var{architecture-type}
8989 Tune to @var{architecture-type} everything applicable about the generated
8990 code, except for the ABI and the set of available instructions. The
8991 choices for @var{architecture-type} are the same as for
8992 @option{-march=@var{architecture-type}}.
8994 @item -mmax-stack-frame=@var{n}
8995 @opindex mmax-stack-frame
8996 Warn when the stack frame of a function exceeds @var{n} bytes.
8998 @item -melinux-stacksize=@var{n}
8999 @opindex melinux-stacksize
9000 Only available with the @samp{cris-axis-aout} target. Arranges for
9001 indications in the program to the kernel loader that the stack of the
9002 program should be set to @var{n} bytes.
9008 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9009 @option{-march=v3} and @option{-march=v8} respectively.
9011 @item -mmul-bug-workaround
9012 @itemx -mno-mul-bug-workaround
9013 @opindex mmul-bug-workaround
9014 @opindex mno-mul-bug-workaround
9015 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9016 models where it applies. This option is active by default.
9020 Enable CRIS-specific verbose debug-related information in the assembly
9021 code. This option also has the effect to turn off the @samp{#NO_APP}
9022 formatted-code indicator to the assembler at the beginning of the
9027 Do not use condition-code results from previous instruction; always emit
9028 compare and test instructions before use of condition codes.
9030 @item -mno-side-effects
9031 @opindex mno-side-effects
9032 Do not emit instructions with side-effects in addressing modes other than
9036 @itemx -mno-stack-align
9038 @itemx -mno-data-align
9039 @itemx -mconst-align
9040 @itemx -mno-const-align
9041 @opindex mstack-align
9042 @opindex mno-stack-align
9043 @opindex mdata-align
9044 @opindex mno-data-align
9045 @opindex mconst-align
9046 @opindex mno-const-align
9047 These options (no-options) arranges (eliminate arrangements) for the
9048 stack-frame, individual data and constants to be aligned for the maximum
9049 single data access size for the chosen CPU model. The default is to
9050 arrange for 32-bit alignment. ABI details such as structure layout are
9051 not affected by these options.
9059 Similar to the stack- data- and const-align options above, these options
9060 arrange for stack-frame, writable data and constants to all be 32-bit,
9061 16-bit or 8-bit aligned. The default is 32-bit alignment.
9063 @item -mno-prologue-epilogue
9064 @itemx -mprologue-epilogue
9065 @opindex mno-prologue-epilogue
9066 @opindex mprologue-epilogue
9067 With @option{-mno-prologue-epilogue}, the normal function prologue and
9068 epilogue that sets up the stack-frame are omitted and no return
9069 instructions or return sequences are generated in the code. Use this
9070 option only together with visual inspection of the compiled code: no
9071 warnings or errors are generated when call-saved registers must be saved,
9072 or storage for local variable needs to be allocated.
9078 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9079 instruction sequences that load addresses for functions from the PLT part
9080 of the GOT rather than (traditional on other architectures) calls to the
9081 PLT@. The default is @option{-mgotplt}.
9085 Legacy no-op option only recognized with the cris-axis-aout target.
9089 Legacy no-op option only recognized with the cris-axis-elf and
9090 cris-axis-linux-gnu targets.
9094 Only recognized with the cris-axis-aout target, where it selects a
9095 GNU/linux-like multilib, include files and instruction set for
9100 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9104 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9105 to link with input-output functions from a simulator library. Code,
9106 initialized data and zero-initialized data are allocated consecutively.
9110 Like @option{-sim}, but pass linker options to locate initialized data at
9111 0x40000000 and zero-initialized data at 0x80000000.
9115 @subsection CRX Options
9118 These options are defined specifically for the CRX ports.
9124 Enable the use of multiply-accumulate instructions. Disabled by default.
9128 Push instructions will be used to pass outgoing arguments when functions
9129 are called. Enabled by default.
9132 @node Darwin Options
9133 @subsection Darwin Options
9134 @cindex Darwin options
9136 These options are defined for all architectures running the Darwin operating
9139 FSF GCC on Darwin does not create ``fat'' object files; it will create
9140 an object file for the single architecture that it was built to
9141 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9142 @option{-arch} options are used; it does so by running the compiler or
9143 linker multiple times and joining the results together with
9146 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9147 @samp{i686}) is determined by the flags that specify the ISA
9148 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9149 @option{-force_cpusubtype_ALL} option can be used to override this.
9151 The Darwin tools vary in their behavior when presented with an ISA
9152 mismatch. The assembler, @file{as}, will only permit instructions to
9153 be used that are valid for the subtype of the file it is generating,
9154 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9155 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9156 and print an error if asked to create a shared library with a less
9157 restrictive subtype than its input files (for instance, trying to put
9158 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9159 for executables, @file{ld}, will quietly give the executable the most
9160 restrictive subtype of any of its input files.
9165 Add the framework directory @var{dir} to the head of the list of
9166 directories to be searched for header files. These directories are
9167 interleaved with those specified by @option{-I} options and are
9168 scanned in a left-to-right order.
9170 A framework directory is a directory with frameworks in it. A
9171 framework is a directory with a @samp{"Headers"} and/or
9172 @samp{"PrivateHeaders"} directory contained directly in it that ends
9173 in @samp{".framework"}. The name of a framework is the name of this
9174 directory excluding the @samp{".framework"}. Headers associated with
9175 the framework are found in one of those two directories, with
9176 @samp{"Headers"} being searched first. A subframework is a framework
9177 directory that is in a framework's @samp{"Frameworks"} directory.
9178 Includes of subframework headers can only appear in a header of a
9179 framework that contains the subframework, or in a sibling subframework
9180 header. Two subframeworks are siblings if they occur in the same
9181 framework. A subframework should not have the same name as a
9182 framework, a warning will be issued if this is violated. Currently a
9183 subframework cannot have subframeworks, in the future, the mechanism
9184 may be extended to support this. The standard frameworks can be found
9185 in @samp{"/System/Library/Frameworks"} and
9186 @samp{"/Library/Frameworks"}. An example include looks like
9187 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9188 the name of the framework and header.h is found in the
9189 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9191 @item -iframework@var{dir}
9193 Like @option{-F} except the directory is a treated as a system
9194 directory. The main difference between this @option{-iframework} and
9195 @option{-F} is that with @option{-iframework} the compiler does not
9196 warn about constructs contained within header files found via
9197 @var{dir}. This option is valid only for the C family of languages.
9201 Emit debugging information for symbols that are used. For STABS
9202 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9203 This is by default ON@.
9207 Emit debugging information for all symbols and types.
9209 @item -mmacosx-version-min=@var{version}
9210 The earliest version of MacOS X that this executable will run on
9211 is @var{version}. Typical values of @var{version} include @code{10.1},
9212 @code{10.2}, and @code{10.3.9}.
9214 If the compiler was built to use the system's headers by default,
9215 then the default for this option is the system version on which the
9216 compiler is running, otherwise the default is to make choices which
9217 are compatible with as many systems and code bases as possible.
9221 Enable kernel development mode. The @option{-mkernel} option sets
9222 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9223 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9224 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9225 applicable. This mode also sets @option{-mno-altivec},
9226 @option{-msoft-float}, @option{-fno-builtin} and
9227 @option{-mlong-branch} for PowerPC targets.
9229 @item -mone-byte-bool
9230 @opindex mone-byte-bool
9231 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9232 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9233 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9234 option has no effect on x86.
9236 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9237 to generate code that is not binary compatible with code generated
9238 without that switch. Using this switch may require recompiling all
9239 other modules in a program, including system libraries. Use this
9240 switch to conform to a non-default data model.
9242 @item -mfix-and-continue
9243 @itemx -ffix-and-continue
9244 @itemx -findirect-data
9245 @opindex mfix-and-continue
9246 @opindex ffix-and-continue
9247 @opindex findirect-data
9248 Generate code suitable for fast turn around development. Needed to
9249 enable gdb to dynamically load @code{.o} files into already running
9250 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9251 are provided for backwards compatibility.
9255 Loads all members of static archive libraries.
9256 See man ld(1) for more information.
9258 @item -arch_errors_fatal
9259 @opindex arch_errors_fatal
9260 Cause the errors having to do with files that have the wrong architecture
9264 @opindex bind_at_load
9265 Causes the output file to be marked such that the dynamic linker will
9266 bind all undefined references when the file is loaded or launched.
9270 Produce a Mach-o bundle format file.
9271 See man ld(1) for more information.
9273 @item -bundle_loader @var{executable}
9274 @opindex bundle_loader
9275 This option specifies the @var{executable} that will be loading the build
9276 output file being linked. See man ld(1) for more information.
9280 When passed this option, GCC will produce a dynamic library instead of
9281 an executable when linking, using the Darwin @file{libtool} command.
9283 @item -force_cpusubtype_ALL
9284 @opindex force_cpusubtype_ALL
9285 This causes GCC's output file to have the @var{ALL} subtype, instead of
9286 one controlled by the @option{-mcpu} or @option{-march} option.
9288 @item -allowable_client @var{client_name}
9290 @itemx -compatibility_version
9291 @itemx -current_version
9293 @itemx -dependency-file
9295 @itemx -dylinker_install_name
9297 @itemx -exported_symbols_list
9299 @itemx -flat_namespace
9300 @itemx -force_flat_namespace
9301 @itemx -headerpad_max_install_names
9304 @itemx -install_name
9305 @itemx -keep_private_externs
9306 @itemx -multi_module
9307 @itemx -multiply_defined
9308 @itemx -multiply_defined_unused
9310 @itemx -no_dead_strip_inits_and_terms
9311 @itemx -nofixprebinding
9314 @itemx -noseglinkedit
9315 @itemx -pagezero_size
9317 @itemx -prebind_all_twolevel_modules
9318 @itemx -private_bundle
9319 @itemx -read_only_relocs
9321 @itemx -sectobjectsymbols
9325 @itemx -sectobjectsymbols
9328 @itemx -segs_read_only_addr
9329 @itemx -segs_read_write_addr
9330 @itemx -seg_addr_table
9331 @itemx -seg_addr_table_filename
9334 @itemx -segs_read_only_addr
9335 @itemx -segs_read_write_addr
9336 @itemx -single_module
9339 @itemx -sub_umbrella
9340 @itemx -twolevel_namespace
9343 @itemx -unexported_symbols_list
9344 @itemx -weak_reference_mismatches
9347 @opindex allowable_client
9348 @opindex client_name
9349 @opindex compatibility_version
9350 @opindex current_version
9352 @opindex dependency-file
9354 @opindex dylinker_install_name
9356 @opindex exported_symbols_list
9358 @opindex flat_namespace
9359 @opindex force_flat_namespace
9360 @opindex headerpad_max_install_names
9363 @opindex install_name
9364 @opindex keep_private_externs
9365 @opindex multi_module
9366 @opindex multiply_defined
9367 @opindex multiply_defined_unused
9369 @opindex no_dead_strip_inits_and_terms
9370 @opindex nofixprebinding
9371 @opindex nomultidefs
9373 @opindex noseglinkedit
9374 @opindex pagezero_size
9376 @opindex prebind_all_twolevel_modules
9377 @opindex private_bundle
9378 @opindex read_only_relocs
9380 @opindex sectobjectsymbols
9384 @opindex sectobjectsymbols
9387 @opindex segs_read_only_addr
9388 @opindex segs_read_write_addr
9389 @opindex seg_addr_table
9390 @opindex seg_addr_table_filename
9391 @opindex seglinkedit
9393 @opindex segs_read_only_addr
9394 @opindex segs_read_write_addr
9395 @opindex single_module
9397 @opindex sub_library
9398 @opindex sub_umbrella
9399 @opindex twolevel_namespace
9402 @opindex unexported_symbols_list
9403 @opindex weak_reference_mismatches
9404 @opindex whatsloaded
9406 These options are passed to the Darwin linker. The Darwin linker man page
9407 describes them in detail.
9410 @node DEC Alpha Options
9411 @subsection DEC Alpha Options
9413 These @samp{-m} options are defined for the DEC Alpha implementations:
9416 @item -mno-soft-float
9418 @opindex mno-soft-float
9419 @opindex msoft-float
9420 Use (do not use) the hardware floating-point instructions for
9421 floating-point operations. When @option{-msoft-float} is specified,
9422 functions in @file{libgcc.a} will be used to perform floating-point
9423 operations. Unless they are replaced by routines that emulate the
9424 floating-point operations, or compiled in such a way as to call such
9425 emulations routines, these routines will issue floating-point
9426 operations. If you are compiling for an Alpha without floating-point
9427 operations, you must ensure that the library is built so as not to call
9430 Note that Alpha implementations without floating-point operations are
9431 required to have floating-point registers.
9436 @opindex mno-fp-regs
9437 Generate code that uses (does not use) the floating-point register set.
9438 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9439 register set is not used, floating point operands are passed in integer
9440 registers as if they were integers and floating-point results are passed
9441 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9442 so any function with a floating-point argument or return value called by code
9443 compiled with @option{-mno-fp-regs} must also be compiled with that
9446 A typical use of this option is building a kernel that does not use,
9447 and hence need not save and restore, any floating-point registers.
9451 The Alpha architecture implements floating-point hardware optimized for
9452 maximum performance. It is mostly compliant with the IEEE floating
9453 point standard. However, for full compliance, software assistance is
9454 required. This option generates code fully IEEE compliant code
9455 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9456 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9457 defined during compilation. The resulting code is less efficient but is
9458 able to correctly support denormalized numbers and exceptional IEEE
9459 values such as not-a-number and plus/minus infinity. Other Alpha
9460 compilers call this option @option{-ieee_with_no_inexact}.
9462 @item -mieee-with-inexact
9463 @opindex mieee-with-inexact
9464 This is like @option{-mieee} except the generated code also maintains
9465 the IEEE @var{inexact-flag}. Turning on this option causes the
9466 generated code to implement fully-compliant IEEE math. In addition to
9467 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9468 macro. On some Alpha implementations the resulting code may execute
9469 significantly slower than the code generated by default. Since there is
9470 very little code that depends on the @var{inexact-flag}, you should
9471 normally not specify this option. Other Alpha compilers call this
9472 option @option{-ieee_with_inexact}.
9474 @item -mfp-trap-mode=@var{trap-mode}
9475 @opindex mfp-trap-mode
9476 This option controls what floating-point related traps are enabled.
9477 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9478 The trap mode can be set to one of four values:
9482 This is the default (normal) setting. The only traps that are enabled
9483 are the ones that cannot be disabled in software (e.g., division by zero
9487 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9491 Like @samp{u}, but the instructions are marked to be safe for software
9492 completion (see Alpha architecture manual for details).
9495 Like @samp{su}, but inexact traps are enabled as well.
9498 @item -mfp-rounding-mode=@var{rounding-mode}
9499 @opindex mfp-rounding-mode
9500 Selects the IEEE rounding mode. Other Alpha compilers call this option
9501 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9506 Normal IEEE rounding mode. Floating point numbers are rounded towards
9507 the nearest machine number or towards the even machine number in case
9511 Round towards minus infinity.
9514 Chopped rounding mode. Floating point numbers are rounded towards zero.
9517 Dynamic rounding mode. A field in the floating point control register
9518 (@var{fpcr}, see Alpha architecture reference manual) controls the
9519 rounding mode in effect. The C library initializes this register for
9520 rounding towards plus infinity. Thus, unless your program modifies the
9521 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9524 @item -mtrap-precision=@var{trap-precision}
9525 @opindex mtrap-precision
9526 In the Alpha architecture, floating point traps are imprecise. This
9527 means without software assistance it is impossible to recover from a
9528 floating trap and program execution normally needs to be terminated.
9529 GCC can generate code that can assist operating system trap handlers
9530 in determining the exact location that caused a floating point trap.
9531 Depending on the requirements of an application, different levels of
9532 precisions can be selected:
9536 Program precision. This option is the default and means a trap handler
9537 can only identify which program caused a floating point exception.
9540 Function precision. The trap handler can determine the function that
9541 caused a floating point exception.
9544 Instruction precision. The trap handler can determine the exact
9545 instruction that caused a floating point exception.
9548 Other Alpha compilers provide the equivalent options called
9549 @option{-scope_safe} and @option{-resumption_safe}.
9551 @item -mieee-conformant
9552 @opindex mieee-conformant
9553 This option marks the generated code as IEEE conformant. You must not
9554 use this option unless you also specify @option{-mtrap-precision=i} and either
9555 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9556 is to emit the line @samp{.eflag 48} in the function prologue of the
9557 generated assembly file. Under DEC Unix, this has the effect that
9558 IEEE-conformant math library routines will be linked in.
9560 @item -mbuild-constants
9561 @opindex mbuild-constants
9562 Normally GCC examines a 32- or 64-bit integer constant to
9563 see if it can construct it from smaller constants in two or three
9564 instructions. If it cannot, it will output the constant as a literal and
9565 generate code to load it from the data segment at runtime.
9567 Use this option to require GCC to construct @emph{all} integer constants
9568 using code, even if it takes more instructions (the maximum is six).
9570 You would typically use this option to build a shared library dynamic
9571 loader. Itself a shared library, it must relocate itself in memory
9572 before it can find the variables and constants in its own data segment.
9578 Select whether to generate code to be assembled by the vendor-supplied
9579 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9597 Indicate whether GCC should generate code to use the optional BWX,
9598 CIX, FIX and MAX instruction sets. The default is to use the instruction
9599 sets supported by the CPU type specified via @option{-mcpu=} option or that
9600 of the CPU on which GCC was built if none was specified.
9605 @opindex mfloat-ieee
9606 Generate code that uses (does not use) VAX F and G floating point
9607 arithmetic instead of IEEE single and double precision.
9609 @item -mexplicit-relocs
9610 @itemx -mno-explicit-relocs
9611 @opindex mexplicit-relocs
9612 @opindex mno-explicit-relocs
9613 Older Alpha assemblers provided no way to generate symbol relocations
9614 except via assembler macros. Use of these macros does not allow
9615 optimal instruction scheduling. GNU binutils as of version 2.12
9616 supports a new syntax that allows the compiler to explicitly mark
9617 which relocations should apply to which instructions. This option
9618 is mostly useful for debugging, as GCC detects the capabilities of
9619 the assembler when it is built and sets the default accordingly.
9623 @opindex msmall-data
9624 @opindex mlarge-data
9625 When @option{-mexplicit-relocs} is in effect, static data is
9626 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9627 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9628 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9629 16-bit relocations off of the @code{$gp} register. This limits the
9630 size of the small data area to 64KB, but allows the variables to be
9631 directly accessed via a single instruction.
9633 The default is @option{-mlarge-data}. With this option the data area
9634 is limited to just below 2GB@. Programs that require more than 2GB of
9635 data must use @code{malloc} or @code{mmap} to allocate the data in the
9636 heap instead of in the program's data segment.
9638 When generating code for shared libraries, @option{-fpic} implies
9639 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9643 @opindex msmall-text
9644 @opindex mlarge-text
9645 When @option{-msmall-text} is used, the compiler assumes that the
9646 code of the entire program (or shared library) fits in 4MB, and is
9647 thus reachable with a branch instruction. When @option{-msmall-data}
9648 is used, the compiler can assume that all local symbols share the
9649 same @code{$gp} value, and thus reduce the number of instructions
9650 required for a function call from 4 to 1.
9652 The default is @option{-mlarge-text}.
9654 @item -mcpu=@var{cpu_type}
9656 Set the instruction set and instruction scheduling parameters for
9657 machine type @var{cpu_type}. You can specify either the @samp{EV}
9658 style name or the corresponding chip number. GCC supports scheduling
9659 parameters for the EV4, EV5 and EV6 family of processors and will
9660 choose the default values for the instruction set from the processor
9661 you specify. If you do not specify a processor type, GCC will default
9662 to the processor on which the compiler was built.
9664 Supported values for @var{cpu_type} are
9670 Schedules as an EV4 and has no instruction set extensions.
9674 Schedules as an EV5 and has no instruction set extensions.
9678 Schedules as an EV5 and supports the BWX extension.
9683 Schedules as an EV5 and supports the BWX and MAX extensions.
9687 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9691 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9694 @item -mtune=@var{cpu_type}
9696 Set only the instruction scheduling parameters for machine type
9697 @var{cpu_type}. The instruction set is not changed.
9699 @item -mmemory-latency=@var{time}
9700 @opindex mmemory-latency
9701 Sets the latency the scheduler should assume for typical memory
9702 references as seen by the application. This number is highly
9703 dependent on the memory access patterns used by the application
9704 and the size of the external cache on the machine.
9706 Valid options for @var{time} are
9710 A decimal number representing clock cycles.
9716 The compiler contains estimates of the number of clock cycles for
9717 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9718 (also called Dcache, Scache, and Bcache), as well as to main memory.
9719 Note that L3 is only valid for EV5.
9724 @node DEC Alpha/VMS Options
9725 @subsection DEC Alpha/VMS Options
9727 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9730 @item -mvms-return-codes
9731 @opindex mvms-return-codes
9732 Return VMS condition codes from main. The default is to return POSIX
9733 style condition (e.g.@: error) codes.
9737 @subsection FRV Options
9744 Only use the first 32 general purpose registers.
9749 Use all 64 general purpose registers.
9754 Use only the first 32 floating point registers.
9759 Use all 64 floating point registers
9762 @opindex mhard-float
9764 Use hardware instructions for floating point operations.
9767 @opindex msoft-float
9769 Use library routines for floating point operations.
9774 Dynamically allocate condition code registers.
9779 Do not try to dynamically allocate condition code registers, only
9780 use @code{icc0} and @code{fcc0}.
9785 Change ABI to use double word insns.
9790 Do not use double word instructions.
9795 Use floating point double instructions.
9800 Do not use floating point double instructions.
9805 Use media instructions.
9810 Do not use media instructions.
9815 Use multiply and add/subtract instructions.
9820 Do not use multiply and add/subtract instructions.
9825 Select the FDPIC ABI, that uses function descriptors to represent
9826 pointers to functions. Without any PIC/PIE-related options, it
9827 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9828 assumes GOT entries and small data are within a 12-bit range from the
9829 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9830 are computed with 32 bits.
9831 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9834 @opindex minline-plt
9836 Enable inlining of PLT entries in function calls to functions that are
9837 not known to bind locally. It has no effect without @option{-mfdpic}.
9838 It's enabled by default if optimizing for speed and compiling for
9839 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9840 optimization option such as @option{-O3} or above is present in the
9846 Assume a large TLS segment when generating thread-local code.
9851 Do not assume a large TLS segment when generating thread-local code.
9856 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9857 that is known to be in read-only sections. It's enabled by default,
9858 except for @option{-fpic} or @option{-fpie}: even though it may help
9859 make the global offset table smaller, it trades 1 instruction for 4.
9860 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9861 one of which may be shared by multiple symbols, and it avoids the need
9862 for a GOT entry for the referenced symbol, so it's more likely to be a
9863 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9865 @item -multilib-library-pic
9866 @opindex multilib-library-pic
9868 Link with the (library, not FD) pic libraries. It's implied by
9869 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9870 @option{-fpic} without @option{-mfdpic}. You should never have to use
9876 Follow the EABI requirement of always creating a frame pointer whenever
9877 a stack frame is allocated. This option is enabled by default and can
9878 be disabled with @option{-mno-linked-fp}.
9881 @opindex mlong-calls
9883 Use indirect addressing to call functions outside the current
9884 compilation unit. This allows the functions to be placed anywhere
9885 within the 32-bit address space.
9887 @item -malign-labels
9888 @opindex malign-labels
9890 Try to align labels to an 8-byte boundary by inserting nops into the
9891 previous packet. This option only has an effect when VLIW packing
9892 is enabled. It doesn't create new packets; it merely adds nops to
9896 @opindex mlibrary-pic
9898 Generate position-independent EABI code.
9903 Use only the first four media accumulator registers.
9908 Use all eight media accumulator registers.
9913 Pack VLIW instructions.
9918 Do not pack VLIW instructions.
9923 Do not mark ABI switches in e_flags.
9928 Enable the use of conditional-move instructions (default).
9930 This switch is mainly for debugging the compiler and will likely be removed
9931 in a future version.
9933 @item -mno-cond-move
9934 @opindex mno-cond-move
9936 Disable the use of conditional-move instructions.
9938 This switch is mainly for debugging the compiler and will likely be removed
9939 in a future version.
9944 Enable the use of conditional set instructions (default).
9946 This switch is mainly for debugging the compiler and will likely be removed
9947 in a future version.
9952 Disable the use of conditional set instructions.
9954 This switch is mainly for debugging the compiler and will likely be removed
9955 in a future version.
9960 Enable the use of conditional execution (default).
9962 This switch is mainly for debugging the compiler and will likely be removed
9963 in a future version.
9965 @item -mno-cond-exec
9966 @opindex mno-cond-exec
9968 Disable the use of conditional execution.
9970 This switch is mainly for debugging the compiler and will likely be removed
9971 in a future version.
9974 @opindex mvliw-branch
9976 Run a pass to pack branches into VLIW instructions (default).
9978 This switch is mainly for debugging the compiler and will likely be removed
9979 in a future version.
9981 @item -mno-vliw-branch
9982 @opindex mno-vliw-branch
9984 Do not run a pass to pack branches into VLIW instructions.
9986 This switch is mainly for debugging the compiler and will likely be removed
9987 in a future version.
9989 @item -mmulti-cond-exec
9990 @opindex mmulti-cond-exec
9992 Enable optimization of @code{&&} and @code{||} in conditional execution
9995 This switch is mainly for debugging the compiler and will likely be removed
9996 in a future version.
9998 @item -mno-multi-cond-exec
9999 @opindex mno-multi-cond-exec
10001 Disable optimization of @code{&&} and @code{||} in conditional execution.
10003 This switch is mainly for debugging the compiler and will likely be removed
10004 in a future version.
10006 @item -mnested-cond-exec
10007 @opindex mnested-cond-exec
10009 Enable nested conditional execution optimizations (default).
10011 This switch is mainly for debugging the compiler and will likely be removed
10012 in a future version.
10014 @item -mno-nested-cond-exec
10015 @opindex mno-nested-cond-exec
10017 Disable nested conditional execution optimizations.
10019 This switch is mainly for debugging the compiler and will likely be removed
10020 in a future version.
10022 @item -moptimize-membar
10023 @opindex moptimize-membar
10025 This switch removes redundant @code{membar} instructions from the
10026 compiler generated code. It is enabled by default.
10028 @item -mno-optimize-membar
10029 @opindex mno-optimize-membar
10031 This switch disables the automatic removal of redundant @code{membar}
10032 instructions from the generated code.
10034 @item -mtomcat-stats
10035 @opindex mtomcat-stats
10037 Cause gas to print out tomcat statistics.
10039 @item -mcpu=@var{cpu}
10042 Select the processor type for which to generate code. Possible values are
10043 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10044 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10048 @node GNU/Linux Options
10049 @subsection GNU/Linux Options
10051 These @samp{-m} options are defined for GNU/Linux targets:
10056 Use the GNU C library instead of uClibc. This is the default except
10057 on @samp{*-*-linux-*uclibc*} targets.
10061 Use uClibc instead of the GNU C library. This is the default on
10062 @samp{*-*-linux-*uclibc*} targets.
10065 @node H8/300 Options
10066 @subsection H8/300 Options
10068 These @samp{-m} options are defined for the H8/300 implementations:
10073 Shorten some address references at link time, when possible; uses the
10074 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10075 ld, Using ld}, for a fuller description.
10079 Generate code for the H8/300H@.
10083 Generate code for the H8S@.
10087 Generate code for the H8S and H8/300H in the normal mode. This switch
10088 must be used either with @option{-mh} or @option{-ms}.
10092 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10096 Make @code{int} data 32 bits by default.
10099 @opindex malign-300
10100 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10101 The default for the H8/300H and H8S is to align longs and floats on 4
10103 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10104 This option has no effect on the H8/300.
10108 @subsection HPPA Options
10109 @cindex HPPA Options
10111 These @samp{-m} options are defined for the HPPA family of computers:
10114 @item -march=@var{architecture-type}
10116 Generate code for the specified architecture. The choices for
10117 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10118 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10119 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10120 architecture option for your machine. Code compiled for lower numbered
10121 architectures will run on higher numbered architectures, but not the
10124 @item -mpa-risc-1-0
10125 @itemx -mpa-risc-1-1
10126 @itemx -mpa-risc-2-0
10127 @opindex mpa-risc-1-0
10128 @opindex mpa-risc-1-1
10129 @opindex mpa-risc-2-0
10130 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10133 @opindex mbig-switch
10134 Generate code suitable for big switch tables. Use this option only if
10135 the assembler/linker complain about out of range branches within a switch
10138 @item -mjump-in-delay
10139 @opindex mjump-in-delay
10140 Fill delay slots of function calls with unconditional jump instructions
10141 by modifying the return pointer for the function call to be the target
10142 of the conditional jump.
10144 @item -mdisable-fpregs
10145 @opindex mdisable-fpregs
10146 Prevent floating point registers from being used in any manner. This is
10147 necessary for compiling kernels which perform lazy context switching of
10148 floating point registers. If you use this option and attempt to perform
10149 floating point operations, the compiler will abort.
10151 @item -mdisable-indexing
10152 @opindex mdisable-indexing
10153 Prevent the compiler from using indexing address modes. This avoids some
10154 rather obscure problems when compiling MIG generated code under MACH@.
10156 @item -mno-space-regs
10157 @opindex mno-space-regs
10158 Generate code that assumes the target has no space registers. This allows
10159 GCC to generate faster indirect calls and use unscaled index address modes.
10161 Such code is suitable for level 0 PA systems and kernels.
10163 @item -mfast-indirect-calls
10164 @opindex mfast-indirect-calls
10165 Generate code that assumes calls never cross space boundaries. This
10166 allows GCC to emit code which performs faster indirect calls.
10168 This option will not work in the presence of shared libraries or nested
10171 @item -mfixed-range=@var{register-range}
10172 @opindex mfixed-range
10173 Generate code treating the given register range as fixed registers.
10174 A fixed register is one that the register allocator can not use. This is
10175 useful when compiling kernel code. A register range is specified as
10176 two registers separated by a dash. Multiple register ranges can be
10177 specified separated by a comma.
10179 @item -mlong-load-store
10180 @opindex mlong-load-store
10181 Generate 3-instruction load and store sequences as sometimes required by
10182 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10185 @item -mportable-runtime
10186 @opindex mportable-runtime
10187 Use the portable calling conventions proposed by HP for ELF systems.
10191 Enable the use of assembler directives only GAS understands.
10193 @item -mschedule=@var{cpu-type}
10195 Schedule code according to the constraints for the machine type
10196 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10197 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10198 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10199 proper scheduling option for your machine. The default scheduling is
10203 @opindex mlinker-opt
10204 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10205 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10206 linkers in which they give bogus error messages when linking some programs.
10209 @opindex msoft-float
10210 Generate output containing library calls for floating point.
10211 @strong{Warning:} the requisite libraries are not available for all HPPA
10212 targets. Normally the facilities of the machine's usual C compiler are
10213 used, but this cannot be done directly in cross-compilation. You must make
10214 your own arrangements to provide suitable library functions for
10215 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
10216 does provide software floating point support.
10218 @option{-msoft-float} changes the calling convention in the output file;
10219 therefore, it is only useful if you compile @emph{all} of a program with
10220 this option. In particular, you need to compile @file{libgcc.a}, the
10221 library that comes with GCC, with @option{-msoft-float} in order for
10226 Generate the predefine, @code{_SIO}, for server IO@. The default is
10227 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10228 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10229 options are available under HP-UX and HI-UX@.
10233 Use GNU ld specific options. This passes @option{-shared} to ld when
10234 building a shared library. It is the default when GCC is configured,
10235 explicitly or implicitly, with the GNU linker. This option does not
10236 have any affect on which ld is called, it only changes what parameters
10237 are passed to that ld. The ld that is called is determined by the
10238 @option{--with-ld} configure option, GCC's program search path, and
10239 finally by the user's @env{PATH}. The linker used by GCC can be printed
10240 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10241 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10245 Use HP ld specific options. This passes @option{-b} to ld when building
10246 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10247 links. It is the default when GCC is configured, explicitly or
10248 implicitly, with the HP linker. This option does not have any affect on
10249 which ld is called, it only changes what parameters are passed to that
10250 ld. The ld that is called is determined by the @option{--with-ld}
10251 configure option, GCC's program search path, and finally by the user's
10252 @env{PATH}. The linker used by GCC can be printed using @samp{which
10253 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10254 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10257 @opindex mno-long-calls
10258 Generate code that uses long call sequences. This ensures that a call
10259 is always able to reach linker generated stubs. The default is to generate
10260 long calls only when the distance from the call site to the beginning
10261 of the function or translation unit, as the case may be, exceeds a
10262 predefined limit set by the branch type being used. The limits for
10263 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10264 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10267 Distances are measured from the beginning of functions when using the
10268 @option{-ffunction-sections} option, or when using the @option{-mgas}
10269 and @option{-mno-portable-runtime} options together under HP-UX with
10272 It is normally not desirable to use this option as it will degrade
10273 performance. However, it may be useful in large applications,
10274 particularly when partial linking is used to build the application.
10276 The types of long calls used depends on the capabilities of the
10277 assembler and linker, and the type of code being generated. The
10278 impact on systems that support long absolute calls, and long pic
10279 symbol-difference or pc-relative calls should be relatively small.
10280 However, an indirect call is used on 32-bit ELF systems in pic code
10281 and it is quite long.
10283 @item -munix=@var{unix-std}
10285 Generate compiler predefines and select a startfile for the specified
10286 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10287 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10288 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10289 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10290 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10293 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10294 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10295 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10296 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10297 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10298 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10300 It is @emph{important} to note that this option changes the interfaces
10301 for various library routines. It also affects the operational behavior
10302 of the C library. Thus, @emph{extreme} care is needed in using this
10305 Library code that is intended to operate with more than one UNIX
10306 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10307 as appropriate. Most GNU software doesn't provide this capability.
10311 Suppress the generation of link options to search libdld.sl when the
10312 @option{-static} option is specified on HP-UX 10 and later.
10316 The HP-UX implementation of setlocale in libc has a dependency on
10317 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10318 when the @option{-static} option is specified, special link options
10319 are needed to resolve this dependency.
10321 On HP-UX 10 and later, the GCC driver adds the necessary options to
10322 link with libdld.sl when the @option{-static} option is specified.
10323 This causes the resulting binary to be dynamic. On the 64-bit port,
10324 the linkers generate dynamic binaries by default in any case. The
10325 @option{-nolibdld} option can be used to prevent the GCC driver from
10326 adding these link options.
10330 Add support for multithreading with the @dfn{dce thread} library
10331 under HP-UX@. This option sets flags for both the preprocessor and
10335 @node i386 and x86-64 Options
10336 @subsection Intel 386 and AMD x86-64 Options
10337 @cindex i386 Options
10338 @cindex x86-64 Options
10339 @cindex Intel 386 Options
10340 @cindex AMD x86-64 Options
10342 These @samp{-m} options are defined for the i386 and x86-64 family of
10346 @item -mtune=@var{cpu-type}
10348 Tune to @var{cpu-type} everything applicable about the generated code, except
10349 for the ABI and the set of available instructions. The choices for
10350 @var{cpu-type} are:
10353 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10354 If you know the CPU on which your code will run, then you should use
10355 the corresponding @option{-mtune} option instead of
10356 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10357 of your application will have, then you should use this option.
10359 As new processors are deployed in the marketplace, the behavior of this
10360 option will change. Therefore, if you upgrade to a newer version of
10361 GCC, the code generated option will change to reflect the processors
10362 that were most common when that version of GCC was released.
10364 There is no @option{-march=generic} option because @option{-march}
10365 indicates the instruction set the compiler can use, and there is no
10366 generic instruction set applicable to all processors. In contrast,
10367 @option{-mtune} indicates the processor (or, in this case, collection of
10368 processors) for which the code is optimized.
10370 This selects the CPU to tune for at compilation time by determining
10371 the processor type of the compiling machine. Using @option{-mtune=native}
10372 will produce code optimized for the local machine under the constraints
10373 of the selected instruction set. Using @option{-march=native} will
10374 enable all instruction subsets supported by the local machine (hence
10375 the result might not run on different machines).
10377 Original Intel's i386 CPU@.
10379 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10380 @item i586, pentium
10381 Intel Pentium CPU with no MMX support.
10383 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10385 Intel PentiumPro CPU@.
10387 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10388 instruction set will be used, so the code will run on all i686 family chips.
10390 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10391 @item pentium3, pentium3m
10392 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10395 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10396 support. Used by Centrino notebooks.
10397 @item pentium4, pentium4m
10398 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10400 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10403 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10404 SSE2 and SSE3 instruction set support.
10406 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10407 instruction set support.
10409 AMD K6 CPU with MMX instruction set support.
10411 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10412 @item athlon, athlon-tbird
10413 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10415 @item athlon-4, athlon-xp, athlon-mp
10416 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10417 instruction set support.
10418 @item k8, opteron, athlon64, athlon-fx
10419 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10420 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10421 @item k8-sse3, opteron-sse3, athlon64-sse3
10422 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10423 @item amdfam10, barcelona
10424 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10425 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10426 instruction set extensions.)
10428 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10431 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
10432 instruction set support.
10434 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
10435 implemented for this chip.)
10437 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10438 implemented for this chip.)
10440 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10443 While picking a specific @var{cpu-type} will schedule things appropriately
10444 for that particular chip, the compiler will not generate any code that
10445 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10448 @item -march=@var{cpu-type}
10450 Generate instructions for the machine type @var{cpu-type}. The choices
10451 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10452 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10454 @item -mcpu=@var{cpu-type}
10456 A deprecated synonym for @option{-mtune}.
10458 @item -mfpmath=@var{unit}
10460 Generate floating point arithmetics for selected unit @var{unit}. The choices
10461 for @var{unit} are:
10465 Use the standard 387 floating point coprocessor present majority of chips and
10466 emulated otherwise. Code compiled with this option will run almost everywhere.
10467 The temporary results are computed in 80bit precision instead of precision
10468 specified by the type resulting in slightly different results compared to most
10469 of other chips. See @option{-ffloat-store} for more detailed description.
10471 This is the default choice for i386 compiler.
10474 Use scalar floating point instructions present in the SSE instruction set.
10475 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10476 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10477 instruction set supports only single precision arithmetics, thus the double and
10478 extended precision arithmetics is still done using 387. Later version, present
10479 only in Pentium4 and the future AMD x86-64 chips supports double precision
10482 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10483 or @option{-msse2} switches to enable SSE extensions and make this option
10484 effective. For the x86-64 compiler, these extensions are enabled by default.
10486 The resulting code should be considerably faster in the majority of cases and avoid
10487 the numerical instability problems of 387 code, but may break some existing
10488 code that expects temporaries to be 80bit.
10490 This is the default choice for the x86-64 compiler.
10493 Attempt to utilize both instruction sets at once. This effectively double the
10494 amount of available registers and on chips with separate execution units for
10495 387 and SSE the execution resources too. Use this option with care, as it is
10496 still experimental, because the GCC register allocator does not model separate
10497 functional units well resulting in instable performance.
10500 @item -masm=@var{dialect}
10501 @opindex masm=@var{dialect}
10502 Output asm instructions using selected @var{dialect}. Supported
10503 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10504 not support @samp{intel}.
10507 @itemx -mno-ieee-fp
10509 @opindex mno-ieee-fp
10510 Control whether or not the compiler uses IEEE floating point
10511 comparisons. These handle correctly the case where the result of a
10512 comparison is unordered.
10515 @opindex msoft-float
10516 Generate output containing library calls for floating point.
10517 @strong{Warning:} the requisite libraries are not part of GCC@.
10518 Normally the facilities of the machine's usual C compiler are used, but
10519 this can't be done directly in cross-compilation. You must make your
10520 own arrangements to provide suitable library functions for
10523 On machines where a function returns floating point results in the 80387
10524 register stack, some floating point opcodes may be emitted even if
10525 @option{-msoft-float} is used.
10527 @item -mno-fp-ret-in-387
10528 @opindex mno-fp-ret-in-387
10529 Do not use the FPU registers for return values of functions.
10531 The usual calling convention has functions return values of types
10532 @code{float} and @code{double} in an FPU register, even if there
10533 is no FPU@. The idea is that the operating system should emulate
10536 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10537 in ordinary CPU registers instead.
10539 @item -mno-fancy-math-387
10540 @opindex mno-fancy-math-387
10541 Some 387 emulators do not support the @code{sin}, @code{cos} and
10542 @code{sqrt} instructions for the 387. Specify this option to avoid
10543 generating those instructions. This option is the default on FreeBSD,
10544 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10545 indicates that the target cpu will always have an FPU and so the
10546 instruction will not need emulation. As of revision 2.6.1, these
10547 instructions are not generated unless you also use the
10548 @option{-funsafe-math-optimizations} switch.
10550 @item -malign-double
10551 @itemx -mno-align-double
10552 @opindex malign-double
10553 @opindex mno-align-double
10554 Control whether GCC aligns @code{double}, @code{long double}, and
10555 @code{long long} variables on a two word boundary or a one word
10556 boundary. Aligning @code{double} variables on a two word boundary will
10557 produce code that runs somewhat faster on a @samp{Pentium} at the
10558 expense of more memory.
10560 On x86-64, @option{-malign-double} is enabled by default.
10562 @strong{Warning:} if you use the @option{-malign-double} switch,
10563 structures containing the above types will be aligned differently than
10564 the published application binary interface specifications for the 386
10565 and will not be binary compatible with structures in code compiled
10566 without that switch.
10568 @item -m96bit-long-double
10569 @itemx -m128bit-long-double
10570 @opindex m96bit-long-double
10571 @opindex m128bit-long-double
10572 These switches control the size of @code{long double} type. The i386
10573 application binary interface specifies the size to be 96 bits,
10574 so @option{-m96bit-long-double} is the default in 32 bit mode.
10576 Modern architectures (Pentium and newer) would prefer @code{long double}
10577 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10578 conforming to the ABI, this would not be possible. So specifying a
10579 @option{-m128bit-long-double} will align @code{long double}
10580 to a 16 byte boundary by padding the @code{long double} with an additional
10583 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10584 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10586 Notice that neither of these options enable any extra precision over the x87
10587 standard of 80 bits for a @code{long double}.
10589 @strong{Warning:} if you override the default value for your target ABI, the
10590 structures and arrays containing @code{long double} variables will change
10591 their size as well as function calling convention for function taking
10592 @code{long double} will be modified. Hence they will not be binary
10593 compatible with arrays or structures in code compiled without that switch.
10595 @item -mmlarge-data-threshold=@var{number}
10596 @opindex mlarge-data-threshold=@var{number}
10597 When @option{-mcmodel=medium} is specified, the data greater than
10598 @var{threshold} are placed in large data section. This value must be the
10599 same across all object linked into the binary and defaults to 65535.
10603 Use a different function-calling convention, in which functions that
10604 take a fixed number of arguments return with the @code{ret} @var{num}
10605 instruction, which pops their arguments while returning. This saves one
10606 instruction in the caller since there is no need to pop the arguments
10609 You can specify that an individual function is called with this calling
10610 sequence with the function attribute @samp{stdcall}. You can also
10611 override the @option{-mrtd} option by using the function attribute
10612 @samp{cdecl}. @xref{Function Attributes}.
10614 @strong{Warning:} this calling convention is incompatible with the one
10615 normally used on Unix, so you cannot use it if you need to call
10616 libraries compiled with the Unix compiler.
10618 Also, you must provide function prototypes for all functions that
10619 take variable numbers of arguments (including @code{printf});
10620 otherwise incorrect code will be generated for calls to those
10623 In addition, seriously incorrect code will result if you call a
10624 function with too many arguments. (Normally, extra arguments are
10625 harmlessly ignored.)
10627 @item -mregparm=@var{num}
10629 Control how many registers are used to pass integer arguments. By
10630 default, no registers are used to pass arguments, and at most 3
10631 registers can be used. You can control this behavior for a specific
10632 function by using the function attribute @samp{regparm}.
10633 @xref{Function Attributes}.
10635 @strong{Warning:} if you use this switch, and
10636 @var{num} is nonzero, then you must build all modules with the same
10637 value, including any libraries. This includes the system libraries and
10641 @opindex msseregparm
10642 Use SSE register passing conventions for float and double arguments
10643 and return values. You can control this behavior for a specific
10644 function by using the function attribute @samp{sseregparm}.
10645 @xref{Function Attributes}.
10647 @strong{Warning:} if you use this switch then you must build all
10648 modules with the same value, including any libraries. This includes
10649 the system libraries and startup modules.
10658 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10659 is specified, the significands of results of floating-point operations are
10660 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10661 significands of results of floating-point operations to 53 bits (double
10662 precision) and @option{-mpc80} rounds the significands of results of
10663 floating-point operations to 64 bits (extended double precision), which is
10664 the default. When this option is used, floating-point operations in higher
10665 precisions are not available to the programmer without setting the FPU
10666 control word explicitly.
10668 Setting the rounding of floating-point operations to less than the default
10669 80 bits can speed some programs by 2% or more. Note that some mathematical
10670 libraries assume that extended precision (80 bit) floating-point operations
10671 are enabled by default; routines in such libraries could suffer significant
10672 loss of accuracy, typically through so-called "catastrophic cancellation",
10673 when this option is used to set the precision to less than extended precision.
10675 @item -mstackrealign
10676 @opindex mstackrealign
10677 Realign the stack at entry. On the Intel x86, the
10678 @option{-mstackrealign} option will generate an alternate prologue and
10679 epilogue that realigns the runtime stack. This supports mixing legacy
10680 codes that keep a 4-byte aligned stack with modern codes that keep a
10681 16-byte stack for SSE compatibility. The alternate prologue and
10682 epilogue are slower and bigger than the regular ones, and the
10683 alternate prologue requires an extra scratch register; this lowers the
10684 number of registers available if used in conjunction with the
10685 @code{regparm} attribute. The @option{-mstackrealign} option is
10686 incompatible with the nested function prologue; this is considered a
10687 hard error. See also the attribute @code{force_align_arg_pointer},
10688 applicable to individual functions.
10690 @item -mpreferred-stack-boundary=@var{num}
10691 @opindex mpreferred-stack-boundary
10692 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10693 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10694 the default is 4 (16 bytes or 128 bits).
10696 On Pentium and PentiumPro, @code{double} and @code{long double} values
10697 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10698 suffer significant run time performance penalties. On Pentium III, the
10699 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10700 properly if it is not 16 byte aligned.
10702 To ensure proper alignment of this values on the stack, the stack boundary
10703 must be as aligned as that required by any value stored on the stack.
10704 Further, every function must be generated such that it keeps the stack
10705 aligned. Thus calling a function compiled with a higher preferred
10706 stack boundary from a function compiled with a lower preferred stack
10707 boundary will most likely misalign the stack. It is recommended that
10708 libraries that use callbacks always use the default setting.
10710 This extra alignment does consume extra stack space, and generally
10711 increases code size. Code that is sensitive to stack space usage, such
10712 as embedded systems and operating system kernels, may want to reduce the
10713 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10747 These switches enable or disable the use of instructions in the MMX,
10748 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, SSE5, ABM or 3DNow!@: extended
10750 These extensions are also available as built-in functions: see
10751 @ref{X86 Built-in Functions}, for details of the functions enabled and
10752 disabled by these switches.
10754 To have SSE/SSE2 instructions generated automatically from floating-point
10755 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10757 These options will enable GCC to use these extended instructions in
10758 generated code, even without @option{-mfpmath=sse}. Applications which
10759 perform runtime CPU detection must compile separate files for each
10760 supported architecture, using the appropriate flags. In particular,
10761 the file containing the CPU detection code should be compiled without
10766 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10767 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10768 data types. This is useful for high resolution counters that could be updated
10769 by multiple processors (or cores). This instruction is generated as part of
10770 atomic built-in functions: see @ref{Atomic Builtins} for details.
10774 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10775 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10776 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10777 SAHF are load and store instructions, respectively, for certain status flags.
10778 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10779 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10783 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10784 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
10785 to increase precision instead of DIVSS and SQRTSS (and their vectorized
10786 variants) for single precision floating point arguments. These instructions
10787 are generated only when @option{-funsafe-math-optimizations} is enabled
10788 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
10789 Note that while the throughput of the sequence is higher than the throughput
10790 of the non-reciprocal instruction, the precision of the sequence can be
10791 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
10793 @item -mveclibabi=@var{type}
10794 @opindex mveclibabi
10795 Specifies the ABI type to use for vectorizing intrinsics using an
10796 external library. Supported types are @code{acml} for the AMD
10797 math core library style of interfacing. GCC will currently emit
10798 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10799 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10800 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10801 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10802 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10803 is enabled. A ACML ABI compatible library will have to be specified
10807 @itemx -mno-push-args
10808 @opindex mpush-args
10809 @opindex mno-push-args
10810 Use PUSH operations to store outgoing parameters. This method is shorter
10811 and usually equally fast as method using SUB/MOV operations and is enabled
10812 by default. In some cases disabling it may improve performance because of
10813 improved scheduling and reduced dependencies.
10815 @item -maccumulate-outgoing-args
10816 @opindex maccumulate-outgoing-args
10817 If enabled, the maximum amount of space required for outgoing arguments will be
10818 computed in the function prologue. This is faster on most modern CPUs
10819 because of reduced dependencies, improved scheduling and reduced stack usage
10820 when preferred stack boundary is not equal to 2. The drawback is a notable
10821 increase in code size. This switch implies @option{-mno-push-args}.
10825 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10826 on thread-safe exception handling must compile and link all code with the
10827 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10828 @option{-D_MT}; when linking, it links in a special thread helper library
10829 @option{-lmingwthrd} which cleans up per thread exception handling data.
10831 @item -mno-align-stringops
10832 @opindex mno-align-stringops
10833 Do not align destination of inlined string operations. This switch reduces
10834 code size and improves performance in case the destination is already aligned,
10835 but GCC doesn't know about it.
10837 @item -minline-all-stringops
10838 @opindex minline-all-stringops
10839 By default GCC inlines string operations only when destination is known to be
10840 aligned at least to 4 byte boundary. This enables more inlining, increase code
10841 size, but may improve performance of code that depends on fast memcpy, strlen
10842 and memset for short lengths.
10844 @item -minline-stringops-dynamically
10845 @opindex minline-stringops-dynamically
10846 For string operation of unknown size, inline runtime checks so for small
10847 blocks inline code is used, while for large blocks library call is used.
10849 @item -mstringop-strategy=@var{alg}
10850 @opindex mstringop-strategy=@var{alg}
10851 Overwrite internal decision heuristic about particular algorithm to inline
10852 string operation with. The allowed values are @code{rep_byte},
10853 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10854 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10855 expanding inline loop, @code{libcall} for always expanding library call.
10857 @item -momit-leaf-frame-pointer
10858 @opindex momit-leaf-frame-pointer
10859 Don't keep the frame pointer in a register for leaf functions. This
10860 avoids the instructions to save, set up and restore frame pointers and
10861 makes an extra register available in leaf functions. The option
10862 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10863 which might make debugging harder.
10865 @item -mtls-direct-seg-refs
10866 @itemx -mno-tls-direct-seg-refs
10867 @opindex mtls-direct-seg-refs
10868 Controls whether TLS variables may be accessed with offsets from the
10869 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10870 or whether the thread base pointer must be added. Whether or not this
10871 is legal depends on the operating system, and whether it maps the
10872 segment to cover the entire TLS area.
10874 For systems that use GNU libc, the default is on.
10877 @itemx -mno-fused-madd
10878 @opindex mfused-madd
10879 Enable automatic generation of fused floating point multiply-add instructions
10880 if the ISA supports such instructions. The -mfused-madd option is on by
10881 default. The fused multiply-add instructions have a different
10882 rounding behavior compared to executing a multiply followed by an add.
10885 These @samp{-m} switches are supported in addition to the above
10886 on AMD x86-64 processors in 64-bit environments.
10893 Generate code for a 32-bit or 64-bit environment.
10894 The 32-bit environment sets int, long and pointer to 32 bits and
10895 generates code that runs on any i386 system.
10896 The 64-bit environment sets int to 32 bits and long and pointer
10897 to 64 bits and generates code for AMD's x86-64 architecture. For
10898 darwin only the -m64 option turns off the @option{-fno-pic} and
10899 @option{-mdynamic-no-pic} options.
10901 @item -mno-red-zone
10902 @opindex no-red-zone
10903 Do not use a so called red zone for x86-64 code. The red zone is mandated
10904 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10905 stack pointer that will not be modified by signal or interrupt handlers
10906 and therefore can be used for temporary data without adjusting the stack
10907 pointer. The flag @option{-mno-red-zone} disables this red zone.
10909 @item -mcmodel=small
10910 @opindex mcmodel=small
10911 Generate code for the small code model: the program and its symbols must
10912 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10913 Programs can be statically or dynamically linked. This is the default
10916 @item -mcmodel=kernel
10917 @opindex mcmodel=kernel
10918 Generate code for the kernel code model. The kernel runs in the
10919 negative 2 GB of the address space.
10920 This model has to be used for Linux kernel code.
10922 @item -mcmodel=medium
10923 @opindex mcmodel=medium
10924 Generate code for the medium model: The program is linked in the lower 2
10925 GB of the address space but symbols can be located anywhere in the
10926 address space. Programs can be statically or dynamically linked, but
10927 building of shared libraries are not supported with the medium model.
10929 @item -mcmodel=large
10930 @opindex mcmodel=large
10931 Generate code for the large model: This model makes no assumptions
10932 about addresses and sizes of sections.
10935 @node IA-64 Options
10936 @subsection IA-64 Options
10937 @cindex IA-64 Options
10939 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10943 @opindex mbig-endian
10944 Generate code for a big endian target. This is the default for HP-UX@.
10946 @item -mlittle-endian
10947 @opindex mlittle-endian
10948 Generate code for a little endian target. This is the default for AIX5
10954 @opindex mno-gnu-as
10955 Generate (or don't) code for the GNU assembler. This is the default.
10956 @c Also, this is the default if the configure option @option{--with-gnu-as}
10962 @opindex mno-gnu-ld
10963 Generate (or don't) code for the GNU linker. This is the default.
10964 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10969 Generate code that does not use a global pointer register. The result
10970 is not position independent code, and violates the IA-64 ABI@.
10972 @item -mvolatile-asm-stop
10973 @itemx -mno-volatile-asm-stop
10974 @opindex mvolatile-asm-stop
10975 @opindex mno-volatile-asm-stop
10976 Generate (or don't) a stop bit immediately before and after volatile asm
10979 @item -mregister-names
10980 @itemx -mno-register-names
10981 @opindex mregister-names
10982 @opindex mno-register-names
10983 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10984 the stacked registers. This may make assembler output more readable.
10990 Disable (or enable) optimizations that use the small data section. This may
10991 be useful for working around optimizer bugs.
10993 @item -mconstant-gp
10994 @opindex mconstant-gp
10995 Generate code that uses a single constant global pointer value. This is
10996 useful when compiling kernel code.
11000 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11001 This is useful when compiling firmware code.
11003 @item -minline-float-divide-min-latency
11004 @opindex minline-float-divide-min-latency
11005 Generate code for inline divides of floating point values
11006 using the minimum latency algorithm.
11008 @item -minline-float-divide-max-throughput
11009 @opindex minline-float-divide-max-throughput
11010 Generate code for inline divides of floating point values
11011 using the maximum throughput algorithm.
11013 @item -minline-int-divide-min-latency
11014 @opindex minline-int-divide-min-latency
11015 Generate code for inline divides of integer values
11016 using the minimum latency algorithm.
11018 @item -minline-int-divide-max-throughput
11019 @opindex minline-int-divide-max-throughput
11020 Generate code for inline divides of integer values
11021 using the maximum throughput algorithm.
11023 @item -minline-sqrt-min-latency
11024 @opindex minline-sqrt-min-latency
11025 Generate code for inline square roots
11026 using the minimum latency algorithm.
11028 @item -minline-sqrt-max-throughput
11029 @opindex minline-sqrt-max-throughput
11030 Generate code for inline square roots
11031 using the maximum throughput algorithm.
11033 @item -mno-dwarf2-asm
11034 @itemx -mdwarf2-asm
11035 @opindex mno-dwarf2-asm
11036 @opindex mdwarf2-asm
11037 Don't (or do) generate assembler code for the DWARF2 line number debugging
11038 info. This may be useful when not using the GNU assembler.
11040 @item -mearly-stop-bits
11041 @itemx -mno-early-stop-bits
11042 @opindex mearly-stop-bits
11043 @opindex mno-early-stop-bits
11044 Allow stop bits to be placed earlier than immediately preceding the
11045 instruction that triggered the stop bit. This can improve instruction
11046 scheduling, but does not always do so.
11048 @item -mfixed-range=@var{register-range}
11049 @opindex mfixed-range
11050 Generate code treating the given register range as fixed registers.
11051 A fixed register is one that the register allocator can not use. This is
11052 useful when compiling kernel code. A register range is specified as
11053 two registers separated by a dash. Multiple register ranges can be
11054 specified separated by a comma.
11056 @item -mtls-size=@var{tls-size}
11058 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11061 @item -mtune=@var{cpu-type}
11063 Tune the instruction scheduling for a particular CPU, Valid values are
11064 itanium, itanium1, merced, itanium2, and mckinley.
11070 Add support for multithreading using the POSIX threads library. This
11071 option sets flags for both the preprocessor and linker. It does
11072 not affect the thread safety of object code produced by the compiler or
11073 that of libraries supplied with it. These are HP-UX specific flags.
11079 Generate code for a 32-bit or 64-bit environment.
11080 The 32-bit environment sets int, long and pointer to 32 bits.
11081 The 64-bit environment sets int to 32 bits and long and pointer
11082 to 64 bits. These are HP-UX specific flags.
11084 @item -mno-sched-br-data-spec
11085 @itemx -msched-br-data-spec
11086 @opindex mno-sched-br-data-spec
11087 @opindex msched-br-data-spec
11088 (Dis/En)able data speculative scheduling before reload.
11089 This will result in generation of the ld.a instructions and
11090 the corresponding check instructions (ld.c / chk.a).
11091 The default is 'disable'.
11093 @item -msched-ar-data-spec
11094 @itemx -mno-sched-ar-data-spec
11095 @opindex msched-ar-data-spec
11096 @opindex mno-sched-ar-data-spec
11097 (En/Dis)able data speculative scheduling after reload.
11098 This will result in generation of the ld.a instructions and
11099 the corresponding check instructions (ld.c / chk.a).
11100 The default is 'enable'.
11102 @item -mno-sched-control-spec
11103 @itemx -msched-control-spec
11104 @opindex mno-sched-control-spec
11105 @opindex msched-control-spec
11106 (Dis/En)able control speculative scheduling. This feature is
11107 available only during region scheduling (i.e.@: before reload).
11108 This will result in generation of the ld.s instructions and
11109 the corresponding check instructions chk.s .
11110 The default is 'disable'.
11112 @item -msched-br-in-data-spec
11113 @itemx -mno-sched-br-in-data-spec
11114 @opindex msched-br-in-data-spec
11115 @opindex mno-sched-br-in-data-spec
11116 (En/Dis)able speculative scheduling of the instructions that
11117 are dependent on the data speculative loads before reload.
11118 This is effective only with @option{-msched-br-data-spec} enabled.
11119 The default is 'enable'.
11121 @item -msched-ar-in-data-spec
11122 @itemx -mno-sched-ar-in-data-spec
11123 @opindex msched-ar-in-data-spec
11124 @opindex mno-sched-ar-in-data-spec
11125 (En/Dis)able speculative scheduling of the instructions that
11126 are dependent on the data speculative loads after reload.
11127 This is effective only with @option{-msched-ar-data-spec} enabled.
11128 The default is 'enable'.
11130 @item -msched-in-control-spec
11131 @itemx -mno-sched-in-control-spec
11132 @opindex msched-in-control-spec
11133 @opindex mno-sched-in-control-spec
11134 (En/Dis)able speculative scheduling of the instructions that
11135 are dependent on the control speculative loads.
11136 This is effective only with @option{-msched-control-spec} enabled.
11137 The default is 'enable'.
11140 @itemx -mno-sched-ldc
11141 @opindex msched-ldc
11142 @opindex mno-sched-ldc
11143 (En/Dis)able use of simple data speculation checks ld.c .
11144 If disabled, only chk.a instructions will be emitted to check
11145 data speculative loads.
11146 The default is 'enable'.
11148 @item -mno-sched-control-ldc
11149 @itemx -msched-control-ldc
11150 @opindex mno-sched-control-ldc
11151 @opindex msched-control-ldc
11152 (Dis/En)able use of ld.c instructions to check control speculative loads.
11153 If enabled, in case of control speculative load with no speculatively
11154 scheduled dependent instructions this load will be emitted as ld.sa and
11155 ld.c will be used to check it.
11156 The default is 'disable'.
11158 @item -mno-sched-spec-verbose
11159 @itemx -msched-spec-verbose
11160 @opindex mno-sched-spec-verbose
11161 @opindex msched-spec-verbose
11162 (Dis/En)able printing of the information about speculative motions.
11164 @item -mno-sched-prefer-non-data-spec-insns
11165 @itemx -msched-prefer-non-data-spec-insns
11166 @opindex mno-sched-prefer-non-data-spec-insns
11167 @opindex msched-prefer-non-data-spec-insns
11168 If enabled, data speculative instructions will be chosen for schedule
11169 only if there are no other choices at the moment. This will make
11170 the use of the data speculation much more conservative.
11171 The default is 'disable'.
11173 @item -mno-sched-prefer-non-control-spec-insns
11174 @itemx -msched-prefer-non-control-spec-insns
11175 @opindex mno-sched-prefer-non-control-spec-insns
11176 @opindex msched-prefer-non-control-spec-insns
11177 If enabled, control speculative instructions will be chosen for schedule
11178 only if there are no other choices at the moment. This will make
11179 the use of the control speculation much more conservative.
11180 The default is 'disable'.
11182 @item -mno-sched-count-spec-in-critical-path
11183 @itemx -msched-count-spec-in-critical-path
11184 @opindex mno-sched-count-spec-in-critical-path
11185 @opindex msched-count-spec-in-critical-path
11186 If enabled, speculative dependencies will be considered during
11187 computation of the instructions priorities. This will make the use of the
11188 speculation a bit more conservative.
11189 The default is 'disable'.
11194 @subsection M32C Options
11195 @cindex M32C options
11198 @item -mcpu=@var{name}
11200 Select the CPU for which code is generated. @var{name} may be one of
11201 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11202 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11203 the M32C/80 series.
11207 Specifies that the program will be run on the simulator. This causes
11208 an alternate runtime library to be linked in which supports, for
11209 example, file I/O@. You must not use this option when generating
11210 programs that will run on real hardware; you must provide your own
11211 runtime library for whatever I/O functions are needed.
11213 @item -memregs=@var{number}
11215 Specifies the number of memory-based pseudo-registers GCC will use
11216 during code generation. These pseudo-registers will be used like real
11217 registers, so there is a tradeoff between GCC's ability to fit the
11218 code into available registers, and the performance penalty of using
11219 memory instead of registers. Note that all modules in a program must
11220 be compiled with the same value for this option. Because of that, you
11221 must not use this option with the default runtime libraries gcc
11226 @node M32R/D Options
11227 @subsection M32R/D Options
11228 @cindex M32R/D options
11230 These @option{-m} options are defined for Renesas M32R/D architectures:
11235 Generate code for the M32R/2@.
11239 Generate code for the M32R/X@.
11243 Generate code for the M32R@. This is the default.
11245 @item -mmodel=small
11246 @opindex mmodel=small
11247 Assume all objects live in the lower 16MB of memory (so that their addresses
11248 can be loaded with the @code{ld24} instruction), and assume all subroutines
11249 are reachable with the @code{bl} instruction.
11250 This is the default.
11252 The addressability of a particular object can be set with the
11253 @code{model} attribute.
11255 @item -mmodel=medium
11256 @opindex mmodel=medium
11257 Assume objects may be anywhere in the 32-bit address space (the compiler
11258 will generate @code{seth/add3} instructions to load their addresses), and
11259 assume all subroutines are reachable with the @code{bl} instruction.
11261 @item -mmodel=large
11262 @opindex mmodel=large
11263 Assume objects may be anywhere in the 32-bit address space (the compiler
11264 will generate @code{seth/add3} instructions to load their addresses), and
11265 assume subroutines may not be reachable with the @code{bl} instruction
11266 (the compiler will generate the much slower @code{seth/add3/jl}
11267 instruction sequence).
11270 @opindex msdata=none
11271 Disable use of the small data area. Variables will be put into
11272 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11273 @code{section} attribute has been specified).
11274 This is the default.
11276 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11277 Objects may be explicitly put in the small data area with the
11278 @code{section} attribute using one of these sections.
11280 @item -msdata=sdata
11281 @opindex msdata=sdata
11282 Put small global and static data in the small data area, but do not
11283 generate special code to reference them.
11286 @opindex msdata=use
11287 Put small global and static data in the small data area, and generate
11288 special instructions to reference them.
11292 @cindex smaller data references
11293 Put global and static objects less than or equal to @var{num} bytes
11294 into the small data or bss sections instead of the normal data or bss
11295 sections. The default value of @var{num} is 8.
11296 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11297 for this option to have any effect.
11299 All modules should be compiled with the same @option{-G @var{num}} value.
11300 Compiling with different values of @var{num} may or may not work; if it
11301 doesn't the linker will give an error message---incorrect code will not be
11306 Makes the M32R specific code in the compiler display some statistics
11307 that might help in debugging programs.
11309 @item -malign-loops
11310 @opindex malign-loops
11311 Align all loops to a 32-byte boundary.
11313 @item -mno-align-loops
11314 @opindex mno-align-loops
11315 Do not enforce a 32-byte alignment for loops. This is the default.
11317 @item -missue-rate=@var{number}
11318 @opindex missue-rate=@var{number}
11319 Issue @var{number} instructions per cycle. @var{number} can only be 1
11322 @item -mbranch-cost=@var{number}
11323 @opindex mbranch-cost=@var{number}
11324 @var{number} can only be 1 or 2. If it is 1 then branches will be
11325 preferred over conditional code, if it is 2, then the opposite will
11328 @item -mflush-trap=@var{number}
11329 @opindex mflush-trap=@var{number}
11330 Specifies the trap number to use to flush the cache. The default is
11331 12. Valid numbers are between 0 and 15 inclusive.
11333 @item -mno-flush-trap
11334 @opindex mno-flush-trap
11335 Specifies that the cache cannot be flushed by using a trap.
11337 @item -mflush-func=@var{name}
11338 @opindex mflush-func=@var{name}
11339 Specifies the name of the operating system function to call to flush
11340 the cache. The default is @emph{_flush_cache}, but a function call
11341 will only be used if a trap is not available.
11343 @item -mno-flush-func
11344 @opindex mno-flush-func
11345 Indicates that there is no OS function for flushing the cache.
11349 @node M680x0 Options
11350 @subsection M680x0 Options
11351 @cindex M680x0 options
11353 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11354 The default settings depend on which architecture was selected when
11355 the compiler was configured; the defaults for the most common choices
11359 @item -march=@var{arch}
11361 Generate code for a specific M680x0 or ColdFire instruction set
11362 architecture. Permissible values of @var{arch} for M680x0
11363 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11364 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11365 architectures are selected according to Freescale's ISA classification
11366 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11367 @samp{isab} and @samp{isac}.
11369 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11370 code for a ColdFire target. The @var{arch} in this macro is one of the
11371 @option{-march} arguments given above.
11373 When used together, @option{-march} and @option{-mtune} select code
11374 that runs on a family of similar processors but that is optimized
11375 for a particular microarchitecture.
11377 @item -mcpu=@var{cpu}
11379 Generate code for a specific M680x0 or ColdFire processor.
11380 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11381 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11382 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11383 below, which also classifies the CPUs into families:
11385 @multitable @columnfractions 0.20 0.80
11386 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11387 @item @samp{51qe} @tab @samp{51qe}
11388 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11389 @item @samp{5206e} @tab @samp{5206e}
11390 @item @samp{5208} @tab @samp{5207} @samp{5208}
11391 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11392 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11393 @item @samp{5216} @tab @samp{5214} @samp{5216}
11394 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11395 @item @samp{5225} @tab @samp{5224} @samp{5225}
11396 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11397 @item @samp{5249} @tab @samp{5249}
11398 @item @samp{5250} @tab @samp{5250}
11399 @item @samp{5271} @tab @samp{5270} @samp{5271}
11400 @item @samp{5272} @tab @samp{5272}
11401 @item @samp{5275} @tab @samp{5274} @samp{5275}
11402 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11403 @item @samp{5307} @tab @samp{5307}
11404 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11405 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11406 @item @samp{5407} @tab @samp{5407}
11407 @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}
11410 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11411 @var{arch} is compatible with @var{cpu}. Other combinations of
11412 @option{-mcpu} and @option{-march} are rejected.
11414 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11415 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11416 where the value of @var{family} is given by the table above.
11418 @item -mtune=@var{tune}
11420 Tune the code for a particular microarchitecture, within the
11421 constraints set by @option{-march} and @option{-mcpu}.
11422 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11423 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11424 and @samp{cpu32}. The ColdFire microarchitectures
11425 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11427 You can also use @option{-mtune=68020-40} for code that needs
11428 to run relatively well on 68020, 68030 and 68040 targets.
11429 @option{-mtune=68020-60} is similar but includes 68060 targets
11430 as well. These two options select the same tuning decisions as
11431 @option{-m68020-40} and @option{-m68020-60} respectively.
11433 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11434 when tuning for 680x0 architecture @var{arch}. It also defines
11435 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11436 option is used. If gcc is tuning for a range of architectures,
11437 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11438 it defines the macros for every architecture in the range.
11440 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11441 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11442 of the arguments given above.
11448 Generate output for a 68000. This is the default
11449 when the compiler is configured for 68000-based systems.
11450 It is equivalent to @option{-march=68000}.
11452 Use this option for microcontrollers with a 68000 or EC000 core,
11453 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11457 Generate output for a 68010. This is the default
11458 when the compiler is configured for 68010-based systems.
11459 It is equivalent to @option{-march=68010}.
11465 Generate output for a 68020. This is the default
11466 when the compiler is configured for 68020-based systems.
11467 It is equivalent to @option{-march=68020}.
11471 Generate output for a 68030. This is the default when the compiler is
11472 configured for 68030-based systems. It is equivalent to
11473 @option{-march=68030}.
11477 Generate output for a 68040. This is the default when the compiler is
11478 configured for 68040-based systems. It is equivalent to
11479 @option{-march=68040}.
11481 This option inhibits the use of 68881/68882 instructions that have to be
11482 emulated by software on the 68040. Use this option if your 68040 does not
11483 have code to emulate those instructions.
11487 Generate output for a 68060. This is the default when the compiler is
11488 configured for 68060-based systems. It is equivalent to
11489 @option{-march=68060}.
11491 This option inhibits the use of 68020 and 68881/68882 instructions that
11492 have to be emulated by software on the 68060. Use this option if your 68060
11493 does not have code to emulate those instructions.
11497 Generate output for a CPU32. This is the default
11498 when the compiler is configured for CPU32-based systems.
11499 It is equivalent to @option{-march=cpu32}.
11501 Use this option for microcontrollers with a
11502 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11503 68336, 68340, 68341, 68349 and 68360.
11507 Generate output for a 520X ColdFire CPU@. This is the default
11508 when the compiler is configured for 520X-based systems.
11509 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11510 in favor of that option.
11512 Use this option for microcontroller with a 5200 core, including
11513 the MCF5202, MCF5203, MCF5204 and MCF5206.
11517 Generate output for a 5206e ColdFire CPU@. The option is now
11518 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11522 Generate output for a member of the ColdFire 528X family.
11523 The option is now deprecated in favor of the equivalent
11524 @option{-mcpu=528x}.
11528 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
11529 in favor of the equivalent @option{-mcpu=5307}.
11533 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
11534 in favor of the equivalent @option{-mcpu=5407}.
11538 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11539 This includes use of hardware floating point instructions.
11540 The option is equivalent to @option{-mcpu=547x}, and is now
11541 deprecated in favor of that option.
11545 Generate output for a 68040, without using any of the new instructions.
11546 This results in code which can run relatively efficiently on either a
11547 68020/68881 or a 68030 or a 68040. The generated code does use the
11548 68881 instructions that are emulated on the 68040.
11550 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11554 Generate output for a 68060, without using any of the new instructions.
11555 This results in code which can run relatively efficiently on either a
11556 68020/68881 or a 68030 or a 68040. The generated code does use the
11557 68881 instructions that are emulated on the 68060.
11559 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11563 @opindex mhard-float
11565 Generate floating-point instructions. This is the default for 68020
11566 and above, and for ColdFire devices that have an FPU@. It defines the
11567 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11568 on ColdFire targets.
11571 @opindex msoft-float
11572 Do not generate floating-point instructions; use library calls instead.
11573 This is the default for 68000, 68010, and 68832 targets. It is also
11574 the default for ColdFire devices that have no FPU.
11580 Generate (do not generate) ColdFire hardware divide and remainder
11581 instructions. If @option{-march} is used without @option{-mcpu},
11582 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11583 architectures. Otherwise, the default is taken from the target CPU
11584 (either the default CPU, or the one specified by @option{-mcpu}). For
11585 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11586 @option{-mcpu=5206e}.
11588 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11592 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11593 Additionally, parameters passed on the stack are also aligned to a
11594 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11598 Do not consider type @code{int} to be 16 bits wide. This is the default.
11601 @itemx -mno-bitfield
11602 @opindex mnobitfield
11603 @opindex mno-bitfield
11604 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11605 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11609 Do use the bit-field instructions. The @option{-m68020} option implies
11610 @option{-mbitfield}. This is the default if you use a configuration
11611 designed for a 68020.
11615 Use a different function-calling convention, in which functions
11616 that take a fixed number of arguments return with the @code{rtd}
11617 instruction, which pops their arguments while returning. This
11618 saves one instruction in the caller since there is no need to pop
11619 the arguments there.
11621 This calling convention is incompatible with the one normally
11622 used on Unix, so you cannot use it if you need to call libraries
11623 compiled with the Unix compiler.
11625 Also, you must provide function prototypes for all functions that
11626 take variable numbers of arguments (including @code{printf});
11627 otherwise incorrect code will be generated for calls to those
11630 In addition, seriously incorrect code will result if you call a
11631 function with too many arguments. (Normally, extra arguments are
11632 harmlessly ignored.)
11634 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11635 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11639 Do not use the calling conventions selected by @option{-mrtd}.
11640 This is the default.
11643 @itemx -mno-align-int
11644 @opindex malign-int
11645 @opindex mno-align-int
11646 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11647 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11648 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11649 Aligning variables on 32-bit boundaries produces code that runs somewhat
11650 faster on processors with 32-bit busses at the expense of more memory.
11652 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11653 align structures containing the above types differently than
11654 most published application binary interface specifications for the m68k.
11658 Use the pc-relative addressing mode of the 68000 directly, instead of
11659 using a global offset table. At present, this option implies @option{-fpic},
11660 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11661 not presently supported with @option{-mpcrel}, though this could be supported for
11662 68020 and higher processors.
11664 @item -mno-strict-align
11665 @itemx -mstrict-align
11666 @opindex mno-strict-align
11667 @opindex mstrict-align
11668 Do not (do) assume that unaligned memory references will be handled by
11672 Generate code that allows the data segment to be located in a different
11673 area of memory from the text segment. This allows for execute in place in
11674 an environment without virtual memory management. This option implies
11677 @item -mno-sep-data
11678 Generate code that assumes that the data segment follows the text segment.
11679 This is the default.
11681 @item -mid-shared-library
11682 Generate code that supports shared libraries via the library ID method.
11683 This allows for execute in place and shared libraries in an environment
11684 without virtual memory management. This option implies @option{-fPIC}.
11686 @item -mno-id-shared-library
11687 Generate code that doesn't assume ID based shared libraries are being used.
11688 This is the default.
11690 @item -mshared-library-id=n
11691 Specified the identification number of the ID based shared library being
11692 compiled. Specifying a value of 0 will generate more compact code, specifying
11693 other values will force the allocation of that number to the current
11694 library but is no more space or time efficient than omitting this option.
11698 @node M68hc1x Options
11699 @subsection M68hc1x Options
11700 @cindex M68hc1x options
11702 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11703 microcontrollers. The default values for these options depends on
11704 which style of microcontroller was selected when the compiler was configured;
11705 the defaults for the most common choices are given below.
11712 Generate output for a 68HC11. This is the default
11713 when the compiler is configured for 68HC11-based systems.
11719 Generate output for a 68HC12. This is the default
11720 when the compiler is configured for 68HC12-based systems.
11726 Generate output for a 68HCS12.
11728 @item -mauto-incdec
11729 @opindex mauto-incdec
11730 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11737 Enable the use of 68HC12 min and max instructions.
11740 @itemx -mno-long-calls
11741 @opindex mlong-calls
11742 @opindex mno-long-calls
11743 Treat all calls as being far away (near). If calls are assumed to be
11744 far away, the compiler will use the @code{call} instruction to
11745 call a function and the @code{rtc} instruction for returning.
11749 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11751 @item -msoft-reg-count=@var{count}
11752 @opindex msoft-reg-count
11753 Specify the number of pseudo-soft registers which are used for the
11754 code generation. The maximum number is 32. Using more pseudo-soft
11755 register may or may not result in better code depending on the program.
11756 The default is 4 for 68HC11 and 2 for 68HC12.
11760 @node MCore Options
11761 @subsection MCore Options
11762 @cindex MCore options
11764 These are the @samp{-m} options defined for the Motorola M*Core
11770 @itemx -mno-hardlit
11772 @opindex mno-hardlit
11773 Inline constants into the code stream if it can be done in two
11774 instructions or less.
11780 Use the divide instruction. (Enabled by default).
11782 @item -mrelax-immediate
11783 @itemx -mno-relax-immediate
11784 @opindex mrelax-immediate
11785 @opindex mno-relax-immediate
11786 Allow arbitrary sized immediates in bit operations.
11788 @item -mwide-bitfields
11789 @itemx -mno-wide-bitfields
11790 @opindex mwide-bitfields
11791 @opindex mno-wide-bitfields
11792 Always treat bit-fields as int-sized.
11794 @item -m4byte-functions
11795 @itemx -mno-4byte-functions
11796 @opindex m4byte-functions
11797 @opindex mno-4byte-functions
11798 Force all functions to be aligned to a four byte boundary.
11800 @item -mcallgraph-data
11801 @itemx -mno-callgraph-data
11802 @opindex mcallgraph-data
11803 @opindex mno-callgraph-data
11804 Emit callgraph information.
11807 @itemx -mno-slow-bytes
11808 @opindex mslow-bytes
11809 @opindex mno-slow-bytes
11810 Prefer word access when reading byte quantities.
11812 @item -mlittle-endian
11813 @itemx -mbig-endian
11814 @opindex mlittle-endian
11815 @opindex mbig-endian
11816 Generate code for a little endian target.
11822 Generate code for the 210 processor.
11826 @subsection MIPS Options
11827 @cindex MIPS options
11833 Generate big-endian code.
11837 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11840 @item -march=@var{arch}
11842 Generate code that will run on @var{arch}, which can be the name of a
11843 generic MIPS ISA, or the name of a particular processor.
11845 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11846 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11847 The processor names are:
11848 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11849 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11850 @samp{5kc}, @samp{5kf},
11852 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11853 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11854 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11855 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11858 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11859 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11860 @samp{rm7000}, @samp{rm9000},
11863 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11864 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11865 The special value @samp{from-abi} selects the
11866 most compatible architecture for the selected ABI (that is,
11867 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11869 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11870 (for example, @samp{-march=r2k}). Prefixes are optional, and
11871 @samp{vr} may be written @samp{r}.
11873 Names of the form @samp{@var{n}f2_1} refer to processors with
11874 FPUs clocked at half the rate of the core, names of the form
11875 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11876 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11877 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11878 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11879 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11880 accepted as synonyms for @samp{@var{n}f1_1}.
11882 GCC defines two macros based on the value of this option. The first
11883 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11884 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11885 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11886 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11887 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11889 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11890 above. In other words, it will have the full prefix and will not
11891 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11892 the macro names the resolved architecture (either @samp{"mips1"} or
11893 @samp{"mips3"}). It names the default architecture when no
11894 @option{-march} option is given.
11896 @item -mtune=@var{arch}
11898 Optimize for @var{arch}. Among other things, this option controls
11899 the way instructions are scheduled, and the perceived cost of arithmetic
11900 operations. The list of @var{arch} values is the same as for
11903 When this option is not used, GCC will optimize for the processor
11904 specified by @option{-march}. By using @option{-march} and
11905 @option{-mtune} together, it is possible to generate code that will
11906 run on a family of processors, but optimize the code for one
11907 particular member of that family.
11909 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11910 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11911 @samp{-march} ones described above.
11915 Equivalent to @samp{-march=mips1}.
11919 Equivalent to @samp{-march=mips2}.
11923 Equivalent to @samp{-march=mips3}.
11927 Equivalent to @samp{-march=mips4}.
11931 Equivalent to @samp{-march=mips32}.
11935 Equivalent to @samp{-march=mips32r2}.
11939 Equivalent to @samp{-march=mips64}.
11944 @opindex mno-mips16
11945 Generate (do not generate) MIPS16 code. If GCC is targetting a
11946 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11948 MIPS16 code generation can also be controlled on a per-function basis
11949 by means of @code{mips16} and @code{nomips16} attributes.
11950 @xref{Function Attributes}, for more information.
11952 @item -mflip-mips16
11953 @opindex mflip-mips16
11954 Generate MIPS16 code on alternating functions. This option is provided
11955 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11956 not intended for ordinary use in compiling user code.
11958 @item -minterlink-mips16
11959 @itemx -mno-interlink-mips16
11960 @opindex minterlink-mips16
11961 @opindex mno-interlink-mips16
11962 Require (do not require) that non-MIPS16 code be link-compatible with
11965 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
11966 it must either use a call or an indirect jump. @option{-minterlink-mips16}
11967 therefore disables direct jumps unless GCC knows that the target of the
11968 jump is not MIPS16.
11980 Generate code for the given ABI@.
11982 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11983 generates 64-bit code when you select a 64-bit architecture, but you
11984 can use @option{-mgp32} to get 32-bit code instead.
11986 For information about the O64 ABI, see
11987 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11989 GCC supports a variant of the o32 ABI in which floating-point registers
11990 are 64 rather than 32 bits wide. You can select this combination with
11991 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11992 and @samp{mfhc1} instructions and is therefore only supported for
11993 MIPS32R2 processors.
11995 The register assignments for arguments and return values remain the
11996 same, but each scalar value is passed in a single 64-bit register
11997 rather than a pair of 32-bit registers. For example, scalar
11998 floating-point values are returned in @samp{$f0} only, not a
11999 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12000 remains the same, but all 64 bits are saved.
12003 @itemx -mno-abicalls
12005 @opindex mno-abicalls
12006 Generate (do not generate) code that is suitable for SVR4-style
12007 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12012 Generate (do not generate) code that is fully position-independent,
12013 and that can therefore be linked into shared libraries. This option
12014 only affects @option{-mabicalls}.
12016 All @option{-mabicalls} code has traditionally been position-independent,
12017 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12018 as an extension, the GNU toolchain allows executables to use absolute
12019 accesses for locally-binding symbols. It can also use shorter GP
12020 initialization sequences and generate direct calls to locally-defined
12021 functions. This mode is selected by @option{-mno-shared}.
12023 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12024 objects that can only be linked by the GNU linker. However, the option
12025 does not affect the ABI of the final executable; it only affects the ABI
12026 of relocatable objects. Using @option{-mno-shared} will generally make
12027 executables both smaller and quicker.
12029 @option{-mshared} is the default.
12035 Lift (do not lift) the usual restrictions on the size of the global
12038 GCC normally uses a single instruction to load values from the GOT@.
12039 While this is relatively efficient, it will only work if the GOT
12040 is smaller than about 64k. Anything larger will cause the linker
12041 to report an error such as:
12043 @cindex relocation truncated to fit (MIPS)
12045 relocation truncated to fit: R_MIPS_GOT16 foobar
12048 If this happens, you should recompile your code with @option{-mxgot}.
12049 It should then work with very large GOTs, although it will also be
12050 less efficient, since it will take three instructions to fetch the
12051 value of a global symbol.
12053 Note that some linkers can create multiple GOTs. If you have such a
12054 linker, you should only need to use @option{-mxgot} when a single object
12055 file accesses more than 64k's worth of GOT entries. Very few do.
12057 These options have no effect unless GCC is generating position
12062 Assume that general-purpose registers are 32 bits wide.
12066 Assume that general-purpose registers are 64 bits wide.
12070 Assume that floating-point registers are 32 bits wide.
12074 Assume that floating-point registers are 64 bits wide.
12077 @opindex mhard-float
12078 Use floating-point coprocessor instructions.
12081 @opindex msoft-float
12082 Do not use floating-point coprocessor instructions. Implement
12083 floating-point calculations using library calls instead.
12085 @item -msingle-float
12086 @opindex msingle-float
12087 Assume that the floating-point coprocessor only supports single-precision
12090 @item -mdouble-float
12091 @opindex mdouble-float
12092 Assume that the floating-point coprocessor supports double-precision
12093 operations. This is the default.
12099 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12100 implement atomic memory built-in functions. When neither option is
12101 specified, GCC will use the instructions if the target architecture
12104 @option{-mllsc} is useful if the runtime environment can emulate the
12105 instructions and @option{-mno-llsc} can be useful when compiling for
12106 nonstandard ISAs. You can make either option the default by
12107 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12108 respectively. @option{--with-llsc} is the default for some
12109 configurations; see the installation documentation for details.
12115 Use (do not use) revision 1 of the MIPS DSP ASE@.
12116 @xref{MIPS DSP Built-in Functions}. This option defines the
12117 preprocessor macro @samp{__mips_dsp}. It also defines
12118 @samp{__mips_dsp_rev} to 1.
12124 Use (do not use) revision 2 of the MIPS DSP ASE@.
12125 @xref{MIPS DSP Built-in Functions}. This option defines the
12126 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12127 It also defines @samp{__mips_dsp_rev} to 2.
12130 @itemx -mno-smartmips
12131 @opindex msmartmips
12132 @opindex mno-smartmips
12133 Use (do not use) the MIPS SmartMIPS ASE.
12135 @item -mpaired-single
12136 @itemx -mno-paired-single
12137 @opindex mpaired-single
12138 @opindex mno-paired-single
12139 Use (do not use) paired-single floating-point instructions.
12140 @xref{MIPS Paired-Single Support}. This option requires
12141 hardware floating-point support to be enabled.
12147 Use (do not use) MIPS Digital Media Extension instructions.
12148 This option can only be used when generating 64-bit code and requires
12149 hardware floating-point support to be enabled.
12154 @opindex mno-mips3d
12155 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12156 The option @option{-mips3d} implies @option{-mpaired-single}.
12162 Use (do not use) MT Multithreading instructions.
12166 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12167 an explanation of the default and the way that the pointer size is
12172 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12174 The default size of @code{int}s, @code{long}s and pointers depends on
12175 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12176 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12177 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12178 or the same size as integer registers, whichever is smaller.
12184 Assume (do not assume) that all symbols have 32-bit values, regardless
12185 of the selected ABI@. This option is useful in combination with
12186 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12187 to generate shorter and faster references to symbolic addresses.
12191 Put definitions of externally-visible data in a small data section
12192 if that data is no bigger than @var{num} bytes. GCC can then access
12193 the data more efficiently; see @option{-mgpopt} for details.
12195 The default @option{-G} option depends on the configuration.
12197 @item -mlocal-sdata
12198 @itemx -mno-local-sdata
12199 @opindex mlocal-sdata
12200 @opindex mno-local-sdata
12201 Extend (do not extend) the @option{-G} behavior to local data too,
12202 such as to static variables in C@. @option{-mlocal-sdata} is the
12203 default for all configurations.
12205 If the linker complains that an application is using too much small data,
12206 you might want to try rebuilding the less performance-critical parts with
12207 @option{-mno-local-sdata}. You might also want to build large
12208 libraries with @option{-mno-local-sdata}, so that the libraries leave
12209 more room for the main program.
12211 @item -mextern-sdata
12212 @itemx -mno-extern-sdata
12213 @opindex mextern-sdata
12214 @opindex mno-extern-sdata
12215 Assume (do not assume) that externally-defined data will be in
12216 a small data section if that data is within the @option{-G} limit.
12217 @option{-mextern-sdata} is the default for all configurations.
12219 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12220 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12221 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12222 is placed in a small data section. If @var{Var} is defined by another
12223 module, you must either compile that module with a high-enough
12224 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12225 definition. If @var{Var} is common, you must link the application
12226 with a high-enough @option{-G} setting.
12228 The easiest way of satisfying these restrictions is to compile
12229 and link every module with the same @option{-G} option. However,
12230 you may wish to build a library that supports several different
12231 small data limits. You can do this by compiling the library with
12232 the highest supported @option{-G} setting and additionally using
12233 @option{-mno-extern-sdata} to stop the library from making assumptions
12234 about externally-defined data.
12240 Use (do not use) GP-relative accesses for symbols that are known to be
12241 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12242 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12245 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12246 might not hold the value of @code{_gp}. For example, if the code is
12247 part of a library that might be used in a boot monitor, programs that
12248 call boot monitor routines will pass an unknown value in @code{$gp}.
12249 (In such situations, the boot monitor itself would usually be compiled
12250 with @option{-G0}.)
12252 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12253 @option{-mno-extern-sdata}.
12255 @item -membedded-data
12256 @itemx -mno-embedded-data
12257 @opindex membedded-data
12258 @opindex mno-embedded-data
12259 Allocate variables to the read-only data section first if possible, then
12260 next in the small data section if possible, otherwise in data. This gives
12261 slightly slower code than the default, but reduces the amount of RAM required
12262 when executing, and thus may be preferred for some embedded systems.
12264 @item -muninit-const-in-rodata
12265 @itemx -mno-uninit-const-in-rodata
12266 @opindex muninit-const-in-rodata
12267 @opindex mno-uninit-const-in-rodata
12268 Put uninitialized @code{const} variables in the read-only data section.
12269 This option is only meaningful in conjunction with @option{-membedded-data}.
12271 @item -mcode-readable=@var{setting}
12272 @opindex mcode-readable
12273 Specify whether GCC may generate code that reads from executable sections.
12274 There are three possible settings:
12277 @item -mcode-readable=yes
12278 Instructions may freely access executable sections. This is the
12281 @item -mcode-readable=pcrel
12282 MIPS16 PC-relative load instructions can access executable sections,
12283 but other instructions must not do so. This option is useful on 4KSc
12284 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12285 It is also useful on processors that can be configured to have a dual
12286 instruction/data SRAM interface and that, like the M4K, automatically
12287 redirect PC-relative loads to the instruction RAM.
12289 @item -mcode-readable=no
12290 Instructions must not access executable sections. This option can be
12291 useful on targets that are configured to have a dual instruction/data
12292 SRAM interface but that (unlike the M4K) do not automatically redirect
12293 PC-relative loads to the instruction RAM.
12296 @item -msplit-addresses
12297 @itemx -mno-split-addresses
12298 @opindex msplit-addresses
12299 @opindex mno-split-addresses
12300 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12301 relocation operators. This option has been superseded by
12302 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12304 @item -mexplicit-relocs
12305 @itemx -mno-explicit-relocs
12306 @opindex mexplicit-relocs
12307 @opindex mno-explicit-relocs
12308 Use (do not use) assembler relocation operators when dealing with symbolic
12309 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12310 is to use assembler macros instead.
12312 @option{-mexplicit-relocs} is the default if GCC was configured
12313 to use an assembler that supports relocation operators.
12315 @item -mcheck-zero-division
12316 @itemx -mno-check-zero-division
12317 @opindex mcheck-zero-division
12318 @opindex mno-check-zero-division
12319 Trap (do not trap) on integer division by zero.
12321 The default is @option{-mcheck-zero-division}.
12323 @item -mdivide-traps
12324 @itemx -mdivide-breaks
12325 @opindex mdivide-traps
12326 @opindex mdivide-breaks
12327 MIPS systems check for division by zero by generating either a
12328 conditional trap or a break instruction. Using traps results in
12329 smaller code, but is only supported on MIPS II and later. Also, some
12330 versions of the Linux kernel have a bug that prevents trap from
12331 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12332 allow conditional traps on architectures that support them and
12333 @option{-mdivide-breaks} to force the use of breaks.
12335 The default is usually @option{-mdivide-traps}, but this can be
12336 overridden at configure time using @option{--with-divide=breaks}.
12337 Divide-by-zero checks can be completely disabled using
12338 @option{-mno-check-zero-division}.
12343 @opindex mno-memcpy
12344 Force (do not force) the use of @code{memcpy()} for non-trivial block
12345 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12346 most constant-sized copies.
12349 @itemx -mno-long-calls
12350 @opindex mlong-calls
12351 @opindex mno-long-calls
12352 Disable (do not disable) use of the @code{jal} instruction. Calling
12353 functions using @code{jal} is more efficient but requires the caller
12354 and callee to be in the same 256 megabyte segment.
12356 This option has no effect on abicalls code. The default is
12357 @option{-mno-long-calls}.
12363 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12364 instructions, as provided by the R4650 ISA@.
12367 @itemx -mno-fused-madd
12368 @opindex mfused-madd
12369 @opindex mno-fused-madd
12370 Enable (disable) use of the floating point multiply-accumulate
12371 instructions, when they are available. The default is
12372 @option{-mfused-madd}.
12374 When multiply-accumulate instructions are used, the intermediate
12375 product is calculated to infinite precision and is not subject to
12376 the FCSR Flush to Zero bit. This may be undesirable in some
12381 Tell the MIPS assembler to not run its preprocessor over user
12382 assembler files (with a @samp{.s} suffix) when assembling them.
12385 @itemx -mno-fix-r4000
12386 @opindex mfix-r4000
12387 @opindex mno-fix-r4000
12388 Work around certain R4000 CPU errata:
12391 A double-word or a variable shift may give an incorrect result if executed
12392 immediately after starting an integer division.
12394 A double-word or a variable shift may give an incorrect result if executed
12395 while an integer multiplication is in progress.
12397 An integer division may give an incorrect result if started in a delay slot
12398 of a taken branch or a jump.
12402 @itemx -mno-fix-r4400
12403 @opindex mfix-r4400
12404 @opindex mno-fix-r4400
12405 Work around certain R4400 CPU errata:
12408 A double-word or a variable shift may give an incorrect result if executed
12409 immediately after starting an integer division.
12413 @itemx -mno-fix-vr4120
12414 @opindex mfix-vr4120
12415 Work around certain VR4120 errata:
12418 @code{dmultu} does not always produce the correct result.
12420 @code{div} and @code{ddiv} do not always produce the correct result if one
12421 of the operands is negative.
12423 The workarounds for the division errata rely on special functions in
12424 @file{libgcc.a}. At present, these functions are only provided by
12425 the @code{mips64vr*-elf} configurations.
12427 Other VR4120 errata require a nop to be inserted between certain pairs of
12428 instructions. These errata are handled by the assembler, not by GCC itself.
12431 @opindex mfix-vr4130
12432 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12433 workarounds are implemented by the assembler rather than by GCC,
12434 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12435 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12436 instructions are available instead.
12439 @itemx -mno-fix-sb1
12441 Work around certain SB-1 CPU core errata.
12442 (This flag currently works around the SB-1 revision 2
12443 ``F1'' and ``F2'' floating point errata.)
12445 @item -mflush-func=@var{func}
12446 @itemx -mno-flush-func
12447 @opindex mflush-func
12448 Specifies the function to call to flush the I and D caches, or to not
12449 call any such function. If called, the function must take the same
12450 arguments as the common @code{_flush_func()}, that is, the address of the
12451 memory range for which the cache is being flushed, the size of the
12452 memory range, and the number 3 (to flush both caches). The default
12453 depends on the target GCC was configured for, but commonly is either
12454 @samp{_flush_func} or @samp{__cpu_flush}.
12456 @item mbranch-cost=@var{num}
12457 @opindex mbranch-cost
12458 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12459 This cost is only a heuristic and is not guaranteed to produce
12460 consistent results across releases. A zero cost redundantly selects
12461 the default, which is based on the @option{-mtune} setting.
12463 @item -mbranch-likely
12464 @itemx -mno-branch-likely
12465 @opindex mbranch-likely
12466 @opindex mno-branch-likely
12467 Enable or disable use of Branch Likely instructions, regardless of the
12468 default for the selected architecture. By default, Branch Likely
12469 instructions may be generated if they are supported by the selected
12470 architecture. An exception is for the MIPS32 and MIPS64 architectures
12471 and processors which implement those architectures; for those, Branch
12472 Likely instructions will not be generated by default because the MIPS32
12473 and MIPS64 architectures specifically deprecate their use.
12475 @item -mfp-exceptions
12476 @itemx -mno-fp-exceptions
12477 @opindex mfp-exceptions
12478 Specifies whether FP exceptions are enabled. This affects how we schedule
12479 FP instructions for some processors. The default is that FP exceptions are
12482 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12483 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12486 @item -mvr4130-align
12487 @itemx -mno-vr4130-align
12488 @opindex mvr4130-align
12489 The VR4130 pipeline is two-way superscalar, but can only issue two
12490 instructions together if the first one is 8-byte aligned. When this
12491 option is enabled, GCC will align pairs of instructions that it
12492 thinks should execute in parallel.
12494 This option only has an effect when optimizing for the VR4130.
12495 It normally makes code faster, but at the expense of making it bigger.
12496 It is enabled by default at optimization level @option{-O3}.
12500 @subsection MMIX Options
12501 @cindex MMIX Options
12503 These options are defined for the MMIX:
12507 @itemx -mno-libfuncs
12509 @opindex mno-libfuncs
12510 Specify that intrinsic library functions are being compiled, passing all
12511 values in registers, no matter the size.
12514 @itemx -mno-epsilon
12516 @opindex mno-epsilon
12517 Generate floating-point comparison instructions that compare with respect
12518 to the @code{rE} epsilon register.
12520 @item -mabi=mmixware
12522 @opindex mabi-mmixware
12524 Generate code that passes function parameters and return values that (in
12525 the called function) are seen as registers @code{$0} and up, as opposed to
12526 the GNU ABI which uses global registers @code{$231} and up.
12528 @item -mzero-extend
12529 @itemx -mno-zero-extend
12530 @opindex mzero-extend
12531 @opindex mno-zero-extend
12532 When reading data from memory in sizes shorter than 64 bits, use (do not
12533 use) zero-extending load instructions by default, rather than
12534 sign-extending ones.
12537 @itemx -mno-knuthdiv
12539 @opindex mno-knuthdiv
12540 Make the result of a division yielding a remainder have the same sign as
12541 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12542 remainder follows the sign of the dividend. Both methods are
12543 arithmetically valid, the latter being almost exclusively used.
12545 @item -mtoplevel-symbols
12546 @itemx -mno-toplevel-symbols
12547 @opindex mtoplevel-symbols
12548 @opindex mno-toplevel-symbols
12549 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12550 code can be used with the @code{PREFIX} assembly directive.
12554 Generate an executable in the ELF format, rather than the default
12555 @samp{mmo} format used by the @command{mmix} simulator.
12557 @item -mbranch-predict
12558 @itemx -mno-branch-predict
12559 @opindex mbranch-predict
12560 @opindex mno-branch-predict
12561 Use (do not use) the probable-branch instructions, when static branch
12562 prediction indicates a probable branch.
12564 @item -mbase-addresses
12565 @itemx -mno-base-addresses
12566 @opindex mbase-addresses
12567 @opindex mno-base-addresses
12568 Generate (do not generate) code that uses @emph{base addresses}. Using a
12569 base address automatically generates a request (handled by the assembler
12570 and the linker) for a constant to be set up in a global register. The
12571 register is used for one or more base address requests within the range 0
12572 to 255 from the value held in the register. The generally leads to short
12573 and fast code, but the number of different data items that can be
12574 addressed is limited. This means that a program that uses lots of static
12575 data may require @option{-mno-base-addresses}.
12577 @item -msingle-exit
12578 @itemx -mno-single-exit
12579 @opindex msingle-exit
12580 @opindex mno-single-exit
12581 Force (do not force) generated code to have a single exit point in each
12585 @node MN10300 Options
12586 @subsection MN10300 Options
12587 @cindex MN10300 options
12589 These @option{-m} options are defined for Matsushita MN10300 architectures:
12594 Generate code to avoid bugs in the multiply instructions for the MN10300
12595 processors. This is the default.
12597 @item -mno-mult-bug
12598 @opindex mno-mult-bug
12599 Do not generate code to avoid bugs in the multiply instructions for the
12600 MN10300 processors.
12604 Generate code which uses features specific to the AM33 processor.
12608 Do not generate code which uses features specific to the AM33 processor. This
12611 @item -mreturn-pointer-on-d0
12612 @opindex mreturn-pointer-on-d0
12613 When generating a function which returns a pointer, return the pointer
12614 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12615 only in a0, and attempts to call such functions without a prototype
12616 would result in errors. Note that this option is on by default; use
12617 @option{-mno-return-pointer-on-d0} to disable it.
12621 Do not link in the C run-time initialization object file.
12625 Indicate to the linker that it should perform a relaxation optimization pass
12626 to shorten branches, calls and absolute memory addresses. This option only
12627 has an effect when used on the command line for the final link step.
12629 This option makes symbolic debugging impossible.
12633 @subsection MT Options
12636 These @option{-m} options are defined for Morpho MT architectures:
12640 @item -march=@var{cpu-type}
12642 Generate code that will run on @var{cpu-type}, which is the name of a system
12643 representing a certain processor type. Possible values for
12644 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12645 @samp{ms1-16-003} and @samp{ms2}.
12647 When this option is not used, the default is @option{-march=ms1-16-002}.
12651 Use byte loads and stores when generating code.
12655 Do not use byte loads and stores when generating code.
12659 Use simulator runtime
12663 Do not link in the C run-time initialization object file
12664 @file{crti.o}. Other run-time initialization and termination files
12665 such as @file{startup.o} and @file{exit.o} are still included on the
12666 linker command line.
12670 @node PDP-11 Options
12671 @subsection PDP-11 Options
12672 @cindex PDP-11 Options
12674 These options are defined for the PDP-11:
12679 Use hardware FPP floating point. This is the default. (FIS floating
12680 point on the PDP-11/40 is not supported.)
12683 @opindex msoft-float
12684 Do not use hardware floating point.
12688 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12692 Return floating-point results in memory. This is the default.
12696 Generate code for a PDP-11/40.
12700 Generate code for a PDP-11/45. This is the default.
12704 Generate code for a PDP-11/10.
12706 @item -mbcopy-builtin
12707 @opindex bcopy-builtin
12708 Use inline @code{movmemhi} patterns for copying memory. This is the
12713 Do not use inline @code{movmemhi} patterns for copying memory.
12719 Use 16-bit @code{int}. This is the default.
12725 Use 32-bit @code{int}.
12728 @itemx -mno-float32
12730 @opindex mno-float32
12731 Use 64-bit @code{float}. This is the default.
12734 @itemx -mno-float64
12736 @opindex mno-float64
12737 Use 32-bit @code{float}.
12741 Use @code{abshi2} pattern. This is the default.
12745 Do not use @code{abshi2} pattern.
12747 @item -mbranch-expensive
12748 @opindex mbranch-expensive
12749 Pretend that branches are expensive. This is for experimenting with
12750 code generation only.
12752 @item -mbranch-cheap
12753 @opindex mbranch-cheap
12754 Do not pretend that branches are expensive. This is the default.
12758 Generate code for a system with split I&D@.
12762 Generate code for a system without split I&D@. This is the default.
12766 Use Unix assembler syntax. This is the default when configured for
12767 @samp{pdp11-*-bsd}.
12771 Use DEC assembler syntax. This is the default when configured for any
12772 PDP-11 target other than @samp{pdp11-*-bsd}.
12775 @node PowerPC Options
12776 @subsection PowerPC Options
12777 @cindex PowerPC options
12779 These are listed under @xref{RS/6000 and PowerPC Options}.
12781 @node RS/6000 and PowerPC Options
12782 @subsection IBM RS/6000 and PowerPC Options
12783 @cindex RS/6000 and PowerPC Options
12784 @cindex IBM RS/6000 and PowerPC Options
12786 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12793 @itemx -mno-powerpc
12794 @itemx -mpowerpc-gpopt
12795 @itemx -mno-powerpc-gpopt
12796 @itemx -mpowerpc-gfxopt
12797 @itemx -mno-powerpc-gfxopt
12799 @itemx -mno-powerpc64
12803 @itemx -mno-popcntb
12811 @itemx -mno-hard-dfp
12815 @opindex mno-power2
12817 @opindex mno-powerpc
12818 @opindex mpowerpc-gpopt
12819 @opindex mno-powerpc-gpopt
12820 @opindex mpowerpc-gfxopt
12821 @opindex mno-powerpc-gfxopt
12822 @opindex mpowerpc64
12823 @opindex mno-powerpc64
12827 @opindex mno-popcntb
12833 @opindex mno-mfpgpr
12835 @opindex mno-hard-dfp
12836 GCC supports two related instruction set architectures for the
12837 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12838 instructions supported by the @samp{rios} chip set used in the original
12839 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12840 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12841 the IBM 4xx, 6xx, and follow-on microprocessors.
12843 Neither architecture is a subset of the other. However there is a
12844 large common subset of instructions supported by both. An MQ
12845 register is included in processors supporting the POWER architecture.
12847 You use these options to specify which instructions are available on the
12848 processor you are using. The default value of these options is
12849 determined when configuring GCC@. Specifying the
12850 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12851 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12852 rather than the options listed above.
12854 The @option{-mpower} option allows GCC to generate instructions that
12855 are found only in the POWER architecture and to use the MQ register.
12856 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12857 to generate instructions that are present in the POWER2 architecture but
12858 not the original POWER architecture.
12860 The @option{-mpowerpc} option allows GCC to generate instructions that
12861 are found only in the 32-bit subset of the PowerPC architecture.
12862 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12863 GCC to use the optional PowerPC architecture instructions in the
12864 General Purpose group, including floating-point square root. Specifying
12865 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12866 use the optional PowerPC architecture instructions in the Graphics
12867 group, including floating-point select.
12869 The @option{-mmfcrf} option allows GCC to generate the move from
12870 condition register field instruction implemented on the POWER4
12871 processor and other processors that support the PowerPC V2.01
12873 The @option{-mpopcntb} option allows GCC to generate the popcount and
12874 double precision FP reciprocal estimate instruction implemented on the
12875 POWER5 processor and other processors that support the PowerPC V2.02
12877 The @option{-mfprnd} option allows GCC to generate the FP round to
12878 integer instructions implemented on the POWER5+ processor and other
12879 processors that support the PowerPC V2.03 architecture.
12880 The @option{-mcmpb} option allows GCC to generate the compare bytes
12881 instruction implemented on the POWER6 processor and other processors
12882 that support the PowerPC V2.05 architecture.
12883 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12884 general purpose register instructions implemented on the POWER6X
12885 processor and other processors that support the extended PowerPC V2.05
12887 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12888 point instructions implemented on some POWER processors.
12890 The @option{-mpowerpc64} option allows GCC to generate the additional
12891 64-bit instructions that are found in the full PowerPC64 architecture
12892 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12893 @option{-mno-powerpc64}.
12895 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12896 will use only the instructions in the common subset of both
12897 architectures plus some special AIX common-mode calls, and will not use
12898 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12899 permits GCC to use any instruction from either architecture and to
12900 allow use of the MQ register; specify this for the Motorola MPC601.
12902 @item -mnew-mnemonics
12903 @itemx -mold-mnemonics
12904 @opindex mnew-mnemonics
12905 @opindex mold-mnemonics
12906 Select which mnemonics to use in the generated assembler code. With
12907 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12908 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12909 assembler mnemonics defined for the POWER architecture. Instructions
12910 defined in only one architecture have only one mnemonic; GCC uses that
12911 mnemonic irrespective of which of these options is specified.
12913 GCC defaults to the mnemonics appropriate for the architecture in
12914 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12915 value of these option. Unless you are building a cross-compiler, you
12916 should normally not specify either @option{-mnew-mnemonics} or
12917 @option{-mold-mnemonics}, but should instead accept the default.
12919 @item -mcpu=@var{cpu_type}
12921 Set architecture type, register usage, choice of mnemonics, and
12922 instruction scheduling parameters for machine type @var{cpu_type}.
12923 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12924 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12925 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12926 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12927 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12928 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12929 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12930 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12931 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12932 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12934 @option{-mcpu=common} selects a completely generic processor. Code
12935 generated under this option will run on any POWER or PowerPC processor.
12936 GCC will use only the instructions in the common subset of both
12937 architectures, and will not use the MQ register. GCC assumes a generic
12938 processor model for scheduling purposes.
12940 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12941 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12942 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12943 types, with an appropriate, generic processor model assumed for
12944 scheduling purposes.
12946 The other options specify a specific processor. Code generated under
12947 those options will run best on that processor, and may not run at all on
12950 The @option{-mcpu} options automatically enable or disable the
12953 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12954 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12955 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12957 The particular options set for any particular CPU will vary between
12958 compiler versions, depending on what setting seems to produce optimal
12959 code for that CPU; it doesn't necessarily reflect the actual hardware's
12960 capabilities. If you wish to set an individual option to a particular
12961 value, you may specify it after the @option{-mcpu} option, like
12962 @samp{-mcpu=970 -mno-altivec}.
12964 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12965 not enabled or disabled by the @option{-mcpu} option at present because
12966 AIX does not have full support for these options. You may still
12967 enable or disable them individually if you're sure it'll work in your
12970 @item -mtune=@var{cpu_type}
12972 Set the instruction scheduling parameters for machine type
12973 @var{cpu_type}, but do not set the architecture type, register usage, or
12974 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12975 values for @var{cpu_type} are used for @option{-mtune} as for
12976 @option{-mcpu}. If both are specified, the code generated will use the
12977 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12978 scheduling parameters set by @option{-mtune}.
12984 Generate code to compute division as reciprocal estimate and iterative
12985 refinement, creating opportunities for increased throughput. This
12986 feature requires: optional PowerPC Graphics instruction set for single
12987 precision and FRE instruction for double precision, assuming divides
12988 cannot generate user-visible traps, and the domain values not include
12989 Infinities, denormals or zero denominator.
12992 @itemx -mno-altivec
12994 @opindex mno-altivec
12995 Generate code that uses (does not use) AltiVec instructions, and also
12996 enable the use of built-in functions that allow more direct access to
12997 the AltiVec instruction set. You may also need to set
12998 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13004 @opindex mno-vrsave
13005 Generate VRSAVE instructions when generating AltiVec code.
13008 @opindex msecure-plt
13009 Generate code that allows ld and ld.so to build executables and shared
13010 libraries with non-exec .plt and .got sections. This is a PowerPC
13011 32-bit SYSV ABI option.
13015 Generate code that uses a BSS .plt section that ld.so fills in, and
13016 requires .plt and .got sections that are both writable and executable.
13017 This is a PowerPC 32-bit SYSV ABI option.
13023 This switch enables or disables the generation of ISEL instructions.
13025 @item -misel=@var{yes/no}
13026 This switch has been deprecated. Use @option{-misel} and
13027 @option{-mno-isel} instead.
13033 This switch enables or disables the generation of SPE simd
13039 @opindex mno-paired
13040 This switch enables or disables the generation of PAIRED simd
13043 @item -mspe=@var{yes/no}
13044 This option has been deprecated. Use @option{-mspe} and
13045 @option{-mno-spe} instead.
13047 @item -mfloat-gprs=@var{yes/single/double/no}
13048 @itemx -mfloat-gprs
13049 @opindex mfloat-gprs
13050 This switch enables or disables the generation of floating point
13051 operations on the general purpose registers for architectures that
13054 The argument @var{yes} or @var{single} enables the use of
13055 single-precision floating point operations.
13057 The argument @var{double} enables the use of single and
13058 double-precision floating point operations.
13060 The argument @var{no} disables floating point operations on the
13061 general purpose registers.
13063 This option is currently only available on the MPC854x.
13069 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13070 targets (including GNU/Linux). The 32-bit environment sets int, long
13071 and pointer to 32 bits and generates code that runs on any PowerPC
13072 variant. The 64-bit environment sets int to 32 bits and long and
13073 pointer to 64 bits, and generates code for PowerPC64, as for
13074 @option{-mpowerpc64}.
13077 @itemx -mno-fp-in-toc
13078 @itemx -mno-sum-in-toc
13079 @itemx -mminimal-toc
13081 @opindex mno-fp-in-toc
13082 @opindex mno-sum-in-toc
13083 @opindex mminimal-toc
13084 Modify generation of the TOC (Table Of Contents), which is created for
13085 every executable file. The @option{-mfull-toc} option is selected by
13086 default. In that case, GCC will allocate at least one TOC entry for
13087 each unique non-automatic variable reference in your program. GCC
13088 will also place floating-point constants in the TOC@. However, only
13089 16,384 entries are available in the TOC@.
13091 If you receive a linker error message that saying you have overflowed
13092 the available TOC space, you can reduce the amount of TOC space used
13093 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13094 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13095 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13096 generate code to calculate the sum of an address and a constant at
13097 run-time instead of putting that sum into the TOC@. You may specify one
13098 or both of these options. Each causes GCC to produce very slightly
13099 slower and larger code at the expense of conserving TOC space.
13101 If you still run out of space in the TOC even when you specify both of
13102 these options, specify @option{-mminimal-toc} instead. This option causes
13103 GCC to make only one TOC entry for every file. When you specify this
13104 option, GCC will produce code that is slower and larger but which
13105 uses extremely little TOC space. You may wish to use this option
13106 only on files that contain less frequently executed code.
13112 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13113 @code{long} type, and the infrastructure needed to support them.
13114 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13115 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13116 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13119 @itemx -mno-xl-compat
13120 @opindex mxl-compat
13121 @opindex mno-xl-compat
13122 Produce code that conforms more closely to IBM XL compiler semantics
13123 when using AIX-compatible ABI@. Pass floating-point arguments to
13124 prototyped functions beyond the register save area (RSA) on the stack
13125 in addition to argument FPRs. Do not assume that most significant
13126 double in 128-bit long double value is properly rounded when comparing
13127 values and converting to double. Use XL symbol names for long double
13130 The AIX calling convention was extended but not initially documented to
13131 handle an obscure K&R C case of calling a function that takes the
13132 address of its arguments with fewer arguments than declared. IBM XL
13133 compilers access floating point arguments which do not fit in the
13134 RSA from the stack when a subroutine is compiled without
13135 optimization. Because always storing floating-point arguments on the
13136 stack is inefficient and rarely needed, this option is not enabled by
13137 default and only is necessary when calling subroutines compiled by IBM
13138 XL compilers without optimization.
13142 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13143 application written to use message passing with special startup code to
13144 enable the application to run. The system must have PE installed in the
13145 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13146 must be overridden with the @option{-specs=} option to specify the
13147 appropriate directory location. The Parallel Environment does not
13148 support threads, so the @option{-mpe} option and the @option{-pthread}
13149 option are incompatible.
13151 @item -malign-natural
13152 @itemx -malign-power
13153 @opindex malign-natural
13154 @opindex malign-power
13155 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13156 @option{-malign-natural} overrides the ABI-defined alignment of larger
13157 types, such as floating-point doubles, on their natural size-based boundary.
13158 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13159 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13161 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13165 @itemx -mhard-float
13166 @opindex msoft-float
13167 @opindex mhard-float
13168 Generate code that does not use (uses) the floating-point register set.
13169 Software floating point emulation is provided if you use the
13170 @option{-msoft-float} option, and pass the option to GCC when linking.
13173 @itemx -mno-multiple
13175 @opindex mno-multiple
13176 Generate code that uses (does not use) the load multiple word
13177 instructions and the store multiple word instructions. These
13178 instructions are generated by default on POWER systems, and not
13179 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13180 endian PowerPC systems, since those instructions do not work when the
13181 processor is in little endian mode. The exceptions are PPC740 and
13182 PPC750 which permit the instructions usage in little endian mode.
13187 @opindex mno-string
13188 Generate code that uses (does not use) the load string instructions
13189 and the store string word instructions to save multiple registers and
13190 do small block moves. These instructions are generated by default on
13191 POWER systems, and not generated on PowerPC systems. Do not use
13192 @option{-mstring} on little endian PowerPC systems, since those
13193 instructions do not work when the processor is in little endian mode.
13194 The exceptions are PPC740 and PPC750 which permit the instructions
13195 usage in little endian mode.
13200 @opindex mno-update
13201 Generate code that uses (does not use) the load or store instructions
13202 that update the base register to the address of the calculated memory
13203 location. These instructions are generated by default. If you use
13204 @option{-mno-update}, there is a small window between the time that the
13205 stack pointer is updated and the address of the previous frame is
13206 stored, which means code that walks the stack frame across interrupts or
13207 signals may get corrupted data.
13210 @itemx -mno-fused-madd
13211 @opindex mfused-madd
13212 @opindex mno-fused-madd
13213 Generate code that uses (does not use) the floating point multiply and
13214 accumulate instructions. These instructions are generated by default if
13215 hardware floating is used.
13221 Generate code that uses (does not use) the half-word multiply and
13222 multiply-accumulate instructions on the IBM 405 and 440 processors.
13223 These instructions are generated by default when targetting those
13230 Generate code that uses (does not use) the string-search @samp{dlmzb}
13231 instruction on the IBM 405 and 440 processors. This instruction is
13232 generated by default when targetting those processors.
13234 @item -mno-bit-align
13236 @opindex mno-bit-align
13237 @opindex mbit-align
13238 On System V.4 and embedded PowerPC systems do not (do) force structures
13239 and unions that contain bit-fields to be aligned to the base type of the
13242 For example, by default a structure containing nothing but 8
13243 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13244 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13245 the structure would be aligned to a 1 byte boundary and be one byte in
13248 @item -mno-strict-align
13249 @itemx -mstrict-align
13250 @opindex mno-strict-align
13251 @opindex mstrict-align
13252 On System V.4 and embedded PowerPC systems do not (do) assume that
13253 unaligned memory references will be handled by the system.
13255 @item -mrelocatable
13256 @itemx -mno-relocatable
13257 @opindex mrelocatable
13258 @opindex mno-relocatable
13259 On embedded PowerPC systems generate code that allows (does not allow)
13260 the program to be relocated to a different address at runtime. If you
13261 use @option{-mrelocatable} on any module, all objects linked together must
13262 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13264 @item -mrelocatable-lib
13265 @itemx -mno-relocatable-lib
13266 @opindex mrelocatable-lib
13267 @opindex mno-relocatable-lib
13268 On embedded PowerPC systems generate code that allows (does not allow)
13269 the program to be relocated to a different address at runtime. Modules
13270 compiled with @option{-mrelocatable-lib} can be linked with either modules
13271 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13272 with modules compiled with the @option{-mrelocatable} options.
13278 On System V.4 and embedded PowerPC systems do not (do) assume that
13279 register 2 contains a pointer to a global area pointing to the addresses
13280 used in the program.
13283 @itemx -mlittle-endian
13285 @opindex mlittle-endian
13286 On System V.4 and embedded PowerPC systems compile code for the
13287 processor in little endian mode. The @option{-mlittle-endian} option is
13288 the same as @option{-mlittle}.
13291 @itemx -mbig-endian
13293 @opindex mbig-endian
13294 On System V.4 and embedded PowerPC systems compile code for the
13295 processor in big endian mode. The @option{-mbig-endian} option is
13296 the same as @option{-mbig}.
13298 @item -mdynamic-no-pic
13299 @opindex mdynamic-no-pic
13300 On Darwin and Mac OS X systems, compile code so that it is not
13301 relocatable, but that its external references are relocatable. The
13302 resulting code is suitable for applications, but not shared
13305 @item -mprioritize-restricted-insns=@var{priority}
13306 @opindex mprioritize-restricted-insns
13307 This option controls the priority that is assigned to
13308 dispatch-slot restricted instructions during the second scheduling
13309 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13310 @var{no/highest/second-highest} priority to dispatch slot restricted
13313 @item -msched-costly-dep=@var{dependence_type}
13314 @opindex msched-costly-dep
13315 This option controls which dependences are considered costly
13316 by the target during instruction scheduling. The argument
13317 @var{dependence_type} takes one of the following values:
13318 @var{no}: no dependence is costly,
13319 @var{all}: all dependences are costly,
13320 @var{true_store_to_load}: a true dependence from store to load is costly,
13321 @var{store_to_load}: any dependence from store to load is costly,
13322 @var{number}: any dependence which latency >= @var{number} is costly.
13324 @item -minsert-sched-nops=@var{scheme}
13325 @opindex minsert-sched-nops
13326 This option controls which nop insertion scheme will be used during
13327 the second scheduling pass. The argument @var{scheme} takes one of the
13329 @var{no}: Don't insert nops.
13330 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13331 according to the scheduler's grouping.
13332 @var{regroup_exact}: Insert nops to force costly dependent insns into
13333 separate groups. Insert exactly as many nops as needed to force an insn
13334 to a new group, according to the estimated processor grouping.
13335 @var{number}: Insert nops to force costly dependent insns into
13336 separate groups. Insert @var{number} nops to force an insn to a new group.
13339 @opindex mcall-sysv
13340 On System V.4 and embedded PowerPC systems compile code using calling
13341 conventions that adheres to the March 1995 draft of the System V
13342 Application Binary Interface, PowerPC processor supplement. This is the
13343 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13345 @item -mcall-sysv-eabi
13346 @opindex mcall-sysv-eabi
13347 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13349 @item -mcall-sysv-noeabi
13350 @opindex mcall-sysv-noeabi
13351 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13353 @item -mcall-solaris
13354 @opindex mcall-solaris
13355 On System V.4 and embedded PowerPC systems compile code for the Solaris
13359 @opindex mcall-linux
13360 On System V.4 and embedded PowerPC systems compile code for the
13361 Linux-based GNU system.
13365 On System V.4 and embedded PowerPC systems compile code for the
13366 Hurd-based GNU system.
13368 @item -mcall-netbsd
13369 @opindex mcall-netbsd
13370 On System V.4 and embedded PowerPC systems compile code for the
13371 NetBSD operating system.
13373 @item -maix-struct-return
13374 @opindex maix-struct-return
13375 Return all structures in memory (as specified by the AIX ABI)@.
13377 @item -msvr4-struct-return
13378 @opindex msvr4-struct-return
13379 Return structures smaller than 8 bytes in registers (as specified by the
13382 @item -mabi=@var{abi-type}
13384 Extend the current ABI with a particular extension, or remove such extension.
13385 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13386 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13390 Extend the current ABI with SPE ABI extensions. This does not change
13391 the default ABI, instead it adds the SPE ABI extensions to the current
13395 @opindex mabi=no-spe
13396 Disable Booke SPE ABI extensions for the current ABI@.
13398 @item -mabi=ibmlongdouble
13399 @opindex mabi=ibmlongdouble
13400 Change the current ABI to use IBM extended precision long double.
13401 This is a PowerPC 32-bit SYSV ABI option.
13403 @item -mabi=ieeelongdouble
13404 @opindex mabi=ieeelongdouble
13405 Change the current ABI to use IEEE extended precision long double.
13406 This is a PowerPC 32-bit Linux ABI option.
13409 @itemx -mno-prototype
13410 @opindex mprototype
13411 @opindex mno-prototype
13412 On System V.4 and embedded PowerPC systems assume that all calls to
13413 variable argument functions are properly prototyped. Otherwise, the
13414 compiler must insert an instruction before every non prototyped call to
13415 set or clear bit 6 of the condition code register (@var{CR}) to
13416 indicate whether floating point values were passed in the floating point
13417 registers in case the function takes a variable arguments. With
13418 @option{-mprototype}, only calls to prototyped variable argument functions
13419 will set or clear the bit.
13423 On embedded PowerPC systems, assume that the startup module is called
13424 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13425 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
13430 On embedded PowerPC systems, assume that the startup module is called
13431 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13436 On embedded PowerPC systems, assume that the startup module is called
13437 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13440 @item -myellowknife
13441 @opindex myellowknife
13442 On embedded PowerPC systems, assume that the startup module is called
13443 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13448 On System V.4 and embedded PowerPC systems, specify that you are
13449 compiling for a VxWorks system.
13453 Specify that you are compiling for the WindISS simulation environment.
13457 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13458 header to indicate that @samp{eabi} extended relocations are used.
13464 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13465 Embedded Applications Binary Interface (eabi) which is a set of
13466 modifications to the System V.4 specifications. Selecting @option{-meabi}
13467 means that the stack is aligned to an 8 byte boundary, a function
13468 @code{__eabi} is called to from @code{main} to set up the eabi
13469 environment, and the @option{-msdata} option can use both @code{r2} and
13470 @code{r13} to point to two separate small data areas. Selecting
13471 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13472 do not call an initialization function from @code{main}, and the
13473 @option{-msdata} option will only use @code{r13} to point to a single
13474 small data area. The @option{-meabi} option is on by default if you
13475 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13478 @opindex msdata=eabi
13479 On System V.4 and embedded PowerPC systems, put small initialized
13480 @code{const} global and static data in the @samp{.sdata2} section, which
13481 is pointed to by register @code{r2}. Put small initialized
13482 non-@code{const} global and static data in the @samp{.sdata} section,
13483 which is pointed to by register @code{r13}. Put small uninitialized
13484 global and static data in the @samp{.sbss} section, which is adjacent to
13485 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13486 incompatible with the @option{-mrelocatable} option. The
13487 @option{-msdata=eabi} option also sets the @option{-memb} option.
13490 @opindex msdata=sysv
13491 On System V.4 and embedded PowerPC systems, put small global and static
13492 data in the @samp{.sdata} section, which is pointed to by register
13493 @code{r13}. Put small uninitialized global and static data in the
13494 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13495 The @option{-msdata=sysv} option is incompatible with the
13496 @option{-mrelocatable} option.
13498 @item -msdata=default
13500 @opindex msdata=default
13502 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13503 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13504 same as @option{-msdata=sysv}.
13507 @opindex msdata-data
13508 On System V.4 and embedded PowerPC systems, put small global
13509 data in the @samp{.sdata} section. Put small uninitialized global
13510 data in the @samp{.sbss} section. Do not use register @code{r13}
13511 to address small data however. This is the default behavior unless
13512 other @option{-msdata} options are used.
13516 @opindex msdata=none
13518 On embedded PowerPC systems, put all initialized global and static data
13519 in the @samp{.data} section, and all uninitialized data in the
13520 @samp{.bss} section.
13524 @cindex smaller data references (PowerPC)
13525 @cindex .sdata/.sdata2 references (PowerPC)
13526 On embedded PowerPC systems, put global and static items less than or
13527 equal to @var{num} bytes into the small data or bss sections instead of
13528 the normal data or bss section. By default, @var{num} is 8. The
13529 @option{-G @var{num}} switch is also passed to the linker.
13530 All modules should be compiled with the same @option{-G @var{num}} value.
13533 @itemx -mno-regnames
13535 @opindex mno-regnames
13536 On System V.4 and embedded PowerPC systems do (do not) emit register
13537 names in the assembly language output using symbolic forms.
13540 @itemx -mno-longcall
13542 @opindex mno-longcall
13543 By default assume that all calls are far away so that a longer more
13544 expensive calling sequence is required. This is required for calls
13545 further than 32 megabytes (33,554,432 bytes) from the current location.
13546 A short call will be generated if the compiler knows
13547 the call cannot be that far away. This setting can be overridden by
13548 the @code{shortcall} function attribute, or by @code{#pragma
13551 Some linkers are capable of detecting out-of-range calls and generating
13552 glue code on the fly. On these systems, long calls are unnecessary and
13553 generate slower code. As of this writing, the AIX linker can do this,
13554 as can the GNU linker for PowerPC/64. It is planned to add this feature
13555 to the GNU linker for 32-bit PowerPC systems as well.
13557 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13558 callee, L42'', plus a ``branch island'' (glue code). The two target
13559 addresses represent the callee and the ``branch island''. The
13560 Darwin/PPC linker will prefer the first address and generate a ``bl
13561 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13562 otherwise, the linker will generate ``bl L42'' to call the ``branch
13563 island''. The ``branch island'' is appended to the body of the
13564 calling function; it computes the full 32-bit address of the callee
13567 On Mach-O (Darwin) systems, this option directs the compiler emit to
13568 the glue for every direct call, and the Darwin linker decides whether
13569 to use or discard it.
13571 In the future, we may cause GCC to ignore all longcall specifications
13572 when the linker is known to generate glue.
13576 Adds support for multithreading with the @dfn{pthreads} library.
13577 This option sets flags for both the preprocessor and linker.
13581 @node S/390 and zSeries Options
13582 @subsection S/390 and zSeries Options
13583 @cindex S/390 and zSeries Options
13585 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13589 @itemx -msoft-float
13590 @opindex mhard-float
13591 @opindex msoft-float
13592 Use (do not use) the hardware floating-point instructions and registers
13593 for floating-point operations. When @option{-msoft-float} is specified,
13594 functions in @file{libgcc.a} will be used to perform floating-point
13595 operations. When @option{-mhard-float} is specified, the compiler
13596 generates IEEE floating-point instructions. This is the default.
13598 @item -mlong-double-64
13599 @itemx -mlong-double-128
13600 @opindex mlong-double-64
13601 @opindex mlong-double-128
13602 These switches control the size of @code{long double} type. A size
13603 of 64bit makes the @code{long double} type equivalent to the @code{double}
13604 type. This is the default.
13607 @itemx -mno-backchain
13608 @opindex mbackchain
13609 @opindex mno-backchain
13610 Store (do not store) the address of the caller's frame as backchain pointer
13611 into the callee's stack frame.
13612 A backchain may be needed to allow debugging using tools that do not understand
13613 DWARF-2 call frame information.
13614 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13615 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13616 the backchain is placed into the topmost word of the 96/160 byte register
13619 In general, code compiled with @option{-mbackchain} is call-compatible with
13620 code compiled with @option{-mmo-backchain}; however, use of the backchain
13621 for debugging purposes usually requires that the whole binary is built with
13622 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13623 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13624 to build a linux kernel use @option{-msoft-float}.
13626 The default is to not maintain the backchain.
13628 @item -mpacked-stack
13629 @item -mno-packed-stack
13630 @opindex mpacked-stack
13631 @opindex mno-packed-stack
13632 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13633 specified, the compiler uses the all fields of the 96/160 byte register save
13634 area only for their default purpose; unused fields still take up stack space.
13635 When @option{-mpacked-stack} is specified, register save slots are densely
13636 packed at the top of the register save area; unused space is reused for other
13637 purposes, allowing for more efficient use of the available stack space.
13638 However, when @option{-mbackchain} is also in effect, the topmost word of
13639 the save area is always used to store the backchain, and the return address
13640 register is always saved two words below the backchain.
13642 As long as the stack frame backchain is not used, code generated with
13643 @option{-mpacked-stack} is call-compatible with code generated with
13644 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13645 S/390 or zSeries generated code that uses the stack frame backchain at run
13646 time, not just for debugging purposes. Such code is not call-compatible
13647 with code compiled with @option{-mpacked-stack}. Also, note that the
13648 combination of @option{-mbackchain},
13649 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13650 to build a linux kernel use @option{-msoft-float}.
13652 The default is to not use the packed stack layout.
13655 @itemx -mno-small-exec
13656 @opindex msmall-exec
13657 @opindex mno-small-exec
13658 Generate (or do not generate) code using the @code{bras} instruction
13659 to do subroutine calls.
13660 This only works reliably if the total executable size does not
13661 exceed 64k. The default is to use the @code{basr} instruction instead,
13662 which does not have this limitation.
13668 When @option{-m31} is specified, generate code compliant to the
13669 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13670 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13671 particular to generate 64-bit instructions. For the @samp{s390}
13672 targets, the default is @option{-m31}, while the @samp{s390x}
13673 targets default to @option{-m64}.
13679 When @option{-mzarch} is specified, generate code using the
13680 instructions available on z/Architecture.
13681 When @option{-mesa} is specified, generate code using the
13682 instructions available on ESA/390. Note that @option{-mesa} is
13683 not possible with @option{-m64}.
13684 When generating code compliant to the GNU/Linux for S/390 ABI,
13685 the default is @option{-mesa}. When generating code compliant
13686 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13692 Generate (or do not generate) code using the @code{mvcle} instruction
13693 to perform block moves. When @option{-mno-mvcle} is specified,
13694 use a @code{mvc} loop instead. This is the default unless optimizing for
13701 Print (or do not print) additional debug information when compiling.
13702 The default is to not print debug information.
13704 @item -march=@var{cpu-type}
13706 Generate code that will run on @var{cpu-type}, which is the name of a system
13707 representing a certain processor type. Possible values for
13708 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13709 When generating code using the instructions available on z/Architecture,
13710 the default is @option{-march=z900}. Otherwise, the default is
13711 @option{-march=g5}.
13713 @item -mtune=@var{cpu-type}
13715 Tune to @var{cpu-type} everything applicable about the generated code,
13716 except for the ABI and the set of available instructions.
13717 The list of @var{cpu-type} values is the same as for @option{-march}.
13718 The default is the value used for @option{-march}.
13721 @itemx -mno-tpf-trace
13722 @opindex mtpf-trace
13723 @opindex mno-tpf-trace
13724 Generate code that adds (does not add) in TPF OS specific branches to trace
13725 routines in the operating system. This option is off by default, even
13726 when compiling for the TPF OS@.
13729 @itemx -mno-fused-madd
13730 @opindex mfused-madd
13731 @opindex mno-fused-madd
13732 Generate code that uses (does not use) the floating point multiply and
13733 accumulate instructions. These instructions are generated by default if
13734 hardware floating point is used.
13736 @item -mwarn-framesize=@var{framesize}
13737 @opindex mwarn-framesize
13738 Emit a warning if the current function exceeds the given frame size. Because
13739 this is a compile time check it doesn't need to be a real problem when the program
13740 runs. It is intended to identify functions which most probably cause
13741 a stack overflow. It is useful to be used in an environment with limited stack
13742 size e.g.@: the linux kernel.
13744 @item -mwarn-dynamicstack
13745 @opindex mwarn-dynamicstack
13746 Emit a warning if the function calls alloca or uses dynamically
13747 sized arrays. This is generally a bad idea with a limited stack size.
13749 @item -mstack-guard=@var{stack-guard}
13750 @item -mstack-size=@var{stack-size}
13751 @opindex mstack-guard
13752 @opindex mstack-size
13753 If these options are provided the s390 back end emits additional instructions in
13754 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13755 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13756 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13757 the frame size of the compiled function is chosen.
13758 These options are intended to be used to help debugging stack overflow problems.
13759 The additionally emitted code causes only little overhead and hence can also be
13760 used in production like systems without greater performance degradation. The given
13761 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13762 @var{stack-guard} without exceeding 64k.
13763 In order to be efficient the extra code makes the assumption that the stack starts
13764 at an address aligned to the value given by @var{stack-size}.
13765 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13768 @node Score Options
13769 @subsection Score Options
13770 @cindex Score Options
13772 These options are defined for Score implementations:
13777 Compile code for big endian mode. This is the default.
13781 Compile code for little endian mode.
13785 Disable generate bcnz instruction.
13789 Enable generate unaligned load and store instruction.
13793 Enable the use of multiply-accumulate instructions. Disabled by default.
13797 Specify the SCORE5 as the target architecture.
13801 Specify the SCORE5U of the target architecture.
13805 Specify the SCORE7 as the target architecture. This is the default.
13809 Specify the SCORE7D as the target architecture.
13813 @subsection SH Options
13815 These @samp{-m} options are defined for the SH implementations:
13820 Generate code for the SH1.
13824 Generate code for the SH2.
13827 Generate code for the SH2e.
13831 Generate code for the SH3.
13835 Generate code for the SH3e.
13839 Generate code for the SH4 without a floating-point unit.
13841 @item -m4-single-only
13842 @opindex m4-single-only
13843 Generate code for the SH4 with a floating-point unit that only
13844 supports single-precision arithmetic.
13848 Generate code for the SH4 assuming the floating-point unit is in
13849 single-precision mode by default.
13853 Generate code for the SH4.
13857 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13858 floating-point unit is not used.
13860 @item -m4a-single-only
13861 @opindex m4a-single-only
13862 Generate code for the SH4a, in such a way that no double-precision
13863 floating point operations are used.
13866 @opindex m4a-single
13867 Generate code for the SH4a assuming the floating-point unit is in
13868 single-precision mode by default.
13872 Generate code for the SH4a.
13876 Same as @option{-m4a-nofpu}, except that it implicitly passes
13877 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13878 instructions at the moment.
13882 Compile code for the processor in big endian mode.
13886 Compile code for the processor in little endian mode.
13890 Align doubles at 64-bit boundaries. Note that this changes the calling
13891 conventions, and thus some functions from the standard C library will
13892 not work unless you recompile it first with @option{-mdalign}.
13896 Shorten some address references at link time, when possible; uses the
13897 linker option @option{-relax}.
13901 Use 32-bit offsets in @code{switch} tables. The default is to use
13906 Enable the use of the instruction @code{fmovd}.
13910 Comply with the calling conventions defined by Renesas.
13914 Comply with the calling conventions defined by Renesas.
13918 Comply with the calling conventions defined for GCC before the Renesas
13919 conventions were available. This option is the default for all
13920 targets of the SH toolchain except for @samp{sh-symbianelf}.
13923 @opindex mnomacsave
13924 Mark the @code{MAC} register as call-clobbered, even if
13925 @option{-mhitachi} is given.
13929 Increase IEEE-compliance of floating-point code.
13930 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13931 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13932 comparisons of NANs / infinities incurs extra overhead in every
13933 floating point comparison, therefore the default is set to
13934 @option{-ffinite-math-only}.
13936 @item -minline-ic_invalidate
13937 @opindex minline-ic_invalidate
13938 Inline code to invalidate instruction cache entries after setting up
13939 nested function trampolines.
13940 This option has no effect if -musermode is in effect and the selected
13941 code generation option (e.g. -m4) does not allow the use of the icbi
13943 If the selected code generation option does not allow the use of the icbi
13944 instruction, and -musermode is not in effect, the inlined code will
13945 manipulate the instruction cache address array directly with an associative
13946 write. This not only requires privileged mode, but it will also
13947 fail if the cache line had been mapped via the TLB and has become unmapped.
13951 Dump instruction size and location in the assembly code.
13954 @opindex mpadstruct
13955 This option is deprecated. It pads structures to multiple of 4 bytes,
13956 which is incompatible with the SH ABI@.
13960 Optimize for space instead of speed. Implied by @option{-Os}.
13963 @opindex mprefergot
13964 When generating position-independent code, emit function calls using
13965 the Global Offset Table instead of the Procedure Linkage Table.
13969 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13970 if the inlined code would not work in user mode.
13971 This is the default when the target is @code{sh-*-linux*}.
13973 @item -multcost=@var{number}
13974 @opindex multcost=@var{number}
13975 Set the cost to assume for a multiply insn.
13977 @item -mdiv=@var{strategy}
13978 @opindex mdiv=@var{strategy}
13979 Set the division strategy to use for SHmedia code. @var{strategy} must be
13980 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13981 inv:call2, inv:fp .
13982 "fp" performs the operation in floating point. This has a very high latency,
13983 but needs only a few instructions, so it might be a good choice if
13984 your code has enough easily exploitable ILP to allow the compiler to
13985 schedule the floating point instructions together with other instructions.
13986 Division by zero causes a floating point exception.
13987 "inv" uses integer operations to calculate the inverse of the divisor,
13988 and then multiplies the dividend with the inverse. This strategy allows
13989 cse and hoisting of the inverse calculation. Division by zero calculates
13990 an unspecified result, but does not trap.
13991 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13992 have been found, or if the entire operation has been hoisted to the same
13993 place, the last stages of the inverse calculation are intertwined with the
13994 final multiply to reduce the overall latency, at the expense of using a few
13995 more instructions, and thus offering fewer scheduling opportunities with
13997 "call" calls a library function that usually implements the inv:minlat
13999 This gives high code density for m5-*media-nofpu compilations.
14000 "call2" uses a different entry point of the same library function, where it
14001 assumes that a pointer to a lookup table has already been set up, which
14002 exposes the pointer load to cse / code hoisting optimizations.
14003 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14004 code generation, but if the code stays unoptimized, revert to the "call",
14005 "call2", or "fp" strategies, respectively. Note that the
14006 potentially-trapping side effect of division by zero is carried by a
14007 separate instruction, so it is possible that all the integer instructions
14008 are hoisted out, but the marker for the side effect stays where it is.
14009 A recombination to fp operations or a call is not possible in that case.
14010 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14011 that the inverse calculation was nor separated from the multiply, they speed
14012 up division where the dividend fits into 20 bits (plus sign where applicable),
14013 by inserting a test to skip a number of operations in this case; this test
14014 slows down the case of larger dividends. inv20u assumes the case of a such
14015 a small dividend to be unlikely, and inv20l assumes it to be likely.
14017 @item -mdivsi3_libfunc=@var{name}
14018 @opindex mdivsi3_libfunc=@var{name}
14019 Set the name of the library function used for 32 bit signed division to
14020 @var{name}. This only affect the name used in the call and inv:call
14021 division strategies, and the compiler will still expect the same
14022 sets of input/output/clobbered registers as if this option was not present.
14024 @item -madjust-unroll
14025 @opindex madjust-unroll
14026 Throttle unrolling to avoid thrashing target registers.
14027 This option only has an effect if the gcc code base supports the
14028 TARGET_ADJUST_UNROLL_MAX target hook.
14030 @item -mindexed-addressing
14031 @opindex mindexed-addressing
14032 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14033 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14034 semantics for the indexed addressing mode. The architecture allows the
14035 implementation of processors with 64 bit MMU, which the OS could use to
14036 get 32 bit addressing, but since no current hardware implementation supports
14037 this or any other way to make the indexed addressing mode safe to use in
14038 the 32 bit ABI, the default is -mno-indexed-addressing.
14040 @item -mgettrcost=@var{number}
14041 @opindex mgettrcost=@var{number}
14042 Set the cost assumed for the gettr instruction to @var{number}.
14043 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14047 Assume pt* instructions won't trap. This will generally generate better
14048 scheduled code, but is unsafe on current hardware. The current architecture
14049 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14050 This has the unintentional effect of making it unsafe to schedule ptabs /
14051 ptrel before a branch, or hoist it out of a loop. For example,
14052 __do_global_ctors, a part of libgcc that runs constructors at program
14053 startup, calls functions in a list which is delimited by @minus{}1. With the
14054 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14055 That means that all the constructors will be run a bit quicker, but when
14056 the loop comes to the end of the list, the program crashes because ptabs
14057 loads @minus{}1 into a target register. Since this option is unsafe for any
14058 hardware implementing the current architecture specification, the default
14059 is -mno-pt-fixed. Unless the user specifies a specific cost with
14060 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14061 this deters register allocation using target registers for storing
14064 @item -minvalid-symbols
14065 @opindex minvalid-symbols
14066 Assume symbols might be invalid. Ordinary function symbols generated by
14067 the compiler will always be valid to load with movi/shori/ptabs or
14068 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14069 to generate symbols that will cause ptabs / ptrel to trap.
14070 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14071 It will then prevent cross-basic-block cse, hoisting and most scheduling
14072 of symbol loads. The default is @option{-mno-invalid-symbols}.
14075 @node SPARC Options
14076 @subsection SPARC Options
14077 @cindex SPARC options
14079 These @samp{-m} options are supported on the SPARC:
14082 @item -mno-app-regs
14084 @opindex mno-app-regs
14086 Specify @option{-mapp-regs} to generate output using the global registers
14087 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14090 To be fully SVR4 ABI compliant at the cost of some performance loss,
14091 specify @option{-mno-app-regs}. You should compile libraries and system
14092 software with this option.
14095 @itemx -mhard-float
14097 @opindex mhard-float
14098 Generate output containing floating point instructions. This is the
14102 @itemx -msoft-float
14104 @opindex msoft-float
14105 Generate output containing library calls for floating point.
14106 @strong{Warning:} the requisite libraries are not available for all SPARC
14107 targets. Normally the facilities of the machine's usual C compiler are
14108 used, but this cannot be done directly in cross-compilation. You must make
14109 your own arrangements to provide suitable library functions for
14110 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14111 @samp{sparclite-*-*} do provide software floating point support.
14113 @option{-msoft-float} changes the calling convention in the output file;
14114 therefore, it is only useful if you compile @emph{all} of a program with
14115 this option. In particular, you need to compile @file{libgcc.a}, the
14116 library that comes with GCC, with @option{-msoft-float} in order for
14119 @item -mhard-quad-float
14120 @opindex mhard-quad-float
14121 Generate output containing quad-word (long double) floating point
14124 @item -msoft-quad-float
14125 @opindex msoft-quad-float
14126 Generate output containing library calls for quad-word (long double)
14127 floating point instructions. The functions called are those specified
14128 in the SPARC ABI@. This is the default.
14130 As of this writing, there are no SPARC implementations that have hardware
14131 support for the quad-word floating point instructions. They all invoke
14132 a trap handler for one of these instructions, and then the trap handler
14133 emulates the effect of the instruction. Because of the trap handler overhead,
14134 this is much slower than calling the ABI library routines. Thus the
14135 @option{-msoft-quad-float} option is the default.
14137 @item -mno-unaligned-doubles
14138 @itemx -munaligned-doubles
14139 @opindex mno-unaligned-doubles
14140 @opindex munaligned-doubles
14141 Assume that doubles have 8 byte alignment. This is the default.
14143 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14144 alignment only if they are contained in another type, or if they have an
14145 absolute address. Otherwise, it assumes they have 4 byte alignment.
14146 Specifying this option avoids some rare compatibility problems with code
14147 generated by other compilers. It is not the default because it results
14148 in a performance loss, especially for floating point code.
14150 @item -mno-faster-structs
14151 @itemx -mfaster-structs
14152 @opindex mno-faster-structs
14153 @opindex mfaster-structs
14154 With @option{-mfaster-structs}, the compiler assumes that structures
14155 should have 8 byte alignment. This enables the use of pairs of
14156 @code{ldd} and @code{std} instructions for copies in structure
14157 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14158 However, the use of this changed alignment directly violates the SPARC
14159 ABI@. Thus, it's intended only for use on targets where the developer
14160 acknowledges that their resulting code will not be directly in line with
14161 the rules of the ABI@.
14163 @item -mimpure-text
14164 @opindex mimpure-text
14165 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14166 the compiler to not pass @option{-z text} to the linker when linking a
14167 shared object. Using this option, you can link position-dependent
14168 code into a shared object.
14170 @option{-mimpure-text} suppresses the ``relocations remain against
14171 allocatable but non-writable sections'' linker error message.
14172 However, the necessary relocations will trigger copy-on-write, and the
14173 shared object is not actually shared across processes. Instead of
14174 using @option{-mimpure-text}, you should compile all source code with
14175 @option{-fpic} or @option{-fPIC}.
14177 This option is only available on SunOS and Solaris.
14179 @item -mcpu=@var{cpu_type}
14181 Set the instruction set, register set, and instruction scheduling parameters
14182 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14183 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14184 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14185 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14186 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14188 Default instruction scheduling parameters are used for values that select
14189 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14190 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14192 Here is a list of each supported architecture and their supported
14197 v8: supersparc, hypersparc
14198 sparclite: f930, f934, sparclite86x
14200 v9: ultrasparc, ultrasparc3, niagara, niagara2
14203 By default (unless configured otherwise), GCC generates code for the V7
14204 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14205 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14206 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14207 SPARCStation 1, 2, IPX etc.
14209 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14210 architecture. The only difference from V7 code is that the compiler emits
14211 the integer multiply and integer divide instructions which exist in SPARC-V8
14212 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14213 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14216 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14217 the SPARC architecture. This adds the integer multiply, integer divide step
14218 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14219 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14220 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14221 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14222 MB86934 chip, which is the more recent SPARClite with FPU@.
14224 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14225 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14226 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14227 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14228 optimizes it for the TEMIC SPARClet chip.
14230 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14231 architecture. This adds 64-bit integer and floating-point move instructions,
14232 3 additional floating-point condition code registers and conditional move
14233 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14234 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14235 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14236 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14237 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14238 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14239 additionally optimizes it for Sun UltraSPARC T2 chips.
14241 @item -mtune=@var{cpu_type}
14243 Set the instruction scheduling parameters for machine type
14244 @var{cpu_type}, but do not set the instruction set or register set that the
14245 option @option{-mcpu=@var{cpu_type}} would.
14247 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14248 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14249 that select a particular cpu implementation. Those are @samp{cypress},
14250 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14251 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14252 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14257 @opindex mno-v8plus
14258 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14259 difference from the V8 ABI is that the global and out registers are
14260 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14261 mode for all SPARC-V9 processors.
14267 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14268 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14271 These @samp{-m} options are supported in addition to the above
14272 on SPARC-V9 processors in 64-bit environments:
14275 @item -mlittle-endian
14276 @opindex mlittle-endian
14277 Generate code for a processor running in little-endian mode. It is only
14278 available for a few configurations and most notably not on Solaris and Linux.
14284 Generate code for a 32-bit or 64-bit environment.
14285 The 32-bit environment sets int, long and pointer to 32 bits.
14286 The 64-bit environment sets int to 32 bits and long and pointer
14289 @item -mcmodel=medlow
14290 @opindex mcmodel=medlow
14291 Generate code for the Medium/Low code model: 64-bit addresses, programs
14292 must be linked in the low 32 bits of memory. Programs can be statically
14293 or dynamically linked.
14295 @item -mcmodel=medmid
14296 @opindex mcmodel=medmid
14297 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14298 must be linked in the low 44 bits of memory, the text and data segments must
14299 be less than 2GB in size and the data segment must be located within 2GB of
14302 @item -mcmodel=medany
14303 @opindex mcmodel=medany
14304 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14305 may be linked anywhere in memory, the text and data segments must be less
14306 than 2GB in size and the data segment must be located within 2GB of the
14309 @item -mcmodel=embmedany
14310 @opindex mcmodel=embmedany
14311 Generate code for the Medium/Anywhere code model for embedded systems:
14312 64-bit addresses, the text and data segments must be less than 2GB in
14313 size, both starting anywhere in memory (determined at link time). The
14314 global register %g4 points to the base of the data segment. Programs
14315 are statically linked and PIC is not supported.
14318 @itemx -mno-stack-bias
14319 @opindex mstack-bias
14320 @opindex mno-stack-bias
14321 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14322 frame pointer if present, are offset by @minus{}2047 which must be added back
14323 when making stack frame references. This is the default in 64-bit mode.
14324 Otherwise, assume no such offset is present.
14327 These switches are supported in addition to the above on Solaris:
14332 Add support for multithreading using the Solaris threads library. This
14333 option sets flags for both the preprocessor and linker. This option does
14334 not affect the thread safety of object code produced by the compiler or
14335 that of libraries supplied with it.
14339 Add support for multithreading using the POSIX threads library. This
14340 option sets flags for both the preprocessor and linker. This option does
14341 not affect the thread safety of object code produced by the compiler or
14342 that of libraries supplied with it.
14346 This is a synonym for @option{-pthreads}.
14350 @subsection SPU Options
14351 @cindex SPU options
14353 These @samp{-m} options are supported on the SPU:
14357 @itemx -merror-reloc
14358 @opindex mwarn-reloc
14359 @opindex merror-reloc
14361 The loader for SPU does not handle dynamic relocations. By default, GCC
14362 will give an error when it generates code that requires a dynamic
14363 relocation. @option{-mno-error-reloc} disables the error,
14364 @option{-mwarn-reloc} will generate a warning instead.
14367 @itemx -munsafe-dma
14369 @opindex munsafe-dma
14371 Instructions which initiate or test completion of DMA must not be
14372 reordered with respect to loads and stores of the memory which is being
14373 accessed. Users typically address this problem using the volatile
14374 keyword, but that can lead to inefficient code in places where the
14375 memory is known to not change. Rather than mark the memory as volatile
14376 we treat the DMA instructions as potentially effecting all memory. With
14377 @option{-munsafe-dma} users must use the volatile keyword to protect
14380 @item -mbranch-hints
14381 @opindex mbranch-hints
14383 By default, GCC will generate a branch hint instruction to avoid
14384 pipeline stalls for always taken or probably taken branches. A hint
14385 will not be generated closer than 8 instructions away from its branch.
14386 There is little reason to disable them, except for debugging purposes,
14387 or to make an object a little bit smaller.
14391 @opindex msmall-mem
14392 @opindex mlarge-mem
14394 By default, GCC generates code assuming that addresses are never larger
14395 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14396 a full 32 bit address.
14401 By default, GCC links against startup code that assumes the SPU-style
14402 main function interface (which has an unconventional parameter list).
14403 With @option{-mstdmain}, GCC will link your program against startup
14404 code that assumes a C99-style interface to @code{main}, including a
14405 local copy of @code{argv} strings.
14407 @item -mfixed-range=@var{register-range}
14408 @opindex mfixed-range
14409 Generate code treating the given register range as fixed registers.
14410 A fixed register is one that the register allocator can not use. This is
14411 useful when compiling kernel code. A register range is specified as
14412 two registers separated by a dash. Multiple register ranges can be
14413 specified separated by a comma.
14417 @node System V Options
14418 @subsection Options for System V
14420 These additional options are available on System V Release 4 for
14421 compatibility with other compilers on those systems:
14426 Create a shared object.
14427 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14431 Identify the versions of each tool used by the compiler, in a
14432 @code{.ident} assembler directive in the output.
14436 Refrain from adding @code{.ident} directives to the output file (this is
14439 @item -YP,@var{dirs}
14441 Search the directories @var{dirs}, and no others, for libraries
14442 specified with @option{-l}.
14444 @item -Ym,@var{dir}
14446 Look in the directory @var{dir} to find the M4 preprocessor.
14447 The assembler uses this option.
14448 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14449 @c the generic assembler that comes with Solaris takes just -Ym.
14453 @subsection V850 Options
14454 @cindex V850 Options
14456 These @samp{-m} options are defined for V850 implementations:
14460 @itemx -mno-long-calls
14461 @opindex mlong-calls
14462 @opindex mno-long-calls
14463 Treat all calls as being far away (near). If calls are assumed to be
14464 far away, the compiler will always load the functions address up into a
14465 register, and call indirect through the pointer.
14471 Do not optimize (do optimize) basic blocks that use the same index
14472 pointer 4 or more times to copy pointer into the @code{ep} register, and
14473 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14474 option is on by default if you optimize.
14476 @item -mno-prolog-function
14477 @itemx -mprolog-function
14478 @opindex mno-prolog-function
14479 @opindex mprolog-function
14480 Do not use (do use) external functions to save and restore registers
14481 at the prologue and epilogue of a function. The external functions
14482 are slower, but use less code space if more than one function saves
14483 the same number of registers. The @option{-mprolog-function} option
14484 is on by default if you optimize.
14488 Try to make the code as small as possible. At present, this just turns
14489 on the @option{-mep} and @option{-mprolog-function} options.
14491 @item -mtda=@var{n}
14493 Put static or global variables whose size is @var{n} bytes or less into
14494 the tiny data area that register @code{ep} points to. The tiny data
14495 area can hold up to 256 bytes in total (128 bytes for byte references).
14497 @item -msda=@var{n}
14499 Put static or global variables whose size is @var{n} bytes or less into
14500 the small data area that register @code{gp} points to. The small data
14501 area can hold up to 64 kilobytes.
14503 @item -mzda=@var{n}
14505 Put static or global variables whose size is @var{n} bytes or less into
14506 the first 32 kilobytes of memory.
14510 Specify that the target processor is the V850.
14513 @opindex mbig-switch
14514 Generate code suitable for big switch tables. Use this option only if
14515 the assembler/linker complain about out of range branches within a switch
14520 This option will cause r2 and r5 to be used in the code generated by
14521 the compiler. This setting is the default.
14523 @item -mno-app-regs
14524 @opindex mno-app-regs
14525 This option will cause r2 and r5 to be treated as fixed registers.
14529 Specify that the target processor is the V850E1. The preprocessor
14530 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14531 this option is used.
14535 Specify that the target processor is the V850E@. The preprocessor
14536 constant @samp{__v850e__} will be defined if this option is used.
14538 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14539 are defined then a default target processor will be chosen and the
14540 relevant @samp{__v850*__} preprocessor constant will be defined.
14542 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14543 defined, regardless of which processor variant is the target.
14545 @item -mdisable-callt
14546 @opindex mdisable-callt
14547 This option will suppress generation of the CALLT instruction for the
14548 v850e and v850e1 flavors of the v850 architecture. The default is
14549 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14554 @subsection VAX Options
14555 @cindex VAX options
14557 These @samp{-m} options are defined for the VAX:
14562 Do not output certain jump instructions (@code{aobleq} and so on)
14563 that the Unix assembler for the VAX cannot handle across long
14568 Do output those jump instructions, on the assumption that you
14569 will assemble with the GNU assembler.
14573 Output code for g-format floating point numbers instead of d-format.
14576 @node VxWorks Options
14577 @subsection VxWorks Options
14578 @cindex VxWorks Options
14580 The options in this section are defined for all VxWorks targets.
14581 Options specific to the target hardware are listed with the other
14582 options for that target.
14587 GCC can generate code for both VxWorks kernels and real time processes
14588 (RTPs). This option switches from the former to the latter. It also
14589 defines the preprocessor macro @code{__RTP__}.
14592 @opindex non-static
14593 Link an RTP executable against shared libraries rather than static
14594 libraries. The options @option{-static} and @option{-shared} can
14595 also be used for RTPs (@pxref{Link Options}); @option{-static}
14602 These options are passed down to the linker. They are defined for
14603 compatibility with Diab.
14606 @opindex Xbind-lazy
14607 Enable lazy binding of function calls. This option is equivalent to
14608 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14612 Disable lazy binding of function calls. This option is the default and
14613 is defined for compatibility with Diab.
14616 @node x86-64 Options
14617 @subsection x86-64 Options
14618 @cindex x86-64 options
14620 These are listed under @xref{i386 and x86-64 Options}.
14622 @node Xstormy16 Options
14623 @subsection Xstormy16 Options
14624 @cindex Xstormy16 Options
14626 These options are defined for Xstormy16:
14631 Choose startup files and linker script suitable for the simulator.
14634 @node Xtensa Options
14635 @subsection Xtensa Options
14636 @cindex Xtensa Options
14638 These options are supported for Xtensa targets:
14642 @itemx -mno-const16
14644 @opindex mno-const16
14645 Enable or disable use of @code{CONST16} instructions for loading
14646 constant values. The @code{CONST16} instruction is currently not a
14647 standard option from Tensilica. When enabled, @code{CONST16}
14648 instructions are always used in place of the standard @code{L32R}
14649 instructions. The use of @code{CONST16} is enabled by default only if
14650 the @code{L32R} instruction is not available.
14653 @itemx -mno-fused-madd
14654 @opindex mfused-madd
14655 @opindex mno-fused-madd
14656 Enable or disable use of fused multiply/add and multiply/subtract
14657 instructions in the floating-point option. This has no effect if the
14658 floating-point option is not also enabled. Disabling fused multiply/add
14659 and multiply/subtract instructions forces the compiler to use separate
14660 instructions for the multiply and add/subtract operations. This may be
14661 desirable in some cases where strict IEEE 754-compliant results are
14662 required: the fused multiply add/subtract instructions do not round the
14663 intermediate result, thereby producing results with @emph{more} bits of
14664 precision than specified by the IEEE standard. Disabling fused multiply
14665 add/subtract instructions also ensures that the program output is not
14666 sensitive to the compiler's ability to combine multiply and add/subtract
14669 @item -mtext-section-literals
14670 @itemx -mno-text-section-literals
14671 @opindex mtext-section-literals
14672 @opindex mno-text-section-literals
14673 Control the treatment of literal pools. The default is
14674 @option{-mno-text-section-literals}, which places literals in a separate
14675 section in the output file. This allows the literal pool to be placed
14676 in a data RAM/ROM, and it also allows the linker to combine literal
14677 pools from separate object files to remove redundant literals and
14678 improve code size. With @option{-mtext-section-literals}, the literals
14679 are interspersed in the text section in order to keep them as close as
14680 possible to their references. This may be necessary for large assembly
14683 @item -mtarget-align
14684 @itemx -mno-target-align
14685 @opindex mtarget-align
14686 @opindex mno-target-align
14687 When this option is enabled, GCC instructs the assembler to
14688 automatically align instructions to reduce branch penalties at the
14689 expense of some code density. The assembler attempts to widen density
14690 instructions to align branch targets and the instructions following call
14691 instructions. If there are not enough preceding safe density
14692 instructions to align a target, no widening will be performed. The
14693 default is @option{-mtarget-align}. These options do not affect the
14694 treatment of auto-aligned instructions like @code{LOOP}, which the
14695 assembler will always align, either by widening density instructions or
14696 by inserting no-op instructions.
14699 @itemx -mno-longcalls
14700 @opindex mlongcalls
14701 @opindex mno-longcalls
14702 When this option is enabled, GCC instructs the assembler to translate
14703 direct calls to indirect calls unless it can determine that the target
14704 of a direct call is in the range allowed by the call instruction. This
14705 translation typically occurs for calls to functions in other source
14706 files. Specifically, the assembler translates a direct @code{CALL}
14707 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14708 The default is @option{-mno-longcalls}. This option should be used in
14709 programs where the call target can potentially be out of range. This
14710 option is implemented in the assembler, not the compiler, so the
14711 assembly code generated by GCC will still show direct call
14712 instructions---look at the disassembled object code to see the actual
14713 instructions. Note that the assembler will use an indirect call for
14714 every cross-file call, not just those that really will be out of range.
14717 @node zSeries Options
14718 @subsection zSeries Options
14719 @cindex zSeries options
14721 These are listed under @xref{S/390 and zSeries Options}.
14723 @node Code Gen Options
14724 @section Options for Code Generation Conventions
14725 @cindex code generation conventions
14726 @cindex options, code generation
14727 @cindex run-time options
14729 These machine-independent options control the interface conventions
14730 used in code generation.
14732 Most of them have both positive and negative forms; the negative form
14733 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14734 one of the forms is listed---the one which is not the default. You
14735 can figure out the other form by either removing @samp{no-} or adding
14739 @item -fbounds-check
14740 @opindex fbounds-check
14741 For front-ends that support it, generate additional code to check that
14742 indices used to access arrays are within the declared range. This is
14743 currently only supported by the Java and Fortran front-ends, where
14744 this option defaults to true and false respectively.
14748 This option generates traps for signed overflow on addition, subtraction,
14749 multiplication operations.
14753 This option instructs the compiler to assume that signed arithmetic
14754 overflow of addition, subtraction and multiplication wraps around
14755 using twos-complement representation. This flag enables some optimizations
14756 and disables others. This option is enabled by default for the Java
14757 front-end, as required by the Java language specification.
14760 @opindex fexceptions
14761 Enable exception handling. Generates extra code needed to propagate
14762 exceptions. For some targets, this implies GCC will generate frame
14763 unwind information for all functions, which can produce significant data
14764 size overhead, although it does not affect execution. If you do not
14765 specify this option, GCC will enable it by default for languages like
14766 C++ which normally require exception handling, and disable it for
14767 languages like C that do not normally require it. However, you may need
14768 to enable this option when compiling C code that needs to interoperate
14769 properly with exception handlers written in C++. You may also wish to
14770 disable this option if you are compiling older C++ programs that don't
14771 use exception handling.
14773 @item -fnon-call-exceptions
14774 @opindex fnon-call-exceptions
14775 Generate code that allows trapping instructions to throw exceptions.
14776 Note that this requires platform-specific runtime support that does
14777 not exist everywhere. Moreover, it only allows @emph{trapping}
14778 instructions to throw exceptions, i.e.@: memory references or floating
14779 point instructions. It does not allow exceptions to be thrown from
14780 arbitrary signal handlers such as @code{SIGALRM}.
14782 @item -funwind-tables
14783 @opindex funwind-tables
14784 Similar to @option{-fexceptions}, except that it will just generate any needed
14785 static data, but will not affect the generated code in any other way.
14786 You will normally not enable this option; instead, a language processor
14787 that needs this handling would enable it on your behalf.
14789 @item -fasynchronous-unwind-tables
14790 @opindex fasynchronous-unwind-tables
14791 Generate unwind table in dwarf2 format, if supported by target machine. The
14792 table is exact at each instruction boundary, so it can be used for stack
14793 unwinding from asynchronous events (such as debugger or garbage collector).
14795 @item -fpcc-struct-return
14796 @opindex fpcc-struct-return
14797 Return ``short'' @code{struct} and @code{union} values in memory like
14798 longer ones, rather than in registers. This convention is less
14799 efficient, but it has the advantage of allowing intercallability between
14800 GCC-compiled files and files compiled with other compilers, particularly
14801 the Portable C Compiler (pcc).
14803 The precise convention for returning structures in memory depends
14804 on the target configuration macros.
14806 Short structures and unions are those whose size and alignment match
14807 that of some integer type.
14809 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14810 switch is not binary compatible with code compiled with the
14811 @option{-freg-struct-return} switch.
14812 Use it to conform to a non-default application binary interface.
14814 @item -freg-struct-return
14815 @opindex freg-struct-return
14816 Return @code{struct} and @code{union} values in registers when possible.
14817 This is more efficient for small structures than
14818 @option{-fpcc-struct-return}.
14820 If you specify neither @option{-fpcc-struct-return} nor
14821 @option{-freg-struct-return}, GCC defaults to whichever convention is
14822 standard for the target. If there is no standard convention, GCC
14823 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14824 the principal compiler. In those cases, we can choose the standard, and
14825 we chose the more efficient register return alternative.
14827 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14828 switch is not binary compatible with code compiled with the
14829 @option{-fpcc-struct-return} switch.
14830 Use it to conform to a non-default application binary interface.
14832 @item -fshort-enums
14833 @opindex fshort-enums
14834 Allocate to an @code{enum} type only as many bytes as it needs for the
14835 declared range of possible values. Specifically, the @code{enum} type
14836 will be equivalent to the smallest integer type which has enough room.
14838 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14839 code that is not binary compatible with code generated without that switch.
14840 Use it to conform to a non-default application binary interface.
14842 @item -fshort-double
14843 @opindex fshort-double
14844 Use the same size for @code{double} as for @code{float}.
14846 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14847 code that is not binary compatible with code generated without that switch.
14848 Use it to conform to a non-default application binary interface.
14850 @item -fshort-wchar
14851 @opindex fshort-wchar
14852 Override the underlying type for @samp{wchar_t} to be @samp{short
14853 unsigned int} instead of the default for the target. This option is
14854 useful for building programs to run under WINE@.
14856 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14857 code that is not binary compatible with code generated without that switch.
14858 Use it to conform to a non-default application binary interface.
14861 @opindex fno-common
14862 In C, allocate even uninitialized global variables in the data section of the
14863 object file, rather than generating them as common blocks. This has the
14864 effect that if the same variable is declared (without @code{extern}) in
14865 two different compilations, you will get an error when you link them.
14866 The only reason this might be useful is if you wish to verify that the
14867 program will work on other systems which always work this way.
14871 Ignore the @samp{#ident} directive.
14873 @item -finhibit-size-directive
14874 @opindex finhibit-size-directive
14875 Don't output a @code{.size} assembler directive, or anything else that
14876 would cause trouble if the function is split in the middle, and the
14877 two halves are placed at locations far apart in memory. This option is
14878 used when compiling @file{crtstuff.c}; you should not need to use it
14881 @item -fverbose-asm
14882 @opindex fverbose-asm
14883 Put extra commentary information in the generated assembly code to
14884 make it more readable. This option is generally only of use to those
14885 who actually need to read the generated assembly code (perhaps while
14886 debugging the compiler itself).
14888 @option{-fno-verbose-asm}, the default, causes the
14889 extra information to be omitted and is useful when comparing two assembler
14892 @item -frecord-gcc-switches
14893 @opindex frecord-gcc-switches
14894 This switch causes the command line that was used to invoke the
14895 compiler to be recorded into the object file that is being created.
14896 This switch is only implemented on some targets and the exact format
14897 of the recording is target and binary file format dependent, but it
14898 usually takes the form of a section containing ASCII text. This
14899 switch is related to the @option{-fverbose-asm} switch, but that
14900 switch only records information in the assembler output file as
14901 comments, so it never reaches the object file.
14905 @cindex global offset table
14907 Generate position-independent code (PIC) suitable for use in a shared
14908 library, if supported for the target machine. Such code accesses all
14909 constant addresses through a global offset table (GOT)@. The dynamic
14910 loader resolves the GOT entries when the program starts (the dynamic
14911 loader is not part of GCC; it is part of the operating system). If
14912 the GOT size for the linked executable exceeds a machine-specific
14913 maximum size, you get an error message from the linker indicating that
14914 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14915 instead. (These maximums are 8k on the SPARC and 32k
14916 on the m68k and RS/6000. The 386 has no such limit.)
14918 Position-independent code requires special support, and therefore works
14919 only on certain machines. For the 386, GCC supports PIC for System V
14920 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14921 position-independent.
14923 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14928 If supported for the target machine, emit position-independent code,
14929 suitable for dynamic linking and avoiding any limit on the size of the
14930 global offset table. This option makes a difference on the m68k,
14931 PowerPC and SPARC@.
14933 Position-independent code requires special support, and therefore works
14934 only on certain machines.
14936 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14943 These options are similar to @option{-fpic} and @option{-fPIC}, but
14944 generated position independent code can be only linked into executables.
14945 Usually these options are used when @option{-pie} GCC option will be
14946 used during linking.
14948 @option{-fpie} and @option{-fPIE} both define the macros
14949 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14950 for @option{-fpie} and 2 for @option{-fPIE}.
14952 @item -fno-jump-tables
14953 @opindex fno-jump-tables
14954 Do not use jump tables for switch statements even where it would be
14955 more efficient than other code generation strategies. This option is
14956 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14957 building code which forms part of a dynamic linker and cannot
14958 reference the address of a jump table. On some targets, jump tables
14959 do not require a GOT and this option is not needed.
14961 @item -ffixed-@var{reg}
14963 Treat the register named @var{reg} as a fixed register; generated code
14964 should never refer to it (except perhaps as a stack pointer, frame
14965 pointer or in some other fixed role).
14967 @var{reg} must be the name of a register. The register names accepted
14968 are machine-specific and are defined in the @code{REGISTER_NAMES}
14969 macro in the machine description macro file.
14971 This flag does not have a negative form, because it specifies a
14974 @item -fcall-used-@var{reg}
14975 @opindex fcall-used
14976 Treat the register named @var{reg} as an allocable register that is
14977 clobbered by function calls. It may be allocated for temporaries or
14978 variables that do not live across a call. Functions compiled this way
14979 will not save and restore the register @var{reg}.
14981 It is an error to used this flag with the frame pointer or stack pointer.
14982 Use of this flag for other registers that have fixed pervasive roles in
14983 the machine's execution model will produce disastrous results.
14985 This flag does not have a negative form, because it specifies a
14988 @item -fcall-saved-@var{reg}
14989 @opindex fcall-saved
14990 Treat the register named @var{reg} as an allocable register saved by
14991 functions. It may be allocated even for temporaries or variables that
14992 live across a call. Functions compiled this way will save and restore
14993 the register @var{reg} if they use it.
14995 It is an error to used this flag with the frame pointer or stack pointer.
14996 Use of this flag for other registers that have fixed pervasive roles in
14997 the machine's execution model will produce disastrous results.
14999 A different sort of disaster will result from the use of this flag for
15000 a register in which function values may be returned.
15002 This flag does not have a negative form, because it specifies a
15005 @item -fpack-struct[=@var{n}]
15006 @opindex fpack-struct
15007 Without a value specified, pack all structure members together without
15008 holes. When a value is specified (which must be a small power of two), pack
15009 structure members according to this value, representing the maximum
15010 alignment (that is, objects with default alignment requirements larger than
15011 this will be output potentially unaligned at the next fitting location.
15013 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15014 code that is not binary compatible with code generated without that switch.
15015 Additionally, it makes the code suboptimal.
15016 Use it to conform to a non-default application binary interface.
15018 @item -finstrument-functions
15019 @opindex finstrument-functions
15020 Generate instrumentation calls for entry and exit to functions. Just
15021 after function entry and just before function exit, the following
15022 profiling functions will be called with the address of the current
15023 function and its call site. (On some platforms,
15024 @code{__builtin_return_address} does not work beyond the current
15025 function, so the call site information may not be available to the
15026 profiling functions otherwise.)
15029 void __cyg_profile_func_enter (void *this_fn,
15031 void __cyg_profile_func_exit (void *this_fn,
15035 The first argument is the address of the start of the current function,
15036 which may be looked up exactly in the symbol table.
15038 This instrumentation is also done for functions expanded inline in other
15039 functions. The profiling calls will indicate where, conceptually, the
15040 inline function is entered and exited. This means that addressable
15041 versions of such functions must be available. If all your uses of a
15042 function are expanded inline, this may mean an additional expansion of
15043 code size. If you use @samp{extern inline} in your C code, an
15044 addressable version of such functions must be provided. (This is
15045 normally the case anyways, but if you get lucky and the optimizer always
15046 expands the functions inline, you might have gotten away without
15047 providing static copies.)
15049 A function may be given the attribute @code{no_instrument_function}, in
15050 which case this instrumentation will not be done. This can be used, for
15051 example, for the profiling functions listed above, high-priority
15052 interrupt routines, and any functions from which the profiling functions
15053 cannot safely be called (perhaps signal handlers, if the profiling
15054 routines generate output or allocate memory).
15056 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15057 @opindex finstrument-functions-exclude-file-list
15059 Set the list of functions that are excluded from instrumentation (see
15060 the description of @code{-finstrument-functions}). If the file that
15061 contains a function definition matches with one of @var{file}, then
15062 that function is not instrumented. The match is done on substrings:
15063 if the @var{file} parameter is a substring of the file name, it is
15064 considered to be a match.
15067 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15068 will exclude any inline function defined in files whose pathnames
15069 contain @code{/bits/stl} or @code{include/sys}.
15071 If, for some reason, you want to include letter @code{','} in one of
15072 @var{sym}, write @code{'\,'}. For example,
15073 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15074 (note the single quote surrounding the option).
15076 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15077 @opindex finstrument-functions-exclude-function-list
15079 This is similar to @code{-finstrument-functions-exclude-file-list},
15080 but this option sets the list of function names to be excluded from
15081 instrumentation. The function name to be matched is its user-visible
15082 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15083 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15084 match is done on substrings: if the @var{sym} parameter is a substring
15085 of the function name, it is considered to be a match.
15087 @item -fstack-check
15088 @opindex fstack-check
15089 Generate code to verify that you do not go beyond the boundary of the
15090 stack. You should specify this flag if you are running in an
15091 environment with multiple threads, but only rarely need to specify it in
15092 a single-threaded environment since stack overflow is automatically
15093 detected on nearly all systems if there is only one stack.
15095 Note that this switch does not actually cause checking to be done; the
15096 operating system must do that. The switch causes generation of code
15097 to ensure that the operating system sees the stack being extended.
15099 @item -fstack-limit-register=@var{reg}
15100 @itemx -fstack-limit-symbol=@var{sym}
15101 @itemx -fno-stack-limit
15102 @opindex fstack-limit-register
15103 @opindex fstack-limit-symbol
15104 @opindex fno-stack-limit
15105 Generate code to ensure that the stack does not grow beyond a certain value,
15106 either the value of a register or the address of a symbol. If the stack
15107 would grow beyond the value, a signal is raised. For most targets,
15108 the signal is raised before the stack overruns the boundary, so
15109 it is possible to catch the signal without taking special precautions.
15111 For instance, if the stack starts at absolute address @samp{0x80000000}
15112 and grows downwards, you can use the flags
15113 @option{-fstack-limit-symbol=__stack_limit} and
15114 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15115 of 128KB@. Note that this may only work with the GNU linker.
15117 @cindex aliasing of parameters
15118 @cindex parameters, aliased
15119 @item -fargument-alias
15120 @itemx -fargument-noalias
15121 @itemx -fargument-noalias-global
15122 @itemx -fargument-noalias-anything
15123 @opindex fargument-alias
15124 @opindex fargument-noalias
15125 @opindex fargument-noalias-global
15126 @opindex fargument-noalias-anything
15127 Specify the possible relationships among parameters and between
15128 parameters and global data.
15130 @option{-fargument-alias} specifies that arguments (parameters) may
15131 alias each other and may alias global storage.@*
15132 @option{-fargument-noalias} specifies that arguments do not alias
15133 each other, but may alias global storage.@*
15134 @option{-fargument-noalias-global} specifies that arguments do not
15135 alias each other and do not alias global storage.
15136 @option{-fargument-noalias-anything} specifies that arguments do not
15137 alias any other storage.
15139 Each language will automatically use whatever option is required by
15140 the language standard. You should not need to use these options yourself.
15142 @item -fleading-underscore
15143 @opindex fleading-underscore
15144 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15145 change the way C symbols are represented in the object file. One use
15146 is to help link with legacy assembly code.
15148 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15149 generate code that is not binary compatible with code generated without that
15150 switch. Use it to conform to a non-default application binary interface.
15151 Not all targets provide complete support for this switch.
15153 @item -ftls-model=@var{model}
15154 @opindex ftls-model
15155 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15156 The @var{model} argument should be one of @code{global-dynamic},
15157 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15159 The default without @option{-fpic} is @code{initial-exec}; with
15160 @option{-fpic} the default is @code{global-dynamic}.
15162 @item -fvisibility=@var{default|internal|hidden|protected}
15163 @opindex fvisibility
15164 Set the default ELF image symbol visibility to the specified option---all
15165 symbols will be marked with this unless overridden within the code.
15166 Using this feature can very substantially improve linking and
15167 load times of shared object libraries, produce more optimized
15168 code, provide near-perfect API export and prevent symbol clashes.
15169 It is @strong{strongly} recommended that you use this in any shared objects
15172 Despite the nomenclature, @code{default} always means public ie;
15173 available to be linked against from outside the shared object.
15174 @code{protected} and @code{internal} are pretty useless in real-world
15175 usage so the only other commonly used option will be @code{hidden}.
15176 The default if @option{-fvisibility} isn't specified is
15177 @code{default}, i.e., make every
15178 symbol public---this causes the same behavior as previous versions of
15181 A good explanation of the benefits offered by ensuring ELF
15182 symbols have the correct visibility is given by ``How To Write
15183 Shared Libraries'' by Ulrich Drepper (which can be found at
15184 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15185 solution made possible by this option to marking things hidden when
15186 the default is public is to make the default hidden and mark things
15187 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15188 and @code{__attribute__ ((visibility("default")))} instead of
15189 @code{__declspec(dllexport)} you get almost identical semantics with
15190 identical syntax. This is a great boon to those working with
15191 cross-platform projects.
15193 For those adding visibility support to existing code, you may find
15194 @samp{#pragma GCC visibility} of use. This works by you enclosing
15195 the declarations you wish to set visibility for with (for example)
15196 @samp{#pragma GCC visibility push(hidden)} and
15197 @samp{#pragma GCC visibility pop}.
15198 Bear in mind that symbol visibility should be viewed @strong{as
15199 part of the API interface contract} and thus all new code should
15200 always specify visibility when it is not the default ie; declarations
15201 only for use within the local DSO should @strong{always} be marked explicitly
15202 as hidden as so to avoid PLT indirection overheads---making this
15203 abundantly clear also aids readability and self-documentation of the code.
15204 Note that due to ISO C++ specification requirements, operator new and
15205 operator delete must always be of default visibility.
15207 Be aware that headers from outside your project, in particular system
15208 headers and headers from any other library you use, may not be
15209 expecting to be compiled with visibility other than the default. You
15210 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15211 before including any such headers.
15213 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15214 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15215 no modifications. However, this means that calls to @samp{extern}
15216 functions with no explicit visibility will use the PLT, so it is more
15217 effective to use @samp{__attribute ((visibility))} and/or
15218 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15219 declarations should be treated as hidden.
15221 Note that @samp{-fvisibility} does affect C++ vague linkage
15222 entities. This means that, for instance, an exception class that will
15223 be thrown between DSOs must be explicitly marked with default
15224 visibility so that the @samp{type_info} nodes will be unified between
15227 An overview of these techniques, their benefits and how to use them
15228 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15234 @node Environment Variables
15235 @section Environment Variables Affecting GCC
15236 @cindex environment variables
15238 @c man begin ENVIRONMENT
15239 This section describes several environment variables that affect how GCC
15240 operates. Some of them work by specifying directories or prefixes to use
15241 when searching for various kinds of files. Some are used to specify other
15242 aspects of the compilation environment.
15244 Note that you can also specify places to search using options such as
15245 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15246 take precedence over places specified using environment variables, which
15247 in turn take precedence over those specified by the configuration of GCC@.
15248 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15249 GNU Compiler Collection (GCC) Internals}.
15254 @c @itemx LC_COLLATE
15256 @c @itemx LC_MONETARY
15257 @c @itemx LC_NUMERIC
15262 @c @findex LC_COLLATE
15263 @findex LC_MESSAGES
15264 @c @findex LC_MONETARY
15265 @c @findex LC_NUMERIC
15269 These environment variables control the way that GCC uses
15270 localization information that allow GCC to work with different
15271 national conventions. GCC inspects the locale categories
15272 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15273 so. These locale categories can be set to any value supported by your
15274 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15275 Kingdom encoded in UTF-8.
15277 The @env{LC_CTYPE} environment variable specifies character
15278 classification. GCC uses it to determine the character boundaries in
15279 a string; this is needed for some multibyte encodings that contain quote
15280 and escape characters that would otherwise be interpreted as a string
15283 The @env{LC_MESSAGES} environment variable specifies the language to
15284 use in diagnostic messages.
15286 If the @env{LC_ALL} environment variable is set, it overrides the value
15287 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15288 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15289 environment variable. If none of these variables are set, GCC
15290 defaults to traditional C English behavior.
15294 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15295 files. GCC uses temporary files to hold the output of one stage of
15296 compilation which is to be used as input to the next stage: for example,
15297 the output of the preprocessor, which is the input to the compiler
15300 @item GCC_EXEC_PREFIX
15301 @findex GCC_EXEC_PREFIX
15302 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15303 names of the subprograms executed by the compiler. No slash is added
15304 when this prefix is combined with the name of a subprogram, but you can
15305 specify a prefix that ends with a slash if you wish.
15307 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15308 an appropriate prefix to use based on the pathname it was invoked with.
15310 If GCC cannot find the subprogram using the specified prefix, it
15311 tries looking in the usual places for the subprogram.
15313 The default value of @env{GCC_EXEC_PREFIX} is
15314 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15315 the installed compiler. In many cases @var{prefix} is the value
15316 of @code{prefix} when you ran the @file{configure} script.
15318 Other prefixes specified with @option{-B} take precedence over this prefix.
15320 This prefix is also used for finding files such as @file{crt0.o} that are
15323 In addition, the prefix is used in an unusual way in finding the
15324 directories to search for header files. For each of the standard
15325 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15326 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15327 replacing that beginning with the specified prefix to produce an
15328 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15329 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15330 These alternate directories are searched first; the standard directories
15331 come next. If a standard directory begins with the configured
15332 @var{prefix} then the value of @var{prefix} is replaced by
15333 @env{GCC_EXEC_PREFIX} when looking for header files.
15335 @item COMPILER_PATH
15336 @findex COMPILER_PATH
15337 The value of @env{COMPILER_PATH} is a colon-separated list of
15338 directories, much like @env{PATH}. GCC tries the directories thus
15339 specified when searching for subprograms, if it can't find the
15340 subprograms using @env{GCC_EXEC_PREFIX}.
15343 @findex LIBRARY_PATH
15344 The value of @env{LIBRARY_PATH} is a colon-separated list of
15345 directories, much like @env{PATH}. When configured as a native compiler,
15346 GCC tries the directories thus specified when searching for special
15347 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15348 using GCC also uses these directories when searching for ordinary
15349 libraries for the @option{-l} option (but directories specified with
15350 @option{-L} come first).
15354 @cindex locale definition
15355 This variable is used to pass locale information to the compiler. One way in
15356 which this information is used is to determine the character set to be used
15357 when character literals, string literals and comments are parsed in C and C++.
15358 When the compiler is configured to allow multibyte characters,
15359 the following values for @env{LANG} are recognized:
15363 Recognize JIS characters.
15365 Recognize SJIS characters.
15367 Recognize EUCJP characters.
15370 If @env{LANG} is not defined, or if it has some other value, then the
15371 compiler will use mblen and mbtowc as defined by the default locale to
15372 recognize and translate multibyte characters.
15376 Some additional environments variables affect the behavior of the
15379 @include cppenv.texi
15383 @node Precompiled Headers
15384 @section Using Precompiled Headers
15385 @cindex precompiled headers
15386 @cindex speed of compilation
15388 Often large projects have many header files that are included in every
15389 source file. The time the compiler takes to process these header files
15390 over and over again can account for nearly all of the time required to
15391 build the project. To make builds faster, GCC allows users to
15392 `precompile' a header file; then, if builds can use the precompiled
15393 header file they will be much faster.
15395 To create a precompiled header file, simply compile it as you would any
15396 other file, if necessary using the @option{-x} option to make the driver
15397 treat it as a C or C++ header file. You will probably want to use a
15398 tool like @command{make} to keep the precompiled header up-to-date when
15399 the headers it contains change.
15401 A precompiled header file will be searched for when @code{#include} is
15402 seen in the compilation. As it searches for the included file
15403 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15404 compiler looks for a precompiled header in each directory just before it
15405 looks for the include file in that directory. The name searched for is
15406 the name specified in the @code{#include} with @samp{.gch} appended. If
15407 the precompiled header file can't be used, it is ignored.
15409 For instance, if you have @code{#include "all.h"}, and you have
15410 @file{all.h.gch} in the same directory as @file{all.h}, then the
15411 precompiled header file will be used if possible, and the original
15412 header will be used otherwise.
15414 Alternatively, you might decide to put the precompiled header file in a
15415 directory and use @option{-I} to ensure that directory is searched
15416 before (or instead of) the directory containing the original header.
15417 Then, if you want to check that the precompiled header file is always
15418 used, you can put a file of the same name as the original header in this
15419 directory containing an @code{#error} command.
15421 This also works with @option{-include}. So yet another way to use
15422 precompiled headers, good for projects not designed with precompiled
15423 header files in mind, is to simply take most of the header files used by
15424 a project, include them from another header file, precompile that header
15425 file, and @option{-include} the precompiled header. If the header files
15426 have guards against multiple inclusion, they will be skipped because
15427 they've already been included (in the precompiled header).
15429 If you need to precompile the same header file for different
15430 languages, targets, or compiler options, you can instead make a
15431 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15432 header in the directory, perhaps using @option{-o}. It doesn't matter
15433 what you call the files in the directory, every precompiled header in
15434 the directory will be considered. The first precompiled header
15435 encountered in the directory that is valid for this compilation will
15436 be used; they're searched in no particular order.
15438 There are many other possibilities, limited only by your imagination,
15439 good sense, and the constraints of your build system.
15441 A precompiled header file can be used only when these conditions apply:
15445 Only one precompiled header can be used in a particular compilation.
15448 A precompiled header can't be used once the first C token is seen. You
15449 can have preprocessor directives before a precompiled header; you can
15450 even include a precompiled header from inside another header, so long as
15451 there are no C tokens before the @code{#include}.
15454 The precompiled header file must be produced for the same language as
15455 the current compilation. You can't use a C precompiled header for a C++
15459 The precompiled header file must have been produced by the same compiler
15460 binary as the current compilation is using.
15463 Any macros defined before the precompiled header is included must
15464 either be defined in the same way as when the precompiled header was
15465 generated, or must not affect the precompiled header, which usually
15466 means that they don't appear in the precompiled header at all.
15468 The @option{-D} option is one way to define a macro before a
15469 precompiled header is included; using a @code{#define} can also do it.
15470 There are also some options that define macros implicitly, like
15471 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15474 @item If debugging information is output when using the precompiled
15475 header, using @option{-g} or similar, the same kind of debugging information
15476 must have been output when building the precompiled header. However,
15477 a precompiled header built using @option{-g} can be used in a compilation
15478 when no debugging information is being output.
15480 @item The same @option{-m} options must generally be used when building
15481 and using the precompiled header. @xref{Submodel Options},
15482 for any cases where this rule is relaxed.
15484 @item Each of the following options must be the same when building and using
15485 the precompiled header:
15487 @gccoptlist{-fexceptions -funit-at-a-time}
15490 Some other command-line options starting with @option{-f},
15491 @option{-p}, or @option{-O} must be defined in the same way as when
15492 the precompiled header was generated. At present, it's not clear
15493 which options are safe to change and which are not; the safest choice
15494 is to use exactly the same options when generating and using the
15495 precompiled header. The following are known to be safe:
15497 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15498 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15499 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15504 For all of these except the last, the compiler will automatically
15505 ignore the precompiled header if the conditions aren't met. If you
15506 find an option combination that doesn't work and doesn't cause the
15507 precompiled header to be ignored, please consider filing a bug report,
15510 If you do use differing options when generating and using the
15511 precompiled header, the actual behavior will be a mixture of the
15512 behavior for the options. For instance, if you use @option{-g} to
15513 generate the precompiled header but not when using it, you may or may
15514 not get debugging information for routines in the precompiled header.
15516 @node Running Protoize
15517 @section Running Protoize
15519 The program @code{protoize} is an optional part of GCC@. You can use
15520 it to add prototypes to a program, thus converting the program to ISO
15521 C in one respect. The companion program @code{unprotoize} does the
15522 reverse: it removes argument types from any prototypes that are found.
15524 When you run these programs, you must specify a set of source files as
15525 command line arguments. The conversion programs start out by compiling
15526 these files to see what functions they define. The information gathered
15527 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15529 After scanning comes actual conversion. The specified files are all
15530 eligible to be converted; any files they include (whether sources or
15531 just headers) are eligible as well.
15533 But not all the eligible files are converted. By default,
15534 @code{protoize} and @code{unprotoize} convert only source and header
15535 files in the current directory. You can specify additional directories
15536 whose files should be converted with the @option{-d @var{directory}}
15537 option. You can also specify particular files to exclude with the
15538 @option{-x @var{file}} option. A file is converted if it is eligible, its
15539 directory name matches one of the specified directory names, and its
15540 name within the directory has not been excluded.
15542 Basic conversion with @code{protoize} consists of rewriting most
15543 function definitions and function declarations to specify the types of
15544 the arguments. The only ones not rewritten are those for varargs
15547 @code{protoize} optionally inserts prototype declarations at the
15548 beginning of the source file, to make them available for any calls that
15549 precede the function's definition. Or it can insert prototype
15550 declarations with block scope in the blocks where undeclared functions
15553 Basic conversion with @code{unprotoize} consists of rewriting most
15554 function declarations to remove any argument types, and rewriting
15555 function definitions to the old-style pre-ISO form.
15557 Both conversion programs print a warning for any function declaration or
15558 definition that they can't convert. You can suppress these warnings
15561 The output from @code{protoize} or @code{unprotoize} replaces the
15562 original source file. The original file is renamed to a name ending
15563 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15564 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15565 for DOS) file already exists, then the source file is simply discarded.
15567 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15568 scan the program and collect information about the functions it uses.
15569 So neither of these programs will work until GCC is installed.
15571 Here is a table of the options you can use with @code{protoize} and
15572 @code{unprotoize}. Each option works with both programs unless
15576 @item -B @var{directory}
15577 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15578 usual directory (normally @file{/usr/local/lib}). This file contains
15579 prototype information about standard system functions. This option
15580 applies only to @code{protoize}.
15582 @item -c @var{compilation-options}
15583 Use @var{compilation-options} as the options when running @command{gcc} to
15584 produce the @samp{.X} files. The special option @option{-aux-info} is
15585 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15587 Note that the compilation options must be given as a single argument to
15588 @code{protoize} or @code{unprotoize}. If you want to specify several
15589 @command{gcc} options, you must quote the entire set of compilation options
15590 to make them a single word in the shell.
15592 There are certain @command{gcc} arguments that you cannot use, because they
15593 would produce the wrong kind of output. These include @option{-g},
15594 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15595 the @var{compilation-options}, they are ignored.
15598 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15599 systems) instead of @samp{.c}. This is convenient if you are converting
15600 a C program to C++. This option applies only to @code{protoize}.
15603 Add explicit global declarations. This means inserting explicit
15604 declarations at the beginning of each source file for each function
15605 that is called in the file and was not declared. These declarations
15606 precede the first function definition that contains a call to an
15607 undeclared function. This option applies only to @code{protoize}.
15609 @item -i @var{string}
15610 Indent old-style parameter declarations with the string @var{string}.
15611 This option applies only to @code{protoize}.
15613 @code{unprotoize} converts prototyped function definitions to old-style
15614 function definitions, where the arguments are declared between the
15615 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15616 uses five spaces as the indentation. If you want to indent with just
15617 one space instead, use @option{-i " "}.
15620 Keep the @samp{.X} files. Normally, they are deleted after conversion
15624 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15625 a prototype declaration for each function in each block which calls the
15626 function without any declaration. This option applies only to
15630 Make no real changes. This mode just prints information about the conversions
15631 that would have been done without @option{-n}.
15634 Make no @samp{.save} files. The original files are simply deleted.
15635 Use this option with caution.
15637 @item -p @var{program}
15638 Use the program @var{program} as the compiler. Normally, the name
15639 @file{gcc} is used.
15642 Work quietly. Most warnings are suppressed.
15645 Print the version number, just like @option{-v} for @command{gcc}.
15648 If you need special compiler options to compile one of your program's
15649 source files, then you should generate that file's @samp{.X} file
15650 specially, by running @command{gcc} on that source file with the
15651 appropriate options and the option @option{-aux-info}. Then run
15652 @code{protoize} on the entire set of files. @code{protoize} will use
15653 the existing @samp{.X} file because it is newer than the source file.
15657 gcc -Dfoo=bar file1.c -aux-info file1.X
15662 You need to include the special files along with the rest in the
15663 @code{protoize} command, even though their @samp{.X} files already
15664 exist, because otherwise they won't get converted.
15666 @xref{Protoize Caveats}, for more information on how to use
15667 @code{protoize} successfully.