1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
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{[},@dots{}@r{]]} --target-help @gol
167 --version -wrapper@@@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++ -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 -Wno-builtin-macro-redefined @gol
232 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
233 -Wchar-subscripts -Wclobbered -Wcomment @gol
234 -Wconversion -Wcoverage-mismatch -Wno-deprecated @gol
235 -Wno-deprecated-declarations -Wdisabled-optimization @gol
236 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
237 -Werror -Werror=* @gol
238 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
239 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
240 -Wformat-security -Wformat-y2k @gol
241 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
242 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
243 -Winit-self -Winline @gol
244 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
245 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
246 -Wlogical-op -Wlong-long @gol
247 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
248 -Wmissing-format-attribute -Wmissing-include-dirs @gol
249 -Wmissing-noreturn -Wno-mudflap @gol
250 -Wno-multichar -Wnonnull -Wno-overflow @gol
251 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
252 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
253 -Wpointer-arith -Wno-pointer-to-int-cast @gol
254 -Wredundant-decls @gol
255 -Wreturn-type -Wsequence-point -Wshadow @gol
256 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
257 -Wstrict-aliasing -Wstrict-aliasing=n @gol
258 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
259 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
260 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
261 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
262 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
263 -Wunused-value -Wunused-variable @gol
264 -Wvariadic-macros -Wvla @gol
265 -Wvolatile-register-var -Wwrite-strings}
267 @item C and Objective-C-only Warning Options
268 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
269 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
270 -Wold-style-declaration -Wold-style-definition @gol
271 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
272 -Wdeclaration-after-statement -Wpointer-sign}
274 @item Debugging Options
275 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
276 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
277 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
278 -fdump-noaddr -fdump-unnumbered @gol
279 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
280 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
281 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
282 -fdump-statistics @gol
284 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
288 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-nrv -fdump-tree-vect @gol
297 -fdump-tree-sink @gol
298 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
299 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
300 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
301 -ftree-vectorizer-verbose=@var{n} @gol
302 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
303 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
304 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
305 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
306 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
307 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
308 -ftest-coverage -ftime-report -fvar-tracking @gol
309 -g -g@var{level} -gcoff -gdwarf-2 @gol
310 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
311 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
312 -fdebug-prefix-map=@var{old}=@var{new} @gol
313 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
314 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
315 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
316 -print-multi-directory -print-multi-lib @gol
317 -print-prog-name=@var{program} -print-search-dirs -Q @gol
318 -print-sysroot -print-sysroot-headers-suffix @gol
321 @item Optimization Options
322 @xref{Optimize Options,,Options that Control Optimization}.
324 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
325 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
326 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
327 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
328 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
329 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
330 -fdata-sections -fdce -fdce @gol
331 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
332 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
333 -ffinite-math-only -ffloat-store -fforward-propagate @gol
334 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
335 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
336 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
337 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
338 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
339 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
340 -fira-region=@var{region} -fira-coalesce -fno-ira-share-save-slots @gol
341 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
342 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
343 -floop-block -floop-interchange -floop-strip-mine @gol
344 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
345 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
346 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
347 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
348 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
349 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
350 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
351 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
352 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
353 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
354 -fprofile-generate=@var{path} @gol
355 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
356 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
357 -freorder-blocks-and-partition -freorder-functions @gol
358 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
359 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
360 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
361 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
362 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
363 -fselective-scheduling -fselective-scheduling2 @gol
364 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
365 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
366 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
367 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
368 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
369 -ftree-copyrename -ftree-dce @gol
370 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
371 -ftree-loop-distribution @gol
372 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
373 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc @gol
374 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
375 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
376 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
377 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
378 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
380 --param @var{name}=@var{value}
381 -O -O0 -O1 -O2 -O3 -Os}
383 @item Preprocessor Options
384 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
385 @gccoptlist{-A@var{question}=@var{answer} @gol
386 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
387 -C -dD -dI -dM -dN @gol
388 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
389 -idirafter @var{dir} @gol
390 -include @var{file} -imacros @var{file} @gol
391 -iprefix @var{file} -iwithprefix @var{dir} @gol
392 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
393 -imultilib @var{dir} -isysroot @var{dir} @gol
394 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
395 -P -fworking-directory -remap @gol
396 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
397 -Xpreprocessor @var{option}}
399 @item Assembler Option
400 @xref{Assembler Options,,Passing Options to the Assembler}.
401 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
404 @xref{Link Options,,Options for Linking}.
405 @gccoptlist{@var{object-file-name} -l@var{library} @gol
406 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
407 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
408 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
411 @item Directory Options
412 @xref{Directory Options,,Options for Directory Search}.
413 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
414 -specs=@var{file} -I- --sysroot=@var{dir}}
417 @c I wrote this xref this way to avoid overfull hbox. -- rms
418 @xref{Target Options}.
419 @gccoptlist{-V @var{version} -b @var{machine}}
421 @item Machine Dependent Options
422 @xref{Submodel Options,,Hardware Models and Configurations}.
423 @c This list is ordered alphanumerically by subsection name.
424 @c Try and put the significant identifier (CPU or system) first,
425 @c so users have a clue at guessing where the ones they want will be.
428 @gccoptlist{-EB -EL @gol
429 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
430 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
433 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
434 -mabi=@var{name} @gol
435 -mapcs-stack-check -mno-apcs-stack-check @gol
436 -mapcs-float -mno-apcs-float @gol
437 -mapcs-reentrant -mno-apcs-reentrant @gol
438 -msched-prolog -mno-sched-prolog @gol
439 -mlittle-endian -mbig-endian -mwords-little-endian @gol
440 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
441 -mthumb-interwork -mno-thumb-interwork @gol
442 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
443 -mstructure-size-boundary=@var{n} @gol
444 -mabort-on-noreturn @gol
445 -mlong-calls -mno-long-calls @gol
446 -msingle-pic-base -mno-single-pic-base @gol
447 -mpic-register=@var{reg} @gol
448 -mnop-fun-dllimport @gol
449 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
450 -mpoke-function-name @gol
452 -mtpcs-frame -mtpcs-leaf-frame @gol
453 -mcaller-super-interworking -mcallee-super-interworking @gol
455 -mword-relocations @gol
456 -mfix-cortex-m3-ldrd}
459 @gccoptlist{-mmcu=@var{mcu} -msize -mno-interrupts @gol
460 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
462 @emph{Blackfin Options}
463 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
464 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
465 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
466 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
467 -mno-id-shared-library -mshared-library-id=@var{n} @gol
468 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
469 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
470 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
474 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
475 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
476 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
477 -mstack-align -mdata-align -mconst-align @gol
478 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
479 -melf -maout -melinux -mlinux -sim -sim2 @gol
480 -mmul-bug-workaround -mno-mul-bug-workaround}
483 @gccoptlist{-mmac -mpush-args}
485 @emph{Darwin Options}
486 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
487 -arch_only -bind_at_load -bundle -bundle_loader @gol
488 -client_name -compatibility_version -current_version @gol
490 -dependency-file -dylib_file -dylinker_install_name @gol
491 -dynamic -dynamiclib -exported_symbols_list @gol
492 -filelist -flat_namespace -force_cpusubtype_ALL @gol
493 -force_flat_namespace -headerpad_max_install_names @gol
495 -image_base -init -install_name -keep_private_externs @gol
496 -multi_module -multiply_defined -multiply_defined_unused @gol
497 -noall_load -no_dead_strip_inits_and_terms @gol
498 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
499 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
500 -private_bundle -read_only_relocs -sectalign @gol
501 -sectobjectsymbols -whyload -seg1addr @gol
502 -sectcreate -sectobjectsymbols -sectorder @gol
503 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
504 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
505 -segprot -segs_read_only_addr -segs_read_write_addr @gol
506 -single_module -static -sub_library -sub_umbrella @gol
507 -twolevel_namespace -umbrella -undefined @gol
508 -unexported_symbols_list -weak_reference_mismatches @gol
509 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
510 -mkernel -mone-byte-bool}
512 @emph{DEC Alpha Options}
513 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
514 -mieee -mieee-with-inexact -mieee-conformant @gol
515 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
516 -mtrap-precision=@var{mode} -mbuild-constants @gol
517 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
518 -mbwx -mmax -mfix -mcix @gol
519 -mfloat-vax -mfloat-ieee @gol
520 -mexplicit-relocs -msmall-data -mlarge-data @gol
521 -msmall-text -mlarge-text @gol
522 -mmemory-latency=@var{time}}
524 @emph{DEC Alpha/VMS Options}
525 @gccoptlist{-mvms-return-codes}
528 @gccoptlist{-msmall-model -mno-lsim}
531 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
532 -mhard-float -msoft-float @gol
533 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
534 -mdouble -mno-double @gol
535 -mmedia -mno-media -mmuladd -mno-muladd @gol
536 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
537 -mlinked-fp -mlong-calls -malign-labels @gol
538 -mlibrary-pic -macc-4 -macc-8 @gol
539 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
540 -moptimize-membar -mno-optimize-membar @gol
541 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
542 -mvliw-branch -mno-vliw-branch @gol
543 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
544 -mno-nested-cond-exec -mtomcat-stats @gol
548 @emph{GNU/Linux Options}
549 @gccoptlist{-muclibc}
551 @emph{H8/300 Options}
552 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
555 @gccoptlist{-march=@var{architecture-type} @gol
556 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
557 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
558 -mfixed-range=@var{register-range} @gol
559 -mjump-in-delay -mlinker-opt -mlong-calls @gol
560 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
561 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
562 -mno-jump-in-delay -mno-long-load-store @gol
563 -mno-portable-runtime -mno-soft-float @gol
564 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
565 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
566 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
567 -munix=@var{unix-std} -nolibdld -static -threads}
569 @emph{i386 and x86-64 Options}
570 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
571 -mfpmath=@var{unit} @gol
572 -masm=@var{dialect} -mno-fancy-math-387 @gol
573 -mno-fp-ret-in-387 -msoft-float @gol
574 -mno-wide-multiply -mrtd -malign-double @gol
575 -mpreferred-stack-boundary=@var{num}
576 -mincoming-stack-boundary=@var{num}
577 -mcld -mcx16 -msahf -mrecip @gol
578 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
580 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
581 -mthreads -mno-align-stringops -minline-all-stringops @gol
582 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
583 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
584 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
585 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
586 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
587 -mcmodel=@var{code-model} @gol
588 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
589 -mfused-madd -mno-fused-madd -msse2avx}
592 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
593 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
594 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
595 -minline-float-divide-max-throughput @gol
596 -minline-int-divide-min-latency @gol
597 -minline-int-divide-max-throughput @gol
598 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
599 -mno-dwarf2-asm -mearly-stop-bits @gol
600 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
601 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
602 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
603 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
604 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
605 -mno-sched-prefer-non-data-spec-insns @gol
606 -mno-sched-prefer-non-control-spec-insns @gol
607 -mno-sched-count-spec-in-critical-path}
609 @emph{M32R/D Options}
610 @gccoptlist{-m32r2 -m32rx -m32r @gol
612 -malign-loops -mno-align-loops @gol
613 -missue-rate=@var{number} @gol
614 -mbranch-cost=@var{number} @gol
615 -mmodel=@var{code-size-model-type} @gol
616 -msdata=@var{sdata-type} @gol
617 -mno-flush-func -mflush-func=@var{name} @gol
618 -mno-flush-trap -mflush-trap=@var{number} @gol
622 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
624 @emph{M680x0 Options}
625 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
626 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
627 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
628 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
629 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
630 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
631 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
632 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
635 @emph{M68hc1x Options}
636 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
637 -mauto-incdec -minmax -mlong-calls -mshort @gol
638 -msoft-reg-count=@var{count}}
641 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
642 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
643 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
644 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
645 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
648 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
649 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
650 -mips64 -mips64r2 @gol
651 -mips16 -mno-mips16 -mflip-mips16 @gol
652 -minterlink-mips16 -mno-interlink-mips16 @gol
653 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
654 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
655 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
656 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
657 -mfpu=@var{fpu-type} @gol
658 -msmartmips -mno-smartmips @gol
659 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
660 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
661 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
662 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
663 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
664 -membedded-data -mno-embedded-data @gol
665 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
666 -mcode-readable=@var{setting} @gol
667 -msplit-addresses -mno-split-addresses @gol
668 -mexplicit-relocs -mno-explicit-relocs @gol
669 -mcheck-zero-division -mno-check-zero-division @gol
670 -mdivide-traps -mdivide-breaks @gol
671 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
672 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
673 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
674 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
675 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
676 -mflush-func=@var{func} -mno-flush-func @gol
677 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
678 -mfp-exceptions -mno-fp-exceptions @gol
679 -mvr4130-align -mno-vr4130-align}
682 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
683 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
684 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
685 -mno-base-addresses -msingle-exit -mno-single-exit}
687 @emph{MN10300 Options}
688 @gccoptlist{-mmult-bug -mno-mult-bug @gol
689 -mam33 -mno-am33 @gol
690 -mam33-2 -mno-am33-2 @gol
691 -mreturn-pointer-on-d0 @gol
694 @emph{PDP-11 Options}
695 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
696 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
697 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
698 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
699 -mbranch-expensive -mbranch-cheap @gol
700 -msplit -mno-split -munix-asm -mdec-asm}
702 @emph{picoChip Options}
703 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
704 -msymbol-as-address -mno-inefficient-warnings}
706 @emph{PowerPC Options}
707 See RS/6000 and PowerPC Options.
709 @emph{RS/6000 and PowerPC Options}
710 @gccoptlist{-mcpu=@var{cpu-type} @gol
711 -mtune=@var{cpu-type} @gol
712 -mpower -mno-power -mpower2 -mno-power2 @gol
713 -mpowerpc -mpowerpc64 -mno-powerpc @gol
714 -maltivec -mno-altivec @gol
715 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
716 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
717 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
718 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
719 -mnew-mnemonics -mold-mnemonics @gol
720 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
721 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
722 -malign-power -malign-natural @gol
723 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
724 -msingle-float -mdouble-float -msimple-fpu @gol
725 -mstring -mno-string -mupdate -mno-update @gol
726 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
727 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
728 -mstrict-align -mno-strict-align -mrelocatable @gol
729 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
730 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
731 -mdynamic-no-pic -maltivec -mswdiv @gol
732 -mprioritize-restricted-insns=@var{priority} @gol
733 -msched-costly-dep=@var{dependence_type} @gol
734 -minsert-sched-nops=@var{scheme} @gol
735 -mcall-sysv -mcall-netbsd @gol
736 -maix-struct-return -msvr4-struct-return @gol
737 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
738 -misel -mno-isel @gol
739 -misel=yes -misel=no @gol
741 -mspe=yes -mspe=no @gol
743 -mgen-cell-microcode -mwarn-cell-microcode @gol
744 -mvrsave -mno-vrsave @gol
745 -mmulhw -mno-mulhw @gol
746 -mdlmzb -mno-dlmzb @gol
747 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
748 -mprototype -mno-prototype @gol
749 -msim -mmvme -mads -myellowknife -memb -msdata @gol
750 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
752 @emph{S/390 and zSeries Options}
753 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
754 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
755 -mlong-double-64 -mlong-double-128 @gol
756 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
757 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
758 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
759 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
760 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
763 @gccoptlist{-meb -mel @gol
767 -mscore5 -mscore5u -mscore7 -mscore7d}
770 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
771 -m4-nofpu -m4-single-only -m4-single -m4 @gol
772 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
773 -m5-64media -m5-64media-nofpu @gol
774 -m5-32media -m5-32media-nofpu @gol
775 -m5-compact -m5-compact-nofpu @gol
776 -mb -ml -mdalign -mrelax @gol
777 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
778 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
779 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
780 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
781 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
785 @gccoptlist{-mcpu=@var{cpu-type} @gol
786 -mtune=@var{cpu-type} @gol
787 -mcmodel=@var{code-model} @gol
788 -m32 -m64 -mapp-regs -mno-app-regs @gol
789 -mfaster-structs -mno-faster-structs @gol
790 -mfpu -mno-fpu -mhard-float -msoft-float @gol
791 -mhard-quad-float -msoft-quad-float @gol
792 -mimpure-text -mno-impure-text -mlittle-endian @gol
793 -mstack-bias -mno-stack-bias @gol
794 -munaligned-doubles -mno-unaligned-doubles @gol
795 -mv8plus -mno-v8plus -mvis -mno-vis
796 -threads -pthreads -pthread}
799 @gccoptlist{-mwarn-reloc -merror-reloc @gol
800 -msafe-dma -munsafe-dma @gol
802 -msmall-mem -mlarge-mem -mstdmain @gol
803 -mfixed-range=@var{register-range}}
805 @emph{System V Options}
806 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
809 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
810 -mprolog-function -mno-prolog-function -mspace @gol
811 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
812 -mapp-regs -mno-app-regs @gol
813 -mdisable-callt -mno-disable-callt @gol
819 @gccoptlist{-mg -mgnu -munix}
821 @emph{VxWorks Options}
822 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
823 -Xbind-lazy -Xbind-now}
825 @emph{x86-64 Options}
826 See i386 and x86-64 Options.
828 @emph{i386 and x86-64 Windows Options}
829 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
830 -mnop-fun-dllimport -mthread -mwin32 -mwindows}
832 @emph{Xstormy16 Options}
835 @emph{Xtensa Options}
836 @gccoptlist{-mconst16 -mno-const16 @gol
837 -mfused-madd -mno-fused-madd @gol
838 -mserialize-volatile -mno-serialize-volatile @gol
839 -mtext-section-literals -mno-text-section-literals @gol
840 -mtarget-align -mno-target-align @gol
841 -mlongcalls -mno-longcalls}
843 @emph{zSeries Options}
844 See S/390 and zSeries Options.
846 @item Code Generation Options
847 @xref{Code Gen Options,,Options for Code Generation Conventions}.
848 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
849 -ffixed-@var{reg} -fexceptions @gol
850 -fnon-call-exceptions -funwind-tables @gol
851 -fasynchronous-unwind-tables @gol
852 -finhibit-size-directive -finstrument-functions @gol
853 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
854 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
855 -fno-common -fno-ident @gol
856 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
857 -fno-jump-tables @gol
858 -frecord-gcc-switches @gol
859 -freg-struct-return -fshort-enums @gol
860 -fshort-double -fshort-wchar @gol
861 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
862 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
863 -fno-stack-limit -fargument-alias -fargument-noalias @gol
864 -fargument-noalias-global -fargument-noalias-anything @gol
865 -fleading-underscore -ftls-model=@var{model} @gol
866 -ftrapv -fwrapv -fbounds-check @gol
871 * Overall Options:: Controlling the kind of output:
872 an executable, object files, assembler files,
873 or preprocessed source.
874 * C Dialect Options:: Controlling the variant of C language compiled.
875 * C++ Dialect Options:: Variations on C++.
876 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
878 * Language Independent Options:: Controlling how diagnostics should be
880 * Warning Options:: How picky should the compiler be?
881 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
882 * Optimize Options:: How much optimization?
883 * Preprocessor Options:: Controlling header files and macro definitions.
884 Also, getting dependency information for Make.
885 * Assembler Options:: Passing options to the assembler.
886 * Link Options:: Specifying libraries and so on.
887 * Directory Options:: Where to find header files and libraries.
888 Where to find the compiler executable files.
889 * Spec Files:: How to pass switches to sub-processes.
890 * Target Options:: Running a cross-compiler, or an old version of GCC.
893 @node Overall Options
894 @section Options Controlling the Kind of Output
896 Compilation can involve up to four stages: preprocessing, compilation
897 proper, assembly and linking, always in that order. GCC is capable of
898 preprocessing and compiling several files either into several
899 assembler input files, or into one assembler input file; then each
900 assembler input file produces an object file, and linking combines all
901 the object files (those newly compiled, and those specified as input)
902 into an executable file.
904 @cindex file name suffix
905 For any given input file, the file name suffix determines what kind of
910 C source code which must be preprocessed.
913 C source code which should not be preprocessed.
916 C++ source code which should not be preprocessed.
919 Objective-C source code. Note that you must link with the @file{libobjc}
920 library to make an Objective-C program work.
923 Objective-C source code which should not be preprocessed.
927 Objective-C++ source code. Note that you must link with the @file{libobjc}
928 library to make an Objective-C++ program work. Note that @samp{.M} refers
929 to a literal capital M@.
932 Objective-C++ source code which should not be preprocessed.
935 C, C++, Objective-C or Objective-C++ header file to be turned into a
940 @itemx @var{file}.cxx
941 @itemx @var{file}.cpp
942 @itemx @var{file}.CPP
943 @itemx @var{file}.c++
945 C++ source code which must be preprocessed. Note that in @samp{.cxx},
946 the last two letters must both be literally @samp{x}. Likewise,
947 @samp{.C} refers to a literal capital C@.
951 Objective-C++ source code which must be preprocessed.
954 Objective-C++ source code which should not be preprocessed.
959 @itemx @var{file}.hxx
960 @itemx @var{file}.hpp
961 @itemx @var{file}.HPP
962 @itemx @var{file}.h++
963 @itemx @var{file}.tcc
964 C++ header file to be turned into a precompiled header.
967 @itemx @var{file}.for
968 @itemx @var{file}.ftn
969 Fixed form Fortran source code which should not be preprocessed.
972 @itemx @var{file}.FOR
973 @itemx @var{file}.fpp
974 @itemx @var{file}.FPP
975 @itemx @var{file}.FTN
976 Fixed form Fortran source code which must be preprocessed (with the traditional
980 @itemx @var{file}.f95
981 @itemx @var{file}.f03
982 @itemx @var{file}.f08
983 Free form Fortran source code which should not be preprocessed.
986 @itemx @var{file}.F95
987 @itemx @var{file}.F03
988 @itemx @var{file}.F08
989 Free form Fortran source code which must be preprocessed (with the
990 traditional preprocessor).
992 @c FIXME: Descriptions of Java file types.
999 Ada source code file which contains a library unit declaration (a
1000 declaration of a package, subprogram, or generic, or a generic
1001 instantiation), or a library unit renaming declaration (a package,
1002 generic, or subprogram renaming declaration). Such files are also
1005 @item @var{file}.adb
1006 Ada source code file containing a library unit body (a subprogram or
1007 package body). Such files are also called @dfn{bodies}.
1009 @c GCC also knows about some suffixes for languages not yet included:
1020 @itemx @var{file}.sx
1021 Assembler code which must be preprocessed.
1024 An object file to be fed straight into linking.
1025 Any file name with no recognized suffix is treated this way.
1029 You can specify the input language explicitly with the @option{-x} option:
1032 @item -x @var{language}
1033 Specify explicitly the @var{language} for the following input files
1034 (rather than letting the compiler choose a default based on the file
1035 name suffix). This option applies to all following input files until
1036 the next @option{-x} option. Possible values for @var{language} are:
1038 c c-header c-cpp-output
1039 c++ c++-header c++-cpp-output
1040 objective-c objective-c-header objective-c-cpp-output
1041 objective-c++ objective-c++-header objective-c++-cpp-output
1042 assembler assembler-with-cpp
1044 f77 f77-cpp-input f95 f95-cpp-input
1049 Turn off any specification of a language, so that subsequent files are
1050 handled according to their file name suffixes (as they are if @option{-x}
1051 has not been used at all).
1053 @item -pass-exit-codes
1054 @opindex pass-exit-codes
1055 Normally the @command{gcc} program will exit with the code of 1 if any
1056 phase of the compiler returns a non-success return code. If you specify
1057 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1058 numerically highest error produced by any phase that returned an error
1059 indication. The C, C++, and Fortran frontends return 4, if an internal
1060 compiler error is encountered.
1063 If you only want some of the stages of compilation, you can use
1064 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1065 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1066 @command{gcc} is to stop. Note that some combinations (for example,
1067 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1072 Compile or assemble the source files, but do not link. The linking
1073 stage simply is not done. The ultimate output is in the form of an
1074 object file for each source file.
1076 By default, the object file name for a source file is made by replacing
1077 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1079 Unrecognized input files, not requiring compilation or assembly, are
1084 Stop after the stage of compilation proper; do not assemble. The output
1085 is in the form of an assembler code file for each non-assembler input
1088 By default, the assembler file name for a source file is made by
1089 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1091 Input files that don't require compilation are ignored.
1095 Stop after the preprocessing stage; do not run the compiler proper. The
1096 output is in the form of preprocessed source code, which is sent to the
1099 Input files which don't require preprocessing are ignored.
1101 @cindex output file option
1104 Place output in file @var{file}. This applies regardless to whatever
1105 sort of output is being produced, whether it be an executable file,
1106 an object file, an assembler file or preprocessed C code.
1108 If @option{-o} is not specified, the default is to put an executable
1109 file in @file{a.out}, the object file for
1110 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1111 assembler file in @file{@var{source}.s}, a precompiled header file in
1112 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1117 Print (on standard error output) the commands executed to run the stages
1118 of compilation. Also print the version number of the compiler driver
1119 program and of the preprocessor and the compiler proper.
1123 Like @option{-v} except the commands are not executed and all command
1124 arguments are quoted. This is useful for shell scripts to capture the
1125 driver-generated command lines.
1129 Use pipes rather than temporary files for communication between the
1130 various stages of compilation. This fails to work on some systems where
1131 the assembler is unable to read from a pipe; but the GNU assembler has
1136 If you are compiling multiple source files, this option tells the driver
1137 to pass all the source files to the compiler at once (for those
1138 languages for which the compiler can handle this). This will allow
1139 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1140 language for which this is supported is C@. If you pass source files for
1141 multiple languages to the driver, using this option, the driver will invoke
1142 the compiler(s) that support IMA once each, passing each compiler all the
1143 source files appropriate for it. For those languages that do not support
1144 IMA this option will be ignored, and the compiler will be invoked once for
1145 each source file in that language. If you use this option in conjunction
1146 with @option{-save-temps}, the compiler will generate multiple
1148 (one for each source file), but only one (combined) @file{.o} or
1153 Print (on the standard output) a description of the command line options
1154 understood by @command{gcc}. If the @option{-v} option is also specified
1155 then @option{--help} will also be passed on to the various processes
1156 invoked by @command{gcc}, so that they can display the command line options
1157 they accept. If the @option{-Wextra} option has also been specified
1158 (prior to the @option{--help} option), then command line options which
1159 have no documentation associated with them will also be displayed.
1162 @opindex target-help
1163 Print (on the standard output) a description of target-specific command
1164 line options for each tool. For some targets extra target-specific
1165 information may also be printed.
1167 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1168 Print (on the standard output) a description of the command line
1169 options understood by the compiler that fit into all specified classes
1170 and qualifiers. These are the supported classes:
1173 @item @samp{optimizers}
1174 This will display all of the optimization options supported by the
1177 @item @samp{warnings}
1178 This will display all of the options controlling warning messages
1179 produced by the compiler.
1182 This will display target-specific options. Unlike the
1183 @option{--target-help} option however, target-specific options of the
1184 linker and assembler will not be displayed. This is because those
1185 tools do not currently support the extended @option{--help=} syntax.
1188 This will display the values recognized by the @option{--param}
1191 @item @var{language}
1192 This will display the options supported for @var{language}, where
1193 @var{language} is the name of one of the languages supported in this
1197 This will display the options that are common to all languages.
1200 These are the supported qualifiers:
1203 @item @samp{undocumented}
1204 Display only those options which are undocumented.
1207 Display options which take an argument that appears after an equal
1208 sign in the same continuous piece of text, such as:
1209 @samp{--help=target}.
1211 @item @samp{separate}
1212 Display options which take an argument that appears as a separate word
1213 following the original option, such as: @samp{-o output-file}.
1216 Thus for example to display all the undocumented target-specific
1217 switches supported by the compiler the following can be used:
1220 --help=target,undocumented
1223 The sense of a qualifier can be inverted by prefixing it with the
1224 @samp{^} character, so for example to display all binary warning
1225 options (i.e., ones that are either on or off and that do not take an
1226 argument), which have a description the following can be used:
1229 --help=warnings,^joined,^undocumented
1232 The argument to @option{--help=} should not consist solely of inverted
1235 Combining several classes is possible, although this usually
1236 restricts the output by so much that there is nothing to display. One
1237 case where it does work however is when one of the classes is
1238 @var{target}. So for example to display all the target-specific
1239 optimization options the following can be used:
1242 --help=target,optimizers
1245 The @option{--help=} option can be repeated on the command line. Each
1246 successive use will display its requested class of options, skipping
1247 those that have already been displayed.
1249 If the @option{-Q} option appears on the command line before the
1250 @option{--help=} option, then the descriptive text displayed by
1251 @option{--help=} is changed. Instead of describing the displayed
1252 options, an indication is given as to whether the option is enabled,
1253 disabled or set to a specific value (assuming that the compiler
1254 knows this at the point where the @option{--help=} option is used).
1256 Here is a truncated example from the ARM port of @command{gcc}:
1259 % gcc -Q -mabi=2 --help=target -c
1260 The following options are target specific:
1262 -mabort-on-noreturn [disabled]
1266 The output is sensitive to the effects of previous command line
1267 options, so for example it is possible to find out which optimizations
1268 are enabled at @option{-O2} by using:
1271 -Q -O2 --help=optimizers
1274 Alternatively you can discover which binary optimizations are enabled
1275 by @option{-O3} by using:
1278 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1279 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1280 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1285 Display the version number and copyrights of the invoked GCC@.
1289 Invoke all subcommands under a wrapper program. It takes a single
1290 comma separated list as an argument, which will be used to invoke
1294 gcc -c t.c -wrapper gdb,--args
1297 This will invoke all subprograms of gcc under "gdb --args",
1298 thus cc1 invocation will be "gdb --args cc1 ...".
1300 @include @value{srcdir}/../libiberty/at-file.texi
1304 @section Compiling C++ Programs
1306 @cindex suffixes for C++ source
1307 @cindex C++ source file suffixes
1308 C++ source files conventionally use one of the suffixes @samp{.C},
1309 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1310 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1311 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1312 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1313 files with these names and compiles them as C++ programs even if you
1314 call the compiler the same way as for compiling C programs (usually
1315 with the name @command{gcc}).
1319 However, the use of @command{gcc} does not add the C++ library.
1320 @command{g++} is a program that calls GCC and treats @samp{.c},
1321 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1322 files unless @option{-x} is used, and automatically specifies linking
1323 against the C++ library. This program is also useful when
1324 precompiling a C header file with a @samp{.h} extension for use in C++
1325 compilations. On many systems, @command{g++} is also installed with
1326 the name @command{c++}.
1328 @cindex invoking @command{g++}
1329 When you compile C++ programs, you may specify many of the same
1330 command-line options that you use for compiling programs in any
1331 language; or command-line options meaningful for C and related
1332 languages; or options that are meaningful only for C++ programs.
1333 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1334 explanations of options for languages related to C@.
1335 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1336 explanations of options that are meaningful only for C++ programs.
1338 @node C Dialect Options
1339 @section Options Controlling C Dialect
1340 @cindex dialect options
1341 @cindex language dialect options
1342 @cindex options, dialect
1344 The following options control the dialect of C (or languages derived
1345 from C, such as C++, Objective-C and Objective-C++) that the compiler
1349 @cindex ANSI support
1353 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1354 equivalent to @samp{-std=c++98}.
1356 This turns off certain features of GCC that are incompatible with ISO
1357 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1358 such as the @code{asm} and @code{typeof} keywords, and
1359 predefined macros such as @code{unix} and @code{vax} that identify the
1360 type of system you are using. It also enables the undesirable and
1361 rarely used ISO trigraph feature. For the C compiler,
1362 it disables recognition of C++ style @samp{//} comments as well as
1363 the @code{inline} keyword.
1365 The alternate keywords @code{__asm__}, @code{__extension__},
1366 @code{__inline__} and @code{__typeof__} continue to work despite
1367 @option{-ansi}. You would not want to use them in an ISO C program, of
1368 course, but it is useful to put them in header files that might be included
1369 in compilations done with @option{-ansi}. Alternate predefined macros
1370 such as @code{__unix__} and @code{__vax__} are also available, with or
1371 without @option{-ansi}.
1373 The @option{-ansi} option does not cause non-ISO programs to be
1374 rejected gratuitously. For that, @option{-pedantic} is required in
1375 addition to @option{-ansi}. @xref{Warning Options}.
1377 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1378 option is used. Some header files may notice this macro and refrain
1379 from declaring certain functions or defining certain macros that the
1380 ISO standard doesn't call for; this is to avoid interfering with any
1381 programs that might use these names for other things.
1383 Functions that would normally be built in but do not have semantics
1384 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1385 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1386 built-in functions provided by GCC}, for details of the functions
1391 Determine the language standard. @xref{Standards,,Language Standards
1392 Supported by GCC}, for details of these standard versions. This option
1393 is currently only supported when compiling C or C++.
1395 The compiler can accept several base standards, such as @samp{c89} or
1396 @samp{c++98}, and GNU dialects of those standards, such as
1397 @samp{gnu89} or @samp{gnu++98}. By specifying a base standard, the
1398 compiler will accept all programs following that standard and those
1399 using GNU extensions that do not contradict it. For example,
1400 @samp{-std=c89} turns off certain features of GCC that are
1401 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1402 keywords, but not other GNU extensions that do not have a meaning in
1403 ISO C90, such as omitting the middle term of a @code{?:}
1404 expression. On the other hand, by specifying a GNU dialect of a
1405 standard, all features the compiler support are enabled, even when
1406 those features change the meaning of the base standard and some
1407 strict-conforming programs may be rejected. The particular standard
1408 is used by @option{-pedantic} to identify which features are GNU
1409 extensions given that version of the standard. For example
1410 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1411 comments, while @samp{-std=gnu99 -pedantic} would not.
1413 A value for this option must be provided; possible values are
1418 Support all ISO C90 programs (certain GNU extensions that conflict
1419 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1421 @item iso9899:199409
1422 ISO C90 as modified in amendment 1.
1428 ISO C99. Note that this standard is not yet fully supported; see
1429 @w{@uref{http://gcc.gnu.org/gcc-4.4/c99status.html}} for more information. The
1430 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1433 GNU dialect of ISO C90 (including some C99 features). This
1434 is the default for C code.
1438 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1439 this will become the default. The name @samp{gnu9x} is deprecated.
1442 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1446 GNU dialect of @option{-std=c++98}. This is the default for
1450 The working draft of the upcoming ISO C++0x standard. This option
1451 enables experimental features that are likely to be included in
1452 C++0x. The working draft is constantly changing, and any feature that is
1453 enabled by this flag may be removed from future versions of GCC if it is
1454 not part of the C++0x standard.
1457 GNU dialect of @option{-std=c++0x}. This option enables
1458 experimental features that may be removed in future versions of GCC.
1461 @item -fgnu89-inline
1462 @opindex fgnu89-inline
1463 The option @option{-fgnu89-inline} tells GCC to use the traditional
1464 GNU semantics for @code{inline} functions when in C99 mode.
1465 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1466 is accepted and ignored by GCC versions 4.1.3 up to but not including
1467 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1468 C99 mode. Using this option is roughly equivalent to adding the
1469 @code{gnu_inline} function attribute to all inline functions
1470 (@pxref{Function Attributes}).
1472 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1473 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1474 specifies the default behavior). This option was first supported in
1475 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1477 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1478 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1479 in effect for @code{inline} functions. @xref{Common Predefined
1480 Macros,,,cpp,The C Preprocessor}.
1482 @item -aux-info @var{filename}
1484 Output to the given filename prototyped declarations for all functions
1485 declared and/or defined in a translation unit, including those in header
1486 files. This option is silently ignored in any language other than C@.
1488 Besides declarations, the file indicates, in comments, the origin of
1489 each declaration (source file and line), whether the declaration was
1490 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1491 @samp{O} for old, respectively, in the first character after the line
1492 number and the colon), and whether it came from a declaration or a
1493 definition (@samp{C} or @samp{F}, respectively, in the following
1494 character). In the case of function definitions, a K&R-style list of
1495 arguments followed by their declarations is also provided, inside
1496 comments, after the declaration.
1500 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1501 keyword, so that code can use these words as identifiers. You can use
1502 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1503 instead. @option{-ansi} implies @option{-fno-asm}.
1505 In C++, this switch only affects the @code{typeof} keyword, since
1506 @code{asm} and @code{inline} are standard keywords. You may want to
1507 use the @option{-fno-gnu-keywords} flag instead, which has the same
1508 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1509 switch only affects the @code{asm} and @code{typeof} keywords, since
1510 @code{inline} is a standard keyword in ISO C99.
1513 @itemx -fno-builtin-@var{function}
1514 @opindex fno-builtin
1515 @cindex built-in functions
1516 Don't recognize built-in functions that do not begin with
1517 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1518 functions provided by GCC}, for details of the functions affected,
1519 including those which are not built-in functions when @option{-ansi} or
1520 @option{-std} options for strict ISO C conformance are used because they
1521 do not have an ISO standard meaning.
1523 GCC normally generates special code to handle certain built-in functions
1524 more efficiently; for instance, calls to @code{alloca} may become single
1525 instructions that adjust the stack directly, and calls to @code{memcpy}
1526 may become inline copy loops. The resulting code is often both smaller
1527 and faster, but since the function calls no longer appear as such, you
1528 cannot set a breakpoint on those calls, nor can you change the behavior
1529 of the functions by linking with a different library. In addition,
1530 when a function is recognized as a built-in function, GCC may use
1531 information about that function to warn about problems with calls to
1532 that function, or to generate more efficient code, even if the
1533 resulting code still contains calls to that function. For example,
1534 warnings are given with @option{-Wformat} for bad calls to
1535 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1536 known not to modify global memory.
1538 With the @option{-fno-builtin-@var{function}} option
1539 only the built-in function @var{function} is
1540 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1541 function is named that is not built-in in this version of GCC, this
1542 option is ignored. There is no corresponding
1543 @option{-fbuiltin-@var{function}} option; if you wish to enable
1544 built-in functions selectively when using @option{-fno-builtin} or
1545 @option{-ffreestanding}, you may define macros such as:
1548 #define abs(n) __builtin_abs ((n))
1549 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1554 @cindex hosted environment
1556 Assert that compilation takes place in a hosted environment. This implies
1557 @option{-fbuiltin}. A hosted environment is one in which the
1558 entire standard library is available, and in which @code{main} has a return
1559 type of @code{int}. Examples are nearly everything except a kernel.
1560 This is equivalent to @option{-fno-freestanding}.
1562 @item -ffreestanding
1563 @opindex ffreestanding
1564 @cindex hosted environment
1566 Assert that compilation takes place in a freestanding environment. This
1567 implies @option{-fno-builtin}. A freestanding environment
1568 is one in which the standard library may not exist, and program startup may
1569 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1570 This is equivalent to @option{-fno-hosted}.
1572 @xref{Standards,,Language Standards Supported by GCC}, for details of
1573 freestanding and hosted environments.
1577 @cindex openmp parallel
1578 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1579 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1580 compiler generates parallel code according to the OpenMP Application
1581 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1582 implies @option{-pthread}, and thus is only supported on targets that
1583 have support for @option{-pthread}.
1585 @item -fms-extensions
1586 @opindex fms-extensions
1587 Accept some non-standard constructs used in Microsoft header files.
1589 Some cases of unnamed fields in structures and unions are only
1590 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1591 fields within structs/unions}, for details.
1595 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1596 options for strict ISO C conformance) implies @option{-trigraphs}.
1598 @item -no-integrated-cpp
1599 @opindex no-integrated-cpp
1600 Performs a compilation in two passes: preprocessing and compiling. This
1601 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1602 @option{-B} option. The user supplied compilation step can then add in
1603 an additional preprocessing step after normal preprocessing but before
1604 compiling. The default is to use the integrated cpp (internal cpp)
1606 The semantics of this option will change if "cc1", "cc1plus", and
1607 "cc1obj" are merged.
1609 @cindex traditional C language
1610 @cindex C language, traditional
1612 @itemx -traditional-cpp
1613 @opindex traditional-cpp
1614 @opindex traditional
1615 Formerly, these options caused GCC to attempt to emulate a pre-standard
1616 C compiler. They are now only supported with the @option{-E} switch.
1617 The preprocessor continues to support a pre-standard mode. See the GNU
1618 CPP manual for details.
1620 @item -fcond-mismatch
1621 @opindex fcond-mismatch
1622 Allow conditional expressions with mismatched types in the second and
1623 third arguments. The value of such an expression is void. This option
1624 is not supported for C++.
1626 @item -flax-vector-conversions
1627 @opindex flax-vector-conversions
1628 Allow implicit conversions between vectors with differing numbers of
1629 elements and/or incompatible element types. This option should not be
1632 @item -funsigned-char
1633 @opindex funsigned-char
1634 Let the type @code{char} be unsigned, like @code{unsigned char}.
1636 Each kind of machine has a default for what @code{char} should
1637 be. It is either like @code{unsigned char} by default or like
1638 @code{signed char} by default.
1640 Ideally, a portable program should always use @code{signed char} or
1641 @code{unsigned char} when it depends on the signedness of an object.
1642 But many programs have been written to use plain @code{char} and
1643 expect it to be signed, or expect it to be unsigned, depending on the
1644 machines they were written for. This option, and its inverse, let you
1645 make such a program work with the opposite default.
1647 The type @code{char} is always a distinct type from each of
1648 @code{signed char} or @code{unsigned char}, even though its behavior
1649 is always just like one of those two.
1652 @opindex fsigned-char
1653 Let the type @code{char} be signed, like @code{signed char}.
1655 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1656 the negative form of @option{-funsigned-char}. Likewise, the option
1657 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1659 @item -fsigned-bitfields
1660 @itemx -funsigned-bitfields
1661 @itemx -fno-signed-bitfields
1662 @itemx -fno-unsigned-bitfields
1663 @opindex fsigned-bitfields
1664 @opindex funsigned-bitfields
1665 @opindex fno-signed-bitfields
1666 @opindex fno-unsigned-bitfields
1667 These options control whether a bit-field is signed or unsigned, when the
1668 declaration does not use either @code{signed} or @code{unsigned}. By
1669 default, such a bit-field is signed, because this is consistent: the
1670 basic integer types such as @code{int} are signed types.
1673 @node C++ Dialect Options
1674 @section Options Controlling C++ Dialect
1676 @cindex compiler options, C++
1677 @cindex C++ options, command line
1678 @cindex options, C++
1679 This section describes the command-line options that are only meaningful
1680 for C++ programs; but you can also use most of the GNU compiler options
1681 regardless of what language your program is in. For example, you
1682 might compile a file @code{firstClass.C} like this:
1685 g++ -g -frepo -O -c firstClass.C
1689 In this example, only @option{-frepo} is an option meant
1690 only for C++ programs; you can use the other options with any
1691 language supported by GCC@.
1693 Here is a list of options that are @emph{only} for compiling C++ programs:
1697 @item -fabi-version=@var{n}
1698 @opindex fabi-version
1699 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1700 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1701 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1702 the version that conforms most closely to the C++ ABI specification.
1703 Therefore, the ABI obtained using version 0 will change as ABI bugs
1706 The default is version 2.
1708 @item -fno-access-control
1709 @opindex fno-access-control
1710 Turn off all access checking. This switch is mainly useful for working
1711 around bugs in the access control code.
1715 Check that the pointer returned by @code{operator new} is non-null
1716 before attempting to modify the storage allocated. This check is
1717 normally unnecessary because the C++ standard specifies that
1718 @code{operator new} will only return @code{0} if it is declared
1719 @samp{throw()}, in which case the compiler will always check the
1720 return value even without this option. In all other cases, when
1721 @code{operator new} has a non-empty exception specification, memory
1722 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1723 @samp{new (nothrow)}.
1725 @item -fconserve-space
1726 @opindex fconserve-space
1727 Put uninitialized or runtime-initialized global variables into the
1728 common segment, as C does. This saves space in the executable at the
1729 cost of not diagnosing duplicate definitions. If you compile with this
1730 flag and your program mysteriously crashes after @code{main()} has
1731 completed, you may have an object that is being destroyed twice because
1732 two definitions were merged.
1734 This option is no longer useful on most targets, now that support has
1735 been added for putting variables into BSS without making them common.
1737 @item -fno-deduce-init-list
1738 @opindex fno-deduce-init-list
1739 Disable deduction of a template type parameter as
1740 std::initializer_list from a brace-enclosed initializer list, i.e.
1743 template <class T> auto forward(T t) -> decltype (realfn (t))
1750 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1754 This option is present because this deduction is an extension to the
1755 current specification in the C++0x working draft, and there was
1756 some concern about potential overload resolution problems.
1758 @item -ffriend-injection
1759 @opindex ffriend-injection
1760 Inject friend functions into the enclosing namespace, so that they are
1761 visible outside the scope of the class in which they are declared.
1762 Friend functions were documented to work this way in the old Annotated
1763 C++ Reference Manual, and versions of G++ before 4.1 always worked
1764 that way. However, in ISO C++ a friend function which is not declared
1765 in an enclosing scope can only be found using argument dependent
1766 lookup. This option causes friends to be injected as they were in
1769 This option is for compatibility, and may be removed in a future
1772 @item -fno-elide-constructors
1773 @opindex fno-elide-constructors
1774 The C++ standard allows an implementation to omit creating a temporary
1775 which is only used to initialize another object of the same type.
1776 Specifying this option disables that optimization, and forces G++ to
1777 call the copy constructor in all cases.
1779 @item -fno-enforce-eh-specs
1780 @opindex fno-enforce-eh-specs
1781 Don't generate code to check for violation of exception specifications
1782 at runtime. This option violates the C++ standard, but may be useful
1783 for reducing code size in production builds, much like defining
1784 @samp{NDEBUG}. This does not give user code permission to throw
1785 exceptions in violation of the exception specifications; the compiler
1786 will still optimize based on the specifications, so throwing an
1787 unexpected exception will result in undefined behavior.
1790 @itemx -fno-for-scope
1792 @opindex fno-for-scope
1793 If @option{-ffor-scope} is specified, the scope of variables declared in
1794 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1795 as specified by the C++ standard.
1796 If @option{-fno-for-scope} is specified, the scope of variables declared in
1797 a @i{for-init-statement} extends to the end of the enclosing scope,
1798 as was the case in old versions of G++, and other (traditional)
1799 implementations of C++.
1801 The default if neither flag is given to follow the standard,
1802 but to allow and give a warning for old-style code that would
1803 otherwise be invalid, or have different behavior.
1805 @item -fno-gnu-keywords
1806 @opindex fno-gnu-keywords
1807 Do not recognize @code{typeof} as a keyword, so that code can use this
1808 word as an identifier. You can use the keyword @code{__typeof__} instead.
1809 @option{-ansi} implies @option{-fno-gnu-keywords}.
1811 @item -fno-implicit-templates
1812 @opindex fno-implicit-templates
1813 Never emit code for non-inline templates which are instantiated
1814 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1815 @xref{Template Instantiation}, for more information.
1817 @item -fno-implicit-inline-templates
1818 @opindex fno-implicit-inline-templates
1819 Don't emit code for implicit instantiations of inline templates, either.
1820 The default is to handle inlines differently so that compiles with and
1821 without optimization will need the same set of explicit instantiations.
1823 @item -fno-implement-inlines
1824 @opindex fno-implement-inlines
1825 To save space, do not emit out-of-line copies of inline functions
1826 controlled by @samp{#pragma implementation}. This will cause linker
1827 errors if these functions are not inlined everywhere they are called.
1829 @item -fms-extensions
1830 @opindex fms-extensions
1831 Disable pedantic warnings about constructs used in MFC, such as implicit
1832 int and getting a pointer to member function via non-standard syntax.
1834 @item -fno-nonansi-builtins
1835 @opindex fno-nonansi-builtins
1836 Disable built-in declarations of functions that are not mandated by
1837 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1838 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1840 @item -fno-operator-names
1841 @opindex fno-operator-names
1842 Do not treat the operator name keywords @code{and}, @code{bitand},
1843 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1844 synonyms as keywords.
1846 @item -fno-optional-diags
1847 @opindex fno-optional-diags
1848 Disable diagnostics that the standard says a compiler does not need to
1849 issue. Currently, the only such diagnostic issued by G++ is the one for
1850 a name having multiple meanings within a class.
1853 @opindex fpermissive
1854 Downgrade some diagnostics about nonconformant code from errors to
1855 warnings. Thus, using @option{-fpermissive} will allow some
1856 nonconforming code to compile.
1860 Enable automatic template instantiation at link time. This option also
1861 implies @option{-fno-implicit-templates}. @xref{Template
1862 Instantiation}, for more information.
1866 Disable generation of information about every class with virtual
1867 functions for use by the C++ runtime type identification features
1868 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1869 of the language, you can save some space by using this flag. Note that
1870 exception handling uses the same information, but it will generate it as
1871 needed. The @samp{dynamic_cast} operator can still be used for casts that
1872 do not require runtime type information, i.e.@: casts to @code{void *} or to
1873 unambiguous base classes.
1877 Emit statistics about front-end processing at the end of the compilation.
1878 This information is generally only useful to the G++ development team.
1880 @item -ftemplate-depth-@var{n}
1881 @opindex ftemplate-depth
1882 Set the maximum instantiation depth for template classes to @var{n}.
1883 A limit on the template instantiation depth is needed to detect
1884 endless recursions during template class instantiation. ANSI/ISO C++
1885 conforming programs must not rely on a maximum depth greater than 17.
1887 @item -fno-threadsafe-statics
1888 @opindex fno-threadsafe-statics
1889 Do not emit the extra code to use the routines specified in the C++
1890 ABI for thread-safe initialization of local statics. You can use this
1891 option to reduce code size slightly in code that doesn't need to be
1894 @item -fuse-cxa-atexit
1895 @opindex fuse-cxa-atexit
1896 Register destructors for objects with static storage duration with the
1897 @code{__cxa_atexit} function rather than the @code{atexit} function.
1898 This option is required for fully standards-compliant handling of static
1899 destructors, but will only work if your C library supports
1900 @code{__cxa_atexit}.
1902 @item -fno-use-cxa-get-exception-ptr
1903 @opindex fno-use-cxa-get-exception-ptr
1904 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1905 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1906 if the runtime routine is not available.
1908 @item -fvisibility-inlines-hidden
1909 @opindex fvisibility-inlines-hidden
1910 This switch declares that the user does not attempt to compare
1911 pointers to inline methods where the addresses of the two functions
1912 were taken in different shared objects.
1914 The effect of this is that GCC may, effectively, mark inline methods with
1915 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1916 appear in the export table of a DSO and do not require a PLT indirection
1917 when used within the DSO@. Enabling this option can have a dramatic effect
1918 on load and link times of a DSO as it massively reduces the size of the
1919 dynamic export table when the library makes heavy use of templates.
1921 The behavior of this switch is not quite the same as marking the
1922 methods as hidden directly, because it does not affect static variables
1923 local to the function or cause the compiler to deduce that
1924 the function is defined in only one shared object.
1926 You may mark a method as having a visibility explicitly to negate the
1927 effect of the switch for that method. For example, if you do want to
1928 compare pointers to a particular inline method, you might mark it as
1929 having default visibility. Marking the enclosing class with explicit
1930 visibility will have no effect.
1932 Explicitly instantiated inline methods are unaffected by this option
1933 as their linkage might otherwise cross a shared library boundary.
1934 @xref{Template Instantiation}.
1936 @item -fvisibility-ms-compat
1937 @opindex fvisibility-ms-compat
1938 This flag attempts to use visibility settings to make GCC's C++
1939 linkage model compatible with that of Microsoft Visual Studio.
1941 The flag makes these changes to GCC's linkage model:
1945 It sets the default visibility to @code{hidden}, like
1946 @option{-fvisibility=hidden}.
1949 Types, but not their members, are not hidden by default.
1952 The One Definition Rule is relaxed for types without explicit
1953 visibility specifications which are defined in more than one different
1954 shared object: those declarations are permitted if they would have
1955 been permitted when this option was not used.
1958 In new code it is better to use @option{-fvisibility=hidden} and
1959 export those classes which are intended to be externally visible.
1960 Unfortunately it is possible for code to rely, perhaps accidentally,
1961 on the Visual Studio behavior.
1963 Among the consequences of these changes are that static data members
1964 of the same type with the same name but defined in different shared
1965 objects will be different, so changing one will not change the other;
1966 and that pointers to function members defined in different shared
1967 objects may not compare equal. When this flag is given, it is a
1968 violation of the ODR to define types with the same name differently.
1972 Do not use weak symbol support, even if it is provided by the linker.
1973 By default, G++ will use weak symbols if they are available. This
1974 option exists only for testing, and should not be used by end-users;
1975 it will result in inferior code and has no benefits. This option may
1976 be removed in a future release of G++.
1980 Do not search for header files in the standard directories specific to
1981 C++, but do still search the other standard directories. (This option
1982 is used when building the C++ library.)
1985 In addition, these optimization, warning, and code generation options
1986 have meanings only for C++ programs:
1989 @item -fno-default-inline
1990 @opindex fno-default-inline
1991 Do not assume @samp{inline} for functions defined inside a class scope.
1992 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1993 functions will have linkage like inline functions; they just won't be
1996 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
1999 Warn when G++ generates code that is probably not compatible with the
2000 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2001 all such cases, there are probably some cases that are not warned about,
2002 even though G++ is generating incompatible code. There may also be
2003 cases where warnings are emitted even though the code that is generated
2006 You should rewrite your code to avoid these warnings if you are
2007 concerned about the fact that code generated by G++ may not be binary
2008 compatible with code generated by other compilers.
2010 The known incompatibilities at this point include:
2015 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2016 pack data into the same byte as a base class. For example:
2019 struct A @{ virtual void f(); int f1 : 1; @};
2020 struct B : public A @{ int f2 : 1; @};
2024 In this case, G++ will place @code{B::f2} into the same byte
2025 as@code{A::f1}; other compilers will not. You can avoid this problem
2026 by explicitly padding @code{A} so that its size is a multiple of the
2027 byte size on your platform; that will cause G++ and other compilers to
2028 layout @code{B} identically.
2031 Incorrect handling of tail-padding for virtual bases. G++ does not use
2032 tail padding when laying out virtual bases. For example:
2035 struct A @{ virtual void f(); char c1; @};
2036 struct B @{ B(); char c2; @};
2037 struct C : public A, public virtual B @{@};
2041 In this case, G++ will not place @code{B} into the tail-padding for
2042 @code{A}; other compilers will. You can avoid this problem by
2043 explicitly padding @code{A} so that its size is a multiple of its
2044 alignment (ignoring virtual base classes); that will cause G++ and other
2045 compilers to layout @code{C} identically.
2048 Incorrect handling of bit-fields with declared widths greater than that
2049 of their underlying types, when the bit-fields appear in a union. For
2053 union U @{ int i : 4096; @};
2057 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2058 union too small by the number of bits in an @code{int}.
2061 Empty classes can be placed at incorrect offsets. For example:
2071 struct C : public B, public A @{@};
2075 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2076 it should be placed at offset zero. G++ mistakenly believes that the
2077 @code{A} data member of @code{B} is already at offset zero.
2080 Names of template functions whose types involve @code{typename} or
2081 template template parameters can be mangled incorrectly.
2084 template <typename Q>
2085 void f(typename Q::X) @{@}
2087 template <template <typename> class Q>
2088 void f(typename Q<int>::X) @{@}
2092 Instantiations of these templates may be mangled incorrectly.
2096 It also warns psABI related changes. The known psABI changes at this
2102 For SYSV/x86-64, when passing union with long double, it is changed to
2103 pass in memory as specified in psABI. For example:
2113 @code{union U} will always be passed in memory.
2117 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2118 @opindex Wctor-dtor-privacy
2119 @opindex Wno-ctor-dtor-privacy
2120 Warn when a class seems unusable because all the constructors or
2121 destructors in that class are private, and it has neither friends nor
2122 public static member functions.
2124 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2125 @opindex Wnon-virtual-dtor
2126 @opindex Wno-non-virtual-dtor
2127 Warn when a class has virtual functions and accessible non-virtual
2128 destructor, in which case it would be possible but unsafe to delete
2129 an instance of a derived class through a pointer to the base class.
2130 This warning is also enabled if -Weffc++ is specified.
2132 @item -Wreorder @r{(C++ and Objective-C++ only)}
2134 @opindex Wno-reorder
2135 @cindex reordering, warning
2136 @cindex warning for reordering of member initializers
2137 Warn when the order of member initializers given in the code does not
2138 match the order in which they must be executed. For instance:
2144 A(): j (0), i (1) @{ @}
2148 The compiler will rearrange the member initializers for @samp{i}
2149 and @samp{j} to match the declaration order of the members, emitting
2150 a warning to that effect. This warning is enabled by @option{-Wall}.
2153 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2156 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2159 Warn about violations of the following style guidelines from Scott Meyers'
2160 @cite{Effective C++} book:
2164 Item 11: Define a copy constructor and an assignment operator for classes
2165 with dynamically allocated memory.
2168 Item 12: Prefer initialization to assignment in constructors.
2171 Item 14: Make destructors virtual in base classes.
2174 Item 15: Have @code{operator=} return a reference to @code{*this}.
2177 Item 23: Don't try to return a reference when you must return an object.
2181 Also warn about violations of the following style guidelines from
2182 Scott Meyers' @cite{More Effective C++} book:
2186 Item 6: Distinguish between prefix and postfix forms of increment and
2187 decrement operators.
2190 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2194 When selecting this option, be aware that the standard library
2195 headers do not obey all of these guidelines; use @samp{grep -v}
2196 to filter out those warnings.
2198 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2199 @opindex Wstrict-null-sentinel
2200 @opindex Wno-strict-null-sentinel
2201 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2202 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2203 to @code{__null}. Although it is a null pointer constant not a null pointer,
2204 it is guaranteed to be of the same size as a pointer. But this use is
2205 not portable across different compilers.
2207 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2208 @opindex Wno-non-template-friend
2209 @opindex Wnon-template-friend
2210 Disable warnings when non-templatized friend functions are declared
2211 within a template. Since the advent of explicit template specification
2212 support in G++, if the name of the friend is an unqualified-id (i.e.,
2213 @samp{friend foo(int)}), the C++ language specification demands that the
2214 friend declare or define an ordinary, nontemplate function. (Section
2215 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2216 could be interpreted as a particular specialization of a templatized
2217 function. Because this non-conforming behavior is no longer the default
2218 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2219 check existing code for potential trouble spots and is on by default.
2220 This new compiler behavior can be turned off with
2221 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2222 but disables the helpful warning.
2224 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2225 @opindex Wold-style-cast
2226 @opindex Wno-old-style-cast
2227 Warn if an old-style (C-style) cast to a non-void type is used within
2228 a C++ program. The new-style casts (@samp{dynamic_cast},
2229 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2230 less vulnerable to unintended effects and much easier to search for.
2232 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2233 @opindex Woverloaded-virtual
2234 @opindex Wno-overloaded-virtual
2235 @cindex overloaded virtual fn, warning
2236 @cindex warning for overloaded virtual fn
2237 Warn when a function declaration hides virtual functions from a
2238 base class. For example, in:
2245 struct B: public A @{
2250 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2258 will fail to compile.
2260 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2261 @opindex Wno-pmf-conversions
2262 @opindex Wpmf-conversions
2263 Disable the diagnostic for converting a bound pointer to member function
2266 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2267 @opindex Wsign-promo
2268 @opindex Wno-sign-promo
2269 Warn when overload resolution chooses a promotion from unsigned or
2270 enumerated type to a signed type, over a conversion to an unsigned type of
2271 the same size. Previous versions of G++ would try to preserve
2272 unsignedness, but the standard mandates the current behavior.
2277 A& operator = (int);
2287 In this example, G++ will synthesize a default @samp{A& operator =
2288 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2291 @node Objective-C and Objective-C++ Dialect Options
2292 @section Options Controlling Objective-C and Objective-C++ Dialects
2294 @cindex compiler options, Objective-C and Objective-C++
2295 @cindex Objective-C and Objective-C++ options, command line
2296 @cindex options, Objective-C and Objective-C++
2297 (NOTE: This manual does not describe the Objective-C and Objective-C++
2298 languages themselves. See @xref{Standards,,Language Standards
2299 Supported by GCC}, for references.)
2301 This section describes the command-line options that are only meaningful
2302 for Objective-C and Objective-C++ programs, but you can also use most of
2303 the language-independent GNU compiler options.
2304 For example, you might compile a file @code{some_class.m} like this:
2307 gcc -g -fgnu-runtime -O -c some_class.m
2311 In this example, @option{-fgnu-runtime} is an option meant only for
2312 Objective-C and Objective-C++ programs; you can use the other options with
2313 any language supported by GCC@.
2315 Note that since Objective-C is an extension of the C language, Objective-C
2316 compilations may also use options specific to the C front-end (e.g.,
2317 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2318 C++-specific options (e.g., @option{-Wabi}).
2320 Here is a list of options that are @emph{only} for compiling Objective-C
2321 and Objective-C++ programs:
2324 @item -fconstant-string-class=@var{class-name}
2325 @opindex fconstant-string-class
2326 Use @var{class-name} as the name of the class to instantiate for each
2327 literal string specified with the syntax @code{@@"@dots{}"}. The default
2328 class name is @code{NXConstantString} if the GNU runtime is being used, and
2329 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2330 @option{-fconstant-cfstrings} option, if also present, will override the
2331 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2332 to be laid out as constant CoreFoundation strings.
2335 @opindex fgnu-runtime
2336 Generate object code compatible with the standard GNU Objective-C
2337 runtime. This is the default for most types of systems.
2339 @item -fnext-runtime
2340 @opindex fnext-runtime
2341 Generate output compatible with the NeXT runtime. This is the default
2342 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2343 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2346 @item -fno-nil-receivers
2347 @opindex fno-nil-receivers
2348 Assume that all Objective-C message dispatches (e.g.,
2349 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2350 is not @code{nil}. This allows for more efficient entry points in the runtime
2351 to be used. Currently, this option is only available in conjunction with
2352 the NeXT runtime on Mac OS X 10.3 and later.
2354 @item -fobjc-call-cxx-cdtors
2355 @opindex fobjc-call-cxx-cdtors
2356 For each Objective-C class, check if any of its instance variables is a
2357 C++ object with a non-trivial default constructor. If so, synthesize a
2358 special @code{- (id) .cxx_construct} instance method that will run
2359 non-trivial default constructors on any such instance variables, in order,
2360 and then return @code{self}. Similarly, check if any instance variable
2361 is a C++ object with a non-trivial destructor, and if so, synthesize a
2362 special @code{- (void) .cxx_destruct} method that will run
2363 all such default destructors, in reverse order.
2365 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2366 thusly generated will only operate on instance variables declared in the
2367 current Objective-C class, and not those inherited from superclasses. It
2368 is the responsibility of the Objective-C runtime to invoke all such methods
2369 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2370 will be invoked by the runtime immediately after a new object
2371 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2372 be invoked immediately before the runtime deallocates an object instance.
2374 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2375 support for invoking the @code{- (id) .cxx_construct} and
2376 @code{- (void) .cxx_destruct} methods.
2378 @item -fobjc-direct-dispatch
2379 @opindex fobjc-direct-dispatch
2380 Allow fast jumps to the message dispatcher. On Darwin this is
2381 accomplished via the comm page.
2383 @item -fobjc-exceptions
2384 @opindex fobjc-exceptions
2385 Enable syntactic support for structured exception handling in Objective-C,
2386 similar to what is offered by C++ and Java. This option is
2387 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2396 @@catch (AnObjCClass *exc) @{
2403 @@catch (AnotherClass *exc) @{
2406 @@catch (id allOthers) @{
2416 The @code{@@throw} statement may appear anywhere in an Objective-C or
2417 Objective-C++ program; when used inside of a @code{@@catch} block, the
2418 @code{@@throw} may appear without an argument (as shown above), in which case
2419 the object caught by the @code{@@catch} will be rethrown.
2421 Note that only (pointers to) Objective-C objects may be thrown and
2422 caught using this scheme. When an object is thrown, it will be caught
2423 by the nearest @code{@@catch} clause capable of handling objects of that type,
2424 analogously to how @code{catch} blocks work in C++ and Java. A
2425 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2426 any and all Objective-C exceptions not caught by previous @code{@@catch}
2429 The @code{@@finally} clause, if present, will be executed upon exit from the
2430 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2431 regardless of whether any exceptions are thrown, caught or rethrown
2432 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2433 of the @code{finally} clause in Java.
2435 There are several caveats to using the new exception mechanism:
2439 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2440 idioms provided by the @code{NSException} class, the new
2441 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2442 systems, due to additional functionality needed in the (NeXT) Objective-C
2446 As mentioned above, the new exceptions do not support handling
2447 types other than Objective-C objects. Furthermore, when used from
2448 Objective-C++, the Objective-C exception model does not interoperate with C++
2449 exceptions at this time. This means you cannot @code{@@throw} an exception
2450 from Objective-C and @code{catch} it in C++, or vice versa
2451 (i.e., @code{throw @dots{} @@catch}).
2454 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2455 blocks for thread-safe execution:
2458 @@synchronized (ObjCClass *guard) @{
2463 Upon entering the @code{@@synchronized} block, a thread of execution shall
2464 first check whether a lock has been placed on the corresponding @code{guard}
2465 object by another thread. If it has, the current thread shall wait until
2466 the other thread relinquishes its lock. Once @code{guard} becomes available,
2467 the current thread will place its own lock on it, execute the code contained in
2468 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2469 making @code{guard} available to other threads).
2471 Unlike Java, Objective-C does not allow for entire methods to be marked
2472 @code{@@synchronized}. Note that throwing exceptions out of
2473 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2474 to be unlocked properly.
2478 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2480 @item -freplace-objc-classes
2481 @opindex freplace-objc-classes
2482 Emit a special marker instructing @command{ld(1)} not to statically link in
2483 the resulting object file, and allow @command{dyld(1)} to load it in at
2484 run time instead. This is used in conjunction with the Fix-and-Continue
2485 debugging mode, where the object file in question may be recompiled and
2486 dynamically reloaded in the course of program execution, without the need
2487 to restart the program itself. Currently, Fix-and-Continue functionality
2488 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2493 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2494 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2495 compile time) with static class references that get initialized at load time,
2496 which improves run-time performance. Specifying the @option{-fzero-link} flag
2497 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2498 to be retained. This is useful in Zero-Link debugging mode, since it allows
2499 for individual class implementations to be modified during program execution.
2503 Dump interface declarations for all classes seen in the source file to a
2504 file named @file{@var{sourcename}.decl}.
2506 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2507 @opindex Wassign-intercept
2508 @opindex Wno-assign-intercept
2509 Warn whenever an Objective-C assignment is being intercepted by the
2512 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2513 @opindex Wno-protocol
2515 If a class is declared to implement a protocol, a warning is issued for
2516 every method in the protocol that is not implemented by the class. The
2517 default behavior is to issue a warning for every method not explicitly
2518 implemented in the class, even if a method implementation is inherited
2519 from the superclass. If you use the @option{-Wno-protocol} option, then
2520 methods inherited from the superclass are considered to be implemented,
2521 and no warning is issued for them.
2523 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2525 @opindex Wno-selector
2526 Warn if multiple methods of different types for the same selector are
2527 found during compilation. The check is performed on the list of methods
2528 in the final stage of compilation. Additionally, a check is performed
2529 for each selector appearing in a @code{@@selector(@dots{})}
2530 expression, and a corresponding method for that selector has been found
2531 during compilation. Because these checks scan the method table only at
2532 the end of compilation, these warnings are not produced if the final
2533 stage of compilation is not reached, for example because an error is
2534 found during compilation, or because the @option{-fsyntax-only} option is
2537 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2538 @opindex Wstrict-selector-match
2539 @opindex Wno-strict-selector-match
2540 Warn if multiple methods with differing argument and/or return types are
2541 found for a given selector when attempting to send a message using this
2542 selector to a receiver of type @code{id} or @code{Class}. When this flag
2543 is off (which is the default behavior), the compiler will omit such warnings
2544 if any differences found are confined to types which share the same size
2547 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2548 @opindex Wundeclared-selector
2549 @opindex Wno-undeclared-selector
2550 Warn if a @code{@@selector(@dots{})} expression referring to an
2551 undeclared selector is found. A selector is considered undeclared if no
2552 method with that name has been declared before the
2553 @code{@@selector(@dots{})} expression, either explicitly in an
2554 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2555 an @code{@@implementation} section. This option always performs its
2556 checks as soon as a @code{@@selector(@dots{})} expression is found,
2557 while @option{-Wselector} only performs its checks in the final stage of
2558 compilation. This also enforces the coding style convention
2559 that methods and selectors must be declared before being used.
2561 @item -print-objc-runtime-info
2562 @opindex print-objc-runtime-info
2563 Generate C header describing the largest structure that is passed by
2568 @node Language Independent Options
2569 @section Options to Control Diagnostic Messages Formatting
2570 @cindex options to control diagnostics formatting
2571 @cindex diagnostic messages
2572 @cindex message formatting
2574 Traditionally, diagnostic messages have been formatted irrespective of
2575 the output device's aspect (e.g.@: its width, @dots{}). The options described
2576 below can be used to control the diagnostic messages formatting
2577 algorithm, e.g.@: how many characters per line, how often source location
2578 information should be reported. Right now, only the C++ front end can
2579 honor these options. However it is expected, in the near future, that
2580 the remaining front ends would be able to digest them correctly.
2583 @item -fmessage-length=@var{n}
2584 @opindex fmessage-length
2585 Try to format error messages so that they fit on lines of about @var{n}
2586 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2587 the front ends supported by GCC@. If @var{n} is zero, then no
2588 line-wrapping will be done; each error message will appear on a single
2591 @opindex fdiagnostics-show-location
2592 @item -fdiagnostics-show-location=once
2593 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2594 reporter to emit @emph{once} source location information; that is, in
2595 case the message is too long to fit on a single physical line and has to
2596 be wrapped, the source location won't be emitted (as prefix) again,
2597 over and over, in subsequent continuation lines. This is the default
2600 @item -fdiagnostics-show-location=every-line
2601 Only meaningful in line-wrapping mode. Instructs the diagnostic
2602 messages reporter to emit the same source location information (as
2603 prefix) for physical lines that result from the process of breaking
2604 a message which is too long to fit on a single line.
2606 @item -fdiagnostics-show-option
2607 @opindex fdiagnostics-show-option
2608 This option instructs the diagnostic machinery to add text to each
2609 diagnostic emitted, which indicates which command line option directly
2610 controls that diagnostic, when such an option is known to the
2611 diagnostic machinery.
2613 @item -Wcoverage-mismatch
2614 @opindex Wcoverage-mismatch
2615 Warn if feedback profiles do not match when using the
2616 @option{-fprofile-use} option.
2617 If a source file was changed between @option{-fprofile-gen} and
2618 @option{-fprofile-use}, the files with the profile feedback can fail
2619 to match the source file and GCC can not use the profile feedback
2620 information. By default, GCC emits an error message in this case.
2621 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2622 error. GCC does not use appropriate feedback profiles, so using this
2623 option can result in poorly optimized code. This option is useful
2624 only in the case of very minor changes such as bug fixes to an
2629 @node Warning Options
2630 @section Options to Request or Suppress Warnings
2631 @cindex options to control warnings
2632 @cindex warning messages
2633 @cindex messages, warning
2634 @cindex suppressing warnings
2636 Warnings are diagnostic messages that report constructions which
2637 are not inherently erroneous but which are risky or suggest there
2638 may have been an error.
2640 The following language-independent options do not enable specific
2641 warnings but control the kinds of diagnostics produced by GCC.
2644 @cindex syntax checking
2646 @opindex fsyntax-only
2647 Check the code for syntax errors, but don't do anything beyond that.
2651 Inhibit all warning messages.
2656 Make all warnings into errors.
2661 Make the specified warning into an error. The specifier for a warning
2662 is appended, for example @option{-Werror=switch} turns the warnings
2663 controlled by @option{-Wswitch} into errors. This switch takes a
2664 negative form, to be used to negate @option{-Werror} for specific
2665 warnings, for example @option{-Wno-error=switch} makes
2666 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2667 is in effect. You can use the @option{-fdiagnostics-show-option}
2668 option to have each controllable warning amended with the option which
2669 controls it, to determine what to use with this option.
2671 Note that specifying @option{-Werror=}@var{foo} automatically implies
2672 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2675 @item -Wfatal-errors
2676 @opindex Wfatal-errors
2677 @opindex Wno-fatal-errors
2678 This option causes the compiler to abort compilation on the first error
2679 occurred rather than trying to keep going and printing further error
2684 You can request many specific warnings with options beginning
2685 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2686 implicit declarations. Each of these specific warning options also
2687 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2688 example, @option{-Wno-implicit}. This manual lists only one of the
2689 two forms, whichever is not the default. For further,
2690 language-specific options also refer to @ref{C++ Dialect Options} and
2691 @ref{Objective-C and Objective-C++ Dialect Options}.
2696 Issue all the warnings demanded by strict ISO C and ISO C++;
2697 reject all programs that use forbidden extensions, and some other
2698 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2699 version of the ISO C standard specified by any @option{-std} option used.
2701 Valid ISO C and ISO C++ programs should compile properly with or without
2702 this option (though a rare few will require @option{-ansi} or a
2703 @option{-std} option specifying the required version of ISO C)@. However,
2704 without this option, certain GNU extensions and traditional C and C++
2705 features are supported as well. With this option, they are rejected.
2707 @option{-pedantic} does not cause warning messages for use of the
2708 alternate keywords whose names begin and end with @samp{__}. Pedantic
2709 warnings are also disabled in the expression that follows
2710 @code{__extension__}. However, only system header files should use
2711 these escape routes; application programs should avoid them.
2712 @xref{Alternate Keywords}.
2714 Some users try to use @option{-pedantic} to check programs for strict ISO
2715 C conformance. They soon find that it does not do quite what they want:
2716 it finds some non-ISO practices, but not all---only those for which
2717 ISO C @emph{requires} a diagnostic, and some others for which
2718 diagnostics have been added.
2720 A feature to report any failure to conform to ISO C might be useful in
2721 some instances, but would require considerable additional work and would
2722 be quite different from @option{-pedantic}. We don't have plans to
2723 support such a feature in the near future.
2725 Where the standard specified with @option{-std} represents a GNU
2726 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2727 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2728 extended dialect is based. Warnings from @option{-pedantic} are given
2729 where they are required by the base standard. (It would not make sense
2730 for such warnings to be given only for features not in the specified GNU
2731 C dialect, since by definition the GNU dialects of C include all
2732 features the compiler supports with the given option, and there would be
2733 nothing to warn about.)
2735 @item -pedantic-errors
2736 @opindex pedantic-errors
2737 Like @option{-pedantic}, except that errors are produced rather than
2743 This enables all the warnings about constructions that some users
2744 consider questionable, and that are easy to avoid (or modify to
2745 prevent the warning), even in conjunction with macros. This also
2746 enables some language-specific warnings described in @ref{C++ Dialect
2747 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2749 @option{-Wall} turns on the following warning flags:
2751 @gccoptlist{-Waddress @gol
2752 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2754 -Wchar-subscripts @gol
2756 -Wimplicit-function-declaration @gol
2759 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2760 -Wmissing-braces @gol
2766 -Wsequence-point @gol
2767 -Wsign-compare @r{(only in C++)} @gol
2768 -Wstrict-aliasing @gol
2769 -Wstrict-overflow=1 @gol
2772 -Wuninitialized @gol
2773 -Wunknown-pragmas @gol
2774 -Wunused-function @gol
2777 -Wunused-variable @gol
2778 -Wvolatile-register-var @gol
2781 Note that some warning flags are not implied by @option{-Wall}. Some of
2782 them warn about constructions that users generally do not consider
2783 questionable, but which occasionally you might wish to check for;
2784 others warn about constructions that are necessary or hard to avoid in
2785 some cases, and there is no simple way to modify the code to suppress
2786 the warning. Some of them are enabled by @option{-Wextra} but many of
2787 them must be enabled individually.
2793 This enables some extra warning flags that are not enabled by
2794 @option{-Wall}. (This option used to be called @option{-W}. The older
2795 name is still supported, but the newer name is more descriptive.)
2797 @gccoptlist{-Wclobbered @gol
2799 -Wignored-qualifiers @gol
2800 -Wmissing-field-initializers @gol
2801 -Wmissing-parameter-type @r{(C only)} @gol
2802 -Wold-style-declaration @r{(C only)} @gol
2803 -Woverride-init @gol
2806 -Wuninitialized @gol
2807 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2810 The option @option{-Wextra} also prints warning messages for the
2816 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2817 @samp{>}, or @samp{>=}.
2820 (C++ only) An enumerator and a non-enumerator both appear in a
2821 conditional expression.
2824 (C++ only) Ambiguous virtual bases.
2827 (C++ only) Subscripting an array which has been declared @samp{register}.
2830 (C++ only) Taking the address of a variable which has been declared
2834 (C++ only) A base class is not initialized in a derived class' copy
2839 @item -Wchar-subscripts
2840 @opindex Wchar-subscripts
2841 @opindex Wno-char-subscripts
2842 Warn if an array subscript has type @code{char}. This is a common cause
2843 of error, as programmers often forget that this type is signed on some
2845 This warning is enabled by @option{-Wall}.
2849 @opindex Wno-comment
2850 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2851 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2852 This warning is enabled by @option{-Wall}.
2857 @opindex ffreestanding
2858 @opindex fno-builtin
2859 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2860 the arguments supplied have types appropriate to the format string
2861 specified, and that the conversions specified in the format string make
2862 sense. This includes standard functions, and others specified by format
2863 attributes (@pxref{Function Attributes}), in the @code{printf},
2864 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2865 not in the C standard) families (or other target-specific families).
2866 Which functions are checked without format attributes having been
2867 specified depends on the standard version selected, and such checks of
2868 functions without the attribute specified are disabled by
2869 @option{-ffreestanding} or @option{-fno-builtin}.
2871 The formats are checked against the format features supported by GNU
2872 libc version 2.2. These include all ISO C90 and C99 features, as well
2873 as features from the Single Unix Specification and some BSD and GNU
2874 extensions. Other library implementations may not support all these
2875 features; GCC does not support warning about features that go beyond a
2876 particular library's limitations. However, if @option{-pedantic} is used
2877 with @option{-Wformat}, warnings will be given about format features not
2878 in the selected standard version (but not for @code{strfmon} formats,
2879 since those are not in any version of the C standard). @xref{C Dialect
2880 Options,,Options Controlling C Dialect}.
2882 Since @option{-Wformat} also checks for null format arguments for
2883 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2885 @option{-Wformat} is included in @option{-Wall}. For more control over some
2886 aspects of format checking, the options @option{-Wformat-y2k},
2887 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2888 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2889 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2892 @opindex Wformat-y2k
2893 @opindex Wno-format-y2k
2894 If @option{-Wformat} is specified, also warn about @code{strftime}
2895 formats which may yield only a two-digit year.
2897 @item -Wno-format-contains-nul
2898 @opindex Wno-format-contains-nul
2899 @opindex Wformat-contains-nul
2900 If @option{-Wformat} is specified, do not warn about format strings that
2903 @item -Wno-format-extra-args
2904 @opindex Wno-format-extra-args
2905 @opindex Wformat-extra-args
2906 If @option{-Wformat} is specified, do not warn about excess arguments to a
2907 @code{printf} or @code{scanf} format function. The C standard specifies
2908 that such arguments are ignored.
2910 Where the unused arguments lie between used arguments that are
2911 specified with @samp{$} operand number specifications, normally
2912 warnings are still given, since the implementation could not know what
2913 type to pass to @code{va_arg} to skip the unused arguments. However,
2914 in the case of @code{scanf} formats, this option will suppress the
2915 warning if the unused arguments are all pointers, since the Single
2916 Unix Specification says that such unused arguments are allowed.
2918 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2919 @opindex Wno-format-zero-length
2920 @opindex Wformat-zero-length
2921 If @option{-Wformat} is specified, do not warn about zero-length formats.
2922 The C standard specifies that zero-length formats are allowed.
2924 @item -Wformat-nonliteral
2925 @opindex Wformat-nonliteral
2926 @opindex Wno-format-nonliteral
2927 If @option{-Wformat} is specified, also warn if the format string is not a
2928 string literal and so cannot be checked, unless the format function
2929 takes its format arguments as a @code{va_list}.
2931 @item -Wformat-security
2932 @opindex Wformat-security
2933 @opindex Wno-format-security
2934 If @option{-Wformat} is specified, also warn about uses of format
2935 functions that represent possible security problems. At present, this
2936 warns about calls to @code{printf} and @code{scanf} functions where the
2937 format string is not a string literal and there are no format arguments,
2938 as in @code{printf (foo);}. This may be a security hole if the format
2939 string came from untrusted input and contains @samp{%n}. (This is
2940 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2941 in future warnings may be added to @option{-Wformat-security} that are not
2942 included in @option{-Wformat-nonliteral}.)
2946 @opindex Wno-format=2
2947 Enable @option{-Wformat} plus format checks not included in
2948 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2949 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2951 @item -Wnonnull @r{(C and Objective-C only)}
2953 @opindex Wno-nonnull
2954 Warn about passing a null pointer for arguments marked as
2955 requiring a non-null value by the @code{nonnull} function attribute.
2957 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2958 can be disabled with the @option{-Wno-nonnull} option.
2960 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2962 @opindex Wno-init-self
2963 Warn about uninitialized variables which are initialized with themselves.
2964 Note this option can only be used with the @option{-Wuninitialized} option.
2966 For example, GCC will warn about @code{i} being uninitialized in the
2967 following snippet only when @option{-Winit-self} has been specified:
2978 @item -Wimplicit-int @r{(C and Objective-C only)}
2979 @opindex Wimplicit-int
2980 @opindex Wno-implicit-int
2981 Warn when a declaration does not specify a type.
2982 This warning is enabled by @option{-Wall}.
2984 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2985 @opindex Wimplicit-function-declaration
2986 @opindex Wno-implicit-function-declaration
2987 Give a warning whenever a function is used before being declared. In
2988 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2989 enabled by default and it is made into an error by
2990 @option{-pedantic-errors}. This warning is also enabled by
2995 @opindex Wno-implicit
2996 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2997 This warning is enabled by @option{-Wall}.
2999 @item -Wignored-qualifiers @r{(C and C++ only)}
3000 @opindex Wignored-qualifiers
3001 @opindex Wno-ignored-qualifiers
3002 Warn if the return type of a function has a type qualifier
3003 such as @code{const}. For ISO C such a type qualifier has no effect,
3004 since the value returned by a function is not an lvalue.
3005 For C++, the warning is only emitted for scalar types or @code{void}.
3006 ISO C prohibits qualified @code{void} return types on function
3007 definitions, so such return types always receive a warning
3008 even without this option.
3010 This warning is also enabled by @option{-Wextra}.
3015 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3016 a function with external linkage, returning int, taking either zero
3017 arguments, two, or three arguments of appropriate types. This warning
3018 is enabled by default in C++ and is enabled by either @option{-Wall}
3019 or @option{-pedantic}.
3021 @item -Wmissing-braces
3022 @opindex Wmissing-braces
3023 @opindex Wno-missing-braces
3024 Warn if an aggregate or union initializer is not fully bracketed. In
3025 the following example, the initializer for @samp{a} is not fully
3026 bracketed, but that for @samp{b} is fully bracketed.
3029 int a[2][2] = @{ 0, 1, 2, 3 @};
3030 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3033 This warning is enabled by @option{-Wall}.
3035 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3036 @opindex Wmissing-include-dirs
3037 @opindex Wno-missing-include-dirs
3038 Warn if a user-supplied include directory does not exist.
3041 @opindex Wparentheses
3042 @opindex Wno-parentheses
3043 Warn if parentheses are omitted in certain contexts, such
3044 as when there is an assignment in a context where a truth value
3045 is expected, or when operators are nested whose precedence people
3046 often get confused about.
3048 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3049 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3050 interpretation from that of ordinary mathematical notation.
3052 Also warn about constructions where there may be confusion to which
3053 @code{if} statement an @code{else} branch belongs. Here is an example of
3068 In C/C++, every @code{else} branch belongs to the innermost possible
3069 @code{if} statement, which in this example is @code{if (b)}. This is
3070 often not what the programmer expected, as illustrated in the above
3071 example by indentation the programmer chose. When there is the
3072 potential for this confusion, GCC will issue a warning when this flag
3073 is specified. To eliminate the warning, add explicit braces around
3074 the innermost @code{if} statement so there is no way the @code{else}
3075 could belong to the enclosing @code{if}. The resulting code would
3092 This warning is enabled by @option{-Wall}.
3094 @item -Wsequence-point
3095 @opindex Wsequence-point
3096 @opindex Wno-sequence-point
3097 Warn about code that may have undefined semantics because of violations
3098 of sequence point rules in the C and C++ standards.
3100 The C and C++ standards defines the order in which expressions in a C/C++
3101 program are evaluated in terms of @dfn{sequence points}, which represent
3102 a partial ordering between the execution of parts of the program: those
3103 executed before the sequence point, and those executed after it. These
3104 occur after the evaluation of a full expression (one which is not part
3105 of a larger expression), after the evaluation of the first operand of a
3106 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3107 function is called (but after the evaluation of its arguments and the
3108 expression denoting the called function), and in certain other places.
3109 Other than as expressed by the sequence point rules, the order of
3110 evaluation of subexpressions of an expression is not specified. All
3111 these rules describe only a partial order rather than a total order,
3112 since, for example, if two functions are called within one expression
3113 with no sequence point between them, the order in which the functions
3114 are called is not specified. However, the standards committee have
3115 ruled that function calls do not overlap.
3117 It is not specified when between sequence points modifications to the
3118 values of objects take effect. Programs whose behavior depends on this
3119 have undefined behavior; the C and C++ standards specify that ``Between
3120 the previous and next sequence point an object shall have its stored
3121 value modified at most once by the evaluation of an expression.
3122 Furthermore, the prior value shall be read only to determine the value
3123 to be stored.''. If a program breaks these rules, the results on any
3124 particular implementation are entirely unpredictable.
3126 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3127 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3128 diagnosed by this option, and it may give an occasional false positive
3129 result, but in general it has been found fairly effective at detecting
3130 this sort of problem in programs.
3132 The standard is worded confusingly, therefore there is some debate
3133 over the precise meaning of the sequence point rules in subtle cases.
3134 Links to discussions of the problem, including proposed formal
3135 definitions, may be found on the GCC readings page, at
3136 @w{@uref{http://gcc.gnu.org/readings.html}}.
3138 This warning is enabled by @option{-Wall} for C and C++.
3141 @opindex Wreturn-type
3142 @opindex Wno-return-type
3143 Warn whenever a function is defined with a return-type that defaults
3144 to @code{int}. Also warn about any @code{return} statement with no
3145 return-value in a function whose return-type is not @code{void}
3146 (falling off the end of the function body is considered returning
3147 without a value), and about a @code{return} statement with a
3148 expression in a function whose return-type is @code{void}.
3150 For C++, a function without return type always produces a diagnostic
3151 message, even when @option{-Wno-return-type} is specified. The only
3152 exceptions are @samp{main} and functions defined in system headers.
3154 This warning is enabled by @option{-Wall}.
3159 Warn whenever a @code{switch} statement has an index of enumerated type
3160 and lacks a @code{case} for one or more of the named codes of that
3161 enumeration. (The presence of a @code{default} label prevents this
3162 warning.) @code{case} labels outside the enumeration range also
3163 provoke warnings when this option is used.
3164 This warning is enabled by @option{-Wall}.
3166 @item -Wswitch-default
3167 @opindex Wswitch-default
3168 @opindex Wno-switch-default
3169 Warn whenever a @code{switch} statement does not have a @code{default}
3173 @opindex Wswitch-enum
3174 @opindex Wno-switch-enum
3175 Warn whenever a @code{switch} statement has an index of enumerated type
3176 and lacks a @code{case} for one or more of the named codes of that
3177 enumeration. @code{case} labels outside the enumeration range also
3178 provoke warnings when this option is used.
3180 @item -Wsync-nand @r{(C and C++ only)}
3182 @opindex Wno-sync-nand
3183 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3184 built-in functions are used. These functions changed semantics in GCC 4.4.
3188 @opindex Wno-trigraphs
3189 Warn if any trigraphs are encountered that might change the meaning of
3190 the program (trigraphs within comments are not warned about).
3191 This warning is enabled by @option{-Wall}.
3193 @item -Wunused-function
3194 @opindex Wunused-function
3195 @opindex Wno-unused-function
3196 Warn whenever a static function is declared but not defined or a
3197 non-inline static function is unused.
3198 This warning is enabled by @option{-Wall}.
3200 @item -Wunused-label
3201 @opindex Wunused-label
3202 @opindex Wno-unused-label
3203 Warn whenever a label is declared but not used.
3204 This warning is enabled by @option{-Wall}.
3206 To suppress this warning use the @samp{unused} attribute
3207 (@pxref{Variable Attributes}).
3209 @item -Wunused-parameter
3210 @opindex Wunused-parameter
3211 @opindex Wno-unused-parameter
3212 Warn whenever a function parameter is unused aside from its declaration.
3214 To suppress this warning use the @samp{unused} attribute
3215 (@pxref{Variable Attributes}).
3217 @item -Wunused-variable
3218 @opindex Wunused-variable
3219 @opindex Wno-unused-variable
3220 Warn whenever a local variable or non-constant static variable is unused
3221 aside from its declaration.
3222 This warning is enabled by @option{-Wall}.
3224 To suppress this warning use the @samp{unused} attribute
3225 (@pxref{Variable Attributes}).
3227 @item -Wunused-value
3228 @opindex Wunused-value
3229 @opindex Wno-unused-value
3230 Warn whenever a statement computes a result that is explicitly not
3231 used. To suppress this warning cast the unused expression to
3232 @samp{void}. This includes an expression-statement or the left-hand
3233 side of a comma expression that contains no side effects. For example,
3234 an expression such as @samp{x[i,j]} will cause a warning, while
3235 @samp{x[(void)i,j]} will not.
3237 This warning is enabled by @option{-Wall}.
3242 All the above @option{-Wunused} options combined.
3244 In order to get a warning about an unused function parameter, you must
3245 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3246 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3248 @item -Wuninitialized
3249 @opindex Wuninitialized
3250 @opindex Wno-uninitialized
3251 Warn if an automatic variable is used without first being initialized
3252 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3253 warn if a non-static reference or non-static @samp{const} member
3254 appears in a class without constructors.
3256 If you want to warn about code which uses the uninitialized value of the
3257 variable in its own initializer, use the @option{-Winit-self} option.
3259 These warnings occur for individual uninitialized or clobbered
3260 elements of structure, union or array variables as well as for
3261 variables which are uninitialized or clobbered as a whole. They do
3262 not occur for variables or elements declared @code{volatile}. Because
3263 these warnings depend on optimization, the exact variables or elements
3264 for which there are warnings will depend on the precise optimization
3265 options and version of GCC used.
3267 Note that there may be no warning about a variable that is used only
3268 to compute a value that itself is never used, because such
3269 computations may be deleted by data flow analysis before the warnings
3272 These warnings are made optional because GCC is not smart
3273 enough to see all the reasons why the code might be correct
3274 despite appearing to have an error. Here is one example of how
3295 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3296 always initialized, but GCC doesn't know this. Here is
3297 another common case:
3302 if (change_y) save_y = y, y = new_y;
3304 if (change_y) y = save_y;
3309 This has no bug because @code{save_y} is used only if it is set.
3311 @cindex @code{longjmp} warnings
3312 This option also warns when a non-volatile automatic variable might be
3313 changed by a call to @code{longjmp}. These warnings as well are possible
3314 only in optimizing compilation.
3316 The compiler sees only the calls to @code{setjmp}. It cannot know
3317 where @code{longjmp} will be called; in fact, a signal handler could
3318 call it at any point in the code. As a result, you may get a warning
3319 even when there is in fact no problem because @code{longjmp} cannot
3320 in fact be called at the place which would cause a problem.
3322 Some spurious warnings can be avoided if you declare all the functions
3323 you use that never return as @code{noreturn}. @xref{Function
3326 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3328 @item -Wunknown-pragmas
3329 @opindex Wunknown-pragmas
3330 @opindex Wno-unknown-pragmas
3331 @cindex warning for unknown pragmas
3332 @cindex unknown pragmas, warning
3333 @cindex pragmas, warning of unknown
3334 Warn when a #pragma directive is encountered which is not understood by
3335 GCC@. If this command line option is used, warnings will even be issued
3336 for unknown pragmas in system header files. This is not the case if
3337 the warnings were only enabled by the @option{-Wall} command line option.
3340 @opindex Wno-pragmas
3342 Do not warn about misuses of pragmas, such as incorrect parameters,
3343 invalid syntax, or conflicts between pragmas. See also
3344 @samp{-Wunknown-pragmas}.
3346 @item -Wstrict-aliasing
3347 @opindex Wstrict-aliasing
3348 @opindex Wno-strict-aliasing
3349 This option is only active when @option{-fstrict-aliasing} is active.
3350 It warns about code which might break the strict aliasing rules that the
3351 compiler is using for optimization. The warning does not catch all
3352 cases, but does attempt to catch the more common pitfalls. It is
3353 included in @option{-Wall}.
3354 It is equivalent to @option{-Wstrict-aliasing=3}
3356 @item -Wstrict-aliasing=n
3357 @opindex Wstrict-aliasing=n
3358 @opindex Wno-strict-aliasing=n
3359 This option is only active when @option{-fstrict-aliasing} is active.
3360 It warns about code which might break the strict aliasing rules that the
3361 compiler is using for optimization.
3362 Higher levels correspond to higher accuracy (fewer false positives).
3363 Higher levels also correspond to more effort, similar to the way -O works.
3364 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3367 Level 1: Most aggressive, quick, least accurate.
3368 Possibly useful when higher levels
3369 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3370 false negatives. However, it has many false positives.
3371 Warns for all pointer conversions between possibly incompatible types,
3372 even if never dereferenced. Runs in the frontend only.
3374 Level 2: Aggressive, quick, not too precise.
3375 May still have many false positives (not as many as level 1 though),
3376 and few false negatives (but possibly more than level 1).
3377 Unlike level 1, it only warns when an address is taken. Warns about
3378 incomplete types. Runs in the frontend only.
3380 Level 3 (default for @option{-Wstrict-aliasing}):
3381 Should have very few false positives and few false
3382 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3383 Takes care of the common punn+dereference pattern in the frontend:
3384 @code{*(int*)&some_float}.
3385 If optimization is enabled, it also runs in the backend, where it deals
3386 with multiple statement cases using flow-sensitive points-to information.
3387 Only warns when the converted pointer is dereferenced.
3388 Does not warn about incomplete types.
3390 @item -Wstrict-overflow
3391 @itemx -Wstrict-overflow=@var{n}
3392 @opindex Wstrict-overflow
3393 @opindex Wno-strict-overflow
3394 This option is only active when @option{-fstrict-overflow} is active.
3395 It warns about cases where the compiler optimizes based on the
3396 assumption that signed overflow does not occur. Note that it does not
3397 warn about all cases where the code might overflow: it only warns
3398 about cases where the compiler implements some optimization. Thus
3399 this warning depends on the optimization level.
3401 An optimization which assumes that signed overflow does not occur is
3402 perfectly safe if the values of the variables involved are such that
3403 overflow never does, in fact, occur. Therefore this warning can
3404 easily give a false positive: a warning about code which is not
3405 actually a problem. To help focus on important issues, several
3406 warning levels are defined. No warnings are issued for the use of
3407 undefined signed overflow when estimating how many iterations a loop
3408 will require, in particular when determining whether a loop will be
3412 @item -Wstrict-overflow=1
3413 Warn about cases which are both questionable and easy to avoid. For
3414 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3415 compiler will simplify this to @code{1}. This level of
3416 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3417 are not, and must be explicitly requested.
3419 @item -Wstrict-overflow=2
3420 Also warn about other cases where a comparison is simplified to a
3421 constant. For example: @code{abs (x) >= 0}. This can only be
3422 simplified when @option{-fstrict-overflow} is in effect, because
3423 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3424 zero. @option{-Wstrict-overflow} (with no level) is the same as
3425 @option{-Wstrict-overflow=2}.
3427 @item -Wstrict-overflow=3
3428 Also warn about other cases where a comparison is simplified. For
3429 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3431 @item -Wstrict-overflow=4
3432 Also warn about other simplifications not covered by the above cases.
3433 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3435 @item -Wstrict-overflow=5
3436 Also warn about cases where the compiler reduces the magnitude of a
3437 constant involved in a comparison. For example: @code{x + 2 > y} will
3438 be simplified to @code{x + 1 >= y}. This is reported only at the
3439 highest warning level because this simplification applies to many
3440 comparisons, so this warning level will give a very large number of
3444 @item -Warray-bounds
3445 @opindex Wno-array-bounds
3446 @opindex Warray-bounds
3447 This option is only active when @option{-ftree-vrp} is active
3448 (default for -O2 and above). It warns about subscripts to arrays
3449 that are always out of bounds. This warning is enabled by @option{-Wall}.
3451 @item -Wno-div-by-zero
3452 @opindex Wno-div-by-zero
3453 @opindex Wdiv-by-zero
3454 Do not warn about compile-time integer division by zero. Floating point
3455 division by zero is not warned about, as it can be a legitimate way of
3456 obtaining infinities and NaNs.
3458 @item -Wsystem-headers
3459 @opindex Wsystem-headers
3460 @opindex Wno-system-headers
3461 @cindex warnings from system headers
3462 @cindex system headers, warnings from
3463 Print warning messages for constructs found in system header files.
3464 Warnings from system headers are normally suppressed, on the assumption
3465 that they usually do not indicate real problems and would only make the
3466 compiler output harder to read. Using this command line option tells
3467 GCC to emit warnings from system headers as if they occurred in user
3468 code. However, note that using @option{-Wall} in conjunction with this
3469 option will @emph{not} warn about unknown pragmas in system
3470 headers---for that, @option{-Wunknown-pragmas} must also be used.
3473 @opindex Wfloat-equal
3474 @opindex Wno-float-equal
3475 Warn if floating point values are used in equality comparisons.
3477 The idea behind this is that sometimes it is convenient (for the
3478 programmer) to consider floating-point values as approximations to
3479 infinitely precise real numbers. If you are doing this, then you need
3480 to compute (by analyzing the code, or in some other way) the maximum or
3481 likely maximum error that the computation introduces, and allow for it
3482 when performing comparisons (and when producing output, but that's a
3483 different problem). In particular, instead of testing for equality, you
3484 would check to see whether the two values have ranges that overlap; and
3485 this is done with the relational operators, so equality comparisons are
3488 @item -Wtraditional @r{(C and Objective-C only)}
3489 @opindex Wtraditional
3490 @opindex Wno-traditional
3491 Warn about certain constructs that behave differently in traditional and
3492 ISO C@. Also warn about ISO C constructs that have no traditional C
3493 equivalent, and/or problematic constructs which should be avoided.
3497 Macro parameters that appear within string literals in the macro body.
3498 In traditional C macro replacement takes place within string literals,
3499 but does not in ISO C@.
3502 In traditional C, some preprocessor directives did not exist.
3503 Traditional preprocessors would only consider a line to be a directive
3504 if the @samp{#} appeared in column 1 on the line. Therefore
3505 @option{-Wtraditional} warns about directives that traditional C
3506 understands but would ignore because the @samp{#} does not appear as the
3507 first character on the line. It also suggests you hide directives like
3508 @samp{#pragma} not understood by traditional C by indenting them. Some
3509 traditional implementations would not recognize @samp{#elif}, so it
3510 suggests avoiding it altogether.
3513 A function-like macro that appears without arguments.
3516 The unary plus operator.
3519 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3520 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3521 constants.) Note, these suffixes appear in macros defined in the system
3522 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3523 Use of these macros in user code might normally lead to spurious
3524 warnings, however GCC's integrated preprocessor has enough context to
3525 avoid warning in these cases.
3528 A function declared external in one block and then used after the end of
3532 A @code{switch} statement has an operand of type @code{long}.
3535 A non-@code{static} function declaration follows a @code{static} one.
3536 This construct is not accepted by some traditional C compilers.
3539 The ISO type of an integer constant has a different width or
3540 signedness from its traditional type. This warning is only issued if
3541 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3542 typically represent bit patterns, are not warned about.
3545 Usage of ISO string concatenation is detected.
3548 Initialization of automatic aggregates.
3551 Identifier conflicts with labels. Traditional C lacks a separate
3552 namespace for labels.
3555 Initialization of unions. If the initializer is zero, the warning is
3556 omitted. This is done under the assumption that the zero initializer in
3557 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3558 initializer warnings and relies on default initialization to zero in the
3562 Conversions by prototypes between fixed/floating point values and vice
3563 versa. The absence of these prototypes when compiling with traditional
3564 C would cause serious problems. This is a subset of the possible
3565 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3568 Use of ISO C style function definitions. This warning intentionally is
3569 @emph{not} issued for prototype declarations or variadic functions
3570 because these ISO C features will appear in your code when using
3571 libiberty's traditional C compatibility macros, @code{PARAMS} and
3572 @code{VPARAMS}. This warning is also bypassed for nested functions
3573 because that feature is already a GCC extension and thus not relevant to
3574 traditional C compatibility.
3577 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3578 @opindex Wtraditional-conversion
3579 @opindex Wno-traditional-conversion
3580 Warn if a prototype causes a type conversion that is different from what
3581 would happen to the same argument in the absence of a prototype. This
3582 includes conversions of fixed point to floating and vice versa, and
3583 conversions changing the width or signedness of a fixed point argument
3584 except when the same as the default promotion.
3586 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3587 @opindex Wdeclaration-after-statement
3588 @opindex Wno-declaration-after-statement
3589 Warn when a declaration is found after a statement in a block. This
3590 construct, known from C++, was introduced with ISO C99 and is by default
3591 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3592 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3597 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3599 @item -Wno-endif-labels
3600 @opindex Wno-endif-labels
3601 @opindex Wendif-labels
3602 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3607 Warn whenever a local variable shadows another local variable, parameter or
3608 global variable or whenever a built-in function is shadowed.
3610 @item -Wlarger-than=@var{len}
3611 @opindex Wlarger-than=@var{len}
3612 @opindex Wlarger-than-@var{len}
3613 Warn whenever an object of larger than @var{len} bytes is defined.
3615 @item -Wframe-larger-than=@var{len}
3616 @opindex Wframe-larger-than
3617 Warn if the size of a function frame is larger than @var{len} bytes.
3618 The computation done to determine the stack frame size is approximate
3619 and not conservative.
3620 The actual requirements may be somewhat greater than @var{len}
3621 even if you do not get a warning. In addition, any space allocated
3622 via @code{alloca}, variable-length arrays, or related constructs
3623 is not included by the compiler when determining
3624 whether or not to issue a warning.
3626 @item -Wunsafe-loop-optimizations
3627 @opindex Wunsafe-loop-optimizations
3628 @opindex Wno-unsafe-loop-optimizations
3629 Warn if the loop cannot be optimized because the compiler could not
3630 assume anything on the bounds of the loop indices. With
3631 @option{-funsafe-loop-optimizations} warn if the compiler made
3634 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3635 @opindex Wno-pedantic-ms-format
3636 @opindex Wpedantic-ms-format
3637 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3638 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3639 depending on the MS runtime, when you are using the options @option{-Wformat}
3640 and @option{-pedantic} without gnu-extensions.
3642 @item -Wpointer-arith
3643 @opindex Wpointer-arith
3644 @opindex Wno-pointer-arith
3645 Warn about anything that depends on the ``size of'' a function type or
3646 of @code{void}. GNU C assigns these types a size of 1, for
3647 convenience in calculations with @code{void *} pointers and pointers
3648 to functions. In C++, warn also when an arithmetic operation involves
3649 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3652 @opindex Wtype-limits
3653 @opindex Wno-type-limits
3654 Warn if a comparison is always true or always false due to the limited
3655 range of the data type, but do not warn for constant expressions. For
3656 example, warn if an unsigned variable is compared against zero with
3657 @samp{<} or @samp{>=}. This warning is also enabled by
3660 @item -Wbad-function-cast @r{(C and Objective-C only)}
3661 @opindex Wbad-function-cast
3662 @opindex Wno-bad-function-cast
3663 Warn whenever a function call is cast to a non-matching type.
3664 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3666 @item -Wc++-compat @r{(C and Objective-C only)}
3667 Warn about ISO C constructs that are outside of the common subset of
3668 ISO C and ISO C++, e.g.@: request for implicit conversion from
3669 @code{void *} to a pointer to non-@code{void} type.
3671 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3672 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3673 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3674 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3678 @opindex Wno-cast-qual
3679 Warn whenever a pointer is cast so as to remove a type qualifier from
3680 the target type. For example, warn if a @code{const char *} is cast
3681 to an ordinary @code{char *}.
3684 @opindex Wcast-align
3685 @opindex Wno-cast-align
3686 Warn whenever a pointer is cast such that the required alignment of the
3687 target is increased. For example, warn if a @code{char *} is cast to
3688 an @code{int *} on machines where integers can only be accessed at
3689 two- or four-byte boundaries.
3691 @item -Wwrite-strings
3692 @opindex Wwrite-strings
3693 @opindex Wno-write-strings
3694 When compiling C, give string constants the type @code{const
3695 char[@var{length}]} so that copying the address of one into a
3696 non-@code{const} @code{char *} pointer will get a warning. These
3697 warnings will help you find at compile time code that can try to write
3698 into a string constant, but only if you have been very careful about
3699 using @code{const} in declarations and prototypes. Otherwise, it will
3700 just be a nuisance. This is why we did not make @option{-Wall} request
3703 When compiling C++, warn about the deprecated conversion from string
3704 literals to @code{char *}. This warning is enabled by default for C++
3709 @opindex Wno-clobbered
3710 Warn for variables that might be changed by @samp{longjmp} or
3711 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3714 @opindex Wconversion
3715 @opindex Wno-conversion
3716 Warn for implicit conversions that may alter a value. This includes
3717 conversions between real and integer, like @code{abs (x)} when
3718 @code{x} is @code{double}; conversions between signed and unsigned,
3719 like @code{unsigned ui = -1}; and conversions to smaller types, like
3720 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3721 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3722 changed by the conversion like in @code{abs (2.0)}. Warnings about
3723 conversions between signed and unsigned integers can be disabled by
3724 using @option{-Wno-sign-conversion}.
3726 For C++, also warn for conversions between @code{NULL} and non-pointer
3727 types; confusing overload resolution for user-defined conversions; and
3728 conversions that will never use a type conversion operator:
3729 conversions to @code{void}, the same type, a base class or a reference
3730 to them. Warnings about conversions between signed and unsigned
3731 integers are disabled by default in C++ unless
3732 @option{-Wsign-conversion} is explicitly enabled.
3735 @opindex Wempty-body
3736 @opindex Wno-empty-body
3737 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3738 while} statement. This warning is also enabled by @option{-Wextra}.
3740 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3741 @opindex Wenum-compare
3742 @opindex Wno-enum-compare
3743 Warn about a comparison between values of different enum types. This
3744 warning is enabled by default.
3746 @item -Wsign-compare
3747 @opindex Wsign-compare
3748 @opindex Wno-sign-compare
3749 @cindex warning for comparison of signed and unsigned values
3750 @cindex comparison of signed and unsigned values, warning
3751 @cindex signed and unsigned values, comparison warning
3752 Warn when a comparison between signed and unsigned values could produce
3753 an incorrect result when the signed value is converted to unsigned.
3754 This warning is also enabled by @option{-Wextra}; to get the other warnings
3755 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3757 @item -Wsign-conversion
3758 @opindex Wsign-conversion
3759 @opindex Wno-sign-conversion
3760 Warn for implicit conversions that may change the sign of an integer
3761 value, like assigning a signed integer expression to an unsigned
3762 integer variable. An explicit cast silences the warning. In C, this
3763 option is enabled also by @option{-Wconversion}.
3767 @opindex Wno-address
3768 Warn about suspicious uses of memory addresses. These include using
3769 the address of a function in a conditional expression, such as
3770 @code{void func(void); if (func)}, and comparisons against the memory
3771 address of a string literal, such as @code{if (x == "abc")}. Such
3772 uses typically indicate a programmer error: the address of a function
3773 always evaluates to true, so their use in a conditional usually
3774 indicate that the programmer forgot the parentheses in a function
3775 call; and comparisons against string literals result in unspecified
3776 behavior and are not portable in C, so they usually indicate that the
3777 programmer intended to use @code{strcmp}. This warning is enabled by
3781 @opindex Wlogical-op
3782 @opindex Wno-logical-op
3783 Warn about suspicious uses of logical operators in expressions.
3784 This includes using logical operators in contexts where a
3785 bit-wise operator is likely to be expected.
3787 @item -Waggregate-return
3788 @opindex Waggregate-return
3789 @opindex Wno-aggregate-return
3790 Warn if any functions that return structures or unions are defined or
3791 called. (In languages where you can return an array, this also elicits
3794 @item -Wno-attributes
3795 @opindex Wno-attributes
3796 @opindex Wattributes
3797 Do not warn if an unexpected @code{__attribute__} is used, such as
3798 unrecognized attributes, function attributes applied to variables,
3799 etc. This will not stop errors for incorrect use of supported
3802 @item -Wno-builtin-macro-redefined
3803 @opindex Wno-builtin-macro-redefined
3804 @opindex Wbuiltin-macro-redefined
3805 Do not warn if certain built-in macros are redefined. This suppresses
3806 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3807 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3809 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3810 @opindex Wstrict-prototypes
3811 @opindex Wno-strict-prototypes
3812 Warn if a function is declared or defined without specifying the
3813 argument types. (An old-style function definition is permitted without
3814 a warning if preceded by a declaration which specifies the argument
3817 @item -Wold-style-declaration @r{(C and Objective-C only)}
3818 @opindex Wold-style-declaration
3819 @opindex Wno-old-style-declaration
3820 Warn for obsolescent usages, according to the C Standard, in a
3821 declaration. For example, warn if storage-class specifiers like
3822 @code{static} are not the first things in a declaration. This warning
3823 is also enabled by @option{-Wextra}.
3825 @item -Wold-style-definition @r{(C and Objective-C only)}
3826 @opindex Wold-style-definition
3827 @opindex Wno-old-style-definition
3828 Warn if an old-style function definition is used. A warning is given
3829 even if there is a previous prototype.
3831 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3832 @opindex Wmissing-parameter-type
3833 @opindex Wno-missing-parameter-type
3834 A function parameter is declared without a type specifier in K&R-style
3841 This warning is also enabled by @option{-Wextra}.
3843 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3844 @opindex Wmissing-prototypes
3845 @opindex Wno-missing-prototypes
3846 Warn if a global function is defined without a previous prototype
3847 declaration. This warning is issued even if the definition itself
3848 provides a prototype. The aim is to detect global functions that fail
3849 to be declared in header files.
3851 @item -Wmissing-declarations
3852 @opindex Wmissing-declarations
3853 @opindex Wno-missing-declarations
3854 Warn if a global function is defined without a previous declaration.
3855 Do so even if the definition itself provides a prototype.
3856 Use this option to detect global functions that are not declared in
3857 header files. In C++, no warnings are issued for function templates,
3858 or for inline functions, or for functions in anonymous namespaces.
3860 @item -Wmissing-field-initializers
3861 @opindex Wmissing-field-initializers
3862 @opindex Wno-missing-field-initializers
3866 Warn if a structure's initializer has some fields missing. For
3867 example, the following code would cause such a warning, because
3868 @code{x.h} is implicitly zero:
3871 struct s @{ int f, g, h; @};
3872 struct s x = @{ 3, 4 @};
3875 This option does not warn about designated initializers, so the following
3876 modification would not trigger a warning:
3879 struct s @{ int f, g, h; @};
3880 struct s x = @{ .f = 3, .g = 4 @};
3883 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3884 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3886 @item -Wmissing-noreturn
3887 @opindex Wmissing-noreturn
3888 @opindex Wno-missing-noreturn
3889 Warn about functions which might be candidates for attribute @code{noreturn}.
3890 Note these are only possible candidates, not absolute ones. Care should
3891 be taken to manually verify functions actually do not ever return before
3892 adding the @code{noreturn} attribute, otherwise subtle code generation
3893 bugs could be introduced. You will not get a warning for @code{main} in
3894 hosted C environments.
3896 @item -Wmissing-format-attribute
3897 @opindex Wmissing-format-attribute
3898 @opindex Wno-missing-format-attribute
3901 Warn about function pointers which might be candidates for @code{format}
3902 attributes. Note these are only possible candidates, not absolute ones.
3903 GCC will guess that function pointers with @code{format} attributes that
3904 are used in assignment, initialization, parameter passing or return
3905 statements should have a corresponding @code{format} attribute in the
3906 resulting type. I.e.@: the left-hand side of the assignment or
3907 initialization, the type of the parameter variable, or the return type
3908 of the containing function respectively should also have a @code{format}
3909 attribute to avoid the warning.
3911 GCC will also warn about function definitions which might be
3912 candidates for @code{format} attributes. Again, these are only
3913 possible candidates. GCC will guess that @code{format} attributes
3914 might be appropriate for any function that calls a function like
3915 @code{vprintf} or @code{vscanf}, but this might not always be the
3916 case, and some functions for which @code{format} attributes are
3917 appropriate may not be detected.
3919 @item -Wno-multichar
3920 @opindex Wno-multichar
3922 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3923 Usually they indicate a typo in the user's code, as they have
3924 implementation-defined values, and should not be used in portable code.
3926 @item -Wnormalized=<none|id|nfc|nfkc>
3927 @opindex Wnormalized=
3930 @cindex character set, input normalization
3931 In ISO C and ISO C++, two identifiers are different if they are
3932 different sequences of characters. However, sometimes when characters
3933 outside the basic ASCII character set are used, you can have two
3934 different character sequences that look the same. To avoid confusion,
3935 the ISO 10646 standard sets out some @dfn{normalization rules} which
3936 when applied ensure that two sequences that look the same are turned into
3937 the same sequence. GCC can warn you if you are using identifiers which
3938 have not been normalized; this option controls that warning.
3940 There are four levels of warning that GCC supports. The default is
3941 @option{-Wnormalized=nfc}, which warns about any identifier which is
3942 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3943 recommended form for most uses.
3945 Unfortunately, there are some characters which ISO C and ISO C++ allow
3946 in identifiers that when turned into NFC aren't allowable as
3947 identifiers. That is, there's no way to use these symbols in portable
3948 ISO C or C++ and have all your identifiers in NFC@.
3949 @option{-Wnormalized=id} suppresses the warning for these characters.
3950 It is hoped that future versions of the standards involved will correct
3951 this, which is why this option is not the default.
3953 You can switch the warning off for all characters by writing
3954 @option{-Wnormalized=none}. You would only want to do this if you
3955 were using some other normalization scheme (like ``D''), because
3956 otherwise you can easily create bugs that are literally impossible to see.
3958 Some characters in ISO 10646 have distinct meanings but look identical
3959 in some fonts or display methodologies, especially once formatting has
3960 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3961 LETTER N'', will display just like a regular @code{n} which has been
3962 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3963 normalization scheme to convert all these into a standard form as
3964 well, and GCC will warn if your code is not in NFKC if you use
3965 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3966 about every identifier that contains the letter O because it might be
3967 confused with the digit 0, and so is not the default, but may be
3968 useful as a local coding convention if the programming environment is
3969 unable to be fixed to display these characters distinctly.
3971 @item -Wno-deprecated
3972 @opindex Wno-deprecated
3973 @opindex Wdeprecated
3974 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
3976 @item -Wno-deprecated-declarations
3977 @opindex Wno-deprecated-declarations
3978 @opindex Wdeprecated-declarations
3979 Do not warn about uses of functions (@pxref{Function Attributes}),
3980 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3981 Attributes}) marked as deprecated by using the @code{deprecated}
3985 @opindex Wno-overflow
3987 Do not warn about compile-time overflow in constant expressions.
3989 @item -Woverride-init @r{(C and Objective-C only)}
3990 @opindex Woverride-init
3991 @opindex Wno-override-init
3995 Warn if an initialized field without side effects is overridden when
3996 using designated initializers (@pxref{Designated Inits, , Designated
3999 This warning is included in @option{-Wextra}. To get other
4000 @option{-Wextra} warnings without this one, use @samp{-Wextra
4001 -Wno-override-init}.
4006 Warn if a structure is given the packed attribute, but the packed
4007 attribute has no effect on the layout or size of the structure.
4008 Such structures may be mis-aligned for little benefit. For
4009 instance, in this code, the variable @code{f.x} in @code{struct bar}
4010 will be misaligned even though @code{struct bar} does not itself
4011 have the packed attribute:
4018 @} __attribute__((packed));
4026 @item -Wpacked-bitfield-compat
4027 @opindex Wpacked-bitfield-compat
4028 @opindex Wno-packed-bitfield-compat
4029 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4030 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4031 the change can lead to differences in the structure layout. GCC
4032 informs you when the offset of such a field has changed in GCC 4.4.
4033 For example there is no longer a 4-bit padding between field @code{a}
4034 and @code{b} in this structure:
4041 @} __attribute__ ((packed));
4044 This warning is enabled by default. Use
4045 @option{-Wno-packed-bitfield-compat} to disable this warning.
4050 Warn if padding is included in a structure, either to align an element
4051 of the structure or to align the whole structure. Sometimes when this
4052 happens it is possible to rearrange the fields of the structure to
4053 reduce the padding and so make the structure smaller.
4055 @item -Wredundant-decls
4056 @opindex Wredundant-decls
4057 @opindex Wno-redundant-decls
4058 Warn if anything is declared more than once in the same scope, even in
4059 cases where multiple declaration is valid and changes nothing.
4061 @item -Wnested-externs @r{(C and Objective-C only)}
4062 @opindex Wnested-externs
4063 @opindex Wno-nested-externs
4064 Warn if an @code{extern} declaration is encountered within a function.
4066 @item -Wunreachable-code
4067 @opindex Wunreachable-code
4068 @opindex Wno-unreachable-code
4069 Warn if the compiler detects that code will never be executed.
4071 This option is intended to warn when the compiler detects that at
4072 least a whole line of source code will never be executed, because
4073 some condition is never satisfied or because it is after a
4074 procedure that never returns.
4076 It is possible for this option to produce a warning even though there
4077 are circumstances under which part of the affected line can be executed,
4078 so care should be taken when removing apparently-unreachable code.
4080 For instance, when a function is inlined, a warning may mean that the
4081 line is unreachable in only one inlined copy of the function.
4083 This option is not made part of @option{-Wall} because in a debugging
4084 version of a program there is often substantial code which checks
4085 correct functioning of the program and is, hopefully, unreachable
4086 because the program does work. Another common use of unreachable
4087 code is to provide behavior which is selectable at compile-time.
4092 Warn if a function can not be inlined and it was declared as inline.
4093 Even with this option, the compiler will not warn about failures to
4094 inline functions declared in system headers.
4096 The compiler uses a variety of heuristics to determine whether or not
4097 to inline a function. For example, the compiler takes into account
4098 the size of the function being inlined and the amount of inlining
4099 that has already been done in the current function. Therefore,
4100 seemingly insignificant changes in the source program can cause the
4101 warnings produced by @option{-Winline} to appear or disappear.
4103 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4104 @opindex Wno-invalid-offsetof
4105 @opindex Winvalid-offsetof
4106 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4107 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4108 to a non-POD type is undefined. In existing C++ implementations,
4109 however, @samp{offsetof} typically gives meaningful results even when
4110 applied to certain kinds of non-POD types. (Such as a simple
4111 @samp{struct} that fails to be a POD type only by virtue of having a
4112 constructor.) This flag is for users who are aware that they are
4113 writing nonportable code and who have deliberately chosen to ignore the
4116 The restrictions on @samp{offsetof} may be relaxed in a future version
4117 of the C++ standard.
4119 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4120 @opindex Wno-int-to-pointer-cast
4121 @opindex Wint-to-pointer-cast
4122 Suppress warnings from casts to pointer type of an integer of a
4125 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4126 @opindex Wno-pointer-to-int-cast
4127 @opindex Wpointer-to-int-cast
4128 Suppress warnings from casts from a pointer to an integer type of a
4132 @opindex Winvalid-pch
4133 @opindex Wno-invalid-pch
4134 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4135 the search path but can't be used.
4139 @opindex Wno-long-long
4140 Warn if @samp{long long} type is used. This is default. To inhibit
4141 the warning messages, use @option{-Wno-long-long}. Flags
4142 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4143 only when @option{-pedantic} flag is used.
4145 @item -Wvariadic-macros
4146 @opindex Wvariadic-macros
4147 @opindex Wno-variadic-macros
4148 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4149 alternate syntax when in pedantic ISO C99 mode. This is default.
4150 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4155 Warn if variable length array is used in the code.
4156 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4157 the variable length array.
4159 @item -Wvolatile-register-var
4160 @opindex Wvolatile-register-var
4161 @opindex Wno-volatile-register-var
4162 Warn if a register variable is declared volatile. The volatile
4163 modifier does not inhibit all optimizations that may eliminate reads
4164 and/or writes to register variables. This warning is enabled by
4167 @item -Wdisabled-optimization
4168 @opindex Wdisabled-optimization
4169 @opindex Wno-disabled-optimization
4170 Warn if a requested optimization pass is disabled. This warning does
4171 not generally indicate that there is anything wrong with your code; it
4172 merely indicates that GCC's optimizers were unable to handle the code
4173 effectively. Often, the problem is that your code is too big or too
4174 complex; GCC will refuse to optimize programs when the optimization
4175 itself is likely to take inordinate amounts of time.
4177 @item -Wpointer-sign @r{(C and Objective-C only)}
4178 @opindex Wpointer-sign
4179 @opindex Wno-pointer-sign
4180 Warn for pointer argument passing or assignment with different signedness.
4181 This option is only supported for C and Objective-C@. It is implied by
4182 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4183 @option{-Wno-pointer-sign}.
4185 @item -Wstack-protector
4186 @opindex Wstack-protector
4187 @opindex Wno-stack-protector
4188 This option is only active when @option{-fstack-protector} is active. It
4189 warns about functions that will not be protected against stack smashing.
4192 @opindex Wno-mudflap
4193 Suppress warnings about constructs that cannot be instrumented by
4196 @item -Woverlength-strings
4197 @opindex Woverlength-strings
4198 @opindex Wno-overlength-strings
4199 Warn about string constants which are longer than the ``minimum
4200 maximum'' length specified in the C standard. Modern compilers
4201 generally allow string constants which are much longer than the
4202 standard's minimum limit, but very portable programs should avoid
4203 using longer strings.
4205 The limit applies @emph{after} string constant concatenation, and does
4206 not count the trailing NUL@. In C89, the limit was 509 characters; in
4207 C99, it was raised to 4095. C++98 does not specify a normative
4208 minimum maximum, so we do not diagnose overlength strings in C++@.
4210 This option is implied by @option{-pedantic}, and can be disabled with
4211 @option{-Wno-overlength-strings}.
4214 @node Debugging Options
4215 @section Options for Debugging Your Program or GCC
4216 @cindex options, debugging
4217 @cindex debugging information options
4219 GCC has various special options that are used for debugging
4220 either your program or GCC:
4225 Produce debugging information in the operating system's native format
4226 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4229 On most systems that use stabs format, @option{-g} enables use of extra
4230 debugging information that only GDB can use; this extra information
4231 makes debugging work better in GDB but will probably make other debuggers
4233 refuse to read the program. If you want to control for certain whether
4234 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4235 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4237 GCC allows you to use @option{-g} with
4238 @option{-O}. The shortcuts taken by optimized code may occasionally
4239 produce surprising results: some variables you declared may not exist
4240 at all; flow of control may briefly move where you did not expect it;
4241 some statements may not be executed because they compute constant
4242 results or their values were already at hand; some statements may
4243 execute in different places because they were moved out of loops.
4245 Nevertheless it proves possible to debug optimized output. This makes
4246 it reasonable to use the optimizer for programs that might have bugs.
4248 The following options are useful when GCC is generated with the
4249 capability for more than one debugging format.
4253 Produce debugging information for use by GDB@. This means to use the
4254 most expressive format available (DWARF 2, stabs, or the native format
4255 if neither of those are supported), including GDB extensions if at all
4260 Produce debugging information in stabs format (if that is supported),
4261 without GDB extensions. This is the format used by DBX on most BSD
4262 systems. On MIPS, Alpha and System V Release 4 systems this option
4263 produces stabs debugging output which is not understood by DBX or SDB@.
4264 On System V Release 4 systems this option requires the GNU assembler.
4266 @item -feliminate-unused-debug-symbols
4267 @opindex feliminate-unused-debug-symbols
4268 Produce debugging information in stabs format (if that is supported),
4269 for only symbols that are actually used.
4271 @item -femit-class-debug-always
4272 Instead of emitting debugging information for a C++ class in only one
4273 object file, emit it in all object files using the class. This option
4274 should be used only with debuggers that are unable to handle the way GCC
4275 normally emits debugging information for classes because using this
4276 option will increase the size of debugging information by as much as a
4281 Produce debugging information in stabs format (if that is supported),
4282 using GNU extensions understood only by the GNU debugger (GDB)@. The
4283 use of these extensions is likely to make other debuggers crash or
4284 refuse to read the program.
4288 Produce debugging information in COFF format (if that is supported).
4289 This is the format used by SDB on most System V systems prior to
4294 Produce debugging information in XCOFF format (if that is supported).
4295 This is the format used by the DBX debugger on IBM RS/6000 systems.
4299 Produce debugging information in XCOFF format (if that is supported),
4300 using GNU extensions understood only by the GNU debugger (GDB)@. The
4301 use of these extensions is likely to make other debuggers crash or
4302 refuse to read the program, and may cause assemblers other than the GNU
4303 assembler (GAS) to fail with an error.
4307 Produce debugging information in DWARF version 2 format (if that is
4308 supported). This is the format used by DBX on IRIX 6. With this
4309 option, GCC uses features of DWARF version 3 when they are useful;
4310 version 3 is upward compatible with version 2, but may still cause
4311 problems for older debuggers.
4315 Produce debugging information in VMS debug format (if that is
4316 supported). This is the format used by DEBUG on VMS systems.
4319 @itemx -ggdb@var{level}
4320 @itemx -gstabs@var{level}
4321 @itemx -gcoff@var{level}
4322 @itemx -gxcoff@var{level}
4323 @itemx -gvms@var{level}
4324 Request debugging information and also use @var{level} to specify how
4325 much information. The default level is 2.
4327 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4330 Level 1 produces minimal information, enough for making backtraces in
4331 parts of the program that you don't plan to debug. This includes
4332 descriptions of functions and external variables, but no information
4333 about local variables and no line numbers.
4335 Level 3 includes extra information, such as all the macro definitions
4336 present in the program. Some debuggers support macro expansion when
4337 you use @option{-g3}.
4339 @option{-gdwarf-2} does not accept a concatenated debug level, because
4340 GCC used to support an option @option{-gdwarf} that meant to generate
4341 debug information in version 1 of the DWARF format (which is very
4342 different from version 2), and it would have been too confusing. That
4343 debug format is long obsolete, but the option cannot be changed now.
4344 Instead use an additional @option{-g@var{level}} option to change the
4345 debug level for DWARF2.
4347 @item -feliminate-dwarf2-dups
4348 @opindex feliminate-dwarf2-dups
4349 Compress DWARF2 debugging information by eliminating duplicated
4350 information about each symbol. This option only makes sense when
4351 generating DWARF2 debugging information with @option{-gdwarf-2}.
4353 @item -femit-struct-debug-baseonly
4354 Emit debug information for struct-like types
4355 only when the base name of the compilation source file
4356 matches the base name of file in which the struct was defined.
4358 This option substantially reduces the size of debugging information,
4359 but at significant potential loss in type information to the debugger.
4360 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4361 See @option{-femit-struct-debug-detailed} for more detailed control.
4363 This option works only with DWARF 2.
4365 @item -femit-struct-debug-reduced
4366 Emit debug information for struct-like types
4367 only when the base name of the compilation source file
4368 matches the base name of file in which the type was defined,
4369 unless the struct is a template or defined in a system header.
4371 This option significantly reduces the size of debugging information,
4372 with some potential loss in type information to the debugger.
4373 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4374 See @option{-femit-struct-debug-detailed} for more detailed control.
4376 This option works only with DWARF 2.
4378 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4379 Specify the struct-like types
4380 for which the compiler will generate debug information.
4381 The intent is to reduce duplicate struct debug information
4382 between different object files within the same program.
4384 This option is a detailed version of
4385 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4386 which will serve for most needs.
4388 A specification has the syntax
4389 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4391 The optional first word limits the specification to
4392 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4393 A struct type is used directly when it is the type of a variable, member.
4394 Indirect uses arise through pointers to structs.
4395 That is, when use of an incomplete struct would be legal, the use is indirect.
4397 @samp{struct one direct; struct two * indirect;}.
4399 The optional second word limits the specification to
4400 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4401 Generic structs are a bit complicated to explain.
4402 For C++, these are non-explicit specializations of template classes,
4403 or non-template classes within the above.
4404 Other programming languages have generics,
4405 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4407 The third word specifies the source files for those
4408 structs for which the compiler will emit debug information.
4409 The values @samp{none} and @samp{any} have the normal meaning.
4410 The value @samp{base} means that
4411 the base of name of the file in which the type declaration appears
4412 must match the base of the name of the main compilation file.
4413 In practice, this means that
4414 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4415 but types declared in other header will not.
4416 The value @samp{sys} means those types satisfying @samp{base}
4417 or declared in system or compiler headers.
4419 You may need to experiment to determine the best settings for your application.
4421 The default is @samp{-femit-struct-debug-detailed=all}.
4423 This option works only with DWARF 2.
4425 @item -fno-merge-debug-strings
4426 @opindex fmerge-debug-strings
4427 @opindex fno-merge-debug-strings
4428 Direct the linker to not merge together strings in the debugging
4429 information which are identical in different object files. Merging is
4430 not supported by all assemblers or linkers. Merging decreases the size
4431 of the debug information in the output file at the cost of increasing
4432 link processing time. Merging is enabled by default.
4434 @item -fdebug-prefix-map=@var{old}=@var{new}
4435 @opindex fdebug-prefix-map
4436 When compiling files in directory @file{@var{old}}, record debugging
4437 information describing them as in @file{@var{new}} instead.
4439 @item -fno-dwarf2-cfi-asm
4440 @opindex fdwarf2-cfi-asm
4441 @opindex fno-dwarf2-cfi-asm
4442 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4443 instead of using GAS @code{.cfi_*} directives.
4445 @cindex @command{prof}
4448 Generate extra code to write profile information suitable for the
4449 analysis program @command{prof}. You must use this option when compiling
4450 the source files you want data about, and you must also use it when
4453 @cindex @command{gprof}
4456 Generate extra code to write profile information suitable for the
4457 analysis program @command{gprof}. You must use this option when compiling
4458 the source files you want data about, and you must also use it when
4463 Makes the compiler print out each function name as it is compiled, and
4464 print some statistics about each pass when it finishes.
4467 @opindex ftime-report
4468 Makes the compiler print some statistics about the time consumed by each
4469 pass when it finishes.
4472 @opindex fmem-report
4473 Makes the compiler print some statistics about permanent memory
4474 allocation when it finishes.
4476 @item -fpre-ipa-mem-report
4477 @opindex fpre-ipa-mem-report
4478 @item -fpost-ipa-mem-report
4479 @opindex fpost-ipa-mem-report
4480 Makes the compiler print some statistics about permanent memory
4481 allocation before or after interprocedural optimization.
4483 @item -fprofile-arcs
4484 @opindex fprofile-arcs
4485 Add code so that program flow @dfn{arcs} are instrumented. During
4486 execution the program records how many times each branch and call is
4487 executed and how many times it is taken or returns. When the compiled
4488 program exits it saves this data to a file called
4489 @file{@var{auxname}.gcda} for each source file. The data may be used for
4490 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4491 test coverage analysis (@option{-ftest-coverage}). Each object file's
4492 @var{auxname} is generated from the name of the output file, if
4493 explicitly specified and it is not the final executable, otherwise it is
4494 the basename of the source file. In both cases any suffix is removed
4495 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4496 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4497 @xref{Cross-profiling}.
4499 @cindex @command{gcov}
4503 This option is used to compile and link code instrumented for coverage
4504 analysis. The option is a synonym for @option{-fprofile-arcs}
4505 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4506 linking). See the documentation for those options for more details.
4511 Compile the source files with @option{-fprofile-arcs} plus optimization
4512 and code generation options. For test coverage analysis, use the
4513 additional @option{-ftest-coverage} option. You do not need to profile
4514 every source file in a program.
4517 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4518 (the latter implies the former).
4521 Run the program on a representative workload to generate the arc profile
4522 information. This may be repeated any number of times. You can run
4523 concurrent instances of your program, and provided that the file system
4524 supports locking, the data files will be correctly updated. Also
4525 @code{fork} calls are detected and correctly handled (double counting
4529 For profile-directed optimizations, compile the source files again with
4530 the same optimization and code generation options plus
4531 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4532 Control Optimization}).
4535 For test coverage analysis, use @command{gcov} to produce human readable
4536 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4537 @command{gcov} documentation for further information.
4541 With @option{-fprofile-arcs}, for each function of your program GCC
4542 creates a program flow graph, then finds a spanning tree for the graph.
4543 Only arcs that are not on the spanning tree have to be instrumented: the
4544 compiler adds code to count the number of times that these arcs are
4545 executed. When an arc is the only exit or only entrance to a block, the
4546 instrumentation code can be added to the block; otherwise, a new basic
4547 block must be created to hold the instrumentation code.
4550 @item -ftest-coverage
4551 @opindex ftest-coverage
4552 Produce a notes file that the @command{gcov} code-coverage utility
4553 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4554 show program coverage. Each source file's note file is called
4555 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4556 above for a description of @var{auxname} and instructions on how to
4557 generate test coverage data. Coverage data will match the source files
4558 more closely, if you do not optimize.
4560 @item -fdbg-cnt-list
4561 @opindex fdbg-cnt-list
4562 Print the name and the counter upperbound for all debug counters.
4564 @item -fdbg-cnt=@var{counter-value-list}
4566 Set the internal debug counter upperbound. @var{counter-value-list}
4567 is a comma-separated list of @var{name}:@var{value} pairs
4568 which sets the upperbound of each debug counter @var{name} to @var{value}.
4569 All debug counters have the initial upperbound of @var{UINT_MAX},
4570 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4571 e.g. With -fdbg-cnt=dce:10,tail_call:0
4572 dbg_cnt(dce) will return true only for first 10 invocations
4573 and dbg_cnt(tail_call) will return false always.
4575 @item -d@var{letters}
4576 @itemx -fdump-rtl-@var{pass}
4578 Says to make debugging dumps during compilation at times specified by
4579 @var{letters}. This is used for debugging the RTL-based passes of the
4580 compiler. The file names for most of the dumps are made by appending a
4581 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4582 from the name of the output file, if explicitly specified and it is not
4583 an executable, otherwise it is the basename of the source file. These
4584 switches may have different effects when @option{-E} is used for
4587 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4588 @option{-d} option @var{letters}. Here are the possible
4589 letters for use in @var{pass} and @var{letters}, and their meanings:
4593 @item -fdump-rtl-alignments
4594 @opindex fdump-rtl-alignments
4595 Dump after branch alignments have been computed.
4597 @item -fdump-rtl-asmcons
4598 @opindex fdump-rtl-asmcons
4599 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4601 @item -fdump-rtl-auto_inc_dec
4602 @opindex fdump-rtl-auto_inc_dec
4603 Dump after auto-inc-dec discovery. This pass is only run on
4604 architectures that have auto inc or auto dec instructions.
4606 @item -fdump-rtl-barriers
4607 @opindex fdump-rtl-barriers
4608 Dump after cleaning up the barrier instructions.
4610 @item -fdump-rtl-bbpart
4611 @opindex fdump-rtl-bbpart
4612 Dump after partitioning hot and cold basic blocks.
4614 @item -fdump-rtl-bbro
4615 @opindex fdump-rtl-bbro
4616 Dump after block reordering.
4618 @item -fdump-rtl-btl1
4619 @itemx -fdump-rtl-btl2
4620 @opindex fdump-rtl-btl2
4621 @opindex fdump-rtl-btl2
4622 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4623 after the two branch
4624 target load optimization passes.
4626 @item -fdump-rtl-bypass
4627 @opindex fdump-rtl-bypass
4628 Dump after jump bypassing and control flow optimizations.
4630 @item -fdump-rtl-combine
4631 @opindex fdump-rtl-combine
4632 Dump after the RTL instruction combination pass.
4634 @item -fdump-rtl-compgotos
4635 @opindex fdump-rtl-compgotos
4636 Dump after duplicating the computed gotos.
4638 @item -fdump-rtl-ce1
4639 @itemx -fdump-rtl-ce2
4640 @itemx -fdump-rtl-ce3
4641 @opindex fdump-rtl-ce1
4642 @opindex fdump-rtl-ce2
4643 @opindex fdump-rtl-ce3
4644 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4645 @option{-fdump-rtl-ce3} enable dumping after the three
4646 if conversion passes.
4648 @itemx -fdump-rtl-cprop_hardreg
4649 @opindex fdump-rtl-cprop_hardreg
4650 Dump after hard register copy propagation.
4652 @itemx -fdump-rtl-csa
4653 @opindex fdump-rtl-csa
4654 Dump after combining stack adjustments.
4656 @item -fdump-rtl-cse1
4657 @itemx -fdump-rtl-cse2
4658 @opindex fdump-rtl-cse1
4659 @opindex fdump-rtl-cse2
4660 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4661 the two common sub-expression elimination passes.
4663 @itemx -fdump-rtl-dce
4664 @opindex fdump-rtl-dce
4665 Dump after the standalone dead code elimination passes.
4667 @itemx -fdump-rtl-dbr
4668 @opindex fdump-rtl-dbr
4669 Dump after delayed branch scheduling.
4671 @item -fdump-rtl-dce1
4672 @itemx -fdump-rtl-dce2
4673 @opindex fdump-rtl-dce1
4674 @opindex fdump-rtl-dce2
4675 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4676 the two dead store elimination passes.
4679 @opindex fdump-rtl-eh
4680 Dump after finalization of EH handling code.
4682 @item -fdump-rtl-eh_ranges
4683 @opindex fdump-rtl-eh_ranges
4684 Dump after conversion of EH handling range regions.
4686 @item -fdump-rtl-expand
4687 @opindex fdump-rtl-expand
4688 Dump after RTL generation.
4690 @item -fdump-rtl-fwprop1
4691 @itemx -fdump-rtl-fwprop2
4692 @opindex fdump-rtl-fwprop1
4693 @opindex fdump-rtl-fwprop2
4694 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4695 dumping after the two forward propagation passes.
4697 @item -fdump-rtl-gcse1
4698 @itemx -fdump-rtl-gcse2
4699 @opindex fdump-rtl-gcse1
4700 @opindex fdump-rtl-gcse2
4701 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4702 after global common subexpression elimination.
4704 @item -fdump-rtl-init-regs
4705 @opindex fdump-rtl-init-regs
4706 Dump after the initialization of the registers.
4708 @item -fdump-rtl-initvals
4709 @opindex fdump-rtl-initvals
4710 Dump after the computation of the initial value sets.
4712 @itemx -fdump-rtl-into_cfglayout
4713 @opindex fdump-rtl-into_cfglayout
4714 Dump after converting to cfglayout mode.
4716 @item -fdump-rtl-ira
4717 @opindex fdump-rtl-ira
4718 Dump after iterated register allocation.
4720 @item -fdump-rtl-jump
4721 @opindex fdump-rtl-jump
4722 Dump after the second jump optimization.
4724 @item -fdump-rtl-loop2
4725 @opindex fdump-rtl-loop2
4726 @option{-fdump-rtl-loop2} enables dumping after the rtl
4727 loop optimization passes.
4729 @item -fdump-rtl-mach
4730 @opindex fdump-rtl-mach
4731 Dump after performing the machine dependent reorganization pass, if that
4734 @item -fdump-rtl-mode_sw
4735 @opindex fdump-rtl-mode_sw
4736 Dump after removing redundant mode switches.
4738 @item -fdump-rtl-rnreg
4739 @opindex fdump-rtl-rnreg
4740 Dump after register renumbering.
4742 @itemx -fdump-rtl-outof_cfglayout
4743 @opindex fdump-rtl-outof_cfglayout
4744 Dump after converting from cfglayout mode.
4746 @item -fdump-rtl-peephole2
4747 @opindex fdump-rtl-peephole2
4748 Dump after the peephole pass.
4750 @item -fdump-rtl-postreload
4751 @opindex fdump-rtl-postreload
4752 Dump after post-reload optimizations.
4754 @itemx -fdump-rtl-pro_and_epilogue
4755 @opindex fdump-rtl-pro_and_epilogue
4756 Dump after generating the function pro and epilogues.
4758 @item -fdump-rtl-regmove
4759 @opindex fdump-rtl-regmove
4760 Dump after the register move pass.
4762 @item -fdump-rtl-sched1
4763 @itemx -fdump-rtl-sched2
4764 @opindex fdump-rtl-sched1
4765 @opindex fdump-rtl-sched2
4766 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4767 after the basic block scheduling passes.
4769 @item -fdump-rtl-see
4770 @opindex fdump-rtl-see
4771 Dump after sign extension elimination.
4773 @item -fdump-rtl-seqabstr
4774 @opindex fdump-rtl-seqabstr
4775 Dump after common sequence discovery.
4777 @item -fdump-rtl-shorten
4778 @opindex fdump-rtl-shorten
4779 Dump after shortening branches.
4781 @item -fdump-rtl-sibling
4782 @opindex fdump-rtl-sibling
4783 Dump after sibling call optimizations.
4785 @item -fdump-rtl-split1
4786 @itemx -fdump-rtl-split2
4787 @itemx -fdump-rtl-split3
4788 @itemx -fdump-rtl-split4
4789 @itemx -fdump-rtl-split5
4790 @opindex fdump-rtl-split1
4791 @opindex fdump-rtl-split2
4792 @opindex fdump-rtl-split3
4793 @opindex fdump-rtl-split4
4794 @opindex fdump-rtl-split5
4795 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
4796 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
4797 @option{-fdump-rtl-split5} enable dumping after five rounds of
4798 instruction splitting.
4800 @item -fdump-rtl-sms
4801 @opindex fdump-rtl-sms
4802 Dump after modulo scheduling. This pass is only run on some
4805 @item -fdump-rtl-stack
4806 @opindex fdump-rtl-stack
4807 Dump after conversion from GCC's "flat register file" registers to the
4808 x87's stack-like registers. This pass is only run on x86 variants.
4810 @item -fdump-rtl-subreg1
4811 @itemx -fdump-rtl-subreg2
4812 @opindex fdump-rtl-subreg1
4813 @opindex fdump-rtl-subreg2
4814 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
4815 the two subreg expansion passes.
4817 @item -fdump-rtl-unshare
4818 @opindex fdump-rtl-unshare
4819 Dump after all rtl has been unshared.
4821 @item -fdump-rtl-vartrack
4822 @opindex fdump-rtl-vartrack
4823 Dump after variable tracking.
4825 @item -fdump-rtl-vregs
4826 @opindex fdump-rtl-vregs
4827 Dump after converting virtual registers to hard registers.
4829 @item -fdump-rtl-web
4830 @opindex fdump-rtl-web
4831 Dump after live range splitting.
4833 @item -fdump-rtl-regclass
4834 @itemx -fdump-rtl-subregs_of_mode_init
4835 @itemx -fdump-rtl-subregs_of_mode_finish
4836 @itemx -fdump-rtl-dfinit
4837 @itemx -fdump-rtl-dfinish
4838 @opindex fdump-rtl-regclass
4839 @opindex fdump-rtl-subregs_of_mode_init
4840 @opindex fdump-rtl-subregs_of_mode_finish
4841 @opindex fdump-rtl-dfinit
4842 @opindex fdump-rtl-dfinish
4843 These dumps are defined but always produce empty files.
4845 @item -fdump-rtl-all
4846 @opindex fdump-rtl-all
4847 Produce all the dumps listed above.
4851 Annotate the assembler output with miscellaneous debugging information.
4855 Dump all macro definitions, at the end of preprocessing, in addition to
4860 Produce a core dump whenever an error occurs.
4864 Print statistics on memory usage, at the end of the run, to
4869 Annotate the assembler output with a comment indicating which
4870 pattern and alternative was used. The length of each instruction is
4875 Dump the RTL in the assembler output as a comment before each instruction.
4876 Also turns on @option{-dp} annotation.
4880 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
4881 dump a representation of the control flow graph suitable for viewing with VCG
4882 to @file{@var{file}.@var{pass}.vcg}.
4886 Just generate RTL for a function instead of compiling it. Usually used
4887 with @option{-fdump-rtl-expand}.
4891 Dump debugging information during parsing, to standard error.
4895 @opindex fdump-noaddr
4896 When doing debugging dumps, suppress address output. This makes it more
4897 feasible to use diff on debugging dumps for compiler invocations with
4898 different compiler binaries and/or different
4899 text / bss / data / heap / stack / dso start locations.
4901 @item -fdump-unnumbered
4902 @opindex fdump-unnumbered
4903 When doing debugging dumps, suppress instruction numbers and address output.
4904 This makes it more feasible to use diff on debugging dumps for compiler
4905 invocations with different options, in particular with and without
4908 @item -fdump-translation-unit @r{(C++ only)}
4909 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4910 @opindex fdump-translation-unit
4911 Dump a representation of the tree structure for the entire translation
4912 unit to a file. The file name is made by appending @file{.tu} to the
4913 source file name. If the @samp{-@var{options}} form is used, @var{options}
4914 controls the details of the dump as described for the
4915 @option{-fdump-tree} options.
4917 @item -fdump-class-hierarchy @r{(C++ only)}
4918 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4919 @opindex fdump-class-hierarchy
4920 Dump a representation of each class's hierarchy and virtual function
4921 table layout to a file. The file name is made by appending @file{.class}
4922 to the source file name. If the @samp{-@var{options}} form is used,
4923 @var{options} controls the details of the dump as described for the
4924 @option{-fdump-tree} options.
4926 @item -fdump-ipa-@var{switch}
4928 Control the dumping at various stages of inter-procedural analysis
4929 language tree to a file. The file name is generated by appending a switch
4930 specific suffix to the source file name. The following dumps are possible:
4934 Enables all inter-procedural analysis dumps.
4937 Dumps information about call-graph optimization, unused function removal,
4938 and inlining decisions.
4941 Dump after function inlining.
4945 @item -fdump-statistics-@var{option}
4946 @opindex -fdump-statistics
4947 Enable and control dumping of pass statistics in a separate file. The
4948 file name is generated by appending a suffix ending in @samp{.statistics}
4949 to the source file name. If the @samp{-@var{option}} form is used,
4950 @samp{-stats} will cause counters to be summed over the whole compilation unit
4951 while @samp{-details} will dump every event as the passes generate them.
4952 The default with no option is to sum counters for each function compiled.
4954 @item -fdump-tree-@var{switch}
4955 @itemx -fdump-tree-@var{switch}-@var{options}
4957 Control the dumping at various stages of processing the intermediate
4958 language tree to a file. The file name is generated by appending a switch
4959 specific suffix to the source file name. If the @samp{-@var{options}}
4960 form is used, @var{options} is a list of @samp{-} separated options that
4961 control the details of the dump. Not all options are applicable to all
4962 dumps, those which are not meaningful will be ignored. The following
4963 options are available
4967 Print the address of each node. Usually this is not meaningful as it
4968 changes according to the environment and source file. Its primary use
4969 is for tying up a dump file with a debug environment.
4971 Inhibit dumping of members of a scope or body of a function merely
4972 because that scope has been reached. Only dump such items when they
4973 are directly reachable by some other path. When dumping pretty-printed
4974 trees, this option inhibits dumping the bodies of control structures.
4976 Print a raw representation of the tree. By default, trees are
4977 pretty-printed into a C-like representation.
4979 Enable more detailed dumps (not honored by every dump option).
4981 Enable dumping various statistics about the pass (not honored by every dump
4984 Enable showing basic block boundaries (disabled in raw dumps).
4986 Enable showing virtual operands for every statement.
4988 Enable showing line numbers for statements.
4990 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4992 Enable showing the tree dump for each statement.
4994 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4995 and @option{lineno}.
4998 The following tree dumps are possible:
5002 Dump before any tree based optimization, to @file{@var{file}.original}.
5005 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5008 @opindex fdump-tree-gimple
5009 Dump each function before and after the gimplification pass to a file. The
5010 file name is made by appending @file{.gimple} to the source file name.
5013 @opindex fdump-tree-cfg
5014 Dump the control flow graph of each function to a file. The file name is
5015 made by appending @file{.cfg} to the source file name.
5018 @opindex fdump-tree-vcg
5019 Dump the control flow graph of each function to a file in VCG format. The
5020 file name is made by appending @file{.vcg} to the source file name. Note
5021 that if the file contains more than one function, the generated file cannot
5022 be used directly by VCG@. You will need to cut and paste each function's
5023 graph into its own separate file first.
5026 @opindex fdump-tree-ch
5027 Dump each function after copying loop headers. The file name is made by
5028 appending @file{.ch} to the source file name.
5031 @opindex fdump-tree-ssa
5032 Dump SSA related information to a file. The file name is made by appending
5033 @file{.ssa} to the source file name.
5036 @opindex fdump-tree-alias
5037 Dump aliasing information for each function. The file name is made by
5038 appending @file{.alias} to the source file name.
5041 @opindex fdump-tree-ccp
5042 Dump each function after CCP@. The file name is made by appending
5043 @file{.ccp} to the source file name.
5046 @opindex fdump-tree-storeccp
5047 Dump each function after STORE-CCP@. The file name is made by appending
5048 @file{.storeccp} to the source file name.
5051 @opindex fdump-tree-pre
5052 Dump trees after partial redundancy elimination. The file name is made
5053 by appending @file{.pre} to the source file name.
5056 @opindex fdump-tree-fre
5057 Dump trees after full redundancy elimination. The file name is made
5058 by appending @file{.fre} to the source file name.
5061 @opindex fdump-tree-copyprop
5062 Dump trees after copy propagation. The file name is made
5063 by appending @file{.copyprop} to the source file name.
5065 @item store_copyprop
5066 @opindex fdump-tree-store_copyprop
5067 Dump trees after store copy-propagation. The file name is made
5068 by appending @file{.store_copyprop} to the source file name.
5071 @opindex fdump-tree-dce
5072 Dump each function after dead code elimination. The file name is made by
5073 appending @file{.dce} to the source file name.
5076 @opindex fdump-tree-mudflap
5077 Dump each function after adding mudflap instrumentation. The file name is
5078 made by appending @file{.mudflap} to the source file name.
5081 @opindex fdump-tree-sra
5082 Dump each function after performing scalar replacement of aggregates. The
5083 file name is made by appending @file{.sra} to the source file name.
5086 @opindex fdump-tree-sink
5087 Dump each function after performing code sinking. The file name is made
5088 by appending @file{.sink} to the source file name.
5091 @opindex fdump-tree-dom
5092 Dump each function after applying dominator tree optimizations. The file
5093 name is made by appending @file{.dom} to the source file name.
5096 @opindex fdump-tree-dse
5097 Dump each function after applying dead store elimination. The file
5098 name is made by appending @file{.dse} to the source file name.
5101 @opindex fdump-tree-phiopt
5102 Dump each function after optimizing PHI nodes into straightline code. The file
5103 name is made by appending @file{.phiopt} to the source file name.
5106 @opindex fdump-tree-forwprop
5107 Dump each function after forward propagating single use variables. The file
5108 name is made by appending @file{.forwprop} to the source file name.
5111 @opindex fdump-tree-copyrename
5112 Dump each function after applying the copy rename optimization. The file
5113 name is made by appending @file{.copyrename} to the source file name.
5116 @opindex fdump-tree-nrv
5117 Dump each function after applying the named return value optimization on
5118 generic trees. The file name is made by appending @file{.nrv} to the source
5122 @opindex fdump-tree-vect
5123 Dump each function after applying vectorization of loops. The file name is
5124 made by appending @file{.vect} to the source file name.
5127 @opindex fdump-tree-vrp
5128 Dump each function after Value Range Propagation (VRP). The file name
5129 is made by appending @file{.vrp} to the source file name.
5132 @opindex fdump-tree-all
5133 Enable all the available tree dumps with the flags provided in this option.
5136 @item -ftree-vectorizer-verbose=@var{n}
5137 @opindex ftree-vectorizer-verbose
5138 This option controls the amount of debugging output the vectorizer prints.
5139 This information is written to standard error, unless
5140 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5141 in which case it is output to the usual dump listing file, @file{.vect}.
5142 For @var{n}=0 no diagnostic information is reported.
5143 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5144 and the total number of loops that got vectorized.
5145 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5146 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5147 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5148 level that @option{-fdump-tree-vect-stats} uses.
5149 Higher verbosity levels mean either more information dumped for each
5150 reported loop, or same amount of information reported for more loops:
5151 If @var{n}=3, alignment related information is added to the reports.
5152 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5153 memory access-patterns) is added to the reports.
5154 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5155 that did not pass the first analysis phase (i.e., may not be countable, or
5156 may have complicated control-flow).
5157 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5158 For @var{n}=7, all the information the vectorizer generates during its
5159 analysis and transformation is reported. This is the same verbosity level
5160 that @option{-fdump-tree-vect-details} uses.
5162 @item -frandom-seed=@var{string}
5163 @opindex frandom-string
5164 This option provides a seed that GCC uses when it would otherwise use
5165 random numbers. It is used to generate certain symbol names
5166 that have to be different in every compiled file. It is also used to
5167 place unique stamps in coverage data files and the object files that
5168 produce them. You can use the @option{-frandom-seed} option to produce
5169 reproducibly identical object files.
5171 The @var{string} should be different for every file you compile.
5173 @item -fsched-verbose=@var{n}
5174 @opindex fsched-verbose
5175 On targets that use instruction scheduling, this option controls the
5176 amount of debugging output the scheduler prints. This information is
5177 written to standard error, unless @option{-fdump-rtl-sched1} or
5178 @option{-fdump-rtl-sched2} is specified, in which case it is output
5179 to the usual dump listing file, @file{.sched} or @file{.sched2}
5180 respectively. However for @var{n} greater than nine, the output is
5181 always printed to standard error.
5183 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5184 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5185 For @var{n} greater than one, it also output basic block probabilities,
5186 detailed ready list information and unit/insn info. For @var{n} greater
5187 than two, it includes RTL at abort point, control-flow and regions info.
5188 And for @var{n} over four, @option{-fsched-verbose} also includes
5193 Store the usual ``temporary'' intermediate files permanently; place them
5194 in the current directory and name them based on the source file. Thus,
5195 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5196 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5197 preprocessed @file{foo.i} output file even though the compiler now
5198 normally uses an integrated preprocessor.
5200 When used in combination with the @option{-x} command line option,
5201 @option{-save-temps} is sensible enough to avoid over writing an
5202 input source file with the same extension as an intermediate file.
5203 The corresponding intermediate file may be obtained by renaming the
5204 source file before using @option{-save-temps}.
5208 Report the CPU time taken by each subprocess in the compilation
5209 sequence. For C source files, this is the compiler proper and assembler
5210 (plus the linker if linking is done). The output looks like this:
5217 The first number on each line is the ``user time'', that is time spent
5218 executing the program itself. The second number is ``system time'',
5219 time spent executing operating system routines on behalf of the program.
5220 Both numbers are in seconds.
5222 @item -fvar-tracking
5223 @opindex fvar-tracking
5224 Run variable tracking pass. It computes where variables are stored at each
5225 position in code. Better debugging information is then generated
5226 (if the debugging information format supports this information).
5228 It is enabled by default when compiling with optimization (@option{-Os},
5229 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5230 the debug info format supports it.
5232 @item -print-file-name=@var{library}
5233 @opindex print-file-name
5234 Print the full absolute name of the library file @var{library} that
5235 would be used when linking---and don't do anything else. With this
5236 option, GCC does not compile or link anything; it just prints the
5239 @item -print-multi-directory
5240 @opindex print-multi-directory
5241 Print the directory name corresponding to the multilib selected by any
5242 other switches present in the command line. This directory is supposed
5243 to exist in @env{GCC_EXEC_PREFIX}.
5245 @item -print-multi-lib
5246 @opindex print-multi-lib
5247 Print the mapping from multilib directory names to compiler switches
5248 that enable them. The directory name is separated from the switches by
5249 @samp{;}, and each switch starts with an @samp{@@} instead of the
5250 @samp{-}, without spaces between multiple switches. This is supposed to
5251 ease shell-processing.
5253 @item -print-prog-name=@var{program}
5254 @opindex print-prog-name
5255 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5257 @item -print-libgcc-file-name
5258 @opindex print-libgcc-file-name
5259 Same as @option{-print-file-name=libgcc.a}.
5261 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5262 but you do want to link with @file{libgcc.a}. You can do
5265 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5268 @item -print-search-dirs
5269 @opindex print-search-dirs
5270 Print the name of the configured installation directory and a list of
5271 program and library directories @command{gcc} will search---and don't do anything else.
5273 This is useful when @command{gcc} prints the error message
5274 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5275 To resolve this you either need to put @file{cpp0} and the other compiler
5276 components where @command{gcc} expects to find them, or you can set the environment
5277 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5278 Don't forget the trailing @samp{/}.
5279 @xref{Environment Variables}.
5281 @item -print-sysroot
5282 @opindex print-sysroot
5283 Print the target sysroot directory that will be used during
5284 compilation. This is the target sysroot specified either at configure
5285 time or using the @option{--sysroot} option, possibly with an extra
5286 suffix that depends on compilation options. If no target sysroot is
5287 specified, the option prints nothing.
5289 @item -print-sysroot-headers-suffix
5290 @opindex print-sysroot-headers-suffix
5291 Print the suffix added to the target sysroot when searching for
5292 headers, or give an error if the compiler is not configured with such
5293 a suffix---and don't do anything else.
5296 @opindex dumpmachine
5297 Print the compiler's target machine (for example,
5298 @samp{i686-pc-linux-gnu})---and don't do anything else.
5301 @opindex dumpversion
5302 Print the compiler version (for example, @samp{3.0})---and don't do
5307 Print the compiler's built-in specs---and don't do anything else. (This
5308 is used when GCC itself is being built.) @xref{Spec Files}.
5310 @item -feliminate-unused-debug-types
5311 @opindex feliminate-unused-debug-types
5312 Normally, when producing DWARF2 output, GCC will emit debugging
5313 information for all types declared in a compilation
5314 unit, regardless of whether or not they are actually used
5315 in that compilation unit. Sometimes this is useful, such as
5316 if, in the debugger, you want to cast a value to a type that is
5317 not actually used in your program (but is declared). More often,
5318 however, this results in a significant amount of wasted space.
5319 With this option, GCC will avoid producing debug symbol output
5320 for types that are nowhere used in the source file being compiled.
5323 @node Optimize Options
5324 @section Options That Control Optimization
5325 @cindex optimize options
5326 @cindex options, optimization
5328 These options control various sorts of optimizations.
5330 Without any optimization option, the compiler's goal is to reduce the
5331 cost of compilation and to make debugging produce the expected
5332 results. Statements are independent: if you stop the program with a
5333 breakpoint between statements, you can then assign a new value to any
5334 variable or change the program counter to any other statement in the
5335 function and get exactly the results you would expect from the source
5338 Turning on optimization flags makes the compiler attempt to improve
5339 the performance and/or code size at the expense of compilation time
5340 and possibly the ability to debug the program.
5342 The compiler performs optimization based on the knowledge it has of the
5343 program. Compiling multiple files at once to a single output file mode allows
5344 the compiler to use information gained from all of the files when compiling
5347 Not all optimizations are controlled directly by a flag. Only
5348 optimizations that have a flag are listed.
5355 Optimize. Optimizing compilation takes somewhat more time, and a lot
5356 more memory for a large function.
5358 With @option{-O}, the compiler tries to reduce code size and execution
5359 time, without performing any optimizations that take a great deal of
5362 @option{-O} turns on the following optimization flags:
5365 -fcprop-registers @gol
5368 -fdelayed-branch @gol
5370 -fguess-branch-probability @gol
5371 -fif-conversion2 @gol
5372 -fif-conversion @gol
5373 -finline-small-functions @gol
5374 -fipa-pure-const @gol
5375 -fipa-reference @gol
5377 -fsplit-wide-types @gol
5378 -ftree-builtin-call-dce @gol
5381 -ftree-copyrename @gol
5383 -ftree-dominator-opts @gol
5390 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5391 where doing so does not interfere with debugging.
5395 Optimize even more. GCC performs nearly all supported optimizations
5396 that do not involve a space-speed tradeoff.
5397 As compared to @option{-O}, this option increases both compilation time
5398 and the performance of the generated code.
5400 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5401 also turns on the following optimization flags:
5402 @gccoptlist{-fthread-jumps @gol
5403 -falign-functions -falign-jumps @gol
5404 -falign-loops -falign-labels @gol
5407 -fcse-follow-jumps -fcse-skip-blocks @gol
5408 -fdelete-null-pointer-checks @gol
5409 -fexpensive-optimizations @gol
5410 -fgcse -fgcse-lm @gol
5411 -findirect-inlining @gol
5412 -foptimize-sibling-calls @gol
5415 -freorder-blocks -freorder-functions @gol
5416 -frerun-cse-after-loop @gol
5417 -fsched-interblock -fsched-spec @gol
5418 -fschedule-insns -fschedule-insns2 @gol
5419 -fstrict-aliasing -fstrict-overflow @gol
5420 -ftree-switch-conversion @gol
5424 Please note the warning under @option{-fgcse} about
5425 invoking @option{-O2} on programs that use computed gotos.
5429 Optimize yet more. @option{-O3} turns on all optimizations specified
5430 by @option{-O2} and also turns on the @option{-finline-functions},
5431 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5432 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5436 Reduce compilation time and make debugging produce the expected
5437 results. This is the default.
5441 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5442 do not typically increase code size. It also performs further
5443 optimizations designed to reduce code size.
5445 @option{-Os} disables the following optimization flags:
5446 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5447 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5448 -fprefetch-loop-arrays -ftree-vect-loop-version}
5450 If you use multiple @option{-O} options, with or without level numbers,
5451 the last such option is the one that is effective.
5454 Options of the form @option{-f@var{flag}} specify machine-independent
5455 flags. Most flags have both positive and negative forms; the negative
5456 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5457 below, only one of the forms is listed---the one you typically will
5458 use. You can figure out the other form by either removing @samp{no-}
5461 The following options control specific optimizations. They are either
5462 activated by @option{-O} options or are related to ones that are. You
5463 can use the following flags in the rare cases when ``fine-tuning'' of
5464 optimizations to be performed is desired.
5467 @item -fno-default-inline
5468 @opindex fno-default-inline
5469 Do not make member functions inline by default merely because they are
5470 defined inside the class scope (C++ only). Otherwise, when you specify
5471 @w{@option{-O}}, member functions defined inside class scope are compiled
5472 inline by default; i.e., you don't need to add @samp{inline} in front of
5473 the member function name.
5475 @item -fno-defer-pop
5476 @opindex fno-defer-pop
5477 Always pop the arguments to each function call as soon as that function
5478 returns. For machines which must pop arguments after a function call,
5479 the compiler normally lets arguments accumulate on the stack for several
5480 function calls and pops them all at once.
5482 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5484 @item -fforward-propagate
5485 @opindex fforward-propagate
5486 Perform a forward propagation pass on RTL@. The pass tries to combine two
5487 instructions and checks if the result can be simplified. If loop unrolling
5488 is active, two passes are performed and the second is scheduled after
5491 This option is enabled by default at optimization levels @option{-O2},
5492 @option{-O3}, @option{-Os}.
5494 @item -fomit-frame-pointer
5495 @opindex fomit-frame-pointer
5496 Don't keep the frame pointer in a register for functions that
5497 don't need one. This avoids the instructions to save, set up and
5498 restore frame pointers; it also makes an extra register available
5499 in many functions. @strong{It also makes debugging impossible on
5502 On some machines, such as the VAX, this flag has no effect, because
5503 the standard calling sequence automatically handles the frame pointer
5504 and nothing is saved by pretending it doesn't exist. The
5505 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5506 whether a target machine supports this flag. @xref{Registers,,Register
5507 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5509 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5511 @item -foptimize-sibling-calls
5512 @opindex foptimize-sibling-calls
5513 Optimize sibling and tail recursive calls.
5515 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5519 Don't pay attention to the @code{inline} keyword. Normally this option
5520 is used to keep the compiler from expanding any functions inline.
5521 Note that if you are not optimizing, no functions can be expanded inline.
5523 @item -finline-small-functions
5524 @opindex finline-small-functions
5525 Integrate functions into their callers when their body is smaller than expected
5526 function call code (so overall size of program gets smaller). The compiler
5527 heuristically decides which functions are simple enough to be worth integrating
5530 Enabled at level @option{-O2}.
5532 @item -findirect-inlining
5533 @opindex findirect-inlining
5534 Inline also indirect calls that are discovered to be known at compile
5535 time thanks to previous inlining. This option has any effect only
5536 when inlining itself is turned on by the @option{-finline-functions}
5537 or @option{-finline-small-functions} options.
5539 Enabled at level @option{-O2}.
5541 @item -finline-functions
5542 @opindex finline-functions
5543 Integrate all simple functions into their callers. The compiler
5544 heuristically decides which functions are simple enough to be worth
5545 integrating in this way.
5547 If all calls to a given function are integrated, and the function is
5548 declared @code{static}, then the function is normally not output as
5549 assembler code in its own right.
5551 Enabled at level @option{-O3}.
5553 @item -finline-functions-called-once
5554 @opindex finline-functions-called-once
5555 Consider all @code{static} functions called once for inlining into their
5556 caller even if they are not marked @code{inline}. If a call to a given
5557 function is integrated, then the function is not output as assembler code
5560 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5562 @item -fearly-inlining
5563 @opindex fearly-inlining
5564 Inline functions marked by @code{always_inline} and functions whose body seems
5565 smaller than the function call overhead early before doing
5566 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5567 makes profiling significantly cheaper and usually inlining faster on programs
5568 having large chains of nested wrapper functions.
5572 @item -finline-limit=@var{n}
5573 @opindex finline-limit
5574 By default, GCC limits the size of functions that can be inlined. This flag
5575 allows coarse control of this limit. @var{n} is the size of functions that
5576 can be inlined in number of pseudo instructions.
5578 Inlining is actually controlled by a number of parameters, which may be
5579 specified individually by using @option{--param @var{name}=@var{value}}.
5580 The @option{-finline-limit=@var{n}} option sets some of these parameters
5584 @item max-inline-insns-single
5585 is set to @var{n}/2.
5586 @item max-inline-insns-auto
5587 is set to @var{n}/2.
5590 See below for a documentation of the individual
5591 parameters controlling inlining and for the defaults of these parameters.
5593 @emph{Note:} there may be no value to @option{-finline-limit} that results
5594 in default behavior.
5596 @emph{Note:} pseudo instruction represents, in this particular context, an
5597 abstract measurement of function's size. In no way does it represent a count
5598 of assembly instructions and as such its exact meaning might change from one
5599 release to an another.
5601 @item -fkeep-inline-functions
5602 @opindex fkeep-inline-functions
5603 In C, emit @code{static} functions that are declared @code{inline}
5604 into the object file, even if the function has been inlined into all
5605 of its callers. This switch does not affect functions using the
5606 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5607 inline functions into the object file.
5609 @item -fkeep-static-consts
5610 @opindex fkeep-static-consts
5611 Emit variables declared @code{static const} when optimization isn't turned
5612 on, even if the variables aren't referenced.
5614 GCC enables this option by default. If you want to force the compiler to
5615 check if the variable was referenced, regardless of whether or not
5616 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5618 @item -fmerge-constants
5619 @opindex fmerge-constants
5620 Attempt to merge identical constants (string constants and floating point
5621 constants) across compilation units.
5623 This option is the default for optimized compilation if the assembler and
5624 linker support it. Use @option{-fno-merge-constants} to inhibit this
5627 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5629 @item -fmerge-all-constants
5630 @opindex fmerge-all-constants
5631 Attempt to merge identical constants and identical variables.
5633 This option implies @option{-fmerge-constants}. In addition to
5634 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5635 arrays or initialized constant variables with integral or floating point
5636 types. Languages like C or C++ require each variable, including multiple
5637 instances of the same variable in recursive calls, to have distinct locations,
5638 so using this option will result in non-conforming
5641 @item -fmodulo-sched
5642 @opindex fmodulo-sched
5643 Perform swing modulo scheduling immediately before the first scheduling
5644 pass. This pass looks at innermost loops and reorders their
5645 instructions by overlapping different iterations.
5647 @item -fmodulo-sched-allow-regmoves
5648 @opindex fmodulo-sched-allow-regmoves
5649 Perform more aggressive SMS based modulo scheduling with register moves
5650 allowed. By setting this flag certain anti-dependences edges will be
5651 deleted which will trigger the generation of reg-moves based on the
5652 life-range analysis. This option is effective only with
5653 @option{-fmodulo-sched} enabled.
5655 @item -fno-branch-count-reg
5656 @opindex fno-branch-count-reg
5657 Do not use ``decrement and branch'' instructions on a count register,
5658 but instead generate a sequence of instructions that decrement a
5659 register, compare it against zero, then branch based upon the result.
5660 This option is only meaningful on architectures that support such
5661 instructions, which include x86, PowerPC, IA-64 and S/390.
5663 The default is @option{-fbranch-count-reg}.
5665 @item -fno-function-cse
5666 @opindex fno-function-cse
5667 Do not put function addresses in registers; make each instruction that
5668 calls a constant function contain the function's address explicitly.
5670 This option results in less efficient code, but some strange hacks
5671 that alter the assembler output may be confused by the optimizations
5672 performed when this option is not used.
5674 The default is @option{-ffunction-cse}
5676 @item -fno-zero-initialized-in-bss
5677 @opindex fno-zero-initialized-in-bss
5678 If the target supports a BSS section, GCC by default puts variables that
5679 are initialized to zero into BSS@. This can save space in the resulting
5682 This option turns off this behavior because some programs explicitly
5683 rely on variables going to the data section. E.g., so that the
5684 resulting executable can find the beginning of that section and/or make
5685 assumptions based on that.
5687 The default is @option{-fzero-initialized-in-bss}.
5689 @item -fmudflap -fmudflapth -fmudflapir
5693 @cindex bounds checking
5695 For front-ends that support it (C and C++), instrument all risky
5696 pointer/array dereferencing operations, some standard library
5697 string/heap functions, and some other associated constructs with
5698 range/validity tests. Modules so instrumented should be immune to
5699 buffer overflows, invalid heap use, and some other classes of C/C++
5700 programming errors. The instrumentation relies on a separate runtime
5701 library (@file{libmudflap}), which will be linked into a program if
5702 @option{-fmudflap} is given at link time. Run-time behavior of the
5703 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5704 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5707 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5708 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5709 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5710 instrumentation should ignore pointer reads. This produces less
5711 instrumentation (and therefore faster execution) and still provides
5712 some protection against outright memory corrupting writes, but allows
5713 erroneously read data to propagate within a program.
5715 @item -fthread-jumps
5716 @opindex fthread-jumps
5717 Perform optimizations where we check to see if a jump branches to a
5718 location where another comparison subsumed by the first is found. If
5719 so, the first branch is redirected to either the destination of the
5720 second branch or a point immediately following it, depending on whether
5721 the condition is known to be true or false.
5723 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5725 @item -fsplit-wide-types
5726 @opindex fsplit-wide-types
5727 When using a type that occupies multiple registers, such as @code{long
5728 long} on a 32-bit system, split the registers apart and allocate them
5729 independently. This normally generates better code for those types,
5730 but may make debugging more difficult.
5732 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5735 @item -fcse-follow-jumps
5736 @opindex fcse-follow-jumps
5737 In common subexpression elimination (CSE), scan through jump instructions
5738 when the target of the jump is not reached by any other path. For
5739 example, when CSE encounters an @code{if} statement with an
5740 @code{else} clause, CSE will follow the jump when the condition
5743 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5745 @item -fcse-skip-blocks
5746 @opindex fcse-skip-blocks
5747 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5748 follow jumps which conditionally skip over blocks. When CSE
5749 encounters a simple @code{if} statement with no else clause,
5750 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5751 body of the @code{if}.
5753 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5755 @item -frerun-cse-after-loop
5756 @opindex frerun-cse-after-loop
5757 Re-run common subexpression elimination after loop optimizations has been
5760 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5764 Perform a global common subexpression elimination pass.
5765 This pass also performs global constant and copy propagation.
5767 @emph{Note:} When compiling a program using computed gotos, a GCC
5768 extension, you may get better runtime performance if you disable
5769 the global common subexpression elimination pass by adding
5770 @option{-fno-gcse} to the command line.
5772 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5776 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5777 attempt to move loads which are only killed by stores into themselves. This
5778 allows a loop containing a load/store sequence to be changed to a load outside
5779 the loop, and a copy/store within the loop.
5781 Enabled by default when gcse is enabled.
5785 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5786 global common subexpression elimination. This pass will attempt to move
5787 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5788 loops containing a load/store sequence can be changed to a load before
5789 the loop and a store after the loop.
5791 Not enabled at any optimization level.
5795 When @option{-fgcse-las} is enabled, the global common subexpression
5796 elimination pass eliminates redundant loads that come after stores to the
5797 same memory location (both partial and full redundancies).
5799 Not enabled at any optimization level.
5801 @item -fgcse-after-reload
5802 @opindex fgcse-after-reload
5803 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5804 pass is performed after reload. The purpose of this pass is to cleanup
5807 @item -funsafe-loop-optimizations
5808 @opindex funsafe-loop-optimizations
5809 If given, the loop optimizer will assume that loop indices do not
5810 overflow, and that the loops with nontrivial exit condition are not
5811 infinite. This enables a wider range of loop optimizations even if
5812 the loop optimizer itself cannot prove that these assumptions are valid.
5813 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5814 if it finds this kind of loop.
5816 @item -fcrossjumping
5817 @opindex fcrossjumping
5818 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5819 resulting code may or may not perform better than without cross-jumping.
5821 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5823 @item -fauto-inc-dec
5824 @opindex fauto-inc-dec
5825 Combine increments or decrements of addresses with memory accesses.
5826 This pass is always skipped on architectures that do not have
5827 instructions to support this. Enabled by default at @option{-O} and
5828 higher on architectures that support this.
5832 Perform dead code elimination (DCE) on RTL@.
5833 Enabled by default at @option{-O} and higher.
5837 Perform dead store elimination (DSE) on RTL@.
5838 Enabled by default at @option{-O} and higher.
5840 @item -fif-conversion
5841 @opindex fif-conversion
5842 Attempt to transform conditional jumps into branch-less equivalents. This
5843 include use of conditional moves, min, max, set flags and abs instructions, and
5844 some tricks doable by standard arithmetics. The use of conditional execution
5845 on chips where it is available is controlled by @code{if-conversion2}.
5847 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5849 @item -fif-conversion2
5850 @opindex fif-conversion2
5851 Use conditional execution (where available) to transform conditional jumps into
5852 branch-less equivalents.
5854 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5856 @item -fdelete-null-pointer-checks
5857 @opindex fdelete-null-pointer-checks
5858 Use global dataflow analysis to identify and eliminate useless checks
5859 for null pointers. The compiler assumes that dereferencing a null
5860 pointer would have halted the program. If a pointer is checked after
5861 it has already been dereferenced, it cannot be null.
5863 In some environments, this assumption is not true, and programs can
5864 safely dereference null pointers. Use
5865 @option{-fno-delete-null-pointer-checks} to disable this optimization
5866 for programs which depend on that behavior.
5868 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5870 @item -fexpensive-optimizations
5871 @opindex fexpensive-optimizations
5872 Perform a number of minor optimizations that are relatively expensive.
5874 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5876 @item -foptimize-register-move
5878 @opindex foptimize-register-move
5880 Attempt to reassign register numbers in move instructions and as
5881 operands of other simple instructions in order to maximize the amount of
5882 register tying. This is especially helpful on machines with two-operand
5885 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5888 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5890 @item -fira-algorithm=@var{algorithm}
5891 Use specified coloring algorithm for the integrated register
5892 allocator. The @var{algorithm} argument should be @code{priority} or
5893 @code{CB}. The first algorithm specifies Chow's priority coloring,
5894 the second one specifies Chaitin-Briggs coloring. The second
5895 algorithm can be unimplemented for some architectures. If it is
5896 implemented, it is the default because Chaitin-Briggs coloring as a
5897 rule generates a better code.
5899 @item -fira-region=@var{region}
5900 Use specified regions for the integrated register allocator. The
5901 @var{region} argument should be one of @code{all}, @code{mixed}, or
5902 @code{one}. The first value means using all loops as register
5903 allocation regions, the second value which is the default means using
5904 all loops except for loops with small register pressure as the
5905 regions, and third one means using all function as a single region.
5906 The first value can give best result for machines with small size and
5907 irregular register set, the third one results in faster and generates
5908 decent code and the smallest size code, and the default value usually
5909 give the best results in most cases and for most architectures.
5911 @item -fira-coalesce
5912 @opindex fira-coalesce
5913 Do optimistic register coalescing. This option might be profitable for
5914 architectures with big regular register files.
5916 @item -fno-ira-share-save-slots
5917 @opindex fno-ira-share-save-slots
5918 Switch off sharing stack slots used for saving call used hard
5919 registers living through a call. Each hard register will get a
5920 separate stack slot and as a result function stack frame will be
5923 @item -fno-ira-share-spill-slots
5924 @opindex fno-ira-share-spill-slots
5925 Switch off sharing stack slots allocated for pseudo-registers. Each
5926 pseudo-register which did not get a hard register will get a separate
5927 stack slot and as a result function stack frame will be bigger.
5929 @item -fira-verbose=@var{n}
5930 @opindex fira-verbose
5931 Set up how verbose dump file for the integrated register allocator
5932 will be. Default value is 5. If the value is greater or equal to 10,
5933 the dump file will be stderr as if the value were @var{n} minus 10.
5935 @item -fdelayed-branch
5936 @opindex fdelayed-branch
5937 If supported for the target machine, attempt to reorder instructions
5938 to exploit instruction slots available after delayed branch
5941 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5943 @item -fschedule-insns
5944 @opindex fschedule-insns
5945 If supported for the target machine, attempt to reorder instructions to
5946 eliminate execution stalls due to required data being unavailable. This
5947 helps machines that have slow floating point or memory load instructions
5948 by allowing other instructions to be issued until the result of the load
5949 or floating point instruction is required.
5951 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5953 @item -fschedule-insns2
5954 @opindex fschedule-insns2
5955 Similar to @option{-fschedule-insns}, but requests an additional pass of
5956 instruction scheduling after register allocation has been done. This is
5957 especially useful on machines with a relatively small number of
5958 registers and where memory load instructions take more than one cycle.
5960 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5962 @item -fno-sched-interblock
5963 @opindex fno-sched-interblock
5964 Don't schedule instructions across basic blocks. This is normally
5965 enabled by default when scheduling before register allocation, i.e.@:
5966 with @option{-fschedule-insns} or at @option{-O2} or higher.
5968 @item -fno-sched-spec
5969 @opindex fno-sched-spec
5970 Don't allow speculative motion of non-load instructions. This is normally
5971 enabled by default when scheduling before register allocation, i.e.@:
5972 with @option{-fschedule-insns} or at @option{-O2} or higher.
5974 @item -fsched-spec-load
5975 @opindex fsched-spec-load
5976 Allow speculative motion of some load instructions. This only makes
5977 sense when scheduling before register allocation, i.e.@: with
5978 @option{-fschedule-insns} or at @option{-O2} or higher.
5980 @item -fsched-spec-load-dangerous
5981 @opindex fsched-spec-load-dangerous
5982 Allow speculative motion of more load instructions. This only makes
5983 sense when scheduling before register allocation, i.e.@: with
5984 @option{-fschedule-insns} or at @option{-O2} or higher.
5986 @item -fsched-stalled-insns
5987 @itemx -fsched-stalled-insns=@var{n}
5988 @opindex fsched-stalled-insns
5989 Define how many insns (if any) can be moved prematurely from the queue
5990 of stalled insns into the ready list, during the second scheduling pass.
5991 @option{-fno-sched-stalled-insns} means that no insns will be moved
5992 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5993 on how many queued insns can be moved prematurely.
5994 @option{-fsched-stalled-insns} without a value is equivalent to
5995 @option{-fsched-stalled-insns=1}.
5997 @item -fsched-stalled-insns-dep
5998 @itemx -fsched-stalled-insns-dep=@var{n}
5999 @opindex fsched-stalled-insns-dep
6000 Define how many insn groups (cycles) will be examined for a dependency
6001 on a stalled insn that is candidate for premature removal from the queue
6002 of stalled insns. This has an effect only during the second scheduling pass,
6003 and only if @option{-fsched-stalled-insns} is used.
6004 @option{-fno-sched-stalled-insns-dep} is equivalent to
6005 @option{-fsched-stalled-insns-dep=0}.
6006 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6007 @option{-fsched-stalled-insns-dep=1}.
6009 @item -fsched2-use-superblocks
6010 @opindex fsched2-use-superblocks
6011 When scheduling after register allocation, do use superblock scheduling
6012 algorithm. Superblock scheduling allows motion across basic block boundaries
6013 resulting on faster schedules. This option is experimental, as not all machine
6014 descriptions used by GCC model the CPU closely enough to avoid unreliable
6015 results from the algorithm.
6017 This only makes sense when scheduling after register allocation, i.e.@: with
6018 @option{-fschedule-insns2} or at @option{-O2} or higher.
6020 @item -fsched2-use-traces
6021 @opindex fsched2-use-traces
6022 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
6023 allocation and additionally perform code duplication in order to increase the
6024 size of superblocks using tracer pass. See @option{-ftracer} for details on
6027 This mode should produce faster but significantly longer programs. Also
6028 without @option{-fbranch-probabilities} the traces constructed may not
6029 match the reality and hurt the performance. This only makes
6030 sense when scheduling after register allocation, i.e.@: with
6031 @option{-fschedule-insns2} or at @option{-O2} or higher.
6035 Eliminate redundant sign extension instructions and move the non-redundant
6036 ones to optimal placement using lazy code motion (LCM).
6038 @item -freschedule-modulo-scheduled-loops
6039 @opindex freschedule-modulo-scheduled-loops
6040 The modulo scheduling comes before the traditional scheduling, if a loop
6041 was modulo scheduled we may want to prevent the later scheduling passes
6042 from changing its schedule, we use this option to control that.
6044 @item -fselective-scheduling
6045 @opindex fselective-scheduling
6046 Schedule instructions using selective scheduling algorithm. Selective
6047 scheduling runs instead of the first scheduler pass.
6049 @item -fselective-scheduling2
6050 @opindex fselective-scheduling2
6051 Schedule instructions using selective scheduling algorithm. Selective
6052 scheduling runs instead of the second scheduler pass.
6054 @item -fsel-sched-pipelining
6055 @opindex fsel-sched-pipelining
6056 Enable software pipelining of innermost loops during selective scheduling.
6057 This option has no effect until one of @option{-fselective-scheduling} or
6058 @option{-fselective-scheduling2} is turned on.
6060 @item -fsel-sched-pipelining-outer-loops
6061 @opindex fsel-sched-pipelining-outer-loops
6062 When pipelining loops during selective scheduling, also pipeline outer loops.
6063 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6065 @item -fcaller-saves
6066 @opindex fcaller-saves
6067 Enable values to be allocated in registers that will be clobbered by
6068 function calls, by emitting extra instructions to save and restore the
6069 registers around such calls. Such allocation is done only when it
6070 seems to result in better code than would otherwise be produced.
6072 This option is always enabled by default on certain machines, usually
6073 those which have no call-preserved registers to use instead.
6075 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6077 @item -fconserve-stack
6078 @opindex fconserve-stack
6079 Attempt to minimize stack usage. The compiler will attempt to use less
6080 stack space, even if that makes the program slower. This option
6081 implies setting the @option{large-stack-frame} parameter to 100
6082 and the @option{large-stack-frame-growth} parameter to 400.
6084 @item -ftree-reassoc
6085 @opindex ftree-reassoc
6086 Perform reassociation on trees. This flag is enabled by default
6087 at @option{-O} and higher.
6091 Perform partial redundancy elimination (PRE) on trees. This flag is
6092 enabled by default at @option{-O2} and @option{-O3}.
6096 Perform full redundancy elimination (FRE) on trees. The difference
6097 between FRE and PRE is that FRE only considers expressions
6098 that are computed on all paths leading to the redundant computation.
6099 This analysis is faster than PRE, though it exposes fewer redundancies.
6100 This flag is enabled by default at @option{-O} and higher.
6102 @item -ftree-copy-prop
6103 @opindex ftree-copy-prop
6104 Perform copy propagation on trees. This pass eliminates unnecessary
6105 copy operations. This flag is enabled by default at @option{-O} and
6108 @item -fipa-pure-const
6109 @opindex fipa-pure-const
6110 Discover which functions are pure or constant.
6111 Enabled by default at @option{-O} and higher.
6113 @item -fipa-reference
6114 @opindex fipa-reference
6115 Discover which static variables do not escape cannot escape the
6117 Enabled by default at @option{-O} and higher.
6119 @item -fipa-struct-reorg
6120 @opindex fipa-struct-reorg
6121 Perform structure reorganization optimization, that change C-like structures
6122 layout in order to better utilize spatial locality. This transformation is
6123 affective for programs containing arrays of structures. Available in two
6124 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6125 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6126 to provide the safety of this transformation. It works only in whole program
6127 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6128 enabled. Structures considered @samp{cold} by this transformation are not
6129 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6131 With this flag, the program debug info reflects a new structure layout.
6135 Perform interprocedural pointer analysis. This option is experimental
6136 and does not affect generated code.
6140 Perform interprocedural constant propagation.
6141 This optimization analyzes the program to determine when values passed
6142 to functions are constants and then optimizes accordingly.
6143 This optimization can substantially increase performance
6144 if the application has constants passed to functions.
6145 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6147 @item -fipa-cp-clone
6148 @opindex fipa-cp-clone
6149 Perform function cloning to make interprocedural constant propagation stronger.
6150 When enabled, interprocedural constant propagation will perform function cloning
6151 when externally visible function can be called with constant arguments.
6152 Because this optimization can create multiple copies of functions,
6153 it may significantly increase code size
6154 (see @option{--param ipcp-unit-growth=@var{value}}).
6155 This flag is enabled by default at @option{-O3}.
6157 @item -fipa-matrix-reorg
6158 @opindex fipa-matrix-reorg
6159 Perform matrix flattening and transposing.
6160 Matrix flattening tries to replace a m-dimensional matrix
6161 with its equivalent n-dimensional matrix, where n < m.
6162 This reduces the level of indirection needed for accessing the elements
6163 of the matrix. The second optimization is matrix transposing that
6164 attempts to change the order of the matrix's dimensions in order to
6165 improve cache locality.
6166 Both optimizations need the @option{-fwhole-program} flag.
6167 Transposing is enabled only if profiling information is available.
6172 Perform forward store motion on trees. This flag is
6173 enabled by default at @option{-O} and higher.
6177 Perform sparse conditional constant propagation (CCP) on trees. This
6178 pass only operates on local scalar variables and is enabled by default
6179 at @option{-O} and higher.
6181 @item -ftree-switch-conversion
6182 Perform conversion of simple initializations in a switch to
6183 initializations from a scalar array. This flag is enabled by default
6184 at @option{-O2} and higher.
6188 Perform dead code elimination (DCE) on trees. This flag is enabled by
6189 default at @option{-O} and higher.
6191 @item -ftree-builtin-call-dce
6192 @opindex ftree-builtin-call-dce
6193 Perform conditional dead code elimination (DCE) for calls to builtin functions
6194 that may set @code{errno} but are otherwise side-effect free. This flag is
6195 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6198 @item -ftree-dominator-opts
6199 @opindex ftree-dominator-opts
6200 Perform a variety of simple scalar cleanups (constant/copy
6201 propagation, redundancy elimination, range propagation and expression
6202 simplification) based on a dominator tree traversal. This also
6203 performs jump threading (to reduce jumps to jumps). This flag is
6204 enabled by default at @option{-O} and higher.
6208 Perform dead store elimination (DSE) on trees. A dead store is a store into
6209 a memory location which will later be overwritten by another store without
6210 any intervening loads. In this case the earlier store can be deleted. This
6211 flag is enabled by default at @option{-O} and higher.
6215 Perform loop header copying on trees. This is beneficial since it increases
6216 effectiveness of code motion optimizations. It also saves one jump. This flag
6217 is enabled by default at @option{-O} and higher. It is not enabled
6218 for @option{-Os}, since it usually increases code size.
6220 @item -ftree-loop-optimize
6221 @opindex ftree-loop-optimize
6222 Perform loop optimizations on trees. This flag is enabled by default
6223 at @option{-O} and higher.
6225 @item -ftree-loop-linear
6226 @opindex ftree-loop-linear
6227 Perform linear loop transformations on tree. This flag can improve cache
6228 performance and allow further loop optimizations to take place.
6230 @item -floop-interchange
6231 Perform loop interchange transformations on loops. Interchanging two
6232 nested loops switches the inner and outer loops. For example, given a
6237 A(J, I) = A(J, I) * C
6241 loop interchange will transform the loop as if the user had written:
6245 A(J, I) = A(J, I) * C
6249 which can be beneficial when @code{N} is larger than the caches,
6250 because in Fortran, the elements of an array are stored in memory
6251 contiguously by column, and the original loop iterates over rows,
6252 potentially creating at each access a cache miss. This optimization
6253 applies to all the languages supported by GCC and is not limited to
6254 Fortran. To use this code transformation, GCC has to be configured
6255 with @option{--with-ppl} and @option{--with-cloog} to enable the
6256 Graphite loop transformation infrastructure.
6258 @item -floop-strip-mine
6259 Perform loop strip mining transformations on loops. Strip mining
6260 splits a loop into two nested loops. The outer loop has strides
6261 equal to the strip size and the inner loop has strides of the
6262 original loop within a strip. For example, given a loop like:
6268 loop strip mining will transform the loop as if the user had written:
6271 DO I = II, min (II + 3, N)
6276 This optimization applies to all the languages supported by GCC and is
6277 not limited to Fortran. To use this code transformation, GCC has to
6278 be configured with @option{--with-ppl} and @option{--with-cloog} to
6279 enable the Graphite loop transformation infrastructure.
6282 Perform loop blocking transformations on loops. Blocking strip mines
6283 each loop in the loop nest such that the memory accesses of the
6284 element loops fit inside caches. For example, given a loop like:
6288 A(J, I) = B(I) + C(J)
6292 loop blocking will transform the loop as if the user had written:
6296 DO I = II, min (II + 63, N)
6297 DO J = JJ, min (JJ + 63, M)
6298 A(J, I) = B(I) + C(J)
6304 which can be beneficial when @code{M} is larger than the caches,
6305 because the innermost loop will iterate over a smaller amount of data
6306 that can be kept in the caches. This optimization applies to all the
6307 languages supported by GCC and is not limited to Fortran. To use this
6308 code transformation, GCC has to be configured with @option{--with-ppl}
6309 and @option{--with-cloog} to enable the Graphite loop transformation
6312 @item -fcheck-data-deps
6313 @opindex fcheck-data-deps
6314 Compare the results of several data dependence analyzers. This option
6315 is used for debugging the data dependence analyzers.
6317 @item -ftree-loop-distribution
6318 Perform loop distribution. This flag can improve cache performance on
6319 big loop bodies and allow further loop optimizations, like
6320 parallelization or vectorization, to take place. For example, the loop
6337 @item -ftree-loop-im
6338 @opindex ftree-loop-im
6339 Perform loop invariant motion on trees. This pass moves only invariants that
6340 would be hard to handle at RTL level (function calls, operations that expand to
6341 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6342 operands of conditions that are invariant out of the loop, so that we can use
6343 just trivial invariantness analysis in loop unswitching. The pass also includes
6346 @item -ftree-loop-ivcanon
6347 @opindex ftree-loop-ivcanon
6348 Create a canonical counter for number of iterations in the loop for that
6349 determining number of iterations requires complicated analysis. Later
6350 optimizations then may determine the number easily. Useful especially
6351 in connection with unrolling.
6355 Perform induction variable optimizations (strength reduction, induction
6356 variable merging and induction variable elimination) on trees.
6358 @item -ftree-parallelize-loops=n
6359 @opindex ftree-parallelize-loops
6360 Parallelize loops, i.e., split their iteration space to run in n threads.
6361 This is only possible for loops whose iterations are independent
6362 and can be arbitrarily reordered. The optimization is only
6363 profitable on multiprocessor machines, for loops that are CPU-intensive,
6364 rather than constrained e.g.@: by memory bandwidth. This option
6365 implies @option{-pthread}, and thus is only supported on targets
6366 that have support for @option{-pthread}.
6370 Perform scalar replacement of aggregates. This pass replaces structure
6371 references with scalars to prevent committing structures to memory too
6372 early. This flag is enabled by default at @option{-O} and higher.
6374 @item -ftree-copyrename
6375 @opindex ftree-copyrename
6376 Perform copy renaming on trees. This pass attempts to rename compiler
6377 temporaries to other variables at copy locations, usually resulting in
6378 variable names which more closely resemble the original variables. This flag
6379 is enabled by default at @option{-O} and higher.
6383 Perform temporary expression replacement during the SSA->normal phase. Single
6384 use/single def temporaries are replaced at their use location with their
6385 defining expression. This results in non-GIMPLE code, but gives the expanders
6386 much more complex trees to work on resulting in better RTL generation. This is
6387 enabled by default at @option{-O} and higher.
6389 @item -ftree-vectorize
6390 @opindex ftree-vectorize
6391 Perform loop vectorization on trees. This flag is enabled by default at
6394 @item -ftree-vect-loop-version
6395 @opindex ftree-vect-loop-version
6396 Perform loop versioning when doing loop vectorization on trees. When a loop
6397 appears to be vectorizable except that data alignment or data dependence cannot
6398 be determined at compile time then vectorized and non-vectorized versions of
6399 the loop are generated along with runtime checks for alignment or dependence
6400 to control which version is executed. This option is enabled by default
6401 except at level @option{-Os} where it is disabled.
6403 @item -fvect-cost-model
6404 @opindex fvect-cost-model
6405 Enable cost model for vectorization.
6409 Perform Value Range Propagation on trees. This is similar to the
6410 constant propagation pass, but instead of values, ranges of values are
6411 propagated. This allows the optimizers to remove unnecessary range
6412 checks like array bound checks and null pointer checks. This is
6413 enabled by default at @option{-O2} and higher. Null pointer check
6414 elimination is only done if @option{-fdelete-null-pointer-checks} is
6419 Perform tail duplication to enlarge superblock size. This transformation
6420 simplifies the control flow of the function allowing other optimizations to do
6423 @item -funroll-loops
6424 @opindex funroll-loops
6425 Unroll loops whose number of iterations can be determined at compile
6426 time or upon entry to the loop. @option{-funroll-loops} implies
6427 @option{-frerun-cse-after-loop}. This option makes code larger,
6428 and may or may not make it run faster.
6430 @item -funroll-all-loops
6431 @opindex funroll-all-loops
6432 Unroll all loops, even if their number of iterations is uncertain when
6433 the loop is entered. This usually makes programs run more slowly.
6434 @option{-funroll-all-loops} implies the same options as
6435 @option{-funroll-loops},
6437 @item -fsplit-ivs-in-unroller
6438 @opindex fsplit-ivs-in-unroller
6439 Enables expressing of values of induction variables in later iterations
6440 of the unrolled loop using the value in the first iteration. This breaks
6441 long dependency chains, thus improving efficiency of the scheduling passes.
6443 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6444 same effect. However in cases the loop body is more complicated than
6445 a single basic block, this is not reliable. It also does not work at all
6446 on some of the architectures due to restrictions in the CSE pass.
6448 This optimization is enabled by default.
6450 @item -fvariable-expansion-in-unroller
6451 @opindex fvariable-expansion-in-unroller
6452 With this option, the compiler will create multiple copies of some
6453 local variables when unrolling a loop which can result in superior code.
6455 @item -fpredictive-commoning
6456 @opindex fpredictive-commoning
6457 Perform predictive commoning optimization, i.e., reusing computations
6458 (especially memory loads and stores) performed in previous
6459 iterations of loops.
6461 This option is enabled at level @option{-O3}.
6463 @item -fprefetch-loop-arrays
6464 @opindex fprefetch-loop-arrays
6465 If supported by the target machine, generate instructions to prefetch
6466 memory to improve the performance of loops that access large arrays.
6468 This option may generate better or worse code; results are highly
6469 dependent on the structure of loops within the source code.
6471 Disabled at level @option{-Os}.
6474 @itemx -fno-peephole2
6475 @opindex fno-peephole
6476 @opindex fno-peephole2
6477 Disable any machine-specific peephole optimizations. The difference
6478 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6479 are implemented in the compiler; some targets use one, some use the
6480 other, a few use both.
6482 @option{-fpeephole} is enabled by default.
6483 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6485 @item -fno-guess-branch-probability
6486 @opindex fno-guess-branch-probability
6487 Do not guess branch probabilities using heuristics.
6489 GCC will use heuristics to guess branch probabilities if they are
6490 not provided by profiling feedback (@option{-fprofile-arcs}). These
6491 heuristics are based on the control flow graph. If some branch probabilities
6492 are specified by @samp{__builtin_expect}, then the heuristics will be
6493 used to guess branch probabilities for the rest of the control flow graph,
6494 taking the @samp{__builtin_expect} info into account. The interactions
6495 between the heuristics and @samp{__builtin_expect} can be complex, and in
6496 some cases, it may be useful to disable the heuristics so that the effects
6497 of @samp{__builtin_expect} are easier to understand.
6499 The default is @option{-fguess-branch-probability} at levels
6500 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6502 @item -freorder-blocks
6503 @opindex freorder-blocks
6504 Reorder basic blocks in the compiled function in order to reduce number of
6505 taken branches and improve code locality.
6507 Enabled at levels @option{-O2}, @option{-O3}.
6509 @item -freorder-blocks-and-partition
6510 @opindex freorder-blocks-and-partition
6511 In addition to reordering basic blocks in the compiled function, in order
6512 to reduce number of taken branches, partitions hot and cold basic blocks
6513 into separate sections of the assembly and .o files, to improve
6514 paging and cache locality performance.
6516 This optimization is automatically turned off in the presence of
6517 exception handling, for linkonce sections, for functions with a user-defined
6518 section attribute and on any architecture that does not support named
6521 @item -freorder-functions
6522 @opindex freorder-functions
6523 Reorder functions in the object file in order to
6524 improve code locality. This is implemented by using special
6525 subsections @code{.text.hot} for most frequently executed functions and
6526 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6527 the linker so object file format must support named sections and linker must
6528 place them in a reasonable way.
6530 Also profile feedback must be available in to make this option effective. See
6531 @option{-fprofile-arcs} for details.
6533 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6535 @item -fstrict-aliasing
6536 @opindex fstrict-aliasing
6537 Allow the compiler to assume the strictest aliasing rules applicable to
6538 the language being compiled. For C (and C++), this activates
6539 optimizations based on the type of expressions. In particular, an
6540 object of one type is assumed never to reside at the same address as an
6541 object of a different type, unless the types are almost the same. For
6542 example, an @code{unsigned int} can alias an @code{int}, but not a
6543 @code{void*} or a @code{double}. A character type may alias any other
6546 @anchor{Type-punning}Pay special attention to code like this:
6559 The practice of reading from a different union member than the one most
6560 recently written to (called ``type-punning'') is common. Even with
6561 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6562 is accessed through the union type. So, the code above will work as
6563 expected. @xref{Structures unions enumerations and bit-fields
6564 implementation}. However, this code might not:
6575 Similarly, access by taking the address, casting the resulting pointer
6576 and dereferencing the result has undefined behavior, even if the cast
6577 uses a union type, e.g.:
6581 return ((union a_union *) &d)->i;
6585 The @option{-fstrict-aliasing} option is enabled at levels
6586 @option{-O2}, @option{-O3}, @option{-Os}.
6588 @item -fstrict-overflow
6589 @opindex fstrict-overflow
6590 Allow the compiler to assume strict signed overflow rules, depending
6591 on the language being compiled. For C (and C++) this means that
6592 overflow when doing arithmetic with signed numbers is undefined, which
6593 means that the compiler may assume that it will not happen. This
6594 permits various optimizations. For example, the compiler will assume
6595 that an expression like @code{i + 10 > i} will always be true for
6596 signed @code{i}. This assumption is only valid if signed overflow is
6597 undefined, as the expression is false if @code{i + 10} overflows when
6598 using twos complement arithmetic. When this option is in effect any
6599 attempt to determine whether an operation on signed numbers will
6600 overflow must be written carefully to not actually involve overflow.
6602 This option also allows the compiler to assume strict pointer
6603 semantics: given a pointer to an object, if adding an offset to that
6604 pointer does not produce a pointer to the same object, the addition is
6605 undefined. This permits the compiler to conclude that @code{p + u >
6606 p} is always true for a pointer @code{p} and unsigned integer
6607 @code{u}. This assumption is only valid because pointer wraparound is
6608 undefined, as the expression is false if @code{p + u} overflows using
6609 twos complement arithmetic.
6611 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6612 that integer signed overflow is fully defined: it wraps. When
6613 @option{-fwrapv} is used, there is no difference between
6614 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6615 integers. With @option{-fwrapv} certain types of overflow are
6616 permitted. For example, if the compiler gets an overflow when doing
6617 arithmetic on constants, the overflowed value can still be used with
6618 @option{-fwrapv}, but not otherwise.
6620 The @option{-fstrict-overflow} option is enabled at levels
6621 @option{-O2}, @option{-O3}, @option{-Os}.
6623 @item -falign-functions
6624 @itemx -falign-functions=@var{n}
6625 @opindex falign-functions
6626 Align the start of functions to the next power-of-two greater than
6627 @var{n}, skipping up to @var{n} bytes. For instance,
6628 @option{-falign-functions=32} aligns functions to the next 32-byte
6629 boundary, but @option{-falign-functions=24} would align to the next
6630 32-byte boundary only if this can be done by skipping 23 bytes or less.
6632 @option{-fno-align-functions} and @option{-falign-functions=1} are
6633 equivalent and mean that functions will not be aligned.
6635 Some assemblers only support this flag when @var{n} is a power of two;
6636 in that case, it is rounded up.
6638 If @var{n} is not specified or is zero, use a machine-dependent default.
6640 Enabled at levels @option{-O2}, @option{-O3}.
6642 @item -falign-labels
6643 @itemx -falign-labels=@var{n}
6644 @opindex falign-labels
6645 Align all branch targets to a power-of-two boundary, skipping up to
6646 @var{n} bytes like @option{-falign-functions}. This option can easily
6647 make code slower, because it must insert dummy operations for when the
6648 branch target is reached in the usual flow of the code.
6650 @option{-fno-align-labels} and @option{-falign-labels=1} are
6651 equivalent and mean that labels will not be aligned.
6653 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6654 are greater than this value, then their values are used instead.
6656 If @var{n} is not specified or is zero, use a machine-dependent default
6657 which is very likely to be @samp{1}, meaning no alignment.
6659 Enabled at levels @option{-O2}, @option{-O3}.
6662 @itemx -falign-loops=@var{n}
6663 @opindex falign-loops
6664 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6665 like @option{-falign-functions}. The hope is that the loop will be
6666 executed many times, which will make up for any execution of the dummy
6669 @option{-fno-align-loops} and @option{-falign-loops=1} are
6670 equivalent and mean that loops will not be aligned.
6672 If @var{n} is not specified or is zero, use a machine-dependent default.
6674 Enabled at levels @option{-O2}, @option{-O3}.
6677 @itemx -falign-jumps=@var{n}
6678 @opindex falign-jumps
6679 Align branch targets to a power-of-two boundary, for branch targets
6680 where the targets can only be reached by jumping, skipping up to @var{n}
6681 bytes like @option{-falign-functions}. In this case, no dummy operations
6684 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6685 equivalent and mean that loops will not be aligned.
6687 If @var{n} is not specified or is zero, use a machine-dependent default.
6689 Enabled at levels @option{-O2}, @option{-O3}.
6691 @item -funit-at-a-time
6692 @opindex funit-at-a-time
6693 This option is left for compatibility reasons. @option{-funit-at-a-time}
6694 has no effect, while @option{-fno-unit-at-a-time} implies
6695 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6699 @item -fno-toplevel-reorder
6700 @opindex fno-toplevel-reorder
6701 Do not reorder top-level functions, variables, and @code{asm}
6702 statements. Output them in the same order that they appear in the
6703 input file. When this option is used, unreferenced static variables
6704 will not be removed. This option is intended to support existing code
6705 which relies on a particular ordering. For new code, it is better to
6708 Enabled at level @option{-O0}. When disabled explicitly, it also imply
6709 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6714 Constructs webs as commonly used for register allocation purposes and assign
6715 each web individual pseudo register. This allows the register allocation pass
6716 to operate on pseudos directly, but also strengthens several other optimization
6717 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6718 however, make debugging impossible, since variables will no longer stay in a
6721 Enabled by default with @option{-funroll-loops}.
6723 @item -fwhole-program
6724 @opindex fwhole-program
6725 Assume that the current compilation unit represents whole program being
6726 compiled. All public functions and variables with the exception of @code{main}
6727 and those merged by attribute @code{externally_visible} become static functions
6728 and in a affect gets more aggressively optimized by interprocedural optimizers.
6729 While this option is equivalent to proper use of @code{static} keyword for
6730 programs consisting of single file, in combination with option
6731 @option{--combine} this flag can be used to compile most of smaller scale C
6732 programs since the functions and variables become local for the whole combined
6733 compilation unit, not for the single source file itself.
6735 This option is not supported for Fortran programs.
6737 @item -fcprop-registers
6738 @opindex fcprop-registers
6739 After register allocation and post-register allocation instruction splitting,
6740 we perform a copy-propagation pass to try to reduce scheduling dependencies
6741 and occasionally eliminate the copy.
6743 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6745 @item -fprofile-correction
6746 @opindex fprofile-correction
6747 Profiles collected using an instrumented binary for multi-threaded programs may
6748 be inconsistent due to missed counter updates. When this option is specified,
6749 GCC will use heuristics to correct or smooth out such inconsistencies. By
6750 default, GCC will emit an error message when an inconsistent profile is detected.
6752 @item -fprofile-dir=@var{path}
6753 @opindex fprofile-dir
6755 Set the directory to search the profile data files in to @var{path}.
6756 This option affects only the profile data generated by
6757 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6758 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
6759 and its related options.
6760 By default, GCC will use the current directory as @var{path}
6761 thus the profile data file will appear in the same directory as the object file.
6763 @item -fprofile-generate
6764 @itemx -fprofile-generate=@var{path}
6765 @opindex fprofile-generate
6767 Enable options usually used for instrumenting application to produce
6768 profile useful for later recompilation with profile feedback based
6769 optimization. You must use @option{-fprofile-generate} both when
6770 compiling and when linking your program.
6772 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6774 If @var{path} is specified, GCC will look at the @var{path} to find
6775 the profile feedback data files. See @option{-fprofile-dir}.
6778 @itemx -fprofile-use=@var{path}
6779 @opindex fprofile-use
6780 Enable profile feedback directed optimizations, and optimizations
6781 generally profitable only with profile feedback available.
6783 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6784 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6786 By default, GCC emits an error message if the feedback profiles do not
6787 match the source code. This error can be turned into a warning by using
6788 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6791 If @var{path} is specified, GCC will look at the @var{path} to find
6792 the profile feedback data files. See @option{-fprofile-dir}.
6795 The following options control compiler behavior regarding floating
6796 point arithmetic. These options trade off between speed and
6797 correctness. All must be specifically enabled.
6801 @opindex ffloat-store
6802 Do not store floating point variables in registers, and inhibit other
6803 options that might change whether a floating point value is taken from a
6806 @cindex floating point precision
6807 This option prevents undesirable excess precision on machines such as
6808 the 68000 where the floating registers (of the 68881) keep more
6809 precision than a @code{double} is supposed to have. Similarly for the
6810 x86 architecture. For most programs, the excess precision does only
6811 good, but a few programs rely on the precise definition of IEEE floating
6812 point. Use @option{-ffloat-store} for such programs, after modifying
6813 them to store all pertinent intermediate computations into variables.
6817 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6818 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6819 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6821 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6823 This option is not turned on by any @option{-O} option since
6824 it can result in incorrect output for programs which depend on
6825 an exact implementation of IEEE or ISO rules/specifications for
6826 math functions. It may, however, yield faster code for programs
6827 that do not require the guarantees of these specifications.
6829 @item -fno-math-errno
6830 @opindex fno-math-errno
6831 Do not set ERRNO after calling math functions that are executed
6832 with a single instruction, e.g., sqrt. A program that relies on
6833 IEEE exceptions for math error handling may want to use this flag
6834 for speed while maintaining IEEE arithmetic compatibility.
6836 This option is not turned on by any @option{-O} option since
6837 it can result in incorrect output for programs which depend on
6838 an exact implementation of IEEE or ISO rules/specifications for
6839 math functions. It may, however, yield faster code for programs
6840 that do not require the guarantees of these specifications.
6842 The default is @option{-fmath-errno}.
6844 On Darwin systems, the math library never sets @code{errno}. There is
6845 therefore no reason for the compiler to consider the possibility that
6846 it might, and @option{-fno-math-errno} is the default.
6848 @item -funsafe-math-optimizations
6849 @opindex funsafe-math-optimizations
6851 Allow optimizations for floating-point arithmetic that (a) assume
6852 that arguments and results are valid and (b) may violate IEEE or
6853 ANSI standards. When used at link-time, it may include libraries
6854 or startup files that change the default FPU control word or other
6855 similar optimizations.
6857 This option is not turned on by any @option{-O} option since
6858 it can result in incorrect output for programs which depend on
6859 an exact implementation of IEEE or ISO rules/specifications for
6860 math functions. It may, however, yield faster code for programs
6861 that do not require the guarantees of these specifications.
6862 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6863 @option{-fassociative-math} and @option{-freciprocal-math}.
6865 The default is @option{-fno-unsafe-math-optimizations}.
6867 @item -fassociative-math
6868 @opindex fassociative-math
6870 Allow re-association of operands in series of floating-point operations.
6871 This violates the ISO C and C++ language standard by possibly changing
6872 computation result. NOTE: re-ordering may change the sign of zero as
6873 well as ignore NaNs and inhibit or create underflow or overflow (and
6874 thus cannot be used on a code which relies on rounding behavior like
6875 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6876 and thus may not be used when ordered comparisons are required.
6877 This option requires that both @option{-fno-signed-zeros} and
6878 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6879 much sense with @option{-frounding-math}.
6881 The default is @option{-fno-associative-math}.
6883 @item -freciprocal-math
6884 @opindex freciprocal-math
6886 Allow the reciprocal of a value to be used instead of dividing by
6887 the value if this enables optimizations. For example @code{x / y}
6888 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6889 is subject to common subexpression elimination. Note that this loses
6890 precision and increases the number of flops operating on the value.
6892 The default is @option{-fno-reciprocal-math}.
6894 @item -ffinite-math-only
6895 @opindex ffinite-math-only
6896 Allow optimizations for floating-point arithmetic that assume
6897 that arguments and results are not NaNs or +-Infs.
6899 This option is not turned on by any @option{-O} option since
6900 it can result in incorrect output for programs which depend on
6901 an exact implementation of IEEE or ISO rules/specifications for
6902 math functions. It may, however, yield faster code for programs
6903 that do not require the guarantees of these specifications.
6905 The default is @option{-fno-finite-math-only}.
6907 @item -fno-signed-zeros
6908 @opindex fno-signed-zeros
6909 Allow optimizations for floating point arithmetic that ignore the
6910 signedness of zero. IEEE arithmetic specifies the behavior of
6911 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6912 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6913 This option implies that the sign of a zero result isn't significant.
6915 The default is @option{-fsigned-zeros}.
6917 @item -fno-trapping-math
6918 @opindex fno-trapping-math
6919 Compile code assuming that floating-point operations cannot generate
6920 user-visible traps. These traps include division by zero, overflow,
6921 underflow, inexact result and invalid operation. This option requires
6922 that @option{-fno-signaling-nans} be in effect. Setting this option may
6923 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6925 This option should never be turned on by any @option{-O} option since
6926 it can result in incorrect output for programs which depend on
6927 an exact implementation of IEEE or ISO rules/specifications for
6930 The default is @option{-ftrapping-math}.
6932 @item -frounding-math
6933 @opindex frounding-math
6934 Disable transformations and optimizations that assume default floating
6935 point rounding behavior. This is round-to-zero for all floating point
6936 to integer conversions, and round-to-nearest for all other arithmetic
6937 truncations. This option should be specified for programs that change
6938 the FP rounding mode dynamically, or that may be executed with a
6939 non-default rounding mode. This option disables constant folding of
6940 floating point expressions at compile-time (which may be affected by
6941 rounding mode) and arithmetic transformations that are unsafe in the
6942 presence of sign-dependent rounding modes.
6944 The default is @option{-fno-rounding-math}.
6946 This option is experimental and does not currently guarantee to
6947 disable all GCC optimizations that are affected by rounding mode.
6948 Future versions of GCC may provide finer control of this setting
6949 using C99's @code{FENV_ACCESS} pragma. This command line option
6950 will be used to specify the default state for @code{FENV_ACCESS}.
6952 @item -frtl-abstract-sequences
6953 @opindex frtl-abstract-sequences
6954 It is a size optimization method. This option is to find identical
6955 sequences of code, which can be turned into pseudo-procedures and
6956 then replace all occurrences with calls to the newly created
6957 subroutine. It is kind of an opposite of @option{-finline-functions}.
6958 This optimization runs at RTL level.
6960 @item -fsignaling-nans
6961 @opindex fsignaling-nans
6962 Compile code assuming that IEEE signaling NaNs may generate user-visible
6963 traps during floating-point operations. Setting this option disables
6964 optimizations that may change the number of exceptions visible with
6965 signaling NaNs. This option implies @option{-ftrapping-math}.
6967 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6970 The default is @option{-fno-signaling-nans}.
6972 This option is experimental and does not currently guarantee to
6973 disable all GCC optimizations that affect signaling NaN behavior.
6975 @item -fsingle-precision-constant
6976 @opindex fsingle-precision-constant
6977 Treat floating point constant as single precision constant instead of
6978 implicitly converting it to double precision constant.
6980 @item -fcx-limited-range
6981 @opindex fcx-limited-range
6982 When enabled, this option states that a range reduction step is not
6983 needed when performing complex division. Also, there is no checking
6984 whether the result of a complex multiplication or division is @code{NaN
6985 + I*NaN}, with an attempt to rescue the situation in that case. The
6986 default is @option{-fno-cx-limited-range}, but is enabled by
6987 @option{-ffast-math}.
6989 This option controls the default setting of the ISO C99
6990 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6993 @item -fcx-fortran-rules
6994 @opindex fcx-fortran-rules
6995 Complex multiplication and division follow Fortran rules. Range
6996 reduction is done as part of complex division, but there is no checking
6997 whether the result of a complex multiplication or division is @code{NaN
6998 + I*NaN}, with an attempt to rescue the situation in that case.
7000 The default is @option{-fno-cx-fortran-rules}.
7004 The following options control optimizations that may improve
7005 performance, but are not enabled by any @option{-O} options. This
7006 section includes experimental options that may produce broken code.
7009 @item -fbranch-probabilities
7010 @opindex fbranch-probabilities
7011 After running a program compiled with @option{-fprofile-arcs}
7012 (@pxref{Debugging Options,, Options for Debugging Your Program or
7013 @command{gcc}}), you can compile it a second time using
7014 @option{-fbranch-probabilities}, to improve optimizations based on
7015 the number of times each branch was taken. When the program
7016 compiled with @option{-fprofile-arcs} exits it saves arc execution
7017 counts to a file called @file{@var{sourcename}.gcda} for each source
7018 file. The information in this data file is very dependent on the
7019 structure of the generated code, so you must use the same source code
7020 and the same optimization options for both compilations.
7022 With @option{-fbranch-probabilities}, GCC puts a
7023 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7024 These can be used to improve optimization. Currently, they are only
7025 used in one place: in @file{reorg.c}, instead of guessing which path a
7026 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7027 exactly determine which path is taken more often.
7029 @item -fprofile-values
7030 @opindex fprofile-values
7031 If combined with @option{-fprofile-arcs}, it adds code so that some
7032 data about values of expressions in the program is gathered.
7034 With @option{-fbranch-probabilities}, it reads back the data gathered
7035 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7036 notes to instructions for their later usage in optimizations.
7038 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7042 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7043 a code to gather information about values of expressions.
7045 With @option{-fbranch-probabilities}, it reads back the data gathered
7046 and actually performs the optimizations based on them.
7047 Currently the optimizations include specialization of division operation
7048 using the knowledge about the value of the denominator.
7050 @item -frename-registers
7051 @opindex frename-registers
7052 Attempt to avoid false dependencies in scheduled code by making use
7053 of registers left over after register allocation. This optimization
7054 will most benefit processors with lots of registers. Depending on the
7055 debug information format adopted by the target, however, it can
7056 make debugging impossible, since variables will no longer stay in
7057 a ``home register''.
7059 Enabled by default with @option{-funroll-loops}.
7063 Perform tail duplication to enlarge superblock size. This transformation
7064 simplifies the control flow of the function allowing other optimizations to do
7067 Enabled with @option{-fprofile-use}.
7069 @item -funroll-loops
7070 @opindex funroll-loops
7071 Unroll loops whose number of iterations can be determined at compile time or
7072 upon entry to the loop. @option{-funroll-loops} implies
7073 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7074 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7075 small constant number of iterations). This option makes code larger, and may
7076 or may not make it run faster.
7078 Enabled with @option{-fprofile-use}.
7080 @item -funroll-all-loops
7081 @opindex funroll-all-loops
7082 Unroll all loops, even if their number of iterations is uncertain when
7083 the loop is entered. This usually makes programs run more slowly.
7084 @option{-funroll-all-loops} implies the same options as
7085 @option{-funroll-loops}.
7088 @opindex fpeel-loops
7089 Peels the loops for that there is enough information that they do not
7090 roll much (from profile feedback). It also turns on complete loop peeling
7091 (i.e.@: complete removal of loops with small constant number of iterations).
7093 Enabled with @option{-fprofile-use}.
7095 @item -fmove-loop-invariants
7096 @opindex fmove-loop-invariants
7097 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7098 at level @option{-O1}
7100 @item -funswitch-loops
7101 @opindex funswitch-loops
7102 Move branches with loop invariant conditions out of the loop, with duplicates
7103 of the loop on both branches (modified according to result of the condition).
7105 @item -ffunction-sections
7106 @itemx -fdata-sections
7107 @opindex ffunction-sections
7108 @opindex fdata-sections
7109 Place each function or data item into its own section in the output
7110 file if the target supports arbitrary sections. The name of the
7111 function or the name of the data item determines the section's name
7114 Use these options on systems where the linker can perform optimizations
7115 to improve locality of reference in the instruction space. Most systems
7116 using the ELF object format and SPARC processors running Solaris 2 have
7117 linkers with such optimizations. AIX may have these optimizations in
7120 Only use these options when there are significant benefits from doing
7121 so. When you specify these options, the assembler and linker will
7122 create larger object and executable files and will also be slower.
7123 You will not be able to use @code{gprof} on all systems if you
7124 specify this option and you may have problems with debugging if
7125 you specify both this option and @option{-g}.
7127 @item -fbranch-target-load-optimize
7128 @opindex fbranch-target-load-optimize
7129 Perform branch target register load optimization before prologue / epilogue
7131 The use of target registers can typically be exposed only during reload,
7132 thus hoisting loads out of loops and doing inter-block scheduling needs
7133 a separate optimization pass.
7135 @item -fbranch-target-load-optimize2
7136 @opindex fbranch-target-load-optimize2
7137 Perform branch target register load optimization after prologue / epilogue
7140 @item -fbtr-bb-exclusive
7141 @opindex fbtr-bb-exclusive
7142 When performing branch target register load optimization, don't reuse
7143 branch target registers in within any basic block.
7145 @item -fstack-protector
7146 @opindex fstack-protector
7147 Emit extra code to check for buffer overflows, such as stack smashing
7148 attacks. This is done by adding a guard variable to functions with
7149 vulnerable objects. This includes functions that call alloca, and
7150 functions with buffers larger than 8 bytes. The guards are initialized
7151 when a function is entered and then checked when the function exits.
7152 If a guard check fails, an error message is printed and the program exits.
7154 @item -fstack-protector-all
7155 @opindex fstack-protector-all
7156 Like @option{-fstack-protector} except that all functions are protected.
7158 @item -fsection-anchors
7159 @opindex fsection-anchors
7160 Try to reduce the number of symbolic address calculations by using
7161 shared ``anchor'' symbols to address nearby objects. This transformation
7162 can help to reduce the number of GOT entries and GOT accesses on some
7165 For example, the implementation of the following function @code{foo}:
7169 int foo (void) @{ return a + b + c; @}
7172 would usually calculate the addresses of all three variables, but if you
7173 compile it with @option{-fsection-anchors}, it will access the variables
7174 from a common anchor point instead. The effect is similar to the
7175 following pseudocode (which isn't valid C):
7180 register int *xr = &x;
7181 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7185 Not all targets support this option.
7187 @item --param @var{name}=@var{value}
7189 In some places, GCC uses various constants to control the amount of
7190 optimization that is done. For example, GCC will not inline functions
7191 that contain more that a certain number of instructions. You can
7192 control some of these constants on the command-line using the
7193 @option{--param} option.
7195 The names of specific parameters, and the meaning of the values, are
7196 tied to the internals of the compiler, and are subject to change
7197 without notice in future releases.
7199 In each case, the @var{value} is an integer. The allowable choices for
7200 @var{name} are given in the following table:
7203 @item sra-max-structure-size
7204 The maximum structure size, in bytes, at which the scalar replacement
7205 of aggregates (SRA) optimization will perform block copies. The
7206 default value, 0, implies that GCC will select the most appropriate
7209 @item sra-field-structure-ratio
7210 The threshold ratio (as a percentage) between instantiated fields and
7211 the complete structure size. We say that if the ratio of the number
7212 of bytes in instantiated fields to the number of bytes in the complete
7213 structure exceeds this parameter, then block copies are not used. The
7216 @item struct-reorg-cold-struct-ratio
7217 The threshold ratio (as a percentage) between a structure frequency
7218 and the frequency of the hottest structure in the program. This parameter
7219 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7220 We say that if the ratio of a structure frequency, calculated by profiling,
7221 to the hottest structure frequency in the program is less than this
7222 parameter, then structure reorganization is not applied to this structure.
7225 @item predictable-branch-cost-outcome
7226 When branch is predicted to be taken with probability lower than this threshold
7227 (in percent), then it is considered well predictable. The default is 10.
7229 @item max-crossjump-edges
7230 The maximum number of incoming edges to consider for crossjumping.
7231 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7232 the number of edges incoming to each block. Increasing values mean
7233 more aggressive optimization, making the compile time increase with
7234 probably small improvement in executable size.
7236 @item min-crossjump-insns
7237 The minimum number of instructions which must be matched at the end
7238 of two blocks before crossjumping will be performed on them. This
7239 value is ignored in the case where all instructions in the block being
7240 crossjumped from are matched. The default value is 5.
7242 @item max-grow-copy-bb-insns
7243 The maximum code size expansion factor when copying basic blocks
7244 instead of jumping. The expansion is relative to a jump instruction.
7245 The default value is 8.
7247 @item max-goto-duplication-insns
7248 The maximum number of instructions to duplicate to a block that jumps
7249 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7250 passes, GCC factors computed gotos early in the compilation process,
7251 and unfactors them as late as possible. Only computed jumps at the
7252 end of a basic blocks with no more than max-goto-duplication-insns are
7253 unfactored. The default value is 8.
7255 @item max-delay-slot-insn-search
7256 The maximum number of instructions to consider when looking for an
7257 instruction to fill a delay slot. If more than this arbitrary number of
7258 instructions is searched, the time savings from filling the delay slot
7259 will be minimal so stop searching. Increasing values mean more
7260 aggressive optimization, making the compile time increase with probably
7261 small improvement in executable run time.
7263 @item max-delay-slot-live-search
7264 When trying to fill delay slots, the maximum number of instructions to
7265 consider when searching for a block with valid live register
7266 information. Increasing this arbitrarily chosen value means more
7267 aggressive optimization, increasing the compile time. This parameter
7268 should be removed when the delay slot code is rewritten to maintain the
7271 @item max-gcse-memory
7272 The approximate maximum amount of memory that will be allocated in
7273 order to perform the global common subexpression elimination
7274 optimization. If more memory than specified is required, the
7275 optimization will not be done.
7277 @item max-gcse-passes
7278 The maximum number of passes of GCSE to run. The default is 1.
7280 @item max-pending-list-length
7281 The maximum number of pending dependencies scheduling will allow
7282 before flushing the current state and starting over. Large functions
7283 with few branches or calls can create excessively large lists which
7284 needlessly consume memory and resources.
7286 @item max-inline-insns-single
7287 Several parameters control the tree inliner used in gcc.
7288 This number sets the maximum number of instructions (counted in GCC's
7289 internal representation) in a single function that the tree inliner
7290 will consider for inlining. This only affects functions declared
7291 inline and methods implemented in a class declaration (C++).
7292 The default value is 450.
7294 @item max-inline-insns-auto
7295 When you use @option{-finline-functions} (included in @option{-O3}),
7296 a lot of functions that would otherwise not be considered for inlining
7297 by the compiler will be investigated. To those functions, a different
7298 (more restrictive) limit compared to functions declared inline can
7300 The default value is 90.
7302 @item large-function-insns
7303 The limit specifying really large functions. For functions larger than this
7304 limit after inlining, inlining is constrained by
7305 @option{--param large-function-growth}. This parameter is useful primarily
7306 to avoid extreme compilation time caused by non-linear algorithms used by the
7308 The default value is 2700.
7310 @item large-function-growth
7311 Specifies maximal growth of large function caused by inlining in percents.
7312 The default value is 100 which limits large function growth to 2.0 times
7315 @item large-unit-insns
7316 The limit specifying large translation unit. Growth caused by inlining of
7317 units larger than this limit is limited by @option{--param inline-unit-growth}.
7318 For small units this might be too tight (consider unit consisting of function A
7319 that is inline and B that just calls A three time. If B is small relative to
7320 A, the growth of unit is 300\% and yet such inlining is very sane. For very
7321 large units consisting of small inlineable functions however the overall unit
7322 growth limit is needed to avoid exponential explosion of code size. Thus for
7323 smaller units, the size is increased to @option{--param large-unit-insns}
7324 before applying @option{--param inline-unit-growth}. The default is 10000
7326 @item inline-unit-growth
7327 Specifies maximal overall growth of the compilation unit caused by inlining.
7328 The default value is 30 which limits unit growth to 1.3 times the original
7331 @item ipcp-unit-growth
7332 Specifies maximal overall growth of the compilation unit caused by
7333 interprocedural constant propagation. The default value is 10 which limits
7334 unit growth to 1.1 times the original size.
7336 @item large-stack-frame
7337 The limit specifying large stack frames. While inlining the algorithm is trying
7338 to not grow past this limit too much. Default value is 256 bytes.
7340 @item large-stack-frame-growth
7341 Specifies maximal growth of large stack frames caused by inlining in percents.
7342 The default value is 1000 which limits large stack frame growth to 11 times
7345 @item max-inline-insns-recursive
7346 @itemx max-inline-insns-recursive-auto
7347 Specifies maximum number of instructions out-of-line copy of self recursive inline
7348 function can grow into by performing recursive inlining.
7350 For functions declared inline @option{--param max-inline-insns-recursive} is
7351 taken into account. For function not declared inline, recursive inlining
7352 happens only when @option{-finline-functions} (included in @option{-O3}) is
7353 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
7354 default value is 450.
7356 @item max-inline-recursive-depth
7357 @itemx max-inline-recursive-depth-auto
7358 Specifies maximum recursion depth used by the recursive inlining.
7360 For functions declared inline @option{--param max-inline-recursive-depth} is
7361 taken into account. For function not declared inline, recursive inlining
7362 happens only when @option{-finline-functions} (included in @option{-O3}) is
7363 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
7366 @item min-inline-recursive-probability
7367 Recursive inlining is profitable only for function having deep recursion
7368 in average and can hurt for function having little recursion depth by
7369 increasing the prologue size or complexity of function body to other
7372 When profile feedback is available (see @option{-fprofile-generate}) the actual
7373 recursion depth can be guessed from probability that function will recurse via
7374 given call expression. This parameter limits inlining only to call expression
7375 whose probability exceeds given threshold (in percents). The default value is
7378 @item inline-call-cost
7379 Specify cost of call instruction relative to simple arithmetics operations
7380 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
7381 functions and at the same time increases size of leaf function that is believed to
7382 reduce function size by being inlined. In effect it increases amount of
7383 inlining for code having large abstraction penalty (many functions that just
7384 pass the arguments to other functions) and decrease inlining for code with low
7385 abstraction penalty. The default value is 12.
7387 @item min-vect-loop-bound
7388 The minimum number of iterations under which a loop will not get vectorized
7389 when @option{-ftree-vectorize} is used. The number of iterations after
7390 vectorization needs to be greater than the value specified by this option
7391 to allow vectorization. The default value is 0.
7393 @item max-unrolled-insns
7394 The maximum number of instructions that a loop should have if that loop
7395 is unrolled, and if the loop is unrolled, it determines how many times
7396 the loop code is unrolled.
7398 @item max-average-unrolled-insns
7399 The maximum number of instructions biased by probabilities of their execution
7400 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7401 it determines how many times the loop code is unrolled.
7403 @item max-unroll-times
7404 The maximum number of unrollings of a single loop.
7406 @item max-peeled-insns
7407 The maximum number of instructions that a loop should have if that loop
7408 is peeled, and if the loop is peeled, it determines how many times
7409 the loop code is peeled.
7411 @item max-peel-times
7412 The maximum number of peelings of a single loop.
7414 @item max-completely-peeled-insns
7415 The maximum number of insns of a completely peeled loop.
7417 @item max-completely-peel-times
7418 The maximum number of iterations of a loop to be suitable for complete peeling.
7420 @item max-unswitch-insns
7421 The maximum number of insns of an unswitched loop.
7423 @item max-unswitch-level
7424 The maximum number of branches unswitched in a single loop.
7427 The minimum cost of an expensive expression in the loop invariant motion.
7429 @item iv-consider-all-candidates-bound
7430 Bound on number of candidates for induction variables below that
7431 all candidates are considered for each use in induction variable
7432 optimizations. Only the most relevant candidates are considered
7433 if there are more candidates, to avoid quadratic time complexity.
7435 @item iv-max-considered-uses
7436 The induction variable optimizations give up on loops that contain more
7437 induction variable uses.
7439 @item iv-always-prune-cand-set-bound
7440 If number of candidates in the set is smaller than this value,
7441 we always try to remove unnecessary ivs from the set during its
7442 optimization when a new iv is added to the set.
7444 @item scev-max-expr-size
7445 Bound on size of expressions used in the scalar evolutions analyzer.
7446 Large expressions slow the analyzer.
7448 @item omega-max-vars
7449 The maximum number of variables in an Omega constraint system.
7450 The default value is 128.
7452 @item omega-max-geqs
7453 The maximum number of inequalities in an Omega constraint system.
7454 The default value is 256.
7457 The maximum number of equalities in an Omega constraint system.
7458 The default value is 128.
7460 @item omega-max-wild-cards
7461 The maximum number of wildcard variables that the Omega solver will
7462 be able to insert. The default value is 18.
7464 @item omega-hash-table-size
7465 The size of the hash table in the Omega solver. The default value is
7468 @item omega-max-keys
7469 The maximal number of keys used by the Omega solver. The default
7472 @item omega-eliminate-redundant-constraints
7473 When set to 1, use expensive methods to eliminate all redundant
7474 constraints. The default value is 0.
7476 @item vect-max-version-for-alignment-checks
7477 The maximum number of runtime checks that can be performed when
7478 doing loop versioning for alignment in the vectorizer. See option
7479 ftree-vect-loop-version for more information.
7481 @item vect-max-version-for-alias-checks
7482 The maximum number of runtime checks that can be performed when
7483 doing loop versioning for alias in the vectorizer. See option
7484 ftree-vect-loop-version for more information.
7486 @item max-iterations-to-track
7488 The maximum number of iterations of a loop the brute force algorithm
7489 for analysis of # of iterations of the loop tries to evaluate.
7491 @item hot-bb-count-fraction
7492 Select fraction of the maximal count of repetitions of basic block in program
7493 given basic block needs to have to be considered hot.
7495 @item hot-bb-frequency-fraction
7496 Select fraction of the maximal frequency of executions of basic block in
7497 function given basic block needs to have to be considered hot
7499 @item max-predicted-iterations
7500 The maximum number of loop iterations we predict statically. This is useful
7501 in cases where function contain single loop with known bound and other loop
7502 with unknown. We predict the known number of iterations correctly, while
7503 the unknown number of iterations average to roughly 10. This means that the
7504 loop without bounds would appear artificially cold relative to the other one.
7506 @item align-threshold
7508 Select fraction of the maximal frequency of executions of basic block in
7509 function given basic block will get aligned.
7511 @item align-loop-iterations
7513 A loop expected to iterate at lest the selected number of iterations will get
7516 @item tracer-dynamic-coverage
7517 @itemx tracer-dynamic-coverage-feedback
7519 This value is used to limit superblock formation once the given percentage of
7520 executed instructions is covered. This limits unnecessary code size
7523 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7524 feedback is available. The real profiles (as opposed to statically estimated
7525 ones) are much less balanced allowing the threshold to be larger value.
7527 @item tracer-max-code-growth
7528 Stop tail duplication once code growth has reached given percentage. This is
7529 rather hokey argument, as most of the duplicates will be eliminated later in
7530 cross jumping, so it may be set to much higher values than is the desired code
7533 @item tracer-min-branch-ratio
7535 Stop reverse growth when the reverse probability of best edge is less than this
7536 threshold (in percent).
7538 @item tracer-min-branch-ratio
7539 @itemx tracer-min-branch-ratio-feedback
7541 Stop forward growth if the best edge do have probability lower than this
7544 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7545 compilation for profile feedback and one for compilation without. The value
7546 for compilation with profile feedback needs to be more conservative (higher) in
7547 order to make tracer effective.
7549 @item max-cse-path-length
7551 Maximum number of basic blocks on path that cse considers. The default is 10.
7554 The maximum instructions CSE process before flushing. The default is 1000.
7556 @item max-aliased-vops
7558 Maximum number of virtual operands per function allowed to represent
7559 aliases before triggering the alias partitioning heuristic. Alias
7560 partitioning reduces compile times and memory consumption needed for
7561 aliasing at the expense of precision loss in alias information. The
7562 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7565 Notice that if a function contains more memory statements than the
7566 value of this parameter, it is not really possible to achieve this
7567 reduction. In this case, the compiler will use the number of memory
7568 statements as the value for @option{max-aliased-vops}.
7570 @item avg-aliased-vops
7572 Average number of virtual operands per statement allowed to represent
7573 aliases before triggering the alias partitioning heuristic. This
7574 works in conjunction with @option{max-aliased-vops}. If a function
7575 contains more than @option{max-aliased-vops} virtual operators, then
7576 memory symbols will be grouped into memory partitions until either the
7577 total number of virtual operators is below @option{max-aliased-vops}
7578 or the average number of virtual operators per memory statement is
7579 below @option{avg-aliased-vops}. The default value for this parameter
7580 is 1 for -O1 and -O2, and 3 for -O3.
7582 @item ggc-min-expand
7584 GCC uses a garbage collector to manage its own memory allocation. This
7585 parameter specifies the minimum percentage by which the garbage
7586 collector's heap should be allowed to expand between collections.
7587 Tuning this may improve compilation speed; it has no effect on code
7590 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7591 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7592 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7593 GCC is not able to calculate RAM on a particular platform, the lower
7594 bound of 30% is used. Setting this parameter and
7595 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7596 every opportunity. This is extremely slow, but can be useful for
7599 @item ggc-min-heapsize
7601 Minimum size of the garbage collector's heap before it begins bothering
7602 to collect garbage. The first collection occurs after the heap expands
7603 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7604 tuning this may improve compilation speed, and has no effect on code
7607 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7608 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7609 with a lower bound of 4096 (four megabytes) and an upper bound of
7610 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7611 particular platform, the lower bound is used. Setting this parameter
7612 very large effectively disables garbage collection. Setting this
7613 parameter and @option{ggc-min-expand} to zero causes a full collection
7614 to occur at every opportunity.
7616 @item max-reload-search-insns
7617 The maximum number of instruction reload should look backward for equivalent
7618 register. Increasing values mean more aggressive optimization, making the
7619 compile time increase with probably slightly better performance. The default
7622 @item max-cselib-memory-locations
7623 The maximum number of memory locations cselib should take into account.
7624 Increasing values mean more aggressive optimization, making the compile time
7625 increase with probably slightly better performance. The default value is 500.
7627 @item reorder-blocks-duplicate
7628 @itemx reorder-blocks-duplicate-feedback
7630 Used by basic block reordering pass to decide whether to use unconditional
7631 branch or duplicate the code on its destination. Code is duplicated when its
7632 estimated size is smaller than this value multiplied by the estimated size of
7633 unconditional jump in the hot spots of the program.
7635 The @option{reorder-block-duplicate-feedback} is used only when profile
7636 feedback is available and may be set to higher values than
7637 @option{reorder-block-duplicate} since information about the hot spots is more
7640 @item max-sched-ready-insns
7641 The maximum number of instructions ready to be issued the scheduler should
7642 consider at any given time during the first scheduling pass. Increasing
7643 values mean more thorough searches, making the compilation time increase
7644 with probably little benefit. The default value is 100.
7646 @item max-sched-region-blocks
7647 The maximum number of blocks in a region to be considered for
7648 interblock scheduling. The default value is 10.
7650 @item max-pipeline-region-blocks
7651 The maximum number of blocks in a region to be considered for
7652 pipelining in the selective scheduler. The default value is 15.
7654 @item max-sched-region-insns
7655 The maximum number of insns in a region to be considered for
7656 interblock scheduling. The default value is 100.
7658 @item max-pipeline-region-insns
7659 The maximum number of insns in a region to be considered for
7660 pipelining in the selective scheduler. The default value is 200.
7663 The minimum probability (in percents) of reaching a source block
7664 for interblock speculative scheduling. The default value is 40.
7666 @item max-sched-extend-regions-iters
7667 The maximum number of iterations through CFG to extend regions.
7668 0 - disable region extension,
7669 N - do at most N iterations.
7670 The default value is 0.
7672 @item max-sched-insn-conflict-delay
7673 The maximum conflict delay for an insn to be considered for speculative motion.
7674 The default value is 3.
7676 @item sched-spec-prob-cutoff
7677 The minimal probability of speculation success (in percents), so that
7678 speculative insn will be scheduled.
7679 The default value is 40.
7681 @item sched-mem-true-dep-cost
7682 Minimal distance (in CPU cycles) between store and load targeting same
7683 memory locations. The default value is 1.
7685 @item selsched-max-lookahead
7686 The maximum size of the lookahead window of selective scheduling. It is a
7687 depth of search for available instructions.
7688 The default value is 50.
7690 @item selsched-max-sched-times
7691 The maximum number of times that an instruction will be scheduled during
7692 selective scheduling. This is the limit on the number of iterations
7693 through which the instruction may be pipelined. The default value is 2.
7695 @item selsched-max-insns-to-rename
7696 The maximum number of best instructions in the ready list that are considered
7697 for renaming in the selective scheduler. The default value is 2.
7699 @item max-last-value-rtl
7700 The maximum size measured as number of RTLs that can be recorded in an expression
7701 in combiner for a pseudo register as last known value of that register. The default
7704 @item integer-share-limit
7705 Small integer constants can use a shared data structure, reducing the
7706 compiler's memory usage and increasing its speed. This sets the maximum
7707 value of a shared integer constant. The default value is 256.
7709 @item min-virtual-mappings
7710 Specifies the minimum number of virtual mappings in the incremental
7711 SSA updater that should be registered to trigger the virtual mappings
7712 heuristic defined by virtual-mappings-ratio. The default value is
7715 @item virtual-mappings-ratio
7716 If the number of virtual mappings is virtual-mappings-ratio bigger
7717 than the number of virtual symbols to be updated, then the incremental
7718 SSA updater switches to a full update for those symbols. The default
7721 @item ssp-buffer-size
7722 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7723 protection when @option{-fstack-protection} is used.
7725 @item max-jump-thread-duplication-stmts
7726 Maximum number of statements allowed in a block that needs to be
7727 duplicated when threading jumps.
7729 @item max-fields-for-field-sensitive
7730 Maximum number of fields in a structure we will treat in
7731 a field sensitive manner during pointer analysis. The default is zero
7732 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7734 @item prefetch-latency
7735 Estimate on average number of instructions that are executed before
7736 prefetch finishes. The distance we prefetch ahead is proportional
7737 to this constant. Increasing this number may also lead to less
7738 streams being prefetched (see @option{simultaneous-prefetches}).
7740 @item simultaneous-prefetches
7741 Maximum number of prefetches that can run at the same time.
7743 @item l1-cache-line-size
7744 The size of cache line in L1 cache, in bytes.
7747 The size of L1 cache, in kilobytes.
7750 The size of L2 cache, in kilobytes.
7752 @item use-canonical-types
7753 Whether the compiler should use the ``canonical'' type system. By
7754 default, this should always be 1, which uses a more efficient internal
7755 mechanism for comparing types in C++ and Objective-C++. However, if
7756 bugs in the canonical type system are causing compilation failures,
7757 set this value to 0 to disable canonical types.
7759 @item switch-conversion-max-branch-ratio
7760 Switch initialization conversion will refuse to create arrays that are
7761 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7762 branches in the switch.
7764 @item max-partial-antic-length
7765 Maximum length of the partial antic set computed during the tree
7766 partial redundancy elimination optimization (@option{-ftree-pre}) when
7767 optimizing at @option{-O3} and above. For some sorts of source code
7768 the enhanced partial redundancy elimination optimization can run away,
7769 consuming all of the memory available on the host machine. This
7770 parameter sets a limit on the length of the sets that are computed,
7771 which prevents the runaway behavior. Setting a value of 0 for
7772 this parameter will allow an unlimited set length.
7774 @item sccvn-max-scc-size
7775 Maximum size of a strongly connected component (SCC) during SCCVN
7776 processing. If this limit is hit, SCCVN processing for the whole
7777 function will not be done and optimizations depending on it will
7778 be disabled. The default maximum SCC size is 10000.
7780 @item ira-max-loops-num
7781 IRA uses a regional register allocation by default. If a function
7782 contains loops more than number given by the parameter, only at most
7783 given number of the most frequently executed loops will form regions
7784 for the regional register allocation. The default value of the
7787 @item ira-max-conflict-table-size
7788 Although IRA uses a sophisticated algorithm of compression conflict
7789 table, the table can be still big for huge functions. If the conflict
7790 table for a function could be more than size in MB given by the
7791 parameter, the conflict table is not built and faster, simpler, and
7792 lower quality register allocation algorithm will be used. The
7793 algorithm do not use pseudo-register conflicts. The default value of
7794 the parameter is 2000.
7796 @item loop-invariant-max-bbs-in-loop
7797 Loop invariant motion can be very expensive, both in compile time and
7798 in amount of needed compile time memory, with very large loops. Loops
7799 with more basic blocks than this parameter won't have loop invariant
7800 motion optimization performed on them. The default value of the
7801 parameter is 1000 for -O1 and 10000 for -O2 and above.
7806 @node Preprocessor Options
7807 @section Options Controlling the Preprocessor
7808 @cindex preprocessor options
7809 @cindex options, preprocessor
7811 These options control the C preprocessor, which is run on each C source
7812 file before actual compilation.
7814 If you use the @option{-E} option, nothing is done except preprocessing.
7815 Some of these options make sense only together with @option{-E} because
7816 they cause the preprocessor output to be unsuitable for actual
7821 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7822 and pass @var{option} directly through to the preprocessor. If
7823 @var{option} contains commas, it is split into multiple options at the
7824 commas. However, many options are modified, translated or interpreted
7825 by the compiler driver before being passed to the preprocessor, and
7826 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7827 interface is undocumented and subject to change, so whenever possible
7828 you should avoid using @option{-Wp} and let the driver handle the
7831 @item -Xpreprocessor @var{option}
7832 @opindex preprocessor
7833 Pass @var{option} as an option to the preprocessor. You can use this to
7834 supply system-specific preprocessor options which GCC does not know how to
7837 If you want to pass an option that takes an argument, you must use
7838 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7841 @include cppopts.texi
7843 @node Assembler Options
7844 @section Passing Options to the Assembler
7846 @c prevent bad page break with this line
7847 You can pass options to the assembler.
7850 @item -Wa,@var{option}
7852 Pass @var{option} as an option to the assembler. If @var{option}
7853 contains commas, it is split into multiple options at the commas.
7855 @item -Xassembler @var{option}
7857 Pass @var{option} as an option to the assembler. You can use this to
7858 supply system-specific assembler options which GCC does not know how to
7861 If you want to pass an option that takes an argument, you must use
7862 @option{-Xassembler} twice, once for the option and once for the argument.
7867 @section Options for Linking
7868 @cindex link options
7869 @cindex options, linking
7871 These options come into play when the compiler links object files into
7872 an executable output file. They are meaningless if the compiler is
7873 not doing a link step.
7877 @item @var{object-file-name}
7878 A file name that does not end in a special recognized suffix is
7879 considered to name an object file or library. (Object files are
7880 distinguished from libraries by the linker according to the file
7881 contents.) If linking is done, these object files are used as input
7890 If any of these options is used, then the linker is not run, and
7891 object file names should not be used as arguments. @xref{Overall
7895 @item -l@var{library}
7896 @itemx -l @var{library}
7898 Search the library named @var{library} when linking. (The second
7899 alternative with the library as a separate argument is only for
7900 POSIX compliance and is not recommended.)
7902 It makes a difference where in the command you write this option; the
7903 linker searches and processes libraries and object files in the order they
7904 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7905 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7906 to functions in @samp{z}, those functions may not be loaded.
7908 The linker searches a standard list of directories for the library,
7909 which is actually a file named @file{lib@var{library}.a}. The linker
7910 then uses this file as if it had been specified precisely by name.
7912 The directories searched include several standard system directories
7913 plus any that you specify with @option{-L}.
7915 Normally the files found this way are library files---archive files
7916 whose members are object files. The linker handles an archive file by
7917 scanning through it for members which define symbols that have so far
7918 been referenced but not defined. But if the file that is found is an
7919 ordinary object file, it is linked in the usual fashion. The only
7920 difference between using an @option{-l} option and specifying a file name
7921 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7922 and searches several directories.
7926 You need this special case of the @option{-l} option in order to
7927 link an Objective-C or Objective-C++ program.
7930 @opindex nostartfiles
7931 Do not use the standard system startup files when linking.
7932 The standard system libraries are used normally, unless @option{-nostdlib}
7933 or @option{-nodefaultlibs} is used.
7935 @item -nodefaultlibs
7936 @opindex nodefaultlibs
7937 Do not use the standard system libraries when linking.
7938 Only the libraries you specify will be passed to the linker.
7939 The standard startup files are used normally, unless @option{-nostartfiles}
7940 is used. The compiler may generate calls to @code{memcmp},
7941 @code{memset}, @code{memcpy} and @code{memmove}.
7942 These entries are usually resolved by entries in
7943 libc. These entry points should be supplied through some other
7944 mechanism when this option is specified.
7948 Do not use the standard system startup files or libraries when linking.
7949 No startup files and only the libraries you specify will be passed to
7950 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7951 @code{memcpy} and @code{memmove}.
7952 These entries are usually resolved by entries in
7953 libc. These entry points should be supplied through some other
7954 mechanism when this option is specified.
7956 @cindex @option{-lgcc}, use with @option{-nostdlib}
7957 @cindex @option{-nostdlib} and unresolved references
7958 @cindex unresolved references and @option{-nostdlib}
7959 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7960 @cindex @option{-nodefaultlibs} and unresolved references
7961 @cindex unresolved references and @option{-nodefaultlibs}
7962 One of the standard libraries bypassed by @option{-nostdlib} and
7963 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7964 that GCC uses to overcome shortcomings of particular machines, or special
7965 needs for some languages.
7966 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7967 Collection (GCC) Internals},
7968 for more discussion of @file{libgcc.a}.)
7969 In most cases, you need @file{libgcc.a} even when you want to avoid
7970 other standard libraries. In other words, when you specify @option{-nostdlib}
7971 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7972 This ensures that you have no unresolved references to internal GCC
7973 library subroutines. (For example, @samp{__main}, used to ensure C++
7974 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7975 GNU Compiler Collection (GCC) Internals}.)
7979 Produce a position independent executable on targets which support it.
7980 For predictable results, you must also specify the same set of options
7981 that were used to generate code (@option{-fpie}, @option{-fPIE},
7982 or model suboptions) when you specify this option.
7986 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7987 that support it. This instructs the linker to add all symbols, not
7988 only used ones, to the dynamic symbol table. This option is needed
7989 for some uses of @code{dlopen} or to allow obtaining backtraces
7990 from within a program.
7994 Remove all symbol table and relocation information from the executable.
7998 On systems that support dynamic linking, this prevents linking with the shared
7999 libraries. On other systems, this option has no effect.
8003 Produce a shared object which can then be linked with other objects to
8004 form an executable. Not all systems support this option. For predictable
8005 results, you must also specify the same set of options that were used to
8006 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8007 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8008 needs to build supplementary stub code for constructors to work. On
8009 multi-libbed systems, @samp{gcc -shared} must select the correct support
8010 libraries to link against. Failing to supply the correct flags may lead
8011 to subtle defects. Supplying them in cases where they are not necessary
8014 @item -shared-libgcc
8015 @itemx -static-libgcc
8016 @opindex shared-libgcc
8017 @opindex static-libgcc
8018 On systems that provide @file{libgcc} as a shared library, these options
8019 force the use of either the shared or static version respectively.
8020 If no shared version of @file{libgcc} was built when the compiler was
8021 configured, these options have no effect.
8023 There are several situations in which an application should use the
8024 shared @file{libgcc} instead of the static version. The most common
8025 of these is when the application wishes to throw and catch exceptions
8026 across different shared libraries. In that case, each of the libraries
8027 as well as the application itself should use the shared @file{libgcc}.
8029 Therefore, the G++ and GCJ drivers automatically add
8030 @option{-shared-libgcc} whenever you build a shared library or a main
8031 executable, because C++ and Java programs typically use exceptions, so
8032 this is the right thing to do.
8034 If, instead, you use the GCC driver to create shared libraries, you may
8035 find that they will not always be linked with the shared @file{libgcc}.
8036 If GCC finds, at its configuration time, that you have a non-GNU linker
8037 or a GNU linker that does not support option @option{--eh-frame-hdr},
8038 it will link the shared version of @file{libgcc} into shared libraries
8039 by default. Otherwise, it will take advantage of the linker and optimize
8040 away the linking with the shared version of @file{libgcc}, linking with
8041 the static version of libgcc by default. This allows exceptions to
8042 propagate through such shared libraries, without incurring relocation
8043 costs at library load time.
8045 However, if a library or main executable is supposed to throw or catch
8046 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8047 for the languages used in the program, or using the option
8048 @option{-shared-libgcc}, such that it is linked with the shared
8053 Bind references to global symbols when building a shared object. Warn
8054 about any unresolved references (unless overridden by the link editor
8055 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8058 @item -T @var{script}
8060 @cindex linker script
8061 Use @var{script} as the linker script. This option is supported by most
8062 systems using the GNU linker. On some targets, such as bare-board
8063 targets without an operating system, the @option{-T} option may be required
8064 when linking to avoid references to undefined symbols.
8066 @item -Xlinker @var{option}
8068 Pass @var{option} as an option to the linker. You can use this to
8069 supply system-specific linker options which GCC does not know how to
8072 If you want to pass an option that takes a separate argument, you must use
8073 @option{-Xlinker} twice, once for the option and once for the argument.
8074 For example, to pass @option{-assert definitions}, you must write
8075 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8076 @option{-Xlinker "-assert definitions"}, because this passes the entire
8077 string as a single argument, which is not what the linker expects.
8079 When using the GNU linker, it is usually more convenient to pass
8080 arguments to linker options using the @option{@var{option}=@var{value}}
8081 syntax than as separate arguments. For example, you can specify
8082 @samp{-Xlinker -Map=output.map} rather than
8083 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8084 this syntax for command-line options.
8086 @item -Wl,@var{option}
8088 Pass @var{option} as an option to the linker. If @var{option} contains
8089 commas, it is split into multiple options at the commas. You can use this
8090 syntax to pass an argument to the option.
8091 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8092 linker. When using the GNU linker, you can also get the same effect with
8093 @samp{-Wl,-Map=output.map}.
8095 @item -u @var{symbol}
8097 Pretend the symbol @var{symbol} is undefined, to force linking of
8098 library modules to define it. You can use @option{-u} multiple times with
8099 different symbols to force loading of additional library modules.
8102 @node Directory Options
8103 @section Options for Directory Search
8104 @cindex directory options
8105 @cindex options, directory search
8108 These options specify directories to search for header files, for
8109 libraries and for parts of the compiler:
8114 Add the directory @var{dir} to the head of the list of directories to be
8115 searched for header files. This can be used to override a system header
8116 file, substituting your own version, since these directories are
8117 searched before the system header file directories. However, you should
8118 not use this option to add directories that contain vendor-supplied
8119 system header files (use @option{-isystem} for that). If you use more than
8120 one @option{-I} option, the directories are scanned in left-to-right
8121 order; the standard system directories come after.
8123 If a standard system include directory, or a directory specified with
8124 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8125 option will be ignored. The directory will still be searched but as a
8126 system directory at its normal position in the system include chain.
8127 This is to ensure that GCC's procedure to fix buggy system headers and
8128 the ordering for the include_next directive are not inadvertently changed.
8129 If you really need to change the search order for system directories,
8130 use the @option{-nostdinc} and/or @option{-isystem} options.
8132 @item -iquote@var{dir}
8134 Add the directory @var{dir} to the head of the list of directories to
8135 be searched for header files only for the case of @samp{#include
8136 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8137 otherwise just like @option{-I}.
8141 Add directory @var{dir} to the list of directories to be searched
8144 @item -B@var{prefix}
8146 This option specifies where to find the executables, libraries,
8147 include files, and data files of the compiler itself.
8149 The compiler driver program runs one or more of the subprograms
8150 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8151 @var{prefix} as a prefix for each program it tries to run, both with and
8152 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8154 For each subprogram to be run, the compiler driver first tries the
8155 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8156 was not specified, the driver tries two standard prefixes, which are
8157 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8158 those results in a file name that is found, the unmodified program
8159 name is searched for using the directories specified in your
8160 @env{PATH} environment variable.
8162 The compiler will check to see if the path provided by the @option{-B}
8163 refers to a directory, and if necessary it will add a directory
8164 separator character at the end of the path.
8166 @option{-B} prefixes that effectively specify directory names also apply
8167 to libraries in the linker, because the compiler translates these
8168 options into @option{-L} options for the linker. They also apply to
8169 includes files in the preprocessor, because the compiler translates these
8170 options into @option{-isystem} options for the preprocessor. In this case,
8171 the compiler appends @samp{include} to the prefix.
8173 The run-time support file @file{libgcc.a} can also be searched for using
8174 the @option{-B} prefix, if needed. If it is not found there, the two
8175 standard prefixes above are tried, and that is all. The file is left
8176 out of the link if it is not found by those means.
8178 Another way to specify a prefix much like the @option{-B} prefix is to use
8179 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8182 As a special kludge, if the path provided by @option{-B} is
8183 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8184 9, then it will be replaced by @file{[dir/]include}. This is to help
8185 with boot-strapping the compiler.
8187 @item -specs=@var{file}
8189 Process @var{file} after the compiler reads in the standard @file{specs}
8190 file, in order to override the defaults that the @file{gcc} driver
8191 program uses when determining what switches to pass to @file{cc1},
8192 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8193 @option{-specs=@var{file}} can be specified on the command line, and they
8194 are processed in order, from left to right.
8196 @item --sysroot=@var{dir}
8198 Use @var{dir} as the logical root directory for headers and libraries.
8199 For example, if the compiler would normally search for headers in
8200 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8201 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8203 If you use both this option and the @option{-isysroot} option, then
8204 the @option{--sysroot} option will apply to libraries, but the
8205 @option{-isysroot} option will apply to header files.
8207 The GNU linker (beginning with version 2.16) has the necessary support
8208 for this option. If your linker does not support this option, the
8209 header file aspect of @option{--sysroot} will still work, but the
8210 library aspect will not.
8214 This option has been deprecated. Please use @option{-iquote} instead for
8215 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8216 Any directories you specify with @option{-I} options before the @option{-I-}
8217 option are searched only for the case of @samp{#include "@var{file}"};
8218 they are not searched for @samp{#include <@var{file}>}.
8220 If additional directories are specified with @option{-I} options after
8221 the @option{-I-}, these directories are searched for all @samp{#include}
8222 directives. (Ordinarily @emph{all} @option{-I} directories are used
8225 In addition, the @option{-I-} option inhibits the use of the current
8226 directory (where the current input file came from) as the first search
8227 directory for @samp{#include "@var{file}"}. There is no way to
8228 override this effect of @option{-I-}. With @option{-I.} you can specify
8229 searching the directory which was current when the compiler was
8230 invoked. That is not exactly the same as what the preprocessor does
8231 by default, but it is often satisfactory.
8233 @option{-I-} does not inhibit the use of the standard system directories
8234 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8241 @section Specifying subprocesses and the switches to pass to them
8244 @command{gcc} is a driver program. It performs its job by invoking a
8245 sequence of other programs to do the work of compiling, assembling and
8246 linking. GCC interprets its command-line parameters and uses these to
8247 deduce which programs it should invoke, and which command-line options
8248 it ought to place on their command lines. This behavior is controlled
8249 by @dfn{spec strings}. In most cases there is one spec string for each
8250 program that GCC can invoke, but a few programs have multiple spec
8251 strings to control their behavior. The spec strings built into GCC can
8252 be overridden by using the @option{-specs=} command-line switch to specify
8255 @dfn{Spec files} are plaintext files that are used to construct spec
8256 strings. They consist of a sequence of directives separated by blank
8257 lines. The type of directive is determined by the first non-whitespace
8258 character on the line and it can be one of the following:
8261 @item %@var{command}
8262 Issues a @var{command} to the spec file processor. The commands that can
8266 @item %include <@var{file}>
8268 Search for @var{file} and insert its text at the current point in the
8271 @item %include_noerr <@var{file}>
8272 @cindex %include_noerr
8273 Just like @samp{%include}, but do not generate an error message if the include
8274 file cannot be found.
8276 @item %rename @var{old_name} @var{new_name}
8278 Rename the spec string @var{old_name} to @var{new_name}.
8282 @item *[@var{spec_name}]:
8283 This tells the compiler to create, override or delete the named spec
8284 string. All lines after this directive up to the next directive or
8285 blank line are considered to be the text for the spec string. If this
8286 results in an empty string then the spec will be deleted. (Or, if the
8287 spec did not exist, then nothing will happened.) Otherwise, if the spec
8288 does not currently exist a new spec will be created. If the spec does
8289 exist then its contents will be overridden by the text of this
8290 directive, unless the first character of that text is the @samp{+}
8291 character, in which case the text will be appended to the spec.
8293 @item [@var{suffix}]:
8294 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
8295 and up to the next directive or blank line are considered to make up the
8296 spec string for the indicated suffix. When the compiler encounters an
8297 input file with the named suffix, it will processes the spec string in
8298 order to work out how to compile that file. For example:
8305 This says that any input file whose name ends in @samp{.ZZ} should be
8306 passed to the program @samp{z-compile}, which should be invoked with the
8307 command-line switch @option{-input} and with the result of performing the
8308 @samp{%i} substitution. (See below.)
8310 As an alternative to providing a spec string, the text that follows a
8311 suffix directive can be one of the following:
8314 @item @@@var{language}
8315 This says that the suffix is an alias for a known @var{language}. This is
8316 similar to using the @option{-x} command-line switch to GCC to specify a
8317 language explicitly. For example:
8324 Says that .ZZ files are, in fact, C++ source files.
8327 This causes an error messages saying:
8330 @var{name} compiler not installed on this system.
8334 GCC already has an extensive list of suffixes built into it.
8335 This directive will add an entry to the end of the list of suffixes, but
8336 since the list is searched from the end backwards, it is effectively
8337 possible to override earlier entries using this technique.
8341 GCC has the following spec strings built into it. Spec files can
8342 override these strings or create their own. Note that individual
8343 targets can also add their own spec strings to this list.
8346 asm Options to pass to the assembler
8347 asm_final Options to pass to the assembler post-processor
8348 cpp Options to pass to the C preprocessor
8349 cc1 Options to pass to the C compiler
8350 cc1plus Options to pass to the C++ compiler
8351 endfile Object files to include at the end of the link
8352 link Options to pass to the linker
8353 lib Libraries to include on the command line to the linker
8354 libgcc Decides which GCC support library to pass to the linker
8355 linker Sets the name of the linker
8356 predefines Defines to be passed to the C preprocessor
8357 signed_char Defines to pass to CPP to say whether @code{char} is signed
8359 startfile Object files to include at the start of the link
8362 Here is a small example of a spec file:
8368 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
8371 This example renames the spec called @samp{lib} to @samp{old_lib} and
8372 then overrides the previous definition of @samp{lib} with a new one.
8373 The new definition adds in some extra command-line options before
8374 including the text of the old definition.
8376 @dfn{Spec strings} are a list of command-line options to be passed to their
8377 corresponding program. In addition, the spec strings can contain
8378 @samp{%}-prefixed sequences to substitute variable text or to
8379 conditionally insert text into the command line. Using these constructs
8380 it is possible to generate quite complex command lines.
8382 Here is a table of all defined @samp{%}-sequences for spec
8383 strings. Note that spaces are not generated automatically around the
8384 results of expanding these sequences. Therefore you can concatenate them
8385 together or combine them with constant text in a single argument.
8389 Substitute one @samp{%} into the program name or argument.
8392 Substitute the name of the input file being processed.
8395 Substitute the basename of the input file being processed.
8396 This is the substring up to (and not including) the last period
8397 and not including the directory.
8400 This is the same as @samp{%b}, but include the file suffix (text after
8404 Marks the argument containing or following the @samp{%d} as a
8405 temporary file name, so that that file will be deleted if GCC exits
8406 successfully. Unlike @samp{%g}, this contributes no text to the
8409 @item %g@var{suffix}
8410 Substitute a file name that has suffix @var{suffix} and is chosen
8411 once per compilation, and mark the argument in the same way as
8412 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
8413 name is now chosen in a way that is hard to predict even when previously
8414 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
8415 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
8416 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
8417 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
8418 was simply substituted with a file name chosen once per compilation,
8419 without regard to any appended suffix (which was therefore treated
8420 just like ordinary text), making such attacks more likely to succeed.
8422 @item %u@var{suffix}
8423 Like @samp{%g}, but generates a new temporary file name even if
8424 @samp{%u@var{suffix}} was already seen.
8426 @item %U@var{suffix}
8427 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
8428 new one if there is no such last file name. In the absence of any
8429 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
8430 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
8431 would involve the generation of two distinct file names, one
8432 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
8433 simply substituted with a file name chosen for the previous @samp{%u},
8434 without regard to any appended suffix.
8436 @item %j@var{suffix}
8437 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
8438 writable, and if save-temps is off; otherwise, substitute the name
8439 of a temporary file, just like @samp{%u}. This temporary file is not
8440 meant for communication between processes, but rather as a junk
8443 @item %|@var{suffix}
8444 @itemx %m@var{suffix}
8445 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
8446 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
8447 all. These are the two most common ways to instruct a program that it
8448 should read from standard input or write to standard output. If you
8449 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
8450 construct: see for example @file{f/lang-specs.h}.
8452 @item %.@var{SUFFIX}
8453 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
8454 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
8455 terminated by the next space or %.
8458 Marks the argument containing or following the @samp{%w} as the
8459 designated output file of this compilation. This puts the argument
8460 into the sequence of arguments that @samp{%o} will substitute later.
8463 Substitutes the names of all the output files, with spaces
8464 automatically placed around them. You should write spaces
8465 around the @samp{%o} as well or the results are undefined.
8466 @samp{%o} is for use in the specs for running the linker.
8467 Input files whose names have no recognized suffix are not compiled
8468 at all, but they are included among the output files, so they will
8472 Substitutes the suffix for object files. Note that this is
8473 handled specially when it immediately follows @samp{%g, %u, or %U},
8474 because of the need for those to form complete file names. The
8475 handling is such that @samp{%O} is treated exactly as if it had already
8476 been substituted, except that @samp{%g, %u, and %U} do not currently
8477 support additional @var{suffix} characters following @samp{%O} as they would
8478 following, for example, @samp{.o}.
8481 Substitutes the standard macro predefinitions for the
8482 current target machine. Use this when running @code{cpp}.
8485 Like @samp{%p}, but puts @samp{__} before and after the name of each
8486 predefined macro, except for macros that start with @samp{__} or with
8487 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8491 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8492 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8493 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8494 and @option{-imultilib} as necessary.
8497 Current argument is the name of a library or startup file of some sort.
8498 Search for that file in a standard list of directories and substitute
8499 the full name found.
8502 Print @var{str} as an error message. @var{str} is terminated by a newline.
8503 Use this when inconsistent options are detected.
8506 Substitute the contents of spec string @var{name} at this point.
8509 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8511 @item %x@{@var{option}@}
8512 Accumulate an option for @samp{%X}.
8515 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8519 Output the accumulated assembler options specified by @option{-Wa}.
8522 Output the accumulated preprocessor options specified by @option{-Wp}.
8525 Process the @code{asm} spec. This is used to compute the
8526 switches to be passed to the assembler.
8529 Process the @code{asm_final} spec. This is a spec string for
8530 passing switches to an assembler post-processor, if such a program is
8534 Process the @code{link} spec. This is the spec for computing the
8535 command line passed to the linker. Typically it will make use of the
8536 @samp{%L %G %S %D and %E} sequences.
8539 Dump out a @option{-L} option for each directory that GCC believes might
8540 contain startup files. If the target supports multilibs then the
8541 current multilib directory will be prepended to each of these paths.
8544 Process the @code{lib} spec. This is a spec string for deciding which
8545 libraries should be included on the command line to the linker.
8548 Process the @code{libgcc} spec. This is a spec string for deciding
8549 which GCC support library should be included on the command line to the linker.
8552 Process the @code{startfile} spec. This is a spec for deciding which
8553 object files should be the first ones passed to the linker. Typically
8554 this might be a file named @file{crt0.o}.
8557 Process the @code{endfile} spec. This is a spec string that specifies
8558 the last object files that will be passed to the linker.
8561 Process the @code{cpp} spec. This is used to construct the arguments
8562 to be passed to the C preprocessor.
8565 Process the @code{cc1} spec. This is used to construct the options to be
8566 passed to the actual C compiler (@samp{cc1}).
8569 Process the @code{cc1plus} spec. This is used to construct the options to be
8570 passed to the actual C++ compiler (@samp{cc1plus}).
8573 Substitute the variable part of a matched option. See below.
8574 Note that each comma in the substituted string is replaced by
8578 Remove all occurrences of @code{-S} from the command line. Note---this
8579 command is position dependent. @samp{%} commands in the spec string
8580 before this one will see @code{-S}, @samp{%} commands in the spec string
8581 after this one will not.
8583 @item %:@var{function}(@var{args})
8584 Call the named function @var{function}, passing it @var{args}.
8585 @var{args} is first processed as a nested spec string, then split
8586 into an argument vector in the usual fashion. The function returns
8587 a string which is processed as if it had appeared literally as part
8588 of the current spec.
8590 The following built-in spec functions are provided:
8594 The @code{getenv} spec function takes two arguments: an environment
8595 variable name and a string. If the environment variable is not
8596 defined, a fatal error is issued. Otherwise, the return value is the
8597 value of the environment variable concatenated with the string. For
8598 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8601 %:getenv(TOPDIR /include)
8604 expands to @file{/path/to/top/include}.
8606 @item @code{if-exists}
8607 The @code{if-exists} spec function takes one argument, an absolute
8608 pathname to a file. If the file exists, @code{if-exists} returns the
8609 pathname. Here is a small example of its usage:
8613 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8616 @item @code{if-exists-else}
8617 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8618 spec function, except that it takes two arguments. The first argument is
8619 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8620 returns the pathname. If it does not exist, it returns the second argument.
8621 This way, @code{if-exists-else} can be used to select one file or another,
8622 based on the existence of the first. Here is a small example of its usage:
8626 crt0%O%s %:if-exists(crti%O%s) \
8627 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8630 @item @code{replace-outfile}
8631 The @code{replace-outfile} spec function takes two arguments. It looks for the
8632 first argument in the outfiles array and replaces it with the second argument. Here
8633 is a small example of its usage:
8636 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8639 @item @code{print-asm-header}
8640 The @code{print-asm-header} function takes no arguments and simply
8641 prints a banner like:
8647 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8650 It is used to separate compiler options from assembler options
8651 in the @option{--target-help} output.
8655 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8656 If that switch was not specified, this substitutes nothing. Note that
8657 the leading dash is omitted when specifying this option, and it is
8658 automatically inserted if the substitution is performed. Thus the spec
8659 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8660 and would output the command line option @option{-foo}.
8662 @item %W@{@code{S}@}
8663 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8666 @item %@{@code{S}*@}
8667 Substitutes all the switches specified to GCC whose names start
8668 with @code{-S}, but which also take an argument. This is used for
8669 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8670 GCC considers @option{-o foo} as being
8671 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8672 text, including the space. Thus two arguments would be generated.
8674 @item %@{@code{S}*&@code{T}*@}
8675 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8676 (the order of @code{S} and @code{T} in the spec is not significant).
8677 There can be any number of ampersand-separated variables; for each the
8678 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8680 @item %@{@code{S}:@code{X}@}
8681 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8683 @item %@{!@code{S}:@code{X}@}
8684 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8686 @item %@{@code{S}*:@code{X}@}
8687 Substitutes @code{X} if one or more switches whose names start with
8688 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8689 once, no matter how many such switches appeared. However, if @code{%*}
8690 appears somewhere in @code{X}, then @code{X} will be substituted once
8691 for each matching switch, with the @code{%*} replaced by the part of
8692 that switch that matched the @code{*}.
8694 @item %@{.@code{S}:@code{X}@}
8695 Substitutes @code{X}, if processing a file with suffix @code{S}.
8697 @item %@{!.@code{S}:@code{X}@}
8698 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8700 @item %@{,@code{S}:@code{X}@}
8701 Substitutes @code{X}, if processing a file for language @code{S}.
8703 @item %@{!,@code{S}:@code{X}@}
8704 Substitutes @code{X}, if not processing a file for language @code{S}.
8706 @item %@{@code{S}|@code{P}:@code{X}@}
8707 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8708 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8709 @code{*} sequences as well, although they have a stronger binding than
8710 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8711 alternatives must be starred, and only the first matching alternative
8714 For example, a spec string like this:
8717 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8720 will output the following command-line options from the following input
8721 command-line options:
8726 -d fred.c -foo -baz -boggle
8727 -d jim.d -bar -baz -boggle
8730 @item %@{S:X; T:Y; :D@}
8732 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8733 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8734 be as many clauses as you need. This may be combined with @code{.},
8735 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8740 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8741 construct may contain other nested @samp{%} constructs or spaces, or
8742 even newlines. They are processed as usual, as described above.
8743 Trailing white space in @code{X} is ignored. White space may also
8744 appear anywhere on the left side of the colon in these constructs,
8745 except between @code{.} or @code{*} and the corresponding word.
8747 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8748 handled specifically in these constructs. If another value of
8749 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8750 @option{-W} switch is found later in the command line, the earlier
8751 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8752 just one letter, which passes all matching options.
8754 The character @samp{|} at the beginning of the predicate text is used to
8755 indicate that a command should be piped to the following command, but
8756 only if @option{-pipe} is specified.
8758 It is built into GCC which switches take arguments and which do not.
8759 (You might think it would be useful to generalize this to allow each
8760 compiler's spec to say which switches take arguments. But this cannot
8761 be done in a consistent fashion. GCC cannot even decide which input
8762 files have been specified without knowing which switches take arguments,
8763 and it must know which input files to compile in order to tell which
8766 GCC also knows implicitly that arguments starting in @option{-l} are to be
8767 treated as compiler output files, and passed to the linker in their
8768 proper position among the other output files.
8770 @c man begin OPTIONS
8772 @node Target Options
8773 @section Specifying Target Machine and Compiler Version
8774 @cindex target options
8775 @cindex cross compiling
8776 @cindex specifying machine version
8777 @cindex specifying compiler version and target machine
8778 @cindex compiler version, specifying
8779 @cindex target machine, specifying
8781 The usual way to run GCC is to run the executable called @file{gcc}, or
8782 @file{<machine>-gcc} when cross-compiling, or
8783 @file{<machine>-gcc-<version>} to run a version other than the one that
8784 was installed last. Sometimes this is inconvenient, so GCC provides
8785 options that will switch to another cross-compiler or version.
8788 @item -b @var{machine}
8790 The argument @var{machine} specifies the target machine for compilation.
8792 The value to use for @var{machine} is the same as was specified as the
8793 machine type when configuring GCC as a cross-compiler. For
8794 example, if a cross-compiler was configured with @samp{configure
8795 arm-elf}, meaning to compile for an arm processor with elf binaries,
8796 then you would specify @option{-b arm-elf} to run that cross compiler.
8797 Because there are other options beginning with @option{-b}, the
8798 configuration must contain a hyphen, or @option{-b} alone should be one
8799 argument followed by the configuration in the next argument.
8801 @item -V @var{version}
8803 The argument @var{version} specifies which version of GCC to run.
8804 This is useful when multiple versions are installed. For example,
8805 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8808 The @option{-V} and @option{-b} options work by running the
8809 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8810 use them if you can just run that directly.
8812 @node Submodel Options
8813 @section Hardware Models and Configurations
8814 @cindex submodel options
8815 @cindex specifying hardware config
8816 @cindex hardware models and configurations, specifying
8817 @cindex machine dependent options
8819 Earlier we discussed the standard option @option{-b} which chooses among
8820 different installed compilers for completely different target
8821 machines, such as VAX vs.@: 68000 vs.@: 80386.
8823 In addition, each of these target machine types can have its own
8824 special options, starting with @samp{-m}, to choose among various
8825 hardware models or configurations---for example, 68010 vs 68020,
8826 floating coprocessor or none. A single installed version of the
8827 compiler can compile for any model or configuration, according to the
8830 Some configurations of the compiler also support additional special
8831 options, usually for compatibility with other compilers on the same
8834 @c This list is ordered alphanumerically by subsection name.
8835 @c It should be the same order and spelling as these options are listed
8836 @c in Machine Dependent Options
8842 * Blackfin Options::
8846 * DEC Alpha Options::
8847 * DEC Alpha/VMS Options::
8850 * GNU/Linux Options::
8853 * i386 and x86-64 Options::
8854 * i386 and x86-64 Windows Options::
8865 * picoChip Options::
8867 * RS/6000 and PowerPC Options::
8868 * S/390 and zSeries Options::
8873 * System V Options::
8878 * Xstormy16 Options::
8884 @subsection ARC Options
8887 These options are defined for ARC implementations:
8892 Compile code for little endian mode. This is the default.
8896 Compile code for big endian mode.
8899 @opindex mmangle-cpu
8900 Prepend the name of the cpu to all public symbol names.
8901 In multiple-processor systems, there are many ARC variants with different
8902 instruction and register set characteristics. This flag prevents code
8903 compiled for one cpu to be linked with code compiled for another.
8904 No facility exists for handling variants that are ``almost identical''.
8905 This is an all or nothing option.
8907 @item -mcpu=@var{cpu}
8909 Compile code for ARC variant @var{cpu}.
8910 Which variants are supported depend on the configuration.
8911 All variants support @option{-mcpu=base}, this is the default.
8913 @item -mtext=@var{text-section}
8914 @itemx -mdata=@var{data-section}
8915 @itemx -mrodata=@var{readonly-data-section}
8919 Put functions, data, and readonly data in @var{text-section},
8920 @var{data-section}, and @var{readonly-data-section} respectively
8921 by default. This can be overridden with the @code{section} attribute.
8922 @xref{Variable Attributes}.
8924 @item -mfix-cortex-m3-ldrd
8925 @opindex mfix-cortex-m3-ldrd
8926 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
8927 with overlapping destination and base registers are used. This option avoids
8928 generating these instructions. This option is enabled by default when
8929 @option{-mcpu=cortex-m3} is specified.
8934 @subsection ARM Options
8937 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8941 @item -mabi=@var{name}
8943 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8944 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8947 @opindex mapcs-frame
8948 Generate a stack frame that is compliant with the ARM Procedure Call
8949 Standard for all functions, even if this is not strictly necessary for
8950 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8951 with this option will cause the stack frames not to be generated for
8952 leaf functions. The default is @option{-mno-apcs-frame}.
8956 This is a synonym for @option{-mapcs-frame}.
8959 @c not currently implemented
8960 @item -mapcs-stack-check
8961 @opindex mapcs-stack-check
8962 Generate code to check the amount of stack space available upon entry to
8963 every function (that actually uses some stack space). If there is
8964 insufficient space available then either the function
8965 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8966 called, depending upon the amount of stack space required. The run time
8967 system is required to provide these functions. The default is
8968 @option{-mno-apcs-stack-check}, since this produces smaller code.
8970 @c not currently implemented
8972 @opindex mapcs-float
8973 Pass floating point arguments using the float point registers. This is
8974 one of the variants of the APCS@. This option is recommended if the
8975 target hardware has a floating point unit or if a lot of floating point
8976 arithmetic is going to be performed by the code. The default is
8977 @option{-mno-apcs-float}, since integer only code is slightly increased in
8978 size if @option{-mapcs-float} is used.
8980 @c not currently implemented
8981 @item -mapcs-reentrant
8982 @opindex mapcs-reentrant
8983 Generate reentrant, position independent code. The default is
8984 @option{-mno-apcs-reentrant}.
8987 @item -mthumb-interwork
8988 @opindex mthumb-interwork
8989 Generate code which supports calling between the ARM and Thumb
8990 instruction sets. Without this option the two instruction sets cannot
8991 be reliably used inside one program. The default is
8992 @option{-mno-thumb-interwork}, since slightly larger code is generated
8993 when @option{-mthumb-interwork} is specified.
8995 @item -mno-sched-prolog
8996 @opindex mno-sched-prolog
8997 Prevent the reordering of instructions in the function prolog, or the
8998 merging of those instruction with the instructions in the function's
8999 body. This means that all functions will start with a recognizable set
9000 of instructions (or in fact one of a choice from a small set of
9001 different function prologues), and this information can be used to
9002 locate the start if functions inside an executable piece of code. The
9003 default is @option{-msched-prolog}.
9005 @item -mfloat-abi=@var{name}
9007 Specifies which floating-point ABI to use. Permissible values
9008 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9010 Specifying @samp{soft} causes GCC to generate output containing
9011 library calls for floating-point operations.
9012 @samp{softfp} allows the generation of code using hardware floating-point
9013 instructions, but still uses the soft-float calling conventions.
9014 @samp{hard} allows generation of floating-point instructions
9015 and uses FPU-specific calling conventions.
9017 Using @option{-mfloat-abi=hard} with VFP coprocessors is not supported.
9018 Use @option{-mfloat-abi=softfp} with the appropriate @option{-mfpu} option
9019 to allow the compiler to generate code that makes use of the hardware
9020 floating-point capabilities for these CPUs.
9022 The default depends on the specific target configuration. Note that
9023 the hard-float and soft-float ABIs are not link-compatible; you must
9024 compile your entire program with the same ABI, and link with a
9025 compatible set of libraries.
9028 @opindex mhard-float
9029 Equivalent to @option{-mfloat-abi=hard}.
9032 @opindex msoft-float
9033 Equivalent to @option{-mfloat-abi=soft}.
9035 @item -mlittle-endian
9036 @opindex mlittle-endian
9037 Generate code for a processor running in little-endian mode. This is
9038 the default for all standard configurations.
9041 @opindex mbig-endian
9042 Generate code for a processor running in big-endian mode; the default is
9043 to compile code for a little-endian processor.
9045 @item -mwords-little-endian
9046 @opindex mwords-little-endian
9047 This option only applies when generating code for big-endian processors.
9048 Generate code for a little-endian word order but a big-endian byte
9049 order. That is, a byte order of the form @samp{32107654}. Note: this
9050 option should only be used if you require compatibility with code for
9051 big-endian ARM processors generated by versions of the compiler prior to
9054 @item -mcpu=@var{name}
9056 This specifies the name of the target ARM processor. GCC uses this name
9057 to determine what kind of instructions it can emit when generating
9058 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9059 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9060 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9061 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9062 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9064 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9065 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9066 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9067 @samp{strongarm1110},
9068 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9069 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9070 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9071 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9072 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9073 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9074 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9075 @samp{cortex-a8}, @samp{cortex-a9},
9076 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9078 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9080 @item -mtune=@var{name}
9082 This option is very similar to the @option{-mcpu=} option, except that
9083 instead of specifying the actual target processor type, and hence
9084 restricting which instructions can be used, it specifies that GCC should
9085 tune the performance of the code as if the target were of the type
9086 specified in this option, but still choosing the instructions that it
9087 will generate based on the cpu specified by a @option{-mcpu=} option.
9088 For some ARM implementations better performance can be obtained by using
9091 @item -march=@var{name}
9093 This specifies the name of the target ARM architecture. GCC uses this
9094 name to determine what kind of instructions it can emit when generating
9095 assembly code. This option can be used in conjunction with or instead
9096 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9097 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9098 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9099 @samp{armv6}, @samp{armv6j},
9100 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9101 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9102 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9104 @item -mfpu=@var{name}
9105 @itemx -mfpe=@var{number}
9106 @itemx -mfp=@var{number}
9110 This specifies what floating point hardware (or hardware emulation) is
9111 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9112 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16} and
9113 @samp{neon}. @option{-mfp} and @option{-mfpe}
9114 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9115 with older versions of GCC@.
9117 If @option{-msoft-float} is specified this specifies the format of
9118 floating point values.
9120 @item -mstructure-size-boundary=@var{n}
9121 @opindex mstructure-size-boundary
9122 The size of all structures and unions will be rounded up to a multiple
9123 of the number of bits set by this option. Permissible values are 8, 32
9124 and 64. The default value varies for different toolchains. For the COFF
9125 targeted toolchain the default value is 8. A value of 64 is only allowed
9126 if the underlying ABI supports it.
9128 Specifying the larger number can produce faster, more efficient code, but
9129 can also increase the size of the program. Different values are potentially
9130 incompatible. Code compiled with one value cannot necessarily expect to
9131 work with code or libraries compiled with another value, if they exchange
9132 information using structures or unions.
9134 @item -mabort-on-noreturn
9135 @opindex mabort-on-noreturn
9136 Generate a call to the function @code{abort} at the end of a
9137 @code{noreturn} function. It will be executed if the function tries to
9141 @itemx -mno-long-calls
9142 @opindex mlong-calls
9143 @opindex mno-long-calls
9144 Tells the compiler to perform function calls by first loading the
9145 address of the function into a register and then performing a subroutine
9146 call on this register. This switch is needed if the target function
9147 will lie outside of the 64 megabyte addressing range of the offset based
9148 version of subroutine call instruction.
9150 Even if this switch is enabled, not all function calls will be turned
9151 into long calls. The heuristic is that static functions, functions
9152 which have the @samp{short-call} attribute, functions that are inside
9153 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9154 definitions have already been compiled within the current compilation
9155 unit, will not be turned into long calls. The exception to this rule is
9156 that weak function definitions, functions with the @samp{long-call}
9157 attribute or the @samp{section} attribute, and functions that are within
9158 the scope of a @samp{#pragma long_calls} directive, will always be
9159 turned into long calls.
9161 This feature is not enabled by default. Specifying
9162 @option{-mno-long-calls} will restore the default behavior, as will
9163 placing the function calls within the scope of a @samp{#pragma
9164 long_calls_off} directive. Note these switches have no effect on how
9165 the compiler generates code to handle function calls via function
9168 @item -msingle-pic-base
9169 @opindex msingle-pic-base
9170 Treat the register used for PIC addressing as read-only, rather than
9171 loading it in the prologue for each function. The run-time system is
9172 responsible for initializing this register with an appropriate value
9173 before execution begins.
9175 @item -mpic-register=@var{reg}
9176 @opindex mpic-register
9177 Specify the register to be used for PIC addressing. The default is R10
9178 unless stack-checking is enabled, when R9 is used.
9180 @item -mcirrus-fix-invalid-insns
9181 @opindex mcirrus-fix-invalid-insns
9182 @opindex mno-cirrus-fix-invalid-insns
9183 Insert NOPs into the instruction stream to in order to work around
9184 problems with invalid Maverick instruction combinations. This option
9185 is only valid if the @option{-mcpu=ep9312} option has been used to
9186 enable generation of instructions for the Cirrus Maverick floating
9187 point co-processor. This option is not enabled by default, since the
9188 problem is only present in older Maverick implementations. The default
9189 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9192 @item -mpoke-function-name
9193 @opindex mpoke-function-name
9194 Write the name of each function into the text section, directly
9195 preceding the function prologue. The generated code is similar to this:
9199 .ascii "arm_poke_function_name", 0
9202 .word 0xff000000 + (t1 - t0)
9203 arm_poke_function_name
9205 stmfd sp!, @{fp, ip, lr, pc@}
9209 When performing a stack backtrace, code can inspect the value of
9210 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9211 location @code{pc - 12} and the top 8 bits are set, then we know that
9212 there is a function name embedded immediately preceding this location
9213 and has length @code{((pc[-3]) & 0xff000000)}.
9217 Generate code for the Thumb instruction set. The default is to
9218 use the 32-bit ARM instruction set.
9219 This option automatically enables either 16-bit Thumb-1 or
9220 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9221 and @option{-march=@var{name}} options.
9224 @opindex mtpcs-frame
9225 Generate a stack frame that is compliant with the Thumb Procedure Call
9226 Standard for all non-leaf functions. (A leaf function is one that does
9227 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9229 @item -mtpcs-leaf-frame
9230 @opindex mtpcs-leaf-frame
9231 Generate a stack frame that is compliant with the Thumb Procedure Call
9232 Standard for all leaf functions. (A leaf function is one that does
9233 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9235 @item -mcallee-super-interworking
9236 @opindex mcallee-super-interworking
9237 Gives all externally visible functions in the file being compiled an ARM
9238 instruction set header which switches to Thumb mode before executing the
9239 rest of the function. This allows these functions to be called from
9240 non-interworking code.
9242 @item -mcaller-super-interworking
9243 @opindex mcaller-super-interworking
9244 Allows calls via function pointers (including virtual functions) to
9245 execute correctly regardless of whether the target code has been
9246 compiled for interworking or not. There is a small overhead in the cost
9247 of executing a function pointer if this option is enabled.
9249 @item -mtp=@var{name}
9251 Specify the access model for the thread local storage pointer. The valid
9252 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9253 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9254 (supported in the arm6k architecture), and @option{auto}, which uses the
9255 best available method for the selected processor. The default setting is
9258 @item -mword-relocations
9259 @opindex mword-relocations
9260 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9261 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9262 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9268 @subsection AVR Options
9271 These options are defined for AVR implementations:
9274 @item -mmcu=@var{mcu}
9276 Specify ATMEL AVR instruction set or MCU type.
9278 Instruction set avr1 is for the minimal AVR core, not supported by the C
9279 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9280 attiny11, attiny12, attiny15, attiny28).
9282 Instruction set avr2 (default) is for the classic AVR core with up to
9283 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9284 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9285 at90c8534, at90s8535).
9287 Instruction set avr3 is for the classic AVR core with up to 128K program
9288 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9290 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9291 memory space (MCU types: atmega8, atmega83, atmega85).
9293 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9294 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9295 atmega64, atmega128, at43usb355, at94k).
9299 Output instruction sizes to the asm file.
9301 @item -mno-interrupts
9302 @opindex mno-interrupts
9303 Generated code is not compatible with hardware interrupts.
9304 Code size will be smaller.
9306 @item -mcall-prologues
9307 @opindex mcall-prologues
9308 Functions prologues/epilogues expanded as call to appropriate
9309 subroutines. Code size will be smaller.
9311 @item -mno-tablejump
9312 @opindex mno-tablejump
9313 Do not generate tablejump insns which sometimes increase code size.
9314 The option is now deprecated in favor of the equivalent
9315 @option{-fno-jump-tables}
9318 @opindex mtiny-stack
9319 Change only the low 8 bits of the stack pointer.
9323 Assume int to be 8 bit integer. This affects the sizes of all types: A
9324 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
9325 and long long will be 4 bytes. Please note that this option does not
9326 comply to the C standards, but it will provide you with smaller code
9330 @node Blackfin Options
9331 @subsection Blackfin Options
9332 @cindex Blackfin Options
9335 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
9337 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
9338 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
9339 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
9340 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
9341 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
9342 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
9344 The optional @var{sirevision} specifies the silicon revision of the target
9345 Blackfin processor. Any workarounds available for the targeted silicon revision
9346 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
9347 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
9348 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
9349 hexadecimal digits representing the major and minor numbers in the silicon
9350 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
9351 is not defined. If @var{sirevision} is @samp{any}, the
9352 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
9353 If this optional @var{sirevision} is not used, GCC assumes the latest known
9354 silicon revision of the targeted Blackfin processor.
9356 Support for @samp{bf561} is incomplete. For @samp{bf561},
9357 Only the processor macro is defined.
9358 Without this option, @samp{bf532} is used as the processor by default.
9359 The corresponding predefined processor macros for @var{cpu} is to
9360 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
9361 provided by libgloss to be linked in if @option{-msim} is not given.
9365 Specifies that the program will be run on the simulator. This causes
9366 the simulator BSP provided by libgloss to be linked in. This option
9367 has effect only for @samp{bfin-elf} toolchain.
9368 Certain other options, such as @option{-mid-shared-library} and
9369 @option{-mfdpic}, imply @option{-msim}.
9371 @item -momit-leaf-frame-pointer
9372 @opindex momit-leaf-frame-pointer
9373 Don't keep the frame pointer in a register for leaf functions. This
9374 avoids the instructions to save, set up and restore frame pointers and
9375 makes an extra register available in leaf functions. The option
9376 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9377 which might make debugging harder.
9379 @item -mspecld-anomaly
9380 @opindex mspecld-anomaly
9381 When enabled, the compiler will ensure that the generated code does not
9382 contain speculative loads after jump instructions. If this option is used,
9383 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
9385 @item -mno-specld-anomaly
9386 @opindex mno-specld-anomaly
9387 Don't generate extra code to prevent speculative loads from occurring.
9389 @item -mcsync-anomaly
9390 @opindex mcsync-anomaly
9391 When enabled, the compiler will ensure that the generated code does not
9392 contain CSYNC or SSYNC instructions too soon after conditional branches.
9393 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
9395 @item -mno-csync-anomaly
9396 @opindex mno-csync-anomaly
9397 Don't generate extra code to prevent CSYNC or SSYNC instructions from
9398 occurring too soon after a conditional branch.
9402 When enabled, the compiler is free to take advantage of the knowledge that
9403 the entire program fits into the low 64k of memory.
9406 @opindex mno-low-64k
9407 Assume that the program is arbitrarily large. This is the default.
9409 @item -mstack-check-l1
9410 @opindex mstack-check-l1
9411 Do stack checking using information placed into L1 scratchpad memory by the
9414 @item -mid-shared-library
9415 @opindex mid-shared-library
9416 Generate code that supports shared libraries via the library ID method.
9417 This allows for execute in place and shared libraries in an environment
9418 without virtual memory management. This option implies @option{-fPIC}.
9419 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9421 @item -mno-id-shared-library
9422 @opindex mno-id-shared-library
9423 Generate code that doesn't assume ID based shared libraries are being used.
9424 This is the default.
9426 @item -mleaf-id-shared-library
9427 @opindex mleaf-id-shared-library
9428 Generate code that supports shared libraries via the library ID method,
9429 but assumes that this library or executable won't link against any other
9430 ID shared libraries. That allows the compiler to use faster code for jumps
9433 @item -mno-leaf-id-shared-library
9434 @opindex mno-leaf-id-shared-library
9435 Do not assume that the code being compiled won't link against any ID shared
9436 libraries. Slower code will be generated for jump and call insns.
9438 @item -mshared-library-id=n
9439 @opindex mshared-library-id
9440 Specified the identification number of the ID based shared library being
9441 compiled. Specifying a value of 0 will generate more compact code, specifying
9442 other values will force the allocation of that number to the current
9443 library but is no more space or time efficient than omitting this option.
9447 Generate code that allows the data segment to be located in a different
9448 area of memory from the text segment. This allows for execute in place in
9449 an environment without virtual memory management by eliminating relocations
9450 against the text section.
9453 @opindex mno-sep-data
9454 Generate code that assumes that the data segment follows the text segment.
9455 This is the default.
9458 @itemx -mno-long-calls
9459 @opindex mlong-calls
9460 @opindex mno-long-calls
9461 Tells the compiler to perform function calls by first loading the
9462 address of the function into a register and then performing a subroutine
9463 call on this register. This switch is needed if the target function
9464 will lie outside of the 24 bit addressing range of the offset based
9465 version of subroutine call instruction.
9467 This feature is not enabled by default. Specifying
9468 @option{-mno-long-calls} will restore the default behavior. Note these
9469 switches have no effect on how the compiler generates code to handle
9470 function calls via function pointers.
9474 Link with the fast floating-point library. This library relaxes some of
9475 the IEEE floating-point standard's rules for checking inputs against
9476 Not-a-Number (NAN), in the interest of performance.
9479 @opindex minline-plt
9480 Enable inlining of PLT entries in function calls to functions that are
9481 not known to bind locally. It has no effect without @option{-mfdpic}.
9485 Build standalone application for multicore Blackfin processor. Proper
9486 start files and link scripts will be used to support multicore.
9487 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9488 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9489 @option{-mcorea} or @option{-mcoreb}. If it's used without
9490 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9491 programming model is used. In this model, the main function of Core B
9492 should be named as coreb_main. If it's used with @option{-mcorea} or
9493 @option{-mcoreb}, one application per core programming model is used.
9494 If this option is not used, single core application programming
9499 Build standalone application for Core A of BF561 when using
9500 one application per core programming model. Proper start files
9501 and link scripts will be used to support Core A. This option
9502 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9506 Build standalone application for Core B of BF561 when using
9507 one application per core programming model. Proper start files
9508 and link scripts will be used to support Core B. This option
9509 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9510 should be used instead of main. It must be used with
9511 @option{-mmulticore}.
9515 Build standalone application for SDRAM. Proper start files and
9516 link scripts will be used to put the application into SDRAM.
9517 Loader should initialize SDRAM before loading the application
9518 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9522 Assume that ICPLBs are enabled at runtime. This has an effect on certain
9523 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
9524 are enabled; for standalone applications the default is off.
9528 @subsection CRIS Options
9529 @cindex CRIS Options
9531 These options are defined specifically for the CRIS ports.
9534 @item -march=@var{architecture-type}
9535 @itemx -mcpu=@var{architecture-type}
9538 Generate code for the specified architecture. The choices for
9539 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9540 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9541 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9544 @item -mtune=@var{architecture-type}
9546 Tune to @var{architecture-type} everything applicable about the generated
9547 code, except for the ABI and the set of available instructions. The
9548 choices for @var{architecture-type} are the same as for
9549 @option{-march=@var{architecture-type}}.
9551 @item -mmax-stack-frame=@var{n}
9552 @opindex mmax-stack-frame
9553 Warn when the stack frame of a function exceeds @var{n} bytes.
9559 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9560 @option{-march=v3} and @option{-march=v8} respectively.
9562 @item -mmul-bug-workaround
9563 @itemx -mno-mul-bug-workaround
9564 @opindex mmul-bug-workaround
9565 @opindex mno-mul-bug-workaround
9566 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9567 models where it applies. This option is active by default.
9571 Enable CRIS-specific verbose debug-related information in the assembly
9572 code. This option also has the effect to turn off the @samp{#NO_APP}
9573 formatted-code indicator to the assembler at the beginning of the
9578 Do not use condition-code results from previous instruction; always emit
9579 compare and test instructions before use of condition codes.
9581 @item -mno-side-effects
9582 @opindex mno-side-effects
9583 Do not emit instructions with side-effects in addressing modes other than
9587 @itemx -mno-stack-align
9589 @itemx -mno-data-align
9590 @itemx -mconst-align
9591 @itemx -mno-const-align
9592 @opindex mstack-align
9593 @opindex mno-stack-align
9594 @opindex mdata-align
9595 @opindex mno-data-align
9596 @opindex mconst-align
9597 @opindex mno-const-align
9598 These options (no-options) arranges (eliminate arrangements) for the
9599 stack-frame, individual data and constants to be aligned for the maximum
9600 single data access size for the chosen CPU model. The default is to
9601 arrange for 32-bit alignment. ABI details such as structure layout are
9602 not affected by these options.
9610 Similar to the stack- data- and const-align options above, these options
9611 arrange for stack-frame, writable data and constants to all be 32-bit,
9612 16-bit or 8-bit aligned. The default is 32-bit alignment.
9614 @item -mno-prologue-epilogue
9615 @itemx -mprologue-epilogue
9616 @opindex mno-prologue-epilogue
9617 @opindex mprologue-epilogue
9618 With @option{-mno-prologue-epilogue}, the normal function prologue and
9619 epilogue that sets up the stack-frame are omitted and no return
9620 instructions or return sequences are generated in the code. Use this
9621 option only together with visual inspection of the compiled code: no
9622 warnings or errors are generated when call-saved registers must be saved,
9623 or storage for local variable needs to be allocated.
9629 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9630 instruction sequences that load addresses for functions from the PLT part
9631 of the GOT rather than (traditional on other architectures) calls to the
9632 PLT@. The default is @option{-mgotplt}.
9636 Legacy no-op option only recognized with the cris-axis-elf and
9637 cris-axis-linux-gnu targets.
9641 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9645 This option, recognized for the cris-axis-elf arranges
9646 to link with input-output functions from a simulator library. Code,
9647 initialized data and zero-initialized data are allocated consecutively.
9651 Like @option{-sim}, but pass linker options to locate initialized data at
9652 0x40000000 and zero-initialized data at 0x80000000.
9656 @subsection CRX Options
9659 These options are defined specifically for the CRX ports.
9665 Enable the use of multiply-accumulate instructions. Disabled by default.
9669 Push instructions will be used to pass outgoing arguments when functions
9670 are called. Enabled by default.
9673 @node Darwin Options
9674 @subsection Darwin Options
9675 @cindex Darwin options
9677 These options are defined for all architectures running the Darwin operating
9680 FSF GCC on Darwin does not create ``fat'' object files; it will create
9681 an object file for the single architecture that it was built to
9682 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9683 @option{-arch} options are used; it does so by running the compiler or
9684 linker multiple times and joining the results together with
9687 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9688 @samp{i686}) is determined by the flags that specify the ISA
9689 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9690 @option{-force_cpusubtype_ALL} option can be used to override this.
9692 The Darwin tools vary in their behavior when presented with an ISA
9693 mismatch. The assembler, @file{as}, will only permit instructions to
9694 be used that are valid for the subtype of the file it is generating,
9695 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9696 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9697 and print an error if asked to create a shared library with a less
9698 restrictive subtype than its input files (for instance, trying to put
9699 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9700 for executables, @file{ld}, will quietly give the executable the most
9701 restrictive subtype of any of its input files.
9706 Add the framework directory @var{dir} to the head of the list of
9707 directories to be searched for header files. These directories are
9708 interleaved with those specified by @option{-I} options and are
9709 scanned in a left-to-right order.
9711 A framework directory is a directory with frameworks in it. A
9712 framework is a directory with a @samp{"Headers"} and/or
9713 @samp{"PrivateHeaders"} directory contained directly in it that ends
9714 in @samp{".framework"}. The name of a framework is the name of this
9715 directory excluding the @samp{".framework"}. Headers associated with
9716 the framework are found in one of those two directories, with
9717 @samp{"Headers"} being searched first. A subframework is a framework
9718 directory that is in a framework's @samp{"Frameworks"} directory.
9719 Includes of subframework headers can only appear in a header of a
9720 framework that contains the subframework, or in a sibling subframework
9721 header. Two subframeworks are siblings if they occur in the same
9722 framework. A subframework should not have the same name as a
9723 framework, a warning will be issued if this is violated. Currently a
9724 subframework cannot have subframeworks, in the future, the mechanism
9725 may be extended to support this. The standard frameworks can be found
9726 in @samp{"/System/Library/Frameworks"} and
9727 @samp{"/Library/Frameworks"}. An example include looks like
9728 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9729 the name of the framework and header.h is found in the
9730 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9732 @item -iframework@var{dir}
9734 Like @option{-F} except the directory is a treated as a system
9735 directory. The main difference between this @option{-iframework} and
9736 @option{-F} is that with @option{-iframework} the compiler does not
9737 warn about constructs contained within header files found via
9738 @var{dir}. This option is valid only for the C family of languages.
9742 Emit debugging information for symbols that are used. For STABS
9743 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9744 This is by default ON@.
9748 Emit debugging information for all symbols and types.
9750 @item -mmacosx-version-min=@var{version}
9751 The earliest version of MacOS X that this executable will run on
9752 is @var{version}. Typical values of @var{version} include @code{10.1},
9753 @code{10.2}, and @code{10.3.9}.
9755 If the compiler was built to use the system's headers by default,
9756 then the default for this option is the system version on which the
9757 compiler is running, otherwise the default is to make choices which
9758 are compatible with as many systems and code bases as possible.
9762 Enable kernel development mode. The @option{-mkernel} option sets
9763 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9764 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9765 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9766 applicable. This mode also sets @option{-mno-altivec},
9767 @option{-msoft-float}, @option{-fno-builtin} and
9768 @option{-mlong-branch} for PowerPC targets.
9770 @item -mone-byte-bool
9771 @opindex mone-byte-bool
9772 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9773 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9774 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9775 option has no effect on x86.
9777 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9778 to generate code that is not binary compatible with code generated
9779 without that switch. Using this switch may require recompiling all
9780 other modules in a program, including system libraries. Use this
9781 switch to conform to a non-default data model.
9783 @item -mfix-and-continue
9784 @itemx -ffix-and-continue
9785 @itemx -findirect-data
9786 @opindex mfix-and-continue
9787 @opindex ffix-and-continue
9788 @opindex findirect-data
9789 Generate code suitable for fast turn around development. Needed to
9790 enable gdb to dynamically load @code{.o} files into already running
9791 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9792 are provided for backwards compatibility.
9796 Loads all members of static archive libraries.
9797 See man ld(1) for more information.
9799 @item -arch_errors_fatal
9800 @opindex arch_errors_fatal
9801 Cause the errors having to do with files that have the wrong architecture
9805 @opindex bind_at_load
9806 Causes the output file to be marked such that the dynamic linker will
9807 bind all undefined references when the file is loaded or launched.
9811 Produce a Mach-o bundle format file.
9812 See man ld(1) for more information.
9814 @item -bundle_loader @var{executable}
9815 @opindex bundle_loader
9816 This option specifies the @var{executable} that will be loading the build
9817 output file being linked. See man ld(1) for more information.
9821 When passed this option, GCC will produce a dynamic library instead of
9822 an executable when linking, using the Darwin @file{libtool} command.
9824 @item -force_cpusubtype_ALL
9825 @opindex force_cpusubtype_ALL
9826 This causes GCC's output file to have the @var{ALL} subtype, instead of
9827 one controlled by the @option{-mcpu} or @option{-march} option.
9829 @item -allowable_client @var{client_name}
9831 @itemx -compatibility_version
9832 @itemx -current_version
9834 @itemx -dependency-file
9836 @itemx -dylinker_install_name
9838 @itemx -exported_symbols_list
9840 @itemx -flat_namespace
9841 @itemx -force_flat_namespace
9842 @itemx -headerpad_max_install_names
9845 @itemx -install_name
9846 @itemx -keep_private_externs
9847 @itemx -multi_module
9848 @itemx -multiply_defined
9849 @itemx -multiply_defined_unused
9851 @itemx -no_dead_strip_inits_and_terms
9852 @itemx -nofixprebinding
9855 @itemx -noseglinkedit
9856 @itemx -pagezero_size
9858 @itemx -prebind_all_twolevel_modules
9859 @itemx -private_bundle
9860 @itemx -read_only_relocs
9862 @itemx -sectobjectsymbols
9866 @itemx -sectobjectsymbols
9869 @itemx -segs_read_only_addr
9870 @itemx -segs_read_write_addr
9871 @itemx -seg_addr_table
9872 @itemx -seg_addr_table_filename
9875 @itemx -segs_read_only_addr
9876 @itemx -segs_read_write_addr
9877 @itemx -single_module
9880 @itemx -sub_umbrella
9881 @itemx -twolevel_namespace
9884 @itemx -unexported_symbols_list
9885 @itemx -weak_reference_mismatches
9887 @opindex allowable_client
9888 @opindex client_name
9889 @opindex compatibility_version
9890 @opindex current_version
9892 @opindex dependency-file
9894 @opindex dylinker_install_name
9896 @opindex exported_symbols_list
9898 @opindex flat_namespace
9899 @opindex force_flat_namespace
9900 @opindex headerpad_max_install_names
9903 @opindex install_name
9904 @opindex keep_private_externs
9905 @opindex multi_module
9906 @opindex multiply_defined
9907 @opindex multiply_defined_unused
9909 @opindex no_dead_strip_inits_and_terms
9910 @opindex nofixprebinding
9911 @opindex nomultidefs
9913 @opindex noseglinkedit
9914 @opindex pagezero_size
9916 @opindex prebind_all_twolevel_modules
9917 @opindex private_bundle
9918 @opindex read_only_relocs
9920 @opindex sectobjectsymbols
9924 @opindex sectobjectsymbols
9927 @opindex segs_read_only_addr
9928 @opindex segs_read_write_addr
9929 @opindex seg_addr_table
9930 @opindex seg_addr_table_filename
9931 @opindex seglinkedit
9933 @opindex segs_read_only_addr
9934 @opindex segs_read_write_addr
9935 @opindex single_module
9937 @opindex sub_library
9938 @opindex sub_umbrella
9939 @opindex twolevel_namespace
9942 @opindex unexported_symbols_list
9943 @opindex weak_reference_mismatches
9944 @opindex whatsloaded
9945 These options are passed to the Darwin linker. The Darwin linker man page
9946 describes them in detail.
9949 @node DEC Alpha Options
9950 @subsection DEC Alpha Options
9952 These @samp{-m} options are defined for the DEC Alpha implementations:
9955 @item -mno-soft-float
9957 @opindex mno-soft-float
9958 @opindex msoft-float
9959 Use (do not use) the hardware floating-point instructions for
9960 floating-point operations. When @option{-msoft-float} is specified,
9961 functions in @file{libgcc.a} will be used to perform floating-point
9962 operations. Unless they are replaced by routines that emulate the
9963 floating-point operations, or compiled in such a way as to call such
9964 emulations routines, these routines will issue floating-point
9965 operations. If you are compiling for an Alpha without floating-point
9966 operations, you must ensure that the library is built so as not to call
9969 Note that Alpha implementations without floating-point operations are
9970 required to have floating-point registers.
9975 @opindex mno-fp-regs
9976 Generate code that uses (does not use) the floating-point register set.
9977 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9978 register set is not used, floating point operands are passed in integer
9979 registers as if they were integers and floating-point results are passed
9980 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9981 so any function with a floating-point argument or return value called by code
9982 compiled with @option{-mno-fp-regs} must also be compiled with that
9985 A typical use of this option is building a kernel that does not use,
9986 and hence need not save and restore, any floating-point registers.
9990 The Alpha architecture implements floating-point hardware optimized for
9991 maximum performance. It is mostly compliant with the IEEE floating
9992 point standard. However, for full compliance, software assistance is
9993 required. This option generates code fully IEEE compliant code
9994 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9995 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9996 defined during compilation. The resulting code is less efficient but is
9997 able to correctly support denormalized numbers and exceptional IEEE
9998 values such as not-a-number and plus/minus infinity. Other Alpha
9999 compilers call this option @option{-ieee_with_no_inexact}.
10001 @item -mieee-with-inexact
10002 @opindex mieee-with-inexact
10003 This is like @option{-mieee} except the generated code also maintains
10004 the IEEE @var{inexact-flag}. Turning on this option causes the
10005 generated code to implement fully-compliant IEEE math. In addition to
10006 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10007 macro. On some Alpha implementations the resulting code may execute
10008 significantly slower than the code generated by default. Since there is
10009 very little code that depends on the @var{inexact-flag}, you should
10010 normally not specify this option. Other Alpha compilers call this
10011 option @option{-ieee_with_inexact}.
10013 @item -mfp-trap-mode=@var{trap-mode}
10014 @opindex mfp-trap-mode
10015 This option controls what floating-point related traps are enabled.
10016 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10017 The trap mode can be set to one of four values:
10021 This is the default (normal) setting. The only traps that are enabled
10022 are the ones that cannot be disabled in software (e.g., division by zero
10026 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10030 Like @samp{u}, but the instructions are marked to be safe for software
10031 completion (see Alpha architecture manual for details).
10034 Like @samp{su}, but inexact traps are enabled as well.
10037 @item -mfp-rounding-mode=@var{rounding-mode}
10038 @opindex mfp-rounding-mode
10039 Selects the IEEE rounding mode. Other Alpha compilers call this option
10040 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10045 Normal IEEE rounding mode. Floating point numbers are rounded towards
10046 the nearest machine number or towards the even machine number in case
10050 Round towards minus infinity.
10053 Chopped rounding mode. Floating point numbers are rounded towards zero.
10056 Dynamic rounding mode. A field in the floating point control register
10057 (@var{fpcr}, see Alpha architecture reference manual) controls the
10058 rounding mode in effect. The C library initializes this register for
10059 rounding towards plus infinity. Thus, unless your program modifies the
10060 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10063 @item -mtrap-precision=@var{trap-precision}
10064 @opindex mtrap-precision
10065 In the Alpha architecture, floating point traps are imprecise. This
10066 means without software assistance it is impossible to recover from a
10067 floating trap and program execution normally needs to be terminated.
10068 GCC can generate code that can assist operating system trap handlers
10069 in determining the exact location that caused a floating point trap.
10070 Depending on the requirements of an application, different levels of
10071 precisions can be selected:
10075 Program precision. This option is the default and means a trap handler
10076 can only identify which program caused a floating point exception.
10079 Function precision. The trap handler can determine the function that
10080 caused a floating point exception.
10083 Instruction precision. The trap handler can determine the exact
10084 instruction that caused a floating point exception.
10087 Other Alpha compilers provide the equivalent options called
10088 @option{-scope_safe} and @option{-resumption_safe}.
10090 @item -mieee-conformant
10091 @opindex mieee-conformant
10092 This option marks the generated code as IEEE conformant. You must not
10093 use this option unless you also specify @option{-mtrap-precision=i} and either
10094 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10095 is to emit the line @samp{.eflag 48} in the function prologue of the
10096 generated assembly file. Under DEC Unix, this has the effect that
10097 IEEE-conformant math library routines will be linked in.
10099 @item -mbuild-constants
10100 @opindex mbuild-constants
10101 Normally GCC examines a 32- or 64-bit integer constant to
10102 see if it can construct it from smaller constants in two or three
10103 instructions. If it cannot, it will output the constant as a literal and
10104 generate code to load it from the data segment at runtime.
10106 Use this option to require GCC to construct @emph{all} integer constants
10107 using code, even if it takes more instructions (the maximum is six).
10109 You would typically use this option to build a shared library dynamic
10110 loader. Itself a shared library, it must relocate itself in memory
10111 before it can find the variables and constants in its own data segment.
10117 Select whether to generate code to be assembled by the vendor-supplied
10118 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10136 Indicate whether GCC should generate code to use the optional BWX,
10137 CIX, FIX and MAX instruction sets. The default is to use the instruction
10138 sets supported by the CPU type specified via @option{-mcpu=} option or that
10139 of the CPU on which GCC was built if none was specified.
10142 @itemx -mfloat-ieee
10143 @opindex mfloat-vax
10144 @opindex mfloat-ieee
10145 Generate code that uses (does not use) VAX F and G floating point
10146 arithmetic instead of IEEE single and double precision.
10148 @item -mexplicit-relocs
10149 @itemx -mno-explicit-relocs
10150 @opindex mexplicit-relocs
10151 @opindex mno-explicit-relocs
10152 Older Alpha assemblers provided no way to generate symbol relocations
10153 except via assembler macros. Use of these macros does not allow
10154 optimal instruction scheduling. GNU binutils as of version 2.12
10155 supports a new syntax that allows the compiler to explicitly mark
10156 which relocations should apply to which instructions. This option
10157 is mostly useful for debugging, as GCC detects the capabilities of
10158 the assembler when it is built and sets the default accordingly.
10161 @itemx -mlarge-data
10162 @opindex msmall-data
10163 @opindex mlarge-data
10164 When @option{-mexplicit-relocs} is in effect, static data is
10165 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10166 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10167 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10168 16-bit relocations off of the @code{$gp} register. This limits the
10169 size of the small data area to 64KB, but allows the variables to be
10170 directly accessed via a single instruction.
10172 The default is @option{-mlarge-data}. With this option the data area
10173 is limited to just below 2GB@. Programs that require more than 2GB of
10174 data must use @code{malloc} or @code{mmap} to allocate the data in the
10175 heap instead of in the program's data segment.
10177 When generating code for shared libraries, @option{-fpic} implies
10178 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10181 @itemx -mlarge-text
10182 @opindex msmall-text
10183 @opindex mlarge-text
10184 When @option{-msmall-text} is used, the compiler assumes that the
10185 code of the entire program (or shared library) fits in 4MB, and is
10186 thus reachable with a branch instruction. When @option{-msmall-data}
10187 is used, the compiler can assume that all local symbols share the
10188 same @code{$gp} value, and thus reduce the number of instructions
10189 required for a function call from 4 to 1.
10191 The default is @option{-mlarge-text}.
10193 @item -mcpu=@var{cpu_type}
10195 Set the instruction set and instruction scheduling parameters for
10196 machine type @var{cpu_type}. You can specify either the @samp{EV}
10197 style name or the corresponding chip number. GCC supports scheduling
10198 parameters for the EV4, EV5 and EV6 family of processors and will
10199 choose the default values for the instruction set from the processor
10200 you specify. If you do not specify a processor type, GCC will default
10201 to the processor on which the compiler was built.
10203 Supported values for @var{cpu_type} are
10209 Schedules as an EV4 and has no instruction set extensions.
10213 Schedules as an EV5 and has no instruction set extensions.
10217 Schedules as an EV5 and supports the BWX extension.
10222 Schedules as an EV5 and supports the BWX and MAX extensions.
10226 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10230 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10233 Native Linux/GNU toolchains also support the value @samp{native},
10234 which selects the best architecture option for the host processor.
10235 @option{-mcpu=native} has no effect if GCC does not recognize
10238 @item -mtune=@var{cpu_type}
10240 Set only the instruction scheduling parameters for machine type
10241 @var{cpu_type}. The instruction set is not changed.
10243 Native Linux/GNU toolchains also support the value @samp{native},
10244 which selects the best architecture option for the host processor.
10245 @option{-mtune=native} has no effect if GCC does not recognize
10248 @item -mmemory-latency=@var{time}
10249 @opindex mmemory-latency
10250 Sets the latency the scheduler should assume for typical memory
10251 references as seen by the application. This number is highly
10252 dependent on the memory access patterns used by the application
10253 and the size of the external cache on the machine.
10255 Valid options for @var{time} are
10259 A decimal number representing clock cycles.
10265 The compiler contains estimates of the number of clock cycles for
10266 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10267 (also called Dcache, Scache, and Bcache), as well as to main memory.
10268 Note that L3 is only valid for EV5.
10273 @node DEC Alpha/VMS Options
10274 @subsection DEC Alpha/VMS Options
10276 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10279 @item -mvms-return-codes
10280 @opindex mvms-return-codes
10281 Return VMS condition codes from main. The default is to return POSIX
10282 style condition (e.g.@: error) codes.
10286 @subsection FR30 Options
10287 @cindex FR30 Options
10289 These options are defined specifically for the FR30 port.
10293 @item -msmall-model
10294 @opindex msmall-model
10295 Use the small address space model. This can produce smaller code, but
10296 it does assume that all symbolic values and addresses will fit into a
10301 Assume that run-time support has been provided and so there is no need
10302 to include the simulator library (@file{libsim.a}) on the linker
10308 @subsection FRV Options
10309 @cindex FRV Options
10315 Only use the first 32 general purpose registers.
10320 Use all 64 general purpose registers.
10325 Use only the first 32 floating point registers.
10330 Use all 64 floating point registers
10333 @opindex mhard-float
10335 Use hardware instructions for floating point operations.
10338 @opindex msoft-float
10340 Use library routines for floating point operations.
10345 Dynamically allocate condition code registers.
10350 Do not try to dynamically allocate condition code registers, only
10351 use @code{icc0} and @code{fcc0}.
10356 Change ABI to use double word insns.
10361 Do not use double word instructions.
10366 Use floating point double instructions.
10369 @opindex mno-double
10371 Do not use floating point double instructions.
10376 Use media instructions.
10381 Do not use media instructions.
10386 Use multiply and add/subtract instructions.
10389 @opindex mno-muladd
10391 Do not use multiply and add/subtract instructions.
10396 Select the FDPIC ABI, that uses function descriptors to represent
10397 pointers to functions. Without any PIC/PIE-related options, it
10398 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
10399 assumes GOT entries and small data are within a 12-bit range from the
10400 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
10401 are computed with 32 bits.
10402 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10405 @opindex minline-plt
10407 Enable inlining of PLT entries in function calls to functions that are
10408 not known to bind locally. It has no effect without @option{-mfdpic}.
10409 It's enabled by default if optimizing for speed and compiling for
10410 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
10411 optimization option such as @option{-O3} or above is present in the
10417 Assume a large TLS segment when generating thread-local code.
10422 Do not assume a large TLS segment when generating thread-local code.
10427 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
10428 that is known to be in read-only sections. It's enabled by default,
10429 except for @option{-fpic} or @option{-fpie}: even though it may help
10430 make the global offset table smaller, it trades 1 instruction for 4.
10431 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
10432 one of which may be shared by multiple symbols, and it avoids the need
10433 for a GOT entry for the referenced symbol, so it's more likely to be a
10434 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
10436 @item -multilib-library-pic
10437 @opindex multilib-library-pic
10439 Link with the (library, not FD) pic libraries. It's implied by
10440 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
10441 @option{-fpic} without @option{-mfdpic}. You should never have to use
10445 @opindex mlinked-fp
10447 Follow the EABI requirement of always creating a frame pointer whenever
10448 a stack frame is allocated. This option is enabled by default and can
10449 be disabled with @option{-mno-linked-fp}.
10452 @opindex mlong-calls
10454 Use indirect addressing to call functions outside the current
10455 compilation unit. This allows the functions to be placed anywhere
10456 within the 32-bit address space.
10458 @item -malign-labels
10459 @opindex malign-labels
10461 Try to align labels to an 8-byte boundary by inserting nops into the
10462 previous packet. This option only has an effect when VLIW packing
10463 is enabled. It doesn't create new packets; it merely adds nops to
10466 @item -mlibrary-pic
10467 @opindex mlibrary-pic
10469 Generate position-independent EABI code.
10474 Use only the first four media accumulator registers.
10479 Use all eight media accumulator registers.
10484 Pack VLIW instructions.
10489 Do not pack VLIW instructions.
10492 @opindex mno-eflags
10494 Do not mark ABI switches in e_flags.
10497 @opindex mcond-move
10499 Enable the use of conditional-move instructions (default).
10501 This switch is mainly for debugging the compiler and will likely be removed
10502 in a future version.
10504 @item -mno-cond-move
10505 @opindex mno-cond-move
10507 Disable the use of conditional-move instructions.
10509 This switch is mainly for debugging the compiler and will likely be removed
10510 in a future version.
10515 Enable the use of conditional set instructions (default).
10517 This switch is mainly for debugging the compiler and will likely be removed
10518 in a future version.
10523 Disable the use of conditional set instructions.
10525 This switch is mainly for debugging the compiler and will likely be removed
10526 in a future version.
10529 @opindex mcond-exec
10531 Enable the use of conditional execution (default).
10533 This switch is mainly for debugging the compiler and will likely be removed
10534 in a future version.
10536 @item -mno-cond-exec
10537 @opindex mno-cond-exec
10539 Disable the use of conditional execution.
10541 This switch is mainly for debugging the compiler and will likely be removed
10542 in a future version.
10544 @item -mvliw-branch
10545 @opindex mvliw-branch
10547 Run a pass to pack branches into VLIW instructions (default).
10549 This switch is mainly for debugging the compiler and will likely be removed
10550 in a future version.
10552 @item -mno-vliw-branch
10553 @opindex mno-vliw-branch
10555 Do not run a pass to pack branches into VLIW instructions.
10557 This switch is mainly for debugging the compiler and will likely be removed
10558 in a future version.
10560 @item -mmulti-cond-exec
10561 @opindex mmulti-cond-exec
10563 Enable optimization of @code{&&} and @code{||} in conditional execution
10566 This switch is mainly for debugging the compiler and will likely be removed
10567 in a future version.
10569 @item -mno-multi-cond-exec
10570 @opindex mno-multi-cond-exec
10572 Disable optimization of @code{&&} and @code{||} in conditional execution.
10574 This switch is mainly for debugging the compiler and will likely be removed
10575 in a future version.
10577 @item -mnested-cond-exec
10578 @opindex mnested-cond-exec
10580 Enable nested conditional execution optimizations (default).
10582 This switch is mainly for debugging the compiler and will likely be removed
10583 in a future version.
10585 @item -mno-nested-cond-exec
10586 @opindex mno-nested-cond-exec
10588 Disable nested conditional execution optimizations.
10590 This switch is mainly for debugging the compiler and will likely be removed
10591 in a future version.
10593 @item -moptimize-membar
10594 @opindex moptimize-membar
10596 This switch removes redundant @code{membar} instructions from the
10597 compiler generated code. It is enabled by default.
10599 @item -mno-optimize-membar
10600 @opindex mno-optimize-membar
10602 This switch disables the automatic removal of redundant @code{membar}
10603 instructions from the generated code.
10605 @item -mtomcat-stats
10606 @opindex mtomcat-stats
10608 Cause gas to print out tomcat statistics.
10610 @item -mcpu=@var{cpu}
10613 Select the processor type for which to generate code. Possible values are
10614 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10615 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10619 @node GNU/Linux Options
10620 @subsection GNU/Linux Options
10622 These @samp{-m} options are defined for GNU/Linux targets:
10627 Use the GNU C library instead of uClibc. This is the default except
10628 on @samp{*-*-linux-*uclibc*} targets.
10632 Use uClibc instead of the GNU C library. This is the default on
10633 @samp{*-*-linux-*uclibc*} targets.
10636 @node H8/300 Options
10637 @subsection H8/300 Options
10639 These @samp{-m} options are defined for the H8/300 implementations:
10644 Shorten some address references at link time, when possible; uses the
10645 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10646 ld, Using ld}, for a fuller description.
10650 Generate code for the H8/300H@.
10654 Generate code for the H8S@.
10658 Generate code for the H8S and H8/300H in the normal mode. This switch
10659 must be used either with @option{-mh} or @option{-ms}.
10663 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10667 Make @code{int} data 32 bits by default.
10670 @opindex malign-300
10671 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10672 The default for the H8/300H and H8S is to align longs and floats on 4
10674 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10675 This option has no effect on the H8/300.
10679 @subsection HPPA Options
10680 @cindex HPPA Options
10682 These @samp{-m} options are defined for the HPPA family of computers:
10685 @item -march=@var{architecture-type}
10687 Generate code for the specified architecture. The choices for
10688 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10689 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10690 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10691 architecture option for your machine. Code compiled for lower numbered
10692 architectures will run on higher numbered architectures, but not the
10695 @item -mpa-risc-1-0
10696 @itemx -mpa-risc-1-1
10697 @itemx -mpa-risc-2-0
10698 @opindex mpa-risc-1-0
10699 @opindex mpa-risc-1-1
10700 @opindex mpa-risc-2-0
10701 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10704 @opindex mbig-switch
10705 Generate code suitable for big switch tables. Use this option only if
10706 the assembler/linker complain about out of range branches within a switch
10709 @item -mjump-in-delay
10710 @opindex mjump-in-delay
10711 Fill delay slots of function calls with unconditional jump instructions
10712 by modifying the return pointer for the function call to be the target
10713 of the conditional jump.
10715 @item -mdisable-fpregs
10716 @opindex mdisable-fpregs
10717 Prevent floating point registers from being used in any manner. This is
10718 necessary for compiling kernels which perform lazy context switching of
10719 floating point registers. If you use this option and attempt to perform
10720 floating point operations, the compiler will abort.
10722 @item -mdisable-indexing
10723 @opindex mdisable-indexing
10724 Prevent the compiler from using indexing address modes. This avoids some
10725 rather obscure problems when compiling MIG generated code under MACH@.
10727 @item -mno-space-regs
10728 @opindex mno-space-regs
10729 Generate code that assumes the target has no space registers. This allows
10730 GCC to generate faster indirect calls and use unscaled index address modes.
10732 Such code is suitable for level 0 PA systems and kernels.
10734 @item -mfast-indirect-calls
10735 @opindex mfast-indirect-calls
10736 Generate code that assumes calls never cross space boundaries. This
10737 allows GCC to emit code which performs faster indirect calls.
10739 This option will not work in the presence of shared libraries or nested
10742 @item -mfixed-range=@var{register-range}
10743 @opindex mfixed-range
10744 Generate code treating the given register range as fixed registers.
10745 A fixed register is one that the register allocator can not use. This is
10746 useful when compiling kernel code. A register range is specified as
10747 two registers separated by a dash. Multiple register ranges can be
10748 specified separated by a comma.
10750 @item -mlong-load-store
10751 @opindex mlong-load-store
10752 Generate 3-instruction load and store sequences as sometimes required by
10753 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10756 @item -mportable-runtime
10757 @opindex mportable-runtime
10758 Use the portable calling conventions proposed by HP for ELF systems.
10762 Enable the use of assembler directives only GAS understands.
10764 @item -mschedule=@var{cpu-type}
10766 Schedule code according to the constraints for the machine type
10767 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10768 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10769 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10770 proper scheduling option for your machine. The default scheduling is
10774 @opindex mlinker-opt
10775 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10776 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10777 linkers in which they give bogus error messages when linking some programs.
10780 @opindex msoft-float
10781 Generate output containing library calls for floating point.
10782 @strong{Warning:} the requisite libraries are not available for all HPPA
10783 targets. Normally the facilities of the machine's usual C compiler are
10784 used, but this cannot be done directly in cross-compilation. You must make
10785 your own arrangements to provide suitable library functions for
10788 @option{-msoft-float} changes the calling convention in the output file;
10789 therefore, it is only useful if you compile @emph{all} of a program with
10790 this option. In particular, you need to compile @file{libgcc.a}, the
10791 library that comes with GCC, with @option{-msoft-float} in order for
10796 Generate the predefine, @code{_SIO}, for server IO@. The default is
10797 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10798 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10799 options are available under HP-UX and HI-UX@.
10803 Use GNU ld specific options. This passes @option{-shared} to ld when
10804 building a shared library. It is the default when GCC is configured,
10805 explicitly or implicitly, with the GNU linker. This option does not
10806 have any affect on which ld is called, it only changes what parameters
10807 are passed to that ld. The ld that is called is determined by the
10808 @option{--with-ld} configure option, GCC's program search path, and
10809 finally by the user's @env{PATH}. The linker used by GCC can be printed
10810 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10811 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10815 Use HP ld specific options. This passes @option{-b} to ld when building
10816 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10817 links. It is the default when GCC is configured, explicitly or
10818 implicitly, with the HP linker. This option does not have any affect on
10819 which ld is called, it only changes what parameters are passed to that
10820 ld. The ld that is called is determined by the @option{--with-ld}
10821 configure option, GCC's program search path, and finally by the user's
10822 @env{PATH}. The linker used by GCC can be printed using @samp{which
10823 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10824 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10827 @opindex mno-long-calls
10828 Generate code that uses long call sequences. This ensures that a call
10829 is always able to reach linker generated stubs. The default is to generate
10830 long calls only when the distance from the call site to the beginning
10831 of the function or translation unit, as the case may be, exceeds a
10832 predefined limit set by the branch type being used. The limits for
10833 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10834 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10837 Distances are measured from the beginning of functions when using the
10838 @option{-ffunction-sections} option, or when using the @option{-mgas}
10839 and @option{-mno-portable-runtime} options together under HP-UX with
10842 It is normally not desirable to use this option as it will degrade
10843 performance. However, it may be useful in large applications,
10844 particularly when partial linking is used to build the application.
10846 The types of long calls used depends on the capabilities of the
10847 assembler and linker, and the type of code being generated. The
10848 impact on systems that support long absolute calls, and long pic
10849 symbol-difference or pc-relative calls should be relatively small.
10850 However, an indirect call is used on 32-bit ELF systems in pic code
10851 and it is quite long.
10853 @item -munix=@var{unix-std}
10855 Generate compiler predefines and select a startfile for the specified
10856 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10857 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10858 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10859 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10860 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10863 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10864 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10865 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10866 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10867 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10868 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10870 It is @emph{important} to note that this option changes the interfaces
10871 for various library routines. It also affects the operational behavior
10872 of the C library. Thus, @emph{extreme} care is needed in using this
10875 Library code that is intended to operate with more than one UNIX
10876 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10877 as appropriate. Most GNU software doesn't provide this capability.
10881 Suppress the generation of link options to search libdld.sl when the
10882 @option{-static} option is specified on HP-UX 10 and later.
10886 The HP-UX implementation of setlocale in libc has a dependency on
10887 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10888 when the @option{-static} option is specified, special link options
10889 are needed to resolve this dependency.
10891 On HP-UX 10 and later, the GCC driver adds the necessary options to
10892 link with libdld.sl when the @option{-static} option is specified.
10893 This causes the resulting binary to be dynamic. On the 64-bit port,
10894 the linkers generate dynamic binaries by default in any case. The
10895 @option{-nolibdld} option can be used to prevent the GCC driver from
10896 adding these link options.
10900 Add support for multithreading with the @dfn{dce thread} library
10901 under HP-UX@. This option sets flags for both the preprocessor and
10905 @node i386 and x86-64 Options
10906 @subsection Intel 386 and AMD x86-64 Options
10907 @cindex i386 Options
10908 @cindex x86-64 Options
10909 @cindex Intel 386 Options
10910 @cindex AMD x86-64 Options
10912 These @samp{-m} options are defined for the i386 and x86-64 family of
10916 @item -mtune=@var{cpu-type}
10918 Tune to @var{cpu-type} everything applicable about the generated code, except
10919 for the ABI and the set of available instructions. The choices for
10920 @var{cpu-type} are:
10923 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10924 If you know the CPU on which your code will run, then you should use
10925 the corresponding @option{-mtune} option instead of
10926 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10927 of your application will have, then you should use this option.
10929 As new processors are deployed in the marketplace, the behavior of this
10930 option will change. Therefore, if you upgrade to a newer version of
10931 GCC, the code generated option will change to reflect the processors
10932 that were most common when that version of GCC was released.
10934 There is no @option{-march=generic} option because @option{-march}
10935 indicates the instruction set the compiler can use, and there is no
10936 generic instruction set applicable to all processors. In contrast,
10937 @option{-mtune} indicates the processor (or, in this case, collection of
10938 processors) for which the code is optimized.
10940 This selects the CPU to tune for at compilation time by determining
10941 the processor type of the compiling machine. Using @option{-mtune=native}
10942 will produce code optimized for the local machine under the constraints
10943 of the selected instruction set. Using @option{-march=native} will
10944 enable all instruction subsets supported by the local machine (hence
10945 the result might not run on different machines).
10947 Original Intel's i386 CPU@.
10949 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10950 @item i586, pentium
10951 Intel Pentium CPU with no MMX support.
10953 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10955 Intel PentiumPro CPU@.
10957 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10958 instruction set will be used, so the code will run on all i686 family chips.
10960 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10961 @item pentium3, pentium3m
10962 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10965 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10966 support. Used by Centrino notebooks.
10967 @item pentium4, pentium4m
10968 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10970 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10973 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10974 SSE2 and SSE3 instruction set support.
10976 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10977 instruction set support.
10979 AMD K6 CPU with MMX instruction set support.
10981 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10982 @item athlon, athlon-tbird
10983 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10985 @item athlon-4, athlon-xp, athlon-mp
10986 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10987 instruction set support.
10988 @item k8, opteron, athlon64, athlon-fx
10989 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10990 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10991 @item k8-sse3, opteron-sse3, athlon64-sse3
10992 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10993 @item amdfam10, barcelona
10994 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10995 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10996 instruction set extensions.)
10998 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11001 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
11002 instruction set support.
11004 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
11005 implemented for this chip.)
11007 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11008 implemented for this chip.)
11010 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
11013 While picking a specific @var{cpu-type} will schedule things appropriately
11014 for that particular chip, the compiler will not generate any code that
11015 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11018 @item -march=@var{cpu-type}
11020 Generate instructions for the machine type @var{cpu-type}. The choices
11021 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11022 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11024 @item -mcpu=@var{cpu-type}
11026 A deprecated synonym for @option{-mtune}.
11028 @item -mfpmath=@var{unit}
11030 Generate floating point arithmetics for selected unit @var{unit}. The choices
11031 for @var{unit} are:
11035 Use the standard 387 floating point coprocessor present majority of chips and
11036 emulated otherwise. Code compiled with this option will run almost everywhere.
11037 The temporary results are computed in 80bit precision instead of precision
11038 specified by the type resulting in slightly different results compared to most
11039 of other chips. See @option{-ffloat-store} for more detailed description.
11041 This is the default choice for i386 compiler.
11044 Use scalar floating point instructions present in the SSE instruction set.
11045 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11046 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11047 instruction set supports only single precision arithmetics, thus the double and
11048 extended precision arithmetics is still done using 387. Later version, present
11049 only in Pentium4 and the future AMD x86-64 chips supports double precision
11052 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11053 or @option{-msse2} switches to enable SSE extensions and make this option
11054 effective. For the x86-64 compiler, these extensions are enabled by default.
11056 The resulting code should be considerably faster in the majority of cases and avoid
11057 the numerical instability problems of 387 code, but may break some existing
11058 code that expects temporaries to be 80bit.
11060 This is the default choice for the x86-64 compiler.
11065 Attempt to utilize both instruction sets at once. This effectively double the
11066 amount of available registers and on chips with separate execution units for
11067 387 and SSE the execution resources too. Use this option with care, as it is
11068 still experimental, because the GCC register allocator does not model separate
11069 functional units well resulting in instable performance.
11072 @item -masm=@var{dialect}
11073 @opindex masm=@var{dialect}
11074 Output asm instructions using selected @var{dialect}. Supported
11075 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11076 not support @samp{intel}.
11079 @itemx -mno-ieee-fp
11081 @opindex mno-ieee-fp
11082 Control whether or not the compiler uses IEEE floating point
11083 comparisons. These handle correctly the case where the result of a
11084 comparison is unordered.
11087 @opindex msoft-float
11088 Generate output containing library calls for floating point.
11089 @strong{Warning:} the requisite libraries are not part of GCC@.
11090 Normally the facilities of the machine's usual C compiler are used, but
11091 this can't be done directly in cross-compilation. You must make your
11092 own arrangements to provide suitable library functions for
11095 On machines where a function returns floating point results in the 80387
11096 register stack, some floating point opcodes may be emitted even if
11097 @option{-msoft-float} is used.
11099 @item -mno-fp-ret-in-387
11100 @opindex mno-fp-ret-in-387
11101 Do not use the FPU registers for return values of functions.
11103 The usual calling convention has functions return values of types
11104 @code{float} and @code{double} in an FPU register, even if there
11105 is no FPU@. The idea is that the operating system should emulate
11108 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11109 in ordinary CPU registers instead.
11111 @item -mno-fancy-math-387
11112 @opindex mno-fancy-math-387
11113 Some 387 emulators do not support the @code{sin}, @code{cos} and
11114 @code{sqrt} instructions for the 387. Specify this option to avoid
11115 generating those instructions. This option is the default on FreeBSD,
11116 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11117 indicates that the target cpu will always have an FPU and so the
11118 instruction will not need emulation. As of revision 2.6.1, these
11119 instructions are not generated unless you also use the
11120 @option{-funsafe-math-optimizations} switch.
11122 @item -malign-double
11123 @itemx -mno-align-double
11124 @opindex malign-double
11125 @opindex mno-align-double
11126 Control whether GCC aligns @code{double}, @code{long double}, and
11127 @code{long long} variables on a two word boundary or a one word
11128 boundary. Aligning @code{double} variables on a two word boundary will
11129 produce code that runs somewhat faster on a @samp{Pentium} at the
11130 expense of more memory.
11132 On x86-64, @option{-malign-double} is enabled by default.
11134 @strong{Warning:} if you use the @option{-malign-double} switch,
11135 structures containing the above types will be aligned differently than
11136 the published application binary interface specifications for the 386
11137 and will not be binary compatible with structures in code compiled
11138 without that switch.
11140 @item -m96bit-long-double
11141 @itemx -m128bit-long-double
11142 @opindex m96bit-long-double
11143 @opindex m128bit-long-double
11144 These switches control the size of @code{long double} type. The i386
11145 application binary interface specifies the size to be 96 bits,
11146 so @option{-m96bit-long-double} is the default in 32 bit mode.
11148 Modern architectures (Pentium and newer) would prefer @code{long double}
11149 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11150 conforming to the ABI, this would not be possible. So specifying a
11151 @option{-m128bit-long-double} will align @code{long double}
11152 to a 16 byte boundary by padding the @code{long double} with an additional
11155 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11156 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11158 Notice that neither of these options enable any extra precision over the x87
11159 standard of 80 bits for a @code{long double}.
11161 @strong{Warning:} if you override the default value for your target ABI, the
11162 structures and arrays containing @code{long double} variables will change
11163 their size as well as function calling convention for function taking
11164 @code{long double} will be modified. Hence they will not be binary
11165 compatible with arrays or structures in code compiled without that switch.
11167 @item -mlarge-data-threshold=@var{number}
11168 @opindex mlarge-data-threshold=@var{number}
11169 When @option{-mcmodel=medium} is specified, the data greater than
11170 @var{threshold} are placed in large data section. This value must be the
11171 same across all object linked into the binary and defaults to 65535.
11175 Use a different function-calling convention, in which functions that
11176 take a fixed number of arguments return with the @code{ret} @var{num}
11177 instruction, which pops their arguments while returning. This saves one
11178 instruction in the caller since there is no need to pop the arguments
11181 You can specify that an individual function is called with this calling
11182 sequence with the function attribute @samp{stdcall}. You can also
11183 override the @option{-mrtd} option by using the function attribute
11184 @samp{cdecl}. @xref{Function Attributes}.
11186 @strong{Warning:} this calling convention is incompatible with the one
11187 normally used on Unix, so you cannot use it if you need to call
11188 libraries compiled with the Unix compiler.
11190 Also, you must provide function prototypes for all functions that
11191 take variable numbers of arguments (including @code{printf});
11192 otherwise incorrect code will be generated for calls to those
11195 In addition, seriously incorrect code will result if you call a
11196 function with too many arguments. (Normally, extra arguments are
11197 harmlessly ignored.)
11199 @item -mregparm=@var{num}
11201 Control how many registers are used to pass integer arguments. By
11202 default, no registers are used to pass arguments, and at most 3
11203 registers can be used. You can control this behavior for a specific
11204 function by using the function attribute @samp{regparm}.
11205 @xref{Function Attributes}.
11207 @strong{Warning:} if you use this switch, and
11208 @var{num} is nonzero, then you must build all modules with the same
11209 value, including any libraries. This includes the system libraries and
11213 @opindex msseregparm
11214 Use SSE register passing conventions for float and double arguments
11215 and return values. You can control this behavior for a specific
11216 function by using the function attribute @samp{sseregparm}.
11217 @xref{Function Attributes}.
11219 @strong{Warning:} if you use this switch then you must build all
11220 modules with the same value, including any libraries. This includes
11221 the system libraries and startup modules.
11230 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11231 is specified, the significands of results of floating-point operations are
11232 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11233 significands of results of floating-point operations to 53 bits (double
11234 precision) and @option{-mpc80} rounds the significands of results of
11235 floating-point operations to 64 bits (extended double precision), which is
11236 the default. When this option is used, floating-point operations in higher
11237 precisions are not available to the programmer without setting the FPU
11238 control word explicitly.
11240 Setting the rounding of floating-point operations to less than the default
11241 80 bits can speed some programs by 2% or more. Note that some mathematical
11242 libraries assume that extended precision (80 bit) floating-point operations
11243 are enabled by default; routines in such libraries could suffer significant
11244 loss of accuracy, typically through so-called "catastrophic cancellation",
11245 when this option is used to set the precision to less than extended precision.
11247 @item -mstackrealign
11248 @opindex mstackrealign
11249 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
11250 option will generate an alternate prologue and epilogue that realigns the
11251 runtime stack if necessary. This supports mixing legacy codes that keep
11252 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11253 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
11254 applicable to individual functions.
11256 @item -mpreferred-stack-boundary=@var{num}
11257 @opindex mpreferred-stack-boundary
11258 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11259 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
11260 the default is 4 (16 bytes or 128 bits).
11262 @item -mincoming-stack-boundary=@var{num}
11263 @opindex mincoming-stack-boundary
11264 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11265 boundary. If @option{-mincoming-stack-boundary} is not specified,
11266 the one specified by @option{-mpreferred-stack-boundary} will be used.
11268 On Pentium and PentiumPro, @code{double} and @code{long double} values
11269 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11270 suffer significant run time performance penalties. On Pentium III, the
11271 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11272 properly if it is not 16 byte aligned.
11274 To ensure proper alignment of this values on the stack, the stack boundary
11275 must be as aligned as that required by any value stored on the stack.
11276 Further, every function must be generated such that it keeps the stack
11277 aligned. Thus calling a function compiled with a higher preferred
11278 stack boundary from a function compiled with a lower preferred stack
11279 boundary will most likely misalign the stack. It is recommended that
11280 libraries that use callbacks always use the default setting.
11282 This extra alignment does consume extra stack space, and generally
11283 increases code size. Code that is sensitive to stack space usage, such
11284 as embedded systems and operating system kernels, may want to reduce the
11285 preferred alignment to @option{-mpreferred-stack-boundary=2}.
11325 These switches enable or disable the use of instructions in the MMX,
11326 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, SSE5, ABM or
11327 3DNow!@: extended instruction sets.
11328 These extensions are also available as built-in functions: see
11329 @ref{X86 Built-in Functions}, for details of the functions enabled and
11330 disabled by these switches.
11332 To have SSE/SSE2 instructions generated automatically from floating-point
11333 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
11335 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
11336 generates new AVX instructions or AVX equivalence for all SSEx instructions
11339 These options will enable GCC to use these extended instructions in
11340 generated code, even without @option{-mfpmath=sse}. Applications which
11341 perform runtime CPU detection must compile separate files for each
11342 supported architecture, using the appropriate flags. In particular,
11343 the file containing the CPU detection code should be compiled without
11348 This option instructs GCC to emit a @code{cld} instruction in the prologue
11349 of functions that use string instructions. String instructions depend on
11350 the DF flag to select between autoincrement or autodecrement mode. While the
11351 ABI specifies the DF flag to be cleared on function entry, some operating
11352 systems violate this specification by not clearing the DF flag in their
11353 exception dispatchers. The exception handler can be invoked with the DF flag
11354 set which leads to wrong direction mode, when string instructions are used.
11355 This option can be enabled by default on 32-bit x86 targets by configuring
11356 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
11357 instructions can be suppressed with the @option{-mno-cld} compiler option
11362 This option will enable GCC to use CMPXCHG16B instruction in generated code.
11363 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
11364 data types. This is useful for high resolution counters that could be updated
11365 by multiple processors (or cores). This instruction is generated as part of
11366 atomic built-in functions: see @ref{Atomic Builtins} for details.
11370 This option will enable GCC to use SAHF instruction in generated 64-bit code.
11371 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
11372 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
11373 SAHF are load and store instructions, respectively, for certain status flags.
11374 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
11375 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
11379 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
11380 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
11381 to increase precision instead of DIVSS and SQRTSS (and their vectorized
11382 variants) for single precision floating point arguments. These instructions
11383 are generated only when @option{-funsafe-math-optimizations} is enabled
11384 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
11385 Note that while the throughput of the sequence is higher than the throughput
11386 of the non-reciprocal instruction, the precision of the sequence can be
11387 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
11389 @item -mveclibabi=@var{type}
11390 @opindex mveclibabi
11391 Specifies the ABI type to use for vectorizing intrinsics using an
11392 external library. Supported types are @code{svml} for the Intel short
11393 vector math library and @code{acml} for the AMD math core library style
11394 of interfacing. GCC will currently emit calls to @code{vmldExp2},
11395 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
11396 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
11397 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
11398 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
11399 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
11400 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
11401 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
11402 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
11403 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
11404 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
11405 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
11406 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
11407 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
11408 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
11409 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
11410 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
11411 compatible library will have to be specified at link time.
11414 @itemx -mno-push-args
11415 @opindex mpush-args
11416 @opindex mno-push-args
11417 Use PUSH operations to store outgoing parameters. This method is shorter
11418 and usually equally fast as method using SUB/MOV operations and is enabled
11419 by default. In some cases disabling it may improve performance because of
11420 improved scheduling and reduced dependencies.
11422 @item -maccumulate-outgoing-args
11423 @opindex maccumulate-outgoing-args
11424 If enabled, the maximum amount of space required for outgoing arguments will be
11425 computed in the function prologue. This is faster on most modern CPUs
11426 because of reduced dependencies, improved scheduling and reduced stack usage
11427 when preferred stack boundary is not equal to 2. The drawback is a notable
11428 increase in code size. This switch implies @option{-mno-push-args}.
11432 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
11433 on thread-safe exception handling must compile and link all code with the
11434 @option{-mthreads} option. When compiling, @option{-mthreads} defines
11435 @option{-D_MT}; when linking, it links in a special thread helper library
11436 @option{-lmingwthrd} which cleans up per thread exception handling data.
11438 @item -mno-align-stringops
11439 @opindex mno-align-stringops
11440 Do not align destination of inlined string operations. This switch reduces
11441 code size and improves performance in case the destination is already aligned,
11442 but GCC doesn't know about it.
11444 @item -minline-all-stringops
11445 @opindex minline-all-stringops
11446 By default GCC inlines string operations only when destination is known to be
11447 aligned at least to 4 byte boundary. This enables more inlining, increase code
11448 size, but may improve performance of code that depends on fast memcpy, strlen
11449 and memset for short lengths.
11451 @item -minline-stringops-dynamically
11452 @opindex minline-stringops-dynamically
11453 For string operation of unknown size, inline runtime checks so for small
11454 blocks inline code is used, while for large blocks library call is used.
11456 @item -mstringop-strategy=@var{alg}
11457 @opindex mstringop-strategy=@var{alg}
11458 Overwrite internal decision heuristic about particular algorithm to inline
11459 string operation with. The allowed values are @code{rep_byte},
11460 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
11461 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
11462 expanding inline loop, @code{libcall} for always expanding library call.
11464 @item -momit-leaf-frame-pointer
11465 @opindex momit-leaf-frame-pointer
11466 Don't keep the frame pointer in a register for leaf functions. This
11467 avoids the instructions to save, set up and restore frame pointers and
11468 makes an extra register available in leaf functions. The option
11469 @option{-fomit-frame-pointer} removes the frame pointer for all functions
11470 which might make debugging harder.
11472 @item -mtls-direct-seg-refs
11473 @itemx -mno-tls-direct-seg-refs
11474 @opindex mtls-direct-seg-refs
11475 Controls whether TLS variables may be accessed with offsets from the
11476 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
11477 or whether the thread base pointer must be added. Whether or not this
11478 is legal depends on the operating system, and whether it maps the
11479 segment to cover the entire TLS area.
11481 For systems that use GNU libc, the default is on.
11484 @itemx -mno-fused-madd
11485 @opindex mfused-madd
11486 Enable automatic generation of fused floating point multiply-add instructions
11487 if the ISA supports such instructions. The -mfused-madd option is on by
11488 default. The fused multiply-add instructions have a different
11489 rounding behavior compared to executing a multiply followed by an add.
11492 @itemx -mno-sse2avx
11494 Specify that the assembler should encode SSE instructions with VEX
11495 prefix. The option @option{-mavx} turns this on by default.
11498 These @samp{-m} switches are supported in addition to the above
11499 on AMD x86-64 processors in 64-bit environments.
11506 Generate code for a 32-bit or 64-bit environment.
11507 The 32-bit environment sets int, long and pointer to 32 bits and
11508 generates code that runs on any i386 system.
11509 The 64-bit environment sets int to 32 bits and long and pointer
11510 to 64 bits and generates code for AMD's x86-64 architecture. For
11511 darwin only the -m64 option turns off the @option{-fno-pic} and
11512 @option{-mdynamic-no-pic} options.
11514 @item -mno-red-zone
11515 @opindex no-red-zone
11516 Do not use a so called red zone for x86-64 code. The red zone is mandated
11517 by the x86-64 ABI, it is a 128-byte area beyond the location of the
11518 stack pointer that will not be modified by signal or interrupt handlers
11519 and therefore can be used for temporary data without adjusting the stack
11520 pointer. The flag @option{-mno-red-zone} disables this red zone.
11522 @item -mcmodel=small
11523 @opindex mcmodel=small
11524 Generate code for the small code model: the program and its symbols must
11525 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11526 Programs can be statically or dynamically linked. This is the default
11529 @item -mcmodel=kernel
11530 @opindex mcmodel=kernel
11531 Generate code for the kernel code model. The kernel runs in the
11532 negative 2 GB of the address space.
11533 This model has to be used for Linux kernel code.
11535 @item -mcmodel=medium
11536 @opindex mcmodel=medium
11537 Generate code for the medium model: The program is linked in the lower 2
11538 GB of the address space. Small symbols are also placed there. Symbols
11539 with sizes larger than @option{-mlarge-data-threshold} are put into
11540 large data or bss sections and can be located above 2GB. Programs can
11541 be statically or dynamically linked.
11543 @item -mcmodel=large
11544 @opindex mcmodel=large
11545 Generate code for the large model: This model makes no assumptions
11546 about addresses and sizes of sections.
11549 @node IA-64 Options
11550 @subsection IA-64 Options
11551 @cindex IA-64 Options
11553 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11557 @opindex mbig-endian
11558 Generate code for a big endian target. This is the default for HP-UX@.
11560 @item -mlittle-endian
11561 @opindex mlittle-endian
11562 Generate code for a little endian target. This is the default for AIX5
11568 @opindex mno-gnu-as
11569 Generate (or don't) code for the GNU assembler. This is the default.
11570 @c Also, this is the default if the configure option @option{--with-gnu-as}
11576 @opindex mno-gnu-ld
11577 Generate (or don't) code for the GNU linker. This is the default.
11578 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11583 Generate code that does not use a global pointer register. The result
11584 is not position independent code, and violates the IA-64 ABI@.
11586 @item -mvolatile-asm-stop
11587 @itemx -mno-volatile-asm-stop
11588 @opindex mvolatile-asm-stop
11589 @opindex mno-volatile-asm-stop
11590 Generate (or don't) a stop bit immediately before and after volatile asm
11593 @item -mregister-names
11594 @itemx -mno-register-names
11595 @opindex mregister-names
11596 @opindex mno-register-names
11597 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11598 the stacked registers. This may make assembler output more readable.
11604 Disable (or enable) optimizations that use the small data section. This may
11605 be useful for working around optimizer bugs.
11607 @item -mconstant-gp
11608 @opindex mconstant-gp
11609 Generate code that uses a single constant global pointer value. This is
11610 useful when compiling kernel code.
11614 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11615 This is useful when compiling firmware code.
11617 @item -minline-float-divide-min-latency
11618 @opindex minline-float-divide-min-latency
11619 Generate code for inline divides of floating point values
11620 using the minimum latency algorithm.
11622 @item -minline-float-divide-max-throughput
11623 @opindex minline-float-divide-max-throughput
11624 Generate code for inline divides of floating point values
11625 using the maximum throughput algorithm.
11627 @item -minline-int-divide-min-latency
11628 @opindex minline-int-divide-min-latency
11629 Generate code for inline divides of integer values
11630 using the minimum latency algorithm.
11632 @item -minline-int-divide-max-throughput
11633 @opindex minline-int-divide-max-throughput
11634 Generate code for inline divides of integer values
11635 using the maximum throughput algorithm.
11637 @item -minline-sqrt-min-latency
11638 @opindex minline-sqrt-min-latency
11639 Generate code for inline square roots
11640 using the minimum latency algorithm.
11642 @item -minline-sqrt-max-throughput
11643 @opindex minline-sqrt-max-throughput
11644 Generate code for inline square roots
11645 using the maximum throughput algorithm.
11647 @item -mno-dwarf2-asm
11648 @itemx -mdwarf2-asm
11649 @opindex mno-dwarf2-asm
11650 @opindex mdwarf2-asm
11651 Don't (or do) generate assembler code for the DWARF2 line number debugging
11652 info. This may be useful when not using the GNU assembler.
11654 @item -mearly-stop-bits
11655 @itemx -mno-early-stop-bits
11656 @opindex mearly-stop-bits
11657 @opindex mno-early-stop-bits
11658 Allow stop bits to be placed earlier than immediately preceding the
11659 instruction that triggered the stop bit. This can improve instruction
11660 scheduling, but does not always do so.
11662 @item -mfixed-range=@var{register-range}
11663 @opindex mfixed-range
11664 Generate code treating the given register range as fixed registers.
11665 A fixed register is one that the register allocator can not use. This is
11666 useful when compiling kernel code. A register range is specified as
11667 two registers separated by a dash. Multiple register ranges can be
11668 specified separated by a comma.
11670 @item -mtls-size=@var{tls-size}
11672 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11675 @item -mtune=@var{cpu-type}
11677 Tune the instruction scheduling for a particular CPU, Valid values are
11678 itanium, itanium1, merced, itanium2, and mckinley.
11684 Add support for multithreading using the POSIX threads library. This
11685 option sets flags for both the preprocessor and linker. It does
11686 not affect the thread safety of object code produced by the compiler or
11687 that of libraries supplied with it. These are HP-UX specific flags.
11693 Generate code for a 32-bit or 64-bit environment.
11694 The 32-bit environment sets int, long and pointer to 32 bits.
11695 The 64-bit environment sets int to 32 bits and long and pointer
11696 to 64 bits. These are HP-UX specific flags.
11698 @item -mno-sched-br-data-spec
11699 @itemx -msched-br-data-spec
11700 @opindex mno-sched-br-data-spec
11701 @opindex msched-br-data-spec
11702 (Dis/En)able data speculative scheduling before reload.
11703 This will result in generation of the ld.a instructions and
11704 the corresponding check instructions (ld.c / chk.a).
11705 The default is 'disable'.
11707 @item -msched-ar-data-spec
11708 @itemx -mno-sched-ar-data-spec
11709 @opindex msched-ar-data-spec
11710 @opindex mno-sched-ar-data-spec
11711 (En/Dis)able data speculative scheduling after reload.
11712 This will result in generation of the ld.a instructions and
11713 the corresponding check instructions (ld.c / chk.a).
11714 The default is 'enable'.
11716 @item -mno-sched-control-spec
11717 @itemx -msched-control-spec
11718 @opindex mno-sched-control-spec
11719 @opindex msched-control-spec
11720 (Dis/En)able control speculative scheduling. This feature is
11721 available only during region scheduling (i.e.@: before reload).
11722 This will result in generation of the ld.s instructions and
11723 the corresponding check instructions chk.s .
11724 The default is 'disable'.
11726 @item -msched-br-in-data-spec
11727 @itemx -mno-sched-br-in-data-spec
11728 @opindex msched-br-in-data-spec
11729 @opindex mno-sched-br-in-data-spec
11730 (En/Dis)able speculative scheduling of the instructions that
11731 are dependent on the data speculative loads before reload.
11732 This is effective only with @option{-msched-br-data-spec} enabled.
11733 The default is 'enable'.
11735 @item -msched-ar-in-data-spec
11736 @itemx -mno-sched-ar-in-data-spec
11737 @opindex msched-ar-in-data-spec
11738 @opindex mno-sched-ar-in-data-spec
11739 (En/Dis)able speculative scheduling of the instructions that
11740 are dependent on the data speculative loads after reload.
11741 This is effective only with @option{-msched-ar-data-spec} enabled.
11742 The default is 'enable'.
11744 @item -msched-in-control-spec
11745 @itemx -mno-sched-in-control-spec
11746 @opindex msched-in-control-spec
11747 @opindex mno-sched-in-control-spec
11748 (En/Dis)able speculative scheduling of the instructions that
11749 are dependent on the control speculative loads.
11750 This is effective only with @option{-msched-control-spec} enabled.
11751 The default is 'enable'.
11754 @itemx -mno-sched-ldc
11755 @opindex msched-ldc
11756 @opindex mno-sched-ldc
11757 (En/Dis)able use of simple data speculation checks ld.c .
11758 If disabled, only chk.a instructions will be emitted to check
11759 data speculative loads.
11760 The default is 'enable'.
11762 @item -mno-sched-control-ldc
11763 @itemx -msched-control-ldc
11764 @opindex mno-sched-control-ldc
11765 @opindex msched-control-ldc
11766 (Dis/En)able use of ld.c instructions to check control speculative loads.
11767 If enabled, in case of control speculative load with no speculatively
11768 scheduled dependent instructions this load will be emitted as ld.sa and
11769 ld.c will be used to check it.
11770 The default is 'disable'.
11772 @item -mno-sched-spec-verbose
11773 @itemx -msched-spec-verbose
11774 @opindex mno-sched-spec-verbose
11775 @opindex msched-spec-verbose
11776 (Dis/En)able printing of the information about speculative motions.
11778 @item -mno-sched-prefer-non-data-spec-insns
11779 @itemx -msched-prefer-non-data-spec-insns
11780 @opindex mno-sched-prefer-non-data-spec-insns
11781 @opindex msched-prefer-non-data-spec-insns
11782 If enabled, data speculative instructions will be chosen for schedule
11783 only if there are no other choices at the moment. This will make
11784 the use of the data speculation much more conservative.
11785 The default is 'disable'.
11787 @item -mno-sched-prefer-non-control-spec-insns
11788 @itemx -msched-prefer-non-control-spec-insns
11789 @opindex mno-sched-prefer-non-control-spec-insns
11790 @opindex msched-prefer-non-control-spec-insns
11791 If enabled, control speculative instructions will be chosen for schedule
11792 only if there are no other choices at the moment. This will make
11793 the use of the control speculation much more conservative.
11794 The default is 'disable'.
11796 @item -mno-sched-count-spec-in-critical-path
11797 @itemx -msched-count-spec-in-critical-path
11798 @opindex mno-sched-count-spec-in-critical-path
11799 @opindex msched-count-spec-in-critical-path
11800 If enabled, speculative dependencies will be considered during
11801 computation of the instructions priorities. This will make the use of the
11802 speculation a bit more conservative.
11803 The default is 'disable'.
11808 @subsection M32C Options
11809 @cindex M32C options
11812 @item -mcpu=@var{name}
11814 Select the CPU for which code is generated. @var{name} may be one of
11815 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11816 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11817 the M32C/80 series.
11821 Specifies that the program will be run on the simulator. This causes
11822 an alternate runtime library to be linked in which supports, for
11823 example, file I/O@. You must not use this option when generating
11824 programs that will run on real hardware; you must provide your own
11825 runtime library for whatever I/O functions are needed.
11827 @item -memregs=@var{number}
11829 Specifies the number of memory-based pseudo-registers GCC will use
11830 during code generation. These pseudo-registers will be used like real
11831 registers, so there is a tradeoff between GCC's ability to fit the
11832 code into available registers, and the performance penalty of using
11833 memory instead of registers. Note that all modules in a program must
11834 be compiled with the same value for this option. Because of that, you
11835 must not use this option with the default runtime libraries gcc
11840 @node M32R/D Options
11841 @subsection M32R/D Options
11842 @cindex M32R/D options
11844 These @option{-m} options are defined for Renesas M32R/D architectures:
11849 Generate code for the M32R/2@.
11853 Generate code for the M32R/X@.
11857 Generate code for the M32R@. This is the default.
11859 @item -mmodel=small
11860 @opindex mmodel=small
11861 Assume all objects live in the lower 16MB of memory (so that their addresses
11862 can be loaded with the @code{ld24} instruction), and assume all subroutines
11863 are reachable with the @code{bl} instruction.
11864 This is the default.
11866 The addressability of a particular object can be set with the
11867 @code{model} attribute.
11869 @item -mmodel=medium
11870 @opindex mmodel=medium
11871 Assume objects may be anywhere in the 32-bit address space (the compiler
11872 will generate @code{seth/add3} instructions to load their addresses), and
11873 assume all subroutines are reachable with the @code{bl} instruction.
11875 @item -mmodel=large
11876 @opindex mmodel=large
11877 Assume objects may be anywhere in the 32-bit address space (the compiler
11878 will generate @code{seth/add3} instructions to load their addresses), and
11879 assume subroutines may not be reachable with the @code{bl} instruction
11880 (the compiler will generate the much slower @code{seth/add3/jl}
11881 instruction sequence).
11884 @opindex msdata=none
11885 Disable use of the small data area. Variables will be put into
11886 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11887 @code{section} attribute has been specified).
11888 This is the default.
11890 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11891 Objects may be explicitly put in the small data area with the
11892 @code{section} attribute using one of these sections.
11894 @item -msdata=sdata
11895 @opindex msdata=sdata
11896 Put small global and static data in the small data area, but do not
11897 generate special code to reference them.
11900 @opindex msdata=use
11901 Put small global and static data in the small data area, and generate
11902 special instructions to reference them.
11906 @cindex smaller data references
11907 Put global and static objects less than or equal to @var{num} bytes
11908 into the small data or bss sections instead of the normal data or bss
11909 sections. The default value of @var{num} is 8.
11910 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11911 for this option to have any effect.
11913 All modules should be compiled with the same @option{-G @var{num}} value.
11914 Compiling with different values of @var{num} may or may not work; if it
11915 doesn't the linker will give an error message---incorrect code will not be
11920 Makes the M32R specific code in the compiler display some statistics
11921 that might help in debugging programs.
11923 @item -malign-loops
11924 @opindex malign-loops
11925 Align all loops to a 32-byte boundary.
11927 @item -mno-align-loops
11928 @opindex mno-align-loops
11929 Do not enforce a 32-byte alignment for loops. This is the default.
11931 @item -missue-rate=@var{number}
11932 @opindex missue-rate=@var{number}
11933 Issue @var{number} instructions per cycle. @var{number} can only be 1
11936 @item -mbranch-cost=@var{number}
11937 @opindex mbranch-cost=@var{number}
11938 @var{number} can only be 1 or 2. If it is 1 then branches will be
11939 preferred over conditional code, if it is 2, then the opposite will
11942 @item -mflush-trap=@var{number}
11943 @opindex mflush-trap=@var{number}
11944 Specifies the trap number to use to flush the cache. The default is
11945 12. Valid numbers are between 0 and 15 inclusive.
11947 @item -mno-flush-trap
11948 @opindex mno-flush-trap
11949 Specifies that the cache cannot be flushed by using a trap.
11951 @item -mflush-func=@var{name}
11952 @opindex mflush-func=@var{name}
11953 Specifies the name of the operating system function to call to flush
11954 the cache. The default is @emph{_flush_cache}, but a function call
11955 will only be used if a trap is not available.
11957 @item -mno-flush-func
11958 @opindex mno-flush-func
11959 Indicates that there is no OS function for flushing the cache.
11963 @node M680x0 Options
11964 @subsection M680x0 Options
11965 @cindex M680x0 options
11967 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11968 The default settings depend on which architecture was selected when
11969 the compiler was configured; the defaults for the most common choices
11973 @item -march=@var{arch}
11975 Generate code for a specific M680x0 or ColdFire instruction set
11976 architecture. Permissible values of @var{arch} for M680x0
11977 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11978 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11979 architectures are selected according to Freescale's ISA classification
11980 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11981 @samp{isab} and @samp{isac}.
11983 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11984 code for a ColdFire target. The @var{arch} in this macro is one of the
11985 @option{-march} arguments given above.
11987 When used together, @option{-march} and @option{-mtune} select code
11988 that runs on a family of similar processors but that is optimized
11989 for a particular microarchitecture.
11991 @item -mcpu=@var{cpu}
11993 Generate code for a specific M680x0 or ColdFire processor.
11994 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11995 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11996 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11997 below, which also classifies the CPUs into families:
11999 @multitable @columnfractions 0.20 0.80
12000 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12001 @item @samp{51qe} @tab @samp{51qe}
12002 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12003 @item @samp{5206e} @tab @samp{5206e}
12004 @item @samp{5208} @tab @samp{5207} @samp{5208}
12005 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12006 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12007 @item @samp{5216} @tab @samp{5214} @samp{5216}
12008 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12009 @item @samp{5225} @tab @samp{5224} @samp{5225}
12010 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12011 @item @samp{5249} @tab @samp{5249}
12012 @item @samp{5250} @tab @samp{5250}
12013 @item @samp{5271} @tab @samp{5270} @samp{5271}
12014 @item @samp{5272} @tab @samp{5272}
12015 @item @samp{5275} @tab @samp{5274} @samp{5275}
12016 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12017 @item @samp{5307} @tab @samp{5307}
12018 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12019 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12020 @item @samp{5407} @tab @samp{5407}
12021 @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}
12024 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12025 @var{arch} is compatible with @var{cpu}. Other combinations of
12026 @option{-mcpu} and @option{-march} are rejected.
12028 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12029 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12030 where the value of @var{family} is given by the table above.
12032 @item -mtune=@var{tune}
12034 Tune the code for a particular microarchitecture, within the
12035 constraints set by @option{-march} and @option{-mcpu}.
12036 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12037 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12038 and @samp{cpu32}. The ColdFire microarchitectures
12039 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12041 You can also use @option{-mtune=68020-40} for code that needs
12042 to run relatively well on 68020, 68030 and 68040 targets.
12043 @option{-mtune=68020-60} is similar but includes 68060 targets
12044 as well. These two options select the same tuning decisions as
12045 @option{-m68020-40} and @option{-m68020-60} respectively.
12047 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12048 when tuning for 680x0 architecture @var{arch}. It also defines
12049 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12050 option is used. If gcc is tuning for a range of architectures,
12051 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12052 it defines the macros for every architecture in the range.
12054 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12055 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12056 of the arguments given above.
12062 Generate output for a 68000. This is the default
12063 when the compiler is configured for 68000-based systems.
12064 It is equivalent to @option{-march=68000}.
12066 Use this option for microcontrollers with a 68000 or EC000 core,
12067 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12071 Generate output for a 68010. This is the default
12072 when the compiler is configured for 68010-based systems.
12073 It is equivalent to @option{-march=68010}.
12079 Generate output for a 68020. This is the default
12080 when the compiler is configured for 68020-based systems.
12081 It is equivalent to @option{-march=68020}.
12085 Generate output for a 68030. This is the default when the compiler is
12086 configured for 68030-based systems. It is equivalent to
12087 @option{-march=68030}.
12091 Generate output for a 68040. This is the default when the compiler is
12092 configured for 68040-based systems. It is equivalent to
12093 @option{-march=68040}.
12095 This option inhibits the use of 68881/68882 instructions that have to be
12096 emulated by software on the 68040. Use this option if your 68040 does not
12097 have code to emulate those instructions.
12101 Generate output for a 68060. This is the default when the compiler is
12102 configured for 68060-based systems. It is equivalent to
12103 @option{-march=68060}.
12105 This option inhibits the use of 68020 and 68881/68882 instructions that
12106 have to be emulated by software on the 68060. Use this option if your 68060
12107 does not have code to emulate those instructions.
12111 Generate output for a CPU32. This is the default
12112 when the compiler is configured for CPU32-based systems.
12113 It is equivalent to @option{-march=cpu32}.
12115 Use this option for microcontrollers with a
12116 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12117 68336, 68340, 68341, 68349 and 68360.
12121 Generate output for a 520X ColdFire CPU@. This is the default
12122 when the compiler is configured for 520X-based systems.
12123 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12124 in favor of that option.
12126 Use this option for microcontroller with a 5200 core, including
12127 the MCF5202, MCF5203, MCF5204 and MCF5206.
12131 Generate output for a 5206e ColdFire CPU@. The option is now
12132 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12136 Generate output for a member of the ColdFire 528X family.
12137 The option is now deprecated in favor of the equivalent
12138 @option{-mcpu=528x}.
12142 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
12143 in favor of the equivalent @option{-mcpu=5307}.
12147 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
12148 in favor of the equivalent @option{-mcpu=5407}.
12152 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12153 This includes use of hardware floating point instructions.
12154 The option is equivalent to @option{-mcpu=547x}, and is now
12155 deprecated in favor of that option.
12159 Generate output for a 68040, without using any of the new instructions.
12160 This results in code which can run relatively efficiently on either a
12161 68020/68881 or a 68030 or a 68040. The generated code does use the
12162 68881 instructions that are emulated on the 68040.
12164 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12168 Generate output for a 68060, without using any of the new instructions.
12169 This results in code which can run relatively efficiently on either a
12170 68020/68881 or a 68030 or a 68040. The generated code does use the
12171 68881 instructions that are emulated on the 68060.
12173 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12177 @opindex mhard-float
12179 Generate floating-point instructions. This is the default for 68020
12180 and above, and for ColdFire devices that have an FPU@. It defines the
12181 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12182 on ColdFire targets.
12185 @opindex msoft-float
12186 Do not generate floating-point instructions; use library calls instead.
12187 This is the default for 68000, 68010, and 68832 targets. It is also
12188 the default for ColdFire devices that have no FPU.
12194 Generate (do not generate) ColdFire hardware divide and remainder
12195 instructions. If @option{-march} is used without @option{-mcpu},
12196 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12197 architectures. Otherwise, the default is taken from the target CPU
12198 (either the default CPU, or the one specified by @option{-mcpu}). For
12199 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12200 @option{-mcpu=5206e}.
12202 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12206 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12207 Additionally, parameters passed on the stack are also aligned to a
12208 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12212 Do not consider type @code{int} to be 16 bits wide. This is the default.
12215 @itemx -mno-bitfield
12216 @opindex mnobitfield
12217 @opindex mno-bitfield
12218 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
12219 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12223 Do use the bit-field instructions. The @option{-m68020} option implies
12224 @option{-mbitfield}. This is the default if you use a configuration
12225 designed for a 68020.
12229 Use a different function-calling convention, in which functions
12230 that take a fixed number of arguments return with the @code{rtd}
12231 instruction, which pops their arguments while returning. This
12232 saves one instruction in the caller since there is no need to pop
12233 the arguments there.
12235 This calling convention is incompatible with the one normally
12236 used on Unix, so you cannot use it if you need to call libraries
12237 compiled with the Unix compiler.
12239 Also, you must provide function prototypes for all functions that
12240 take variable numbers of arguments (including @code{printf});
12241 otherwise incorrect code will be generated for calls to those
12244 In addition, seriously incorrect code will result if you call a
12245 function with too many arguments. (Normally, extra arguments are
12246 harmlessly ignored.)
12248 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
12249 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
12253 Do not use the calling conventions selected by @option{-mrtd}.
12254 This is the default.
12257 @itemx -mno-align-int
12258 @opindex malign-int
12259 @opindex mno-align-int
12260 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
12261 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
12262 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
12263 Aligning variables on 32-bit boundaries produces code that runs somewhat
12264 faster on processors with 32-bit busses at the expense of more memory.
12266 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
12267 align structures containing the above types differently than
12268 most published application binary interface specifications for the m68k.
12272 Use the pc-relative addressing mode of the 68000 directly, instead of
12273 using a global offset table. At present, this option implies @option{-fpic},
12274 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
12275 not presently supported with @option{-mpcrel}, though this could be supported for
12276 68020 and higher processors.
12278 @item -mno-strict-align
12279 @itemx -mstrict-align
12280 @opindex mno-strict-align
12281 @opindex mstrict-align
12282 Do not (do) assume that unaligned memory references will be handled by
12286 Generate code that allows the data segment to be located in a different
12287 area of memory from the text segment. This allows for execute in place in
12288 an environment without virtual memory management. This option implies
12291 @item -mno-sep-data
12292 Generate code that assumes that the data segment follows the text segment.
12293 This is the default.
12295 @item -mid-shared-library
12296 Generate code that supports shared libraries via the library ID method.
12297 This allows for execute in place and shared libraries in an environment
12298 without virtual memory management. This option implies @option{-fPIC}.
12300 @item -mno-id-shared-library
12301 Generate code that doesn't assume ID based shared libraries are being used.
12302 This is the default.
12304 @item -mshared-library-id=n
12305 Specified the identification number of the ID based shared library being
12306 compiled. Specifying a value of 0 will generate more compact code, specifying
12307 other values will force the allocation of that number to the current
12308 library but is no more space or time efficient than omitting this option.
12314 When generating position-independent code for ColdFire, generate code
12315 that works if the GOT has more than 8192 entries. This code is
12316 larger and slower than code generated without this option. On M680x0
12317 processors, this option is not needed; @option{-fPIC} suffices.
12319 GCC normally uses a single instruction to load values from the GOT@.
12320 While this is relatively efficient, it only works if the GOT
12321 is smaller than about 64k. Anything larger causes the linker
12322 to report an error such as:
12324 @cindex relocation truncated to fit (ColdFire)
12326 relocation truncated to fit: R_68K_GOT16O foobar
12329 If this happens, you should recompile your code with @option{-mxgot}.
12330 It should then work with very large GOTs. However, code generated with
12331 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
12332 the value of a global symbol.
12334 Note that some linkers, including newer versions of the GNU linker,
12335 can create multiple GOTs and sort GOT entries. If you have such a linker,
12336 you should only need to use @option{-mxgot} when compiling a single
12337 object file that accesses more than 8192 GOT entries. Very few do.
12339 These options have no effect unless GCC is generating
12340 position-independent code.
12344 @node M68hc1x Options
12345 @subsection M68hc1x Options
12346 @cindex M68hc1x options
12348 These are the @samp{-m} options defined for the 68hc11 and 68hc12
12349 microcontrollers. The default values for these options depends on
12350 which style of microcontroller was selected when the compiler was configured;
12351 the defaults for the most common choices are given below.
12358 Generate output for a 68HC11. This is the default
12359 when the compiler is configured for 68HC11-based systems.
12365 Generate output for a 68HC12. This is the default
12366 when the compiler is configured for 68HC12-based systems.
12372 Generate output for a 68HCS12.
12374 @item -mauto-incdec
12375 @opindex mauto-incdec
12376 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
12383 Enable the use of 68HC12 min and max instructions.
12386 @itemx -mno-long-calls
12387 @opindex mlong-calls
12388 @opindex mno-long-calls
12389 Treat all calls as being far away (near). If calls are assumed to be
12390 far away, the compiler will use the @code{call} instruction to
12391 call a function and the @code{rtc} instruction for returning.
12395 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12397 @item -msoft-reg-count=@var{count}
12398 @opindex msoft-reg-count
12399 Specify the number of pseudo-soft registers which are used for the
12400 code generation. The maximum number is 32. Using more pseudo-soft
12401 register may or may not result in better code depending on the program.
12402 The default is 4 for 68HC11 and 2 for 68HC12.
12406 @node MCore Options
12407 @subsection MCore Options
12408 @cindex MCore options
12410 These are the @samp{-m} options defined for the Motorola M*Core
12416 @itemx -mno-hardlit
12418 @opindex mno-hardlit
12419 Inline constants into the code stream if it can be done in two
12420 instructions or less.
12426 Use the divide instruction. (Enabled by default).
12428 @item -mrelax-immediate
12429 @itemx -mno-relax-immediate
12430 @opindex mrelax-immediate
12431 @opindex mno-relax-immediate
12432 Allow arbitrary sized immediates in bit operations.
12434 @item -mwide-bitfields
12435 @itemx -mno-wide-bitfields
12436 @opindex mwide-bitfields
12437 @opindex mno-wide-bitfields
12438 Always treat bit-fields as int-sized.
12440 @item -m4byte-functions
12441 @itemx -mno-4byte-functions
12442 @opindex m4byte-functions
12443 @opindex mno-4byte-functions
12444 Force all functions to be aligned to a four byte boundary.
12446 @item -mcallgraph-data
12447 @itemx -mno-callgraph-data
12448 @opindex mcallgraph-data
12449 @opindex mno-callgraph-data
12450 Emit callgraph information.
12453 @itemx -mno-slow-bytes
12454 @opindex mslow-bytes
12455 @opindex mno-slow-bytes
12456 Prefer word access when reading byte quantities.
12458 @item -mlittle-endian
12459 @itemx -mbig-endian
12460 @opindex mlittle-endian
12461 @opindex mbig-endian
12462 Generate code for a little endian target.
12468 Generate code for the 210 processor.
12472 Assume that run-time support has been provided and so omit the
12473 simulator library (@file{libsim.a)} from the linker command line.
12475 @item -mstack-increment=@var{size}
12476 @opindex mstack-increment
12477 Set the maximum amount for a single stack increment operation. Large
12478 values can increase the speed of programs which contain functions
12479 that need a large amount of stack space, but they can also trigger a
12480 segmentation fault if the stack is extended too much. The default
12486 @subsection MIPS Options
12487 @cindex MIPS options
12493 Generate big-endian code.
12497 Generate little-endian code. This is the default for @samp{mips*el-*-*}
12500 @item -march=@var{arch}
12502 Generate code that will run on @var{arch}, which can be the name of a
12503 generic MIPS ISA, or the name of a particular processor.
12505 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
12506 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
12507 The processor names are:
12508 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
12509 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
12510 @samp{5kc}, @samp{5kf},
12512 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
12513 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
12514 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
12515 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
12516 @samp{loongson2e}, @samp{loongson2f},
12520 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
12521 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
12522 @samp{rm7000}, @samp{rm9000},
12523 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
12526 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
12527 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
12529 The special value @samp{from-abi} selects the
12530 most compatible architecture for the selected ABI (that is,
12531 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
12533 Native Linux/GNU toolchains also support the value @samp{native},
12534 which selects the best architecture option for the host processor.
12535 @option{-march=native} has no effect if GCC does not recognize
12538 In processor names, a final @samp{000} can be abbreviated as @samp{k}
12539 (for example, @samp{-march=r2k}). Prefixes are optional, and
12540 @samp{vr} may be written @samp{r}.
12542 Names of the form @samp{@var{n}f2_1} refer to processors with
12543 FPUs clocked at half the rate of the core, names of the form
12544 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12545 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12546 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12547 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12548 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12549 accepted as synonyms for @samp{@var{n}f1_1}.
12551 GCC defines two macros based on the value of this option. The first
12552 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12553 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
12554 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12555 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12556 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12558 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12559 above. In other words, it will have the full prefix and will not
12560 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
12561 the macro names the resolved architecture (either @samp{"mips1"} or
12562 @samp{"mips3"}). It names the default architecture when no
12563 @option{-march} option is given.
12565 @item -mtune=@var{arch}
12567 Optimize for @var{arch}. Among other things, this option controls
12568 the way instructions are scheduled, and the perceived cost of arithmetic
12569 operations. The list of @var{arch} values is the same as for
12572 When this option is not used, GCC will optimize for the processor
12573 specified by @option{-march}. By using @option{-march} and
12574 @option{-mtune} together, it is possible to generate code that will
12575 run on a family of processors, but optimize the code for one
12576 particular member of that family.
12578 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12579 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12580 @samp{-march} ones described above.
12584 Equivalent to @samp{-march=mips1}.
12588 Equivalent to @samp{-march=mips2}.
12592 Equivalent to @samp{-march=mips3}.
12596 Equivalent to @samp{-march=mips4}.
12600 Equivalent to @samp{-march=mips32}.
12604 Equivalent to @samp{-march=mips32r2}.
12608 Equivalent to @samp{-march=mips64}.
12612 Equivalent to @samp{-march=mips64r2}.
12617 @opindex mno-mips16
12618 Generate (do not generate) MIPS16 code. If GCC is targetting a
12619 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12621 MIPS16 code generation can also be controlled on a per-function basis
12622 by means of @code{mips16} and @code{nomips16} attributes.
12623 @xref{Function Attributes}, for more information.
12625 @item -mflip-mips16
12626 @opindex mflip-mips16
12627 Generate MIPS16 code on alternating functions. This option is provided
12628 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12629 not intended for ordinary use in compiling user code.
12631 @item -minterlink-mips16
12632 @itemx -mno-interlink-mips16
12633 @opindex minterlink-mips16
12634 @opindex mno-interlink-mips16
12635 Require (do not require) that non-MIPS16 code be link-compatible with
12638 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12639 it must either use a call or an indirect jump. @option{-minterlink-mips16}
12640 therefore disables direct jumps unless GCC knows that the target of the
12641 jump is not MIPS16.
12653 Generate code for the given ABI@.
12655 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
12656 generates 64-bit code when you select a 64-bit architecture, but you
12657 can use @option{-mgp32} to get 32-bit code instead.
12659 For information about the O64 ABI, see
12660 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12662 GCC supports a variant of the o32 ABI in which floating-point registers
12663 are 64 rather than 32 bits wide. You can select this combination with
12664 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
12665 and @samp{mfhc1} instructions and is therefore only supported for
12666 MIPS32R2 processors.
12668 The register assignments for arguments and return values remain the
12669 same, but each scalar value is passed in a single 64-bit register
12670 rather than a pair of 32-bit registers. For example, scalar
12671 floating-point values are returned in @samp{$f0} only, not a
12672 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12673 remains the same, but all 64 bits are saved.
12676 @itemx -mno-abicalls
12678 @opindex mno-abicalls
12679 Generate (do not generate) code that is suitable for SVR4-style
12680 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12685 Generate (do not generate) code that is fully position-independent,
12686 and that can therefore be linked into shared libraries. This option
12687 only affects @option{-mabicalls}.
12689 All @option{-mabicalls} code has traditionally been position-independent,
12690 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12691 as an extension, the GNU toolchain allows executables to use absolute
12692 accesses for locally-binding symbols. It can also use shorter GP
12693 initialization sequences and generate direct calls to locally-defined
12694 functions. This mode is selected by @option{-mno-shared}.
12696 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12697 objects that can only be linked by the GNU linker. However, the option
12698 does not affect the ABI of the final executable; it only affects the ABI
12699 of relocatable objects. Using @option{-mno-shared} will generally make
12700 executables both smaller and quicker.
12702 @option{-mshared} is the default.
12708 Assume (do not assume) that the static and dynamic linkers
12709 support PLTs and copy relocations. This option only affects
12710 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
12711 has no effect without @samp{-msym32}.
12713 You can make @option{-mplt} the default by configuring
12714 GCC with @option{--with-mips-plt}. The default is
12715 @option{-mno-plt} otherwise.
12721 Lift (do not lift) the usual restrictions on the size of the global
12724 GCC normally uses a single instruction to load values from the GOT@.
12725 While this is relatively efficient, it will only work if the GOT
12726 is smaller than about 64k. Anything larger will cause the linker
12727 to report an error such as:
12729 @cindex relocation truncated to fit (MIPS)
12731 relocation truncated to fit: R_MIPS_GOT16 foobar
12734 If this happens, you should recompile your code with @option{-mxgot}.
12735 It should then work with very large GOTs, although it will also be
12736 less efficient, since it will take three instructions to fetch the
12737 value of a global symbol.
12739 Note that some linkers can create multiple GOTs. If you have such a
12740 linker, you should only need to use @option{-mxgot} when a single object
12741 file accesses more than 64k's worth of GOT entries. Very few do.
12743 These options have no effect unless GCC is generating position
12748 Assume that general-purpose registers are 32 bits wide.
12752 Assume that general-purpose registers are 64 bits wide.
12756 Assume that floating-point registers are 32 bits wide.
12760 Assume that floating-point registers are 64 bits wide.
12763 @opindex mhard-float
12764 Use floating-point coprocessor instructions.
12767 @opindex msoft-float
12768 Do not use floating-point coprocessor instructions. Implement
12769 floating-point calculations using library calls instead.
12771 @item -msingle-float
12772 @opindex msingle-float
12773 Assume that the floating-point coprocessor only supports single-precision
12776 @item -mdouble-float
12777 @opindex mdouble-float
12778 Assume that the floating-point coprocessor supports double-precision
12779 operations. This is the default.
12785 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12786 implement atomic memory built-in functions. When neither option is
12787 specified, GCC will use the instructions if the target architecture
12790 @option{-mllsc} is useful if the runtime environment can emulate the
12791 instructions and @option{-mno-llsc} can be useful when compiling for
12792 nonstandard ISAs. You can make either option the default by
12793 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12794 respectively. @option{--with-llsc} is the default for some
12795 configurations; see the installation documentation for details.
12801 Use (do not use) revision 1 of the MIPS DSP ASE@.
12802 @xref{MIPS DSP Built-in Functions}. This option defines the
12803 preprocessor macro @samp{__mips_dsp}. It also defines
12804 @samp{__mips_dsp_rev} to 1.
12810 Use (do not use) revision 2 of the MIPS DSP ASE@.
12811 @xref{MIPS DSP Built-in Functions}. This option defines the
12812 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12813 It also defines @samp{__mips_dsp_rev} to 2.
12816 @itemx -mno-smartmips
12817 @opindex msmartmips
12818 @opindex mno-smartmips
12819 Use (do not use) the MIPS SmartMIPS ASE.
12821 @item -mpaired-single
12822 @itemx -mno-paired-single
12823 @opindex mpaired-single
12824 @opindex mno-paired-single
12825 Use (do not use) paired-single floating-point instructions.
12826 @xref{MIPS Paired-Single Support}. This option requires
12827 hardware floating-point support to be enabled.
12833 Use (do not use) MIPS Digital Media Extension instructions.
12834 This option can only be used when generating 64-bit code and requires
12835 hardware floating-point support to be enabled.
12840 @opindex mno-mips3d
12841 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12842 The option @option{-mips3d} implies @option{-mpaired-single}.
12848 Use (do not use) MT Multithreading instructions.
12852 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12853 an explanation of the default and the way that the pointer size is
12858 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12860 The default size of @code{int}s, @code{long}s and pointers depends on
12861 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12862 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12863 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12864 or the same size as integer registers, whichever is smaller.
12870 Assume (do not assume) that all symbols have 32-bit values, regardless
12871 of the selected ABI@. This option is useful in combination with
12872 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12873 to generate shorter and faster references to symbolic addresses.
12877 Put definitions of externally-visible data in a small data section
12878 if that data is no bigger than @var{num} bytes. GCC can then access
12879 the data more efficiently; see @option{-mgpopt} for details.
12881 The default @option{-G} option depends on the configuration.
12883 @item -mlocal-sdata
12884 @itemx -mno-local-sdata
12885 @opindex mlocal-sdata
12886 @opindex mno-local-sdata
12887 Extend (do not extend) the @option{-G} behavior to local data too,
12888 such as to static variables in C@. @option{-mlocal-sdata} is the
12889 default for all configurations.
12891 If the linker complains that an application is using too much small data,
12892 you might want to try rebuilding the less performance-critical parts with
12893 @option{-mno-local-sdata}. You might also want to build large
12894 libraries with @option{-mno-local-sdata}, so that the libraries leave
12895 more room for the main program.
12897 @item -mextern-sdata
12898 @itemx -mno-extern-sdata
12899 @opindex mextern-sdata
12900 @opindex mno-extern-sdata
12901 Assume (do not assume) that externally-defined data will be in
12902 a small data section if that data is within the @option{-G} limit.
12903 @option{-mextern-sdata} is the default for all configurations.
12905 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12906 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12907 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12908 is placed in a small data section. If @var{Var} is defined by another
12909 module, you must either compile that module with a high-enough
12910 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12911 definition. If @var{Var} is common, you must link the application
12912 with a high-enough @option{-G} setting.
12914 The easiest way of satisfying these restrictions is to compile
12915 and link every module with the same @option{-G} option. However,
12916 you may wish to build a library that supports several different
12917 small data limits. You can do this by compiling the library with
12918 the highest supported @option{-G} setting and additionally using
12919 @option{-mno-extern-sdata} to stop the library from making assumptions
12920 about externally-defined data.
12926 Use (do not use) GP-relative accesses for symbols that are known to be
12927 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12928 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12931 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12932 might not hold the value of @code{_gp}. For example, if the code is
12933 part of a library that might be used in a boot monitor, programs that
12934 call boot monitor routines will pass an unknown value in @code{$gp}.
12935 (In such situations, the boot monitor itself would usually be compiled
12936 with @option{-G0}.)
12938 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12939 @option{-mno-extern-sdata}.
12941 @item -membedded-data
12942 @itemx -mno-embedded-data
12943 @opindex membedded-data
12944 @opindex mno-embedded-data
12945 Allocate variables to the read-only data section first if possible, then
12946 next in the small data section if possible, otherwise in data. This gives
12947 slightly slower code than the default, but reduces the amount of RAM required
12948 when executing, and thus may be preferred for some embedded systems.
12950 @item -muninit-const-in-rodata
12951 @itemx -mno-uninit-const-in-rodata
12952 @opindex muninit-const-in-rodata
12953 @opindex mno-uninit-const-in-rodata
12954 Put uninitialized @code{const} variables in the read-only data section.
12955 This option is only meaningful in conjunction with @option{-membedded-data}.
12957 @item -mcode-readable=@var{setting}
12958 @opindex mcode-readable
12959 Specify whether GCC may generate code that reads from executable sections.
12960 There are three possible settings:
12963 @item -mcode-readable=yes
12964 Instructions may freely access executable sections. This is the
12967 @item -mcode-readable=pcrel
12968 MIPS16 PC-relative load instructions can access executable sections,
12969 but other instructions must not do so. This option is useful on 4KSc
12970 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12971 It is also useful on processors that can be configured to have a dual
12972 instruction/data SRAM interface and that, like the M4K, automatically
12973 redirect PC-relative loads to the instruction RAM.
12975 @item -mcode-readable=no
12976 Instructions must not access executable sections. This option can be
12977 useful on targets that are configured to have a dual instruction/data
12978 SRAM interface but that (unlike the M4K) do not automatically redirect
12979 PC-relative loads to the instruction RAM.
12982 @item -msplit-addresses
12983 @itemx -mno-split-addresses
12984 @opindex msplit-addresses
12985 @opindex mno-split-addresses
12986 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12987 relocation operators. This option has been superseded by
12988 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12990 @item -mexplicit-relocs
12991 @itemx -mno-explicit-relocs
12992 @opindex mexplicit-relocs
12993 @opindex mno-explicit-relocs
12994 Use (do not use) assembler relocation operators when dealing with symbolic
12995 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12996 is to use assembler macros instead.
12998 @option{-mexplicit-relocs} is the default if GCC was configured
12999 to use an assembler that supports relocation operators.
13001 @item -mcheck-zero-division
13002 @itemx -mno-check-zero-division
13003 @opindex mcheck-zero-division
13004 @opindex mno-check-zero-division
13005 Trap (do not trap) on integer division by zero.
13007 The default is @option{-mcheck-zero-division}.
13009 @item -mdivide-traps
13010 @itemx -mdivide-breaks
13011 @opindex mdivide-traps
13012 @opindex mdivide-breaks
13013 MIPS systems check for division by zero by generating either a
13014 conditional trap or a break instruction. Using traps results in
13015 smaller code, but is only supported on MIPS II and later. Also, some
13016 versions of the Linux kernel have a bug that prevents trap from
13017 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
13018 allow conditional traps on architectures that support them and
13019 @option{-mdivide-breaks} to force the use of breaks.
13021 The default is usually @option{-mdivide-traps}, but this can be
13022 overridden at configure time using @option{--with-divide=breaks}.
13023 Divide-by-zero checks can be completely disabled using
13024 @option{-mno-check-zero-division}.
13029 @opindex mno-memcpy
13030 Force (do not force) the use of @code{memcpy()} for non-trivial block
13031 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
13032 most constant-sized copies.
13035 @itemx -mno-long-calls
13036 @opindex mlong-calls
13037 @opindex mno-long-calls
13038 Disable (do not disable) use of the @code{jal} instruction. Calling
13039 functions using @code{jal} is more efficient but requires the caller
13040 and callee to be in the same 256 megabyte segment.
13042 This option has no effect on abicalls code. The default is
13043 @option{-mno-long-calls}.
13049 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
13050 instructions, as provided by the R4650 ISA@.
13053 @itemx -mno-fused-madd
13054 @opindex mfused-madd
13055 @opindex mno-fused-madd
13056 Enable (disable) use of the floating point multiply-accumulate
13057 instructions, when they are available. The default is
13058 @option{-mfused-madd}.
13060 When multiply-accumulate instructions are used, the intermediate
13061 product is calculated to infinite precision and is not subject to
13062 the FCSR Flush to Zero bit. This may be undesirable in some
13067 Tell the MIPS assembler to not run its preprocessor over user
13068 assembler files (with a @samp{.s} suffix) when assembling them.
13071 @itemx -mno-fix-r4000
13072 @opindex mfix-r4000
13073 @opindex mno-fix-r4000
13074 Work around certain R4000 CPU errata:
13077 A double-word or a variable shift may give an incorrect result if executed
13078 immediately after starting an integer division.
13080 A double-word or a variable shift may give an incorrect result if executed
13081 while an integer multiplication is in progress.
13083 An integer division may give an incorrect result if started in a delay slot
13084 of a taken branch or a jump.
13088 @itemx -mno-fix-r4400
13089 @opindex mfix-r4400
13090 @opindex mno-fix-r4400
13091 Work around certain R4400 CPU errata:
13094 A double-word or a variable shift may give an incorrect result if executed
13095 immediately after starting an integer division.
13099 @itemx -mno-fix-r10000
13100 @opindex mfix-r10000
13101 @opindex mno-fix-r10000
13102 Work around certain R10000 errata:
13105 @code{ll}/@code{sc} sequences may not behave atomically on revisions
13106 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
13109 This option can only be used if the target architecture supports
13110 branch-likely instructions. @option{-mfix-r10000} is the default when
13111 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
13115 @itemx -mno-fix-vr4120
13116 @opindex mfix-vr4120
13117 Work around certain VR4120 errata:
13120 @code{dmultu} does not always produce the correct result.
13122 @code{div} and @code{ddiv} do not always produce the correct result if one
13123 of the operands is negative.
13125 The workarounds for the division errata rely on special functions in
13126 @file{libgcc.a}. At present, these functions are only provided by
13127 the @code{mips64vr*-elf} configurations.
13129 Other VR4120 errata require a nop to be inserted between certain pairs of
13130 instructions. These errata are handled by the assembler, not by GCC itself.
13133 @opindex mfix-vr4130
13134 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
13135 workarounds are implemented by the assembler rather than by GCC,
13136 although GCC will avoid using @code{mflo} and @code{mfhi} if the
13137 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
13138 instructions are available instead.
13141 @itemx -mno-fix-sb1
13143 Work around certain SB-1 CPU core errata.
13144 (This flag currently works around the SB-1 revision 2
13145 ``F1'' and ``F2'' floating point errata.)
13147 @item -mr10k-cache-barrier=@var{setting}
13148 @opindex mr10k-cache-barrier
13149 Specify whether GCC should insert cache barriers to avoid the
13150 side-effects of speculation on R10K processors.
13152 In common with many processors, the R10K tries to predict the outcome
13153 of a conditional branch and speculatively executes instructions from
13154 the ``taken'' branch. It later aborts these instructions if the
13155 predicted outcome was wrong. However, on the R10K, even aborted
13156 instructions can have side effects.
13158 This problem only affects kernel stores and, depending on the system,
13159 kernel loads. As an example, a speculatively-executed store may load
13160 the target memory into cache and mark the cache line as dirty, even if
13161 the store itself is later aborted. If a DMA operation writes to the
13162 same area of memory before the ``dirty'' line is flushed, the cached
13163 data will overwrite the DMA-ed data. See the R10K processor manual
13164 for a full description, including other potential problems.
13166 One workaround is to insert cache barrier instructions before every memory
13167 access that might be speculatively executed and that might have side
13168 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
13169 controls GCC's implementation of this workaround. It assumes that
13170 aborted accesses to any byte in the following regions will not have
13175 the memory occupied by the current function's stack frame;
13178 the memory occupied by an incoming stack argument;
13181 the memory occupied by an object with a link-time-constant address.
13184 It is the kernel's responsibility to ensure that speculative
13185 accesses to these regions are indeed safe.
13187 If the input program contains a function declaration such as:
13193 then the implementation of @code{foo} must allow @code{j foo} and
13194 @code{jal foo} to be executed speculatively. GCC honors this
13195 restriction for functions it compiles itself. It expects non-GCC
13196 functions (such as hand-written assembly code) to do the same.
13198 The option has three forms:
13201 @item -mr10k-cache-barrier=load-store
13202 Insert a cache barrier before a load or store that might be
13203 speculatively executed and that might have side effects even
13206 @item -mr10k-cache-barrier=store
13207 Insert a cache barrier before a store that might be speculatively
13208 executed and that might have side effects even if aborted.
13210 @item -mr10k-cache-barrier=none
13211 Disable the insertion of cache barriers. This is the default setting.
13214 @item -mflush-func=@var{func}
13215 @itemx -mno-flush-func
13216 @opindex mflush-func
13217 Specifies the function to call to flush the I and D caches, or to not
13218 call any such function. If called, the function must take the same
13219 arguments as the common @code{_flush_func()}, that is, the address of the
13220 memory range for which the cache is being flushed, the size of the
13221 memory range, and the number 3 (to flush both caches). The default
13222 depends on the target GCC was configured for, but commonly is either
13223 @samp{_flush_func} or @samp{__cpu_flush}.
13225 @item mbranch-cost=@var{num}
13226 @opindex mbranch-cost
13227 Set the cost of branches to roughly @var{num} ``simple'' instructions.
13228 This cost is only a heuristic and is not guaranteed to produce
13229 consistent results across releases. A zero cost redundantly selects
13230 the default, which is based on the @option{-mtune} setting.
13232 @item -mbranch-likely
13233 @itemx -mno-branch-likely
13234 @opindex mbranch-likely
13235 @opindex mno-branch-likely
13236 Enable or disable use of Branch Likely instructions, regardless of the
13237 default for the selected architecture. By default, Branch Likely
13238 instructions may be generated if they are supported by the selected
13239 architecture. An exception is for the MIPS32 and MIPS64 architectures
13240 and processors which implement those architectures; for those, Branch
13241 Likely instructions will not be generated by default because the MIPS32
13242 and MIPS64 architectures specifically deprecate their use.
13244 @item -mfp-exceptions
13245 @itemx -mno-fp-exceptions
13246 @opindex mfp-exceptions
13247 Specifies whether FP exceptions are enabled. This affects how we schedule
13248 FP instructions for some processors. The default is that FP exceptions are
13251 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
13252 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
13255 @item -mvr4130-align
13256 @itemx -mno-vr4130-align
13257 @opindex mvr4130-align
13258 The VR4130 pipeline is two-way superscalar, but can only issue two
13259 instructions together if the first one is 8-byte aligned. When this
13260 option is enabled, GCC will align pairs of instructions that it
13261 thinks should execute in parallel.
13263 This option only has an effect when optimizing for the VR4130.
13264 It normally makes code faster, but at the expense of making it bigger.
13265 It is enabled by default at optimization level @option{-O3}.
13269 @subsection MMIX Options
13270 @cindex MMIX Options
13272 These options are defined for the MMIX:
13276 @itemx -mno-libfuncs
13278 @opindex mno-libfuncs
13279 Specify that intrinsic library functions are being compiled, passing all
13280 values in registers, no matter the size.
13283 @itemx -mno-epsilon
13285 @opindex mno-epsilon
13286 Generate floating-point comparison instructions that compare with respect
13287 to the @code{rE} epsilon register.
13289 @item -mabi=mmixware
13291 @opindex mabi-mmixware
13293 Generate code that passes function parameters and return values that (in
13294 the called function) are seen as registers @code{$0} and up, as opposed to
13295 the GNU ABI which uses global registers @code{$231} and up.
13297 @item -mzero-extend
13298 @itemx -mno-zero-extend
13299 @opindex mzero-extend
13300 @opindex mno-zero-extend
13301 When reading data from memory in sizes shorter than 64 bits, use (do not
13302 use) zero-extending load instructions by default, rather than
13303 sign-extending ones.
13306 @itemx -mno-knuthdiv
13308 @opindex mno-knuthdiv
13309 Make the result of a division yielding a remainder have the same sign as
13310 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
13311 remainder follows the sign of the dividend. Both methods are
13312 arithmetically valid, the latter being almost exclusively used.
13314 @item -mtoplevel-symbols
13315 @itemx -mno-toplevel-symbols
13316 @opindex mtoplevel-symbols
13317 @opindex mno-toplevel-symbols
13318 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
13319 code can be used with the @code{PREFIX} assembly directive.
13323 Generate an executable in the ELF format, rather than the default
13324 @samp{mmo} format used by the @command{mmix} simulator.
13326 @item -mbranch-predict
13327 @itemx -mno-branch-predict
13328 @opindex mbranch-predict
13329 @opindex mno-branch-predict
13330 Use (do not use) the probable-branch instructions, when static branch
13331 prediction indicates a probable branch.
13333 @item -mbase-addresses
13334 @itemx -mno-base-addresses
13335 @opindex mbase-addresses
13336 @opindex mno-base-addresses
13337 Generate (do not generate) code that uses @emph{base addresses}. Using a
13338 base address automatically generates a request (handled by the assembler
13339 and the linker) for a constant to be set up in a global register. The
13340 register is used for one or more base address requests within the range 0
13341 to 255 from the value held in the register. The generally leads to short
13342 and fast code, but the number of different data items that can be
13343 addressed is limited. This means that a program that uses lots of static
13344 data may require @option{-mno-base-addresses}.
13346 @item -msingle-exit
13347 @itemx -mno-single-exit
13348 @opindex msingle-exit
13349 @opindex mno-single-exit
13350 Force (do not force) generated code to have a single exit point in each
13354 @node MN10300 Options
13355 @subsection MN10300 Options
13356 @cindex MN10300 options
13358 These @option{-m} options are defined for Matsushita MN10300 architectures:
13363 Generate code to avoid bugs in the multiply instructions for the MN10300
13364 processors. This is the default.
13366 @item -mno-mult-bug
13367 @opindex mno-mult-bug
13368 Do not generate code to avoid bugs in the multiply instructions for the
13369 MN10300 processors.
13373 Generate code which uses features specific to the AM33 processor.
13377 Do not generate code which uses features specific to the AM33 processor. This
13380 @item -mreturn-pointer-on-d0
13381 @opindex mreturn-pointer-on-d0
13382 When generating a function which returns a pointer, return the pointer
13383 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
13384 only in a0, and attempts to call such functions without a prototype
13385 would result in errors. Note that this option is on by default; use
13386 @option{-mno-return-pointer-on-d0} to disable it.
13390 Do not link in the C run-time initialization object file.
13394 Indicate to the linker that it should perform a relaxation optimization pass
13395 to shorten branches, calls and absolute memory addresses. This option only
13396 has an effect when used on the command line for the final link step.
13398 This option makes symbolic debugging impossible.
13401 @node PDP-11 Options
13402 @subsection PDP-11 Options
13403 @cindex PDP-11 Options
13405 These options are defined for the PDP-11:
13410 Use hardware FPP floating point. This is the default. (FIS floating
13411 point on the PDP-11/40 is not supported.)
13414 @opindex msoft-float
13415 Do not use hardware floating point.
13419 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
13423 Return floating-point results in memory. This is the default.
13427 Generate code for a PDP-11/40.
13431 Generate code for a PDP-11/45. This is the default.
13435 Generate code for a PDP-11/10.
13437 @item -mbcopy-builtin
13438 @opindex bcopy-builtin
13439 Use inline @code{movmemhi} patterns for copying memory. This is the
13444 Do not use inline @code{movmemhi} patterns for copying memory.
13450 Use 16-bit @code{int}. This is the default.
13456 Use 32-bit @code{int}.
13459 @itemx -mno-float32
13461 @opindex mno-float32
13462 Use 64-bit @code{float}. This is the default.
13465 @itemx -mno-float64
13467 @opindex mno-float64
13468 Use 32-bit @code{float}.
13472 Use @code{abshi2} pattern. This is the default.
13476 Do not use @code{abshi2} pattern.
13478 @item -mbranch-expensive
13479 @opindex mbranch-expensive
13480 Pretend that branches are expensive. This is for experimenting with
13481 code generation only.
13483 @item -mbranch-cheap
13484 @opindex mbranch-cheap
13485 Do not pretend that branches are expensive. This is the default.
13489 Generate code for a system with split I&D@.
13493 Generate code for a system without split I&D@. This is the default.
13497 Use Unix assembler syntax. This is the default when configured for
13498 @samp{pdp11-*-bsd}.
13502 Use DEC assembler syntax. This is the default when configured for any
13503 PDP-11 target other than @samp{pdp11-*-bsd}.
13506 @node picoChip Options
13507 @subsection picoChip Options
13508 @cindex picoChip options
13510 These @samp{-m} options are defined for picoChip implementations:
13514 @item -mae=@var{ae_type}
13516 Set the instruction set, register set, and instruction scheduling
13517 parameters for array element type @var{ae_type}. Supported values
13518 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
13520 @option{-mae=ANY} selects a completely generic AE type. Code
13521 generated with this option will run on any of the other AE types. The
13522 code will not be as efficient as it would be if compiled for a specific
13523 AE type, and some types of operation (e.g., multiplication) will not
13524 work properly on all types of AE.
13526 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
13527 for compiled code, and is the default.
13529 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
13530 option may suffer from poor performance of byte (char) manipulation,
13531 since the DSP AE does not provide hardware support for byte load/stores.
13533 @item -msymbol-as-address
13534 Enable the compiler to directly use a symbol name as an address in a
13535 load/store instruction, without first loading it into a
13536 register. Typically, the use of this option will generate larger
13537 programs, which run faster than when the option isn't used. However, the
13538 results vary from program to program, so it is left as a user option,
13539 rather than being permanently enabled.
13541 @item -mno-inefficient-warnings
13542 Disables warnings about the generation of inefficient code. These
13543 warnings can be generated, for example, when compiling code which
13544 performs byte-level memory operations on the MAC AE type. The MAC AE has
13545 no hardware support for byte-level memory operations, so all byte
13546 load/stores must be synthesized from word load/store operations. This is
13547 inefficient and a warning will be generated indicating to the programmer
13548 that they should rewrite the code to avoid byte operations, or to target
13549 an AE type which has the necessary hardware support. This option enables
13550 the warning to be turned off.
13554 @node PowerPC Options
13555 @subsection PowerPC Options
13556 @cindex PowerPC options
13558 These are listed under @xref{RS/6000 and PowerPC Options}.
13560 @node RS/6000 and PowerPC Options
13561 @subsection IBM RS/6000 and PowerPC Options
13562 @cindex RS/6000 and PowerPC Options
13563 @cindex IBM RS/6000 and PowerPC Options
13565 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
13572 @itemx -mno-powerpc
13573 @itemx -mpowerpc-gpopt
13574 @itemx -mno-powerpc-gpopt
13575 @itemx -mpowerpc-gfxopt
13576 @itemx -mno-powerpc-gfxopt
13578 @itemx -mno-powerpc64
13582 @itemx -mno-popcntb
13590 @itemx -mno-hard-dfp
13594 @opindex mno-power2
13596 @opindex mno-powerpc
13597 @opindex mpowerpc-gpopt
13598 @opindex mno-powerpc-gpopt
13599 @opindex mpowerpc-gfxopt
13600 @opindex mno-powerpc-gfxopt
13601 @opindex mpowerpc64
13602 @opindex mno-powerpc64
13606 @opindex mno-popcntb
13612 @opindex mno-mfpgpr
13614 @opindex mno-hard-dfp
13615 GCC supports two related instruction set architectures for the
13616 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
13617 instructions supported by the @samp{rios} chip set used in the original
13618 RS/6000 systems and the @dfn{PowerPC} instruction set is the
13619 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
13620 the IBM 4xx, 6xx, and follow-on microprocessors.
13622 Neither architecture is a subset of the other. However there is a
13623 large common subset of instructions supported by both. An MQ
13624 register is included in processors supporting the POWER architecture.
13626 You use these options to specify which instructions are available on the
13627 processor you are using. The default value of these options is
13628 determined when configuring GCC@. Specifying the
13629 @option{-mcpu=@var{cpu_type}} overrides the specification of these
13630 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
13631 rather than the options listed above.
13633 The @option{-mpower} option allows GCC to generate instructions that
13634 are found only in the POWER architecture and to use the MQ register.
13635 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
13636 to generate instructions that are present in the POWER2 architecture but
13637 not the original POWER architecture.
13639 The @option{-mpowerpc} option allows GCC to generate instructions that
13640 are found only in the 32-bit subset of the PowerPC architecture.
13641 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
13642 GCC to use the optional PowerPC architecture instructions in the
13643 General Purpose group, including floating-point square root. Specifying
13644 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
13645 use the optional PowerPC architecture instructions in the Graphics
13646 group, including floating-point select.
13648 The @option{-mmfcrf} option allows GCC to generate the move from
13649 condition register field instruction implemented on the POWER4
13650 processor and other processors that support the PowerPC V2.01
13652 The @option{-mpopcntb} option allows GCC to generate the popcount and
13653 double precision FP reciprocal estimate instruction implemented on the
13654 POWER5 processor and other processors that support the PowerPC V2.02
13656 The @option{-mfprnd} option allows GCC to generate the FP round to
13657 integer instructions implemented on the POWER5+ processor and other
13658 processors that support the PowerPC V2.03 architecture.
13659 The @option{-mcmpb} option allows GCC to generate the compare bytes
13660 instruction implemented on the POWER6 processor and other processors
13661 that support the PowerPC V2.05 architecture.
13662 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
13663 general purpose register instructions implemented on the POWER6X
13664 processor and other processors that support the extended PowerPC V2.05
13666 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
13667 point instructions implemented on some POWER processors.
13669 The @option{-mpowerpc64} option allows GCC to generate the additional
13670 64-bit instructions that are found in the full PowerPC64 architecture
13671 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
13672 @option{-mno-powerpc64}.
13674 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
13675 will use only the instructions in the common subset of both
13676 architectures plus some special AIX common-mode calls, and will not use
13677 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
13678 permits GCC to use any instruction from either architecture and to
13679 allow use of the MQ register; specify this for the Motorola MPC601.
13681 @item -mnew-mnemonics
13682 @itemx -mold-mnemonics
13683 @opindex mnew-mnemonics
13684 @opindex mold-mnemonics
13685 Select which mnemonics to use in the generated assembler code. With
13686 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
13687 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
13688 assembler mnemonics defined for the POWER architecture. Instructions
13689 defined in only one architecture have only one mnemonic; GCC uses that
13690 mnemonic irrespective of which of these options is specified.
13692 GCC defaults to the mnemonics appropriate for the architecture in
13693 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13694 value of these option. Unless you are building a cross-compiler, you
13695 should normally not specify either @option{-mnew-mnemonics} or
13696 @option{-mold-mnemonics}, but should instead accept the default.
13698 @item -mcpu=@var{cpu_type}
13700 Set architecture type, register usage, choice of mnemonics, and
13701 instruction scheduling parameters for machine type @var{cpu_type}.
13702 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13703 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13704 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13705 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13706 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13707 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13708 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13709 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13710 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
13711 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13712 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13714 @option{-mcpu=common} selects a completely generic processor. Code
13715 generated under this option will run on any POWER or PowerPC processor.
13716 GCC will use only the instructions in the common subset of both
13717 architectures, and will not use the MQ register. GCC assumes a generic
13718 processor model for scheduling purposes.
13720 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13721 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13722 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13723 types, with an appropriate, generic processor model assumed for
13724 scheduling purposes.
13726 The other options specify a specific processor. Code generated under
13727 those options will run best on that processor, and may not run at all on
13730 The @option{-mcpu} options automatically enable or disable the
13733 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
13734 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
13735 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
13736 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr}
13738 The particular options set for any particular CPU will vary between
13739 compiler versions, depending on what setting seems to produce optimal
13740 code for that CPU; it doesn't necessarily reflect the actual hardware's
13741 capabilities. If you wish to set an individual option to a particular
13742 value, you may specify it after the @option{-mcpu} option, like
13743 @samp{-mcpu=970 -mno-altivec}.
13745 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13746 not enabled or disabled by the @option{-mcpu} option at present because
13747 AIX does not have full support for these options. You may still
13748 enable or disable them individually if you're sure it'll work in your
13751 @item -mtune=@var{cpu_type}
13753 Set the instruction scheduling parameters for machine type
13754 @var{cpu_type}, but do not set the architecture type, register usage, or
13755 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
13756 values for @var{cpu_type} are used for @option{-mtune} as for
13757 @option{-mcpu}. If both are specified, the code generated will use the
13758 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13759 scheduling parameters set by @option{-mtune}.
13765 Generate code to compute division as reciprocal estimate and iterative
13766 refinement, creating opportunities for increased throughput. This
13767 feature requires: optional PowerPC Graphics instruction set for single
13768 precision and FRE instruction for double precision, assuming divides
13769 cannot generate user-visible traps, and the domain values not include
13770 Infinities, denormals or zero denominator.
13773 @itemx -mno-altivec
13775 @opindex mno-altivec
13776 Generate code that uses (does not use) AltiVec instructions, and also
13777 enable the use of built-in functions that allow more direct access to
13778 the AltiVec instruction set. You may also need to set
13779 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13785 @opindex mno-vrsave
13786 Generate VRSAVE instructions when generating AltiVec code.
13788 @item -mgen-cell-microcode
13789 @opindex mgen-cell-microcode
13790 Generate Cell microcode instructions
13792 @item -mwarn-cell-microcode
13793 @opindex mwarn-cell-microcode
13794 Warning when a Cell microcode instruction is going to emitted. An example
13795 of a Cell microcode instruction is a variable shift.
13798 @opindex msecure-plt
13799 Generate code that allows ld and ld.so to build executables and shared
13800 libraries with non-exec .plt and .got sections. This is a PowerPC
13801 32-bit SYSV ABI option.
13805 Generate code that uses a BSS .plt section that ld.so fills in, and
13806 requires .plt and .got sections that are both writable and executable.
13807 This is a PowerPC 32-bit SYSV ABI option.
13813 This switch enables or disables the generation of ISEL instructions.
13815 @item -misel=@var{yes/no}
13816 This switch has been deprecated. Use @option{-misel} and
13817 @option{-mno-isel} instead.
13823 This switch enables or disables the generation of SPE simd
13829 @opindex mno-paired
13830 This switch enables or disables the generation of PAIRED simd
13833 @item -mspe=@var{yes/no}
13834 This option has been deprecated. Use @option{-mspe} and
13835 @option{-mno-spe} instead.
13837 @item -mfloat-gprs=@var{yes/single/double/no}
13838 @itemx -mfloat-gprs
13839 @opindex mfloat-gprs
13840 This switch enables or disables the generation of floating point
13841 operations on the general purpose registers for architectures that
13844 The argument @var{yes} or @var{single} enables the use of
13845 single-precision floating point operations.
13847 The argument @var{double} enables the use of single and
13848 double-precision floating point operations.
13850 The argument @var{no} disables floating point operations on the
13851 general purpose registers.
13853 This option is currently only available on the MPC854x.
13859 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13860 targets (including GNU/Linux). The 32-bit environment sets int, long
13861 and pointer to 32 bits and generates code that runs on any PowerPC
13862 variant. The 64-bit environment sets int to 32 bits and long and
13863 pointer to 64 bits, and generates code for PowerPC64, as for
13864 @option{-mpowerpc64}.
13867 @itemx -mno-fp-in-toc
13868 @itemx -mno-sum-in-toc
13869 @itemx -mminimal-toc
13871 @opindex mno-fp-in-toc
13872 @opindex mno-sum-in-toc
13873 @opindex mminimal-toc
13874 Modify generation of the TOC (Table Of Contents), which is created for
13875 every executable file. The @option{-mfull-toc} option is selected by
13876 default. In that case, GCC will allocate at least one TOC entry for
13877 each unique non-automatic variable reference in your program. GCC
13878 will also place floating-point constants in the TOC@. However, only
13879 16,384 entries are available in the TOC@.
13881 If you receive a linker error message that saying you have overflowed
13882 the available TOC space, you can reduce the amount of TOC space used
13883 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13884 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13885 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13886 generate code to calculate the sum of an address and a constant at
13887 run-time instead of putting that sum into the TOC@. You may specify one
13888 or both of these options. Each causes GCC to produce very slightly
13889 slower and larger code at the expense of conserving TOC space.
13891 If you still run out of space in the TOC even when you specify both of
13892 these options, specify @option{-mminimal-toc} instead. This option causes
13893 GCC to make only one TOC entry for every file. When you specify this
13894 option, GCC will produce code that is slower and larger but which
13895 uses extremely little TOC space. You may wish to use this option
13896 only on files that contain less frequently executed code.
13902 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13903 @code{long} type, and the infrastructure needed to support them.
13904 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13905 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13906 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13909 @itemx -mno-xl-compat
13910 @opindex mxl-compat
13911 @opindex mno-xl-compat
13912 Produce code that conforms more closely to IBM XL compiler semantics
13913 when using AIX-compatible ABI@. Pass floating-point arguments to
13914 prototyped functions beyond the register save area (RSA) on the stack
13915 in addition to argument FPRs. Do not assume that most significant
13916 double in 128-bit long double value is properly rounded when comparing
13917 values and converting to double. Use XL symbol names for long double
13920 The AIX calling convention was extended but not initially documented to
13921 handle an obscure K&R C case of calling a function that takes the
13922 address of its arguments with fewer arguments than declared. IBM XL
13923 compilers access floating point arguments which do not fit in the
13924 RSA from the stack when a subroutine is compiled without
13925 optimization. Because always storing floating-point arguments on the
13926 stack is inefficient and rarely needed, this option is not enabled by
13927 default and only is necessary when calling subroutines compiled by IBM
13928 XL compilers without optimization.
13932 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13933 application written to use message passing with special startup code to
13934 enable the application to run. The system must have PE installed in the
13935 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13936 must be overridden with the @option{-specs=} option to specify the
13937 appropriate directory location. The Parallel Environment does not
13938 support threads, so the @option{-mpe} option and the @option{-pthread}
13939 option are incompatible.
13941 @item -malign-natural
13942 @itemx -malign-power
13943 @opindex malign-natural
13944 @opindex malign-power
13945 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13946 @option{-malign-natural} overrides the ABI-defined alignment of larger
13947 types, such as floating-point doubles, on their natural size-based boundary.
13948 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13949 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13951 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13955 @itemx -mhard-float
13956 @opindex msoft-float
13957 @opindex mhard-float
13958 Generate code that does not use (uses) the floating-point register set.
13959 Software floating point emulation is provided if you use the
13960 @option{-msoft-float} option, and pass the option to GCC when linking.
13962 @item -msingle-float
13963 @itemx -mdouble-float
13964 @opindex msingle-float
13965 @opindex mdouble-float
13966 Generate code for single or double-precision floating point operations.
13967 @option{-mdouble-float} implies @option{-msingle-float}.
13970 @opindex msimple-fpu
13971 Do not generate sqrt and div instructions for hardware floating point unit.
13975 Specify type of floating point unit. Valid values are @var{sp_lite}
13976 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
13977 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
13978 and @var{dp_full} (equivalent to -mdouble-float).
13981 @opindex mxilinx-fpu
13982 Perform optimizations for floating point unit on Xilinx PPC 405/440.
13985 @itemx -mno-multiple
13987 @opindex mno-multiple
13988 Generate code that uses (does not use) the load multiple word
13989 instructions and the store multiple word instructions. These
13990 instructions are generated by default on POWER systems, and not
13991 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13992 endian PowerPC systems, since those instructions do not work when the
13993 processor is in little endian mode. The exceptions are PPC740 and
13994 PPC750 which permit the instructions usage in little endian mode.
13999 @opindex mno-string
14000 Generate code that uses (does not use) the load string instructions
14001 and the store string word instructions to save multiple registers and
14002 do small block moves. These instructions are generated by default on
14003 POWER systems, and not generated on PowerPC systems. Do not use
14004 @option{-mstring} on little endian PowerPC systems, since those
14005 instructions do not work when the processor is in little endian mode.
14006 The exceptions are PPC740 and PPC750 which permit the instructions
14007 usage in little endian mode.
14012 @opindex mno-update
14013 Generate code that uses (does not use) the load or store instructions
14014 that update the base register to the address of the calculated memory
14015 location. These instructions are generated by default. If you use
14016 @option{-mno-update}, there is a small window between the time that the
14017 stack pointer is updated and the address of the previous frame is
14018 stored, which means code that walks the stack frame across interrupts or
14019 signals may get corrupted data.
14021 @item -mavoid-indexed-addresses
14022 @item -mno-avoid-indexed-addresses
14023 @opindex mavoid-indexed-addresses
14024 @opindex mno-avoid-indexed-addresses
14025 Generate code that tries to avoid (not avoid) the use of indexed load
14026 or store instructions. These instructions can incur a performance
14027 penalty on Power6 processors in certain situations, such as when
14028 stepping through large arrays that cross a 16M boundary. This option
14029 is enabled by default when targetting Power6 and disabled otherwise.
14032 @itemx -mno-fused-madd
14033 @opindex mfused-madd
14034 @opindex mno-fused-madd
14035 Generate code that uses (does not use) the floating point multiply and
14036 accumulate instructions. These instructions are generated by default if
14037 hardware floating is used.
14043 Generate code that uses (does not use) the half-word multiply and
14044 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
14045 These instructions are generated by default when targetting those
14052 Generate code that uses (does not use) the string-search @samp{dlmzb}
14053 instruction on the IBM 405, 440 and 464 processors. This instruction is
14054 generated by default when targetting those processors.
14056 @item -mno-bit-align
14058 @opindex mno-bit-align
14059 @opindex mbit-align
14060 On System V.4 and embedded PowerPC systems do not (do) force structures
14061 and unions that contain bit-fields to be aligned to the base type of the
14064 For example, by default a structure containing nothing but 8
14065 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
14066 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
14067 the structure would be aligned to a 1 byte boundary and be one byte in
14070 @item -mno-strict-align
14071 @itemx -mstrict-align
14072 @opindex mno-strict-align
14073 @opindex mstrict-align
14074 On System V.4 and embedded PowerPC systems do not (do) assume that
14075 unaligned memory references will be handled by the system.
14077 @item -mrelocatable
14078 @itemx -mno-relocatable
14079 @opindex mrelocatable
14080 @opindex mno-relocatable
14081 On embedded PowerPC systems generate code that allows (does not allow)
14082 the program to be relocated to a different address at runtime. If you
14083 use @option{-mrelocatable} on any module, all objects linked together must
14084 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
14086 @item -mrelocatable-lib
14087 @itemx -mno-relocatable-lib
14088 @opindex mrelocatable-lib
14089 @opindex mno-relocatable-lib
14090 On embedded PowerPC systems generate code that allows (does not allow)
14091 the program to be relocated to a different address at runtime. Modules
14092 compiled with @option{-mrelocatable-lib} can be linked with either modules
14093 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
14094 with modules compiled with the @option{-mrelocatable} options.
14100 On System V.4 and embedded PowerPC systems do not (do) assume that
14101 register 2 contains a pointer to a global area pointing to the addresses
14102 used in the program.
14105 @itemx -mlittle-endian
14107 @opindex mlittle-endian
14108 On System V.4 and embedded PowerPC systems compile code for the
14109 processor in little endian mode. The @option{-mlittle-endian} option is
14110 the same as @option{-mlittle}.
14113 @itemx -mbig-endian
14115 @opindex mbig-endian
14116 On System V.4 and embedded PowerPC systems compile code for the
14117 processor in big endian mode. The @option{-mbig-endian} option is
14118 the same as @option{-mbig}.
14120 @item -mdynamic-no-pic
14121 @opindex mdynamic-no-pic
14122 On Darwin and Mac OS X systems, compile code so that it is not
14123 relocatable, but that its external references are relocatable. The
14124 resulting code is suitable for applications, but not shared
14127 @item -mprioritize-restricted-insns=@var{priority}
14128 @opindex mprioritize-restricted-insns
14129 This option controls the priority that is assigned to
14130 dispatch-slot restricted instructions during the second scheduling
14131 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
14132 @var{no/highest/second-highest} priority to dispatch slot restricted
14135 @item -msched-costly-dep=@var{dependence_type}
14136 @opindex msched-costly-dep
14137 This option controls which dependences are considered costly
14138 by the target during instruction scheduling. The argument
14139 @var{dependence_type} takes one of the following values:
14140 @var{no}: no dependence is costly,
14141 @var{all}: all dependences are costly,
14142 @var{true_store_to_load}: a true dependence from store to load is costly,
14143 @var{store_to_load}: any dependence from store to load is costly,
14144 @var{number}: any dependence which latency >= @var{number} is costly.
14146 @item -minsert-sched-nops=@var{scheme}
14147 @opindex minsert-sched-nops
14148 This option controls which nop insertion scheme will be used during
14149 the second scheduling pass. The argument @var{scheme} takes one of the
14151 @var{no}: Don't insert nops.
14152 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
14153 according to the scheduler's grouping.
14154 @var{regroup_exact}: Insert nops to force costly dependent insns into
14155 separate groups. Insert exactly as many nops as needed to force an insn
14156 to a new group, according to the estimated processor grouping.
14157 @var{number}: Insert nops to force costly dependent insns into
14158 separate groups. Insert @var{number} nops to force an insn to a new group.
14161 @opindex mcall-sysv
14162 On System V.4 and embedded PowerPC systems compile code using calling
14163 conventions that adheres to the March 1995 draft of the System V
14164 Application Binary Interface, PowerPC processor supplement. This is the
14165 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
14167 @item -mcall-sysv-eabi
14168 @opindex mcall-sysv-eabi
14169 Specify both @option{-mcall-sysv} and @option{-meabi} options.
14171 @item -mcall-sysv-noeabi
14172 @opindex mcall-sysv-noeabi
14173 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
14175 @item -mcall-solaris
14176 @opindex mcall-solaris
14177 On System V.4 and embedded PowerPC systems compile code for the Solaris
14181 @opindex mcall-linux
14182 On System V.4 and embedded PowerPC systems compile code for the
14183 Linux-based GNU system.
14187 On System V.4 and embedded PowerPC systems compile code for the
14188 Hurd-based GNU system.
14190 @item -mcall-netbsd
14191 @opindex mcall-netbsd
14192 On System V.4 and embedded PowerPC systems compile code for the
14193 NetBSD operating system.
14195 @item -maix-struct-return
14196 @opindex maix-struct-return
14197 Return all structures in memory (as specified by the AIX ABI)@.
14199 @item -msvr4-struct-return
14200 @opindex msvr4-struct-return
14201 Return structures smaller than 8 bytes in registers (as specified by the
14204 @item -mabi=@var{abi-type}
14206 Extend the current ABI with a particular extension, or remove such extension.
14207 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
14208 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
14212 Extend the current ABI with SPE ABI extensions. This does not change
14213 the default ABI, instead it adds the SPE ABI extensions to the current
14217 @opindex mabi=no-spe
14218 Disable Booke SPE ABI extensions for the current ABI@.
14220 @item -mabi=ibmlongdouble
14221 @opindex mabi=ibmlongdouble
14222 Change the current ABI to use IBM extended precision long double.
14223 This is a PowerPC 32-bit SYSV ABI option.
14225 @item -mabi=ieeelongdouble
14226 @opindex mabi=ieeelongdouble
14227 Change the current ABI to use IEEE extended precision long double.
14228 This is a PowerPC 32-bit Linux ABI option.
14231 @itemx -mno-prototype
14232 @opindex mprototype
14233 @opindex mno-prototype
14234 On System V.4 and embedded PowerPC systems assume that all calls to
14235 variable argument functions are properly prototyped. Otherwise, the
14236 compiler must insert an instruction before every non prototyped call to
14237 set or clear bit 6 of the condition code register (@var{CR}) to
14238 indicate whether floating point values were passed in the floating point
14239 registers in case the function takes a variable arguments. With
14240 @option{-mprototype}, only calls to prototyped variable argument functions
14241 will set or clear the bit.
14245 On embedded PowerPC systems, assume that the startup module is called
14246 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
14247 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
14252 On embedded PowerPC systems, assume that the startup module is called
14253 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
14258 On embedded PowerPC systems, assume that the startup module is called
14259 @file{crt0.o} and the standard C libraries are @file{libads.a} and
14262 @item -myellowknife
14263 @opindex myellowknife
14264 On embedded PowerPC systems, assume that the startup module is called
14265 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
14270 On System V.4 and embedded PowerPC systems, specify that you are
14271 compiling for a VxWorks system.
14275 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
14276 header to indicate that @samp{eabi} extended relocations are used.
14282 On System V.4 and embedded PowerPC systems do (do not) adhere to the
14283 Embedded Applications Binary Interface (eabi) which is a set of
14284 modifications to the System V.4 specifications. Selecting @option{-meabi}
14285 means that the stack is aligned to an 8 byte boundary, a function
14286 @code{__eabi} is called to from @code{main} to set up the eabi
14287 environment, and the @option{-msdata} option can use both @code{r2} and
14288 @code{r13} to point to two separate small data areas. Selecting
14289 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
14290 do not call an initialization function from @code{main}, and the
14291 @option{-msdata} option will only use @code{r13} to point to a single
14292 small data area. The @option{-meabi} option is on by default if you
14293 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
14296 @opindex msdata=eabi
14297 On System V.4 and embedded PowerPC systems, put small initialized
14298 @code{const} global and static data in the @samp{.sdata2} section, which
14299 is pointed to by register @code{r2}. Put small initialized
14300 non-@code{const} global and static data in the @samp{.sdata} section,
14301 which is pointed to by register @code{r13}. Put small uninitialized
14302 global and static data in the @samp{.sbss} section, which is adjacent to
14303 the @samp{.sdata} section. The @option{-msdata=eabi} option is
14304 incompatible with the @option{-mrelocatable} option. The
14305 @option{-msdata=eabi} option also sets the @option{-memb} option.
14308 @opindex msdata=sysv
14309 On System V.4 and embedded PowerPC systems, put small global and static
14310 data in the @samp{.sdata} section, which is pointed to by register
14311 @code{r13}. Put small uninitialized global and static data in the
14312 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
14313 The @option{-msdata=sysv} option is incompatible with the
14314 @option{-mrelocatable} option.
14316 @item -msdata=default
14318 @opindex msdata=default
14320 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
14321 compile code the same as @option{-msdata=eabi}, otherwise compile code the
14322 same as @option{-msdata=sysv}.
14325 @opindex msdata=data
14326 On System V.4 and embedded PowerPC systems, put small global
14327 data in the @samp{.sdata} section. Put small uninitialized global
14328 data in the @samp{.sbss} section. Do not use register @code{r13}
14329 to address small data however. This is the default behavior unless
14330 other @option{-msdata} options are used.
14334 @opindex msdata=none
14336 On embedded PowerPC systems, put all initialized global and static data
14337 in the @samp{.data} section, and all uninitialized data in the
14338 @samp{.bss} section.
14342 @cindex smaller data references (PowerPC)
14343 @cindex .sdata/.sdata2 references (PowerPC)
14344 On embedded PowerPC systems, put global and static items less than or
14345 equal to @var{num} bytes into the small data or bss sections instead of
14346 the normal data or bss section. By default, @var{num} is 8. The
14347 @option{-G @var{num}} switch is also passed to the linker.
14348 All modules should be compiled with the same @option{-G @var{num}} value.
14351 @itemx -mno-regnames
14353 @opindex mno-regnames
14354 On System V.4 and embedded PowerPC systems do (do not) emit register
14355 names in the assembly language output using symbolic forms.
14358 @itemx -mno-longcall
14360 @opindex mno-longcall
14361 By default assume that all calls are far away so that a longer more
14362 expensive calling sequence is required. This is required for calls
14363 further than 32 megabytes (33,554,432 bytes) from the current location.
14364 A short call will be generated if the compiler knows
14365 the call cannot be that far away. This setting can be overridden by
14366 the @code{shortcall} function attribute, or by @code{#pragma
14369 Some linkers are capable of detecting out-of-range calls and generating
14370 glue code on the fly. On these systems, long calls are unnecessary and
14371 generate slower code. As of this writing, the AIX linker can do this,
14372 as can the GNU linker for PowerPC/64. It is planned to add this feature
14373 to the GNU linker for 32-bit PowerPC systems as well.
14375 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
14376 callee, L42'', plus a ``branch island'' (glue code). The two target
14377 addresses represent the callee and the ``branch island''. The
14378 Darwin/PPC linker will prefer the first address and generate a ``bl
14379 callee'' if the PPC ``bl'' instruction will reach the callee directly;
14380 otherwise, the linker will generate ``bl L42'' to call the ``branch
14381 island''. The ``branch island'' is appended to the body of the
14382 calling function; it computes the full 32-bit address of the callee
14385 On Mach-O (Darwin) systems, this option directs the compiler emit to
14386 the glue for every direct call, and the Darwin linker decides whether
14387 to use or discard it.
14389 In the future, we may cause GCC to ignore all longcall specifications
14390 when the linker is known to generate glue.
14394 Adds support for multithreading with the @dfn{pthreads} library.
14395 This option sets flags for both the preprocessor and linker.
14399 @node S/390 and zSeries Options
14400 @subsection S/390 and zSeries Options
14401 @cindex S/390 and zSeries Options
14403 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
14407 @itemx -msoft-float
14408 @opindex mhard-float
14409 @opindex msoft-float
14410 Use (do not use) the hardware floating-point instructions and registers
14411 for floating-point operations. When @option{-msoft-float} is specified,
14412 functions in @file{libgcc.a} will be used to perform floating-point
14413 operations. When @option{-mhard-float} is specified, the compiler
14414 generates IEEE floating-point instructions. This is the default.
14417 @itemx -mno-hard-dfp
14419 @opindex mno-hard-dfp
14420 Use (do not use) the hardware decimal-floating-point instructions for
14421 decimal-floating-point operations. When @option{-mno-hard-dfp} is
14422 specified, functions in @file{libgcc.a} will be used to perform
14423 decimal-floating-point operations. When @option{-mhard-dfp} is
14424 specified, the compiler generates decimal-floating-point hardware
14425 instructions. This is the default for @option{-march=z9-ec} or higher.
14427 @item -mlong-double-64
14428 @itemx -mlong-double-128
14429 @opindex mlong-double-64
14430 @opindex mlong-double-128
14431 These switches control the size of @code{long double} type. A size
14432 of 64bit makes the @code{long double} type equivalent to the @code{double}
14433 type. This is the default.
14436 @itemx -mno-backchain
14437 @opindex mbackchain
14438 @opindex mno-backchain
14439 Store (do not store) the address of the caller's frame as backchain pointer
14440 into the callee's stack frame.
14441 A backchain may be needed to allow debugging using tools that do not understand
14442 DWARF-2 call frame information.
14443 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
14444 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
14445 the backchain is placed into the topmost word of the 96/160 byte register
14448 In general, code compiled with @option{-mbackchain} is call-compatible with
14449 code compiled with @option{-mmo-backchain}; however, use of the backchain
14450 for debugging purposes usually requires that the whole binary is built with
14451 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
14452 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14453 to build a linux kernel use @option{-msoft-float}.
14455 The default is to not maintain the backchain.
14457 @item -mpacked-stack
14458 @itemx -mno-packed-stack
14459 @opindex mpacked-stack
14460 @opindex mno-packed-stack
14461 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
14462 specified, the compiler uses the all fields of the 96/160 byte register save
14463 area only for their default purpose; unused fields still take up stack space.
14464 When @option{-mpacked-stack} is specified, register save slots are densely
14465 packed at the top of the register save area; unused space is reused for other
14466 purposes, allowing for more efficient use of the available stack space.
14467 However, when @option{-mbackchain} is also in effect, the topmost word of
14468 the save area is always used to store the backchain, and the return address
14469 register is always saved two words below the backchain.
14471 As long as the stack frame backchain is not used, code generated with
14472 @option{-mpacked-stack} is call-compatible with code generated with
14473 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
14474 S/390 or zSeries generated code that uses the stack frame backchain at run
14475 time, not just for debugging purposes. Such code is not call-compatible
14476 with code compiled with @option{-mpacked-stack}. Also, note that the
14477 combination of @option{-mbackchain},
14478 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14479 to build a linux kernel use @option{-msoft-float}.
14481 The default is to not use the packed stack layout.
14484 @itemx -mno-small-exec
14485 @opindex msmall-exec
14486 @opindex mno-small-exec
14487 Generate (or do not generate) code using the @code{bras} instruction
14488 to do subroutine calls.
14489 This only works reliably if the total executable size does not
14490 exceed 64k. The default is to use the @code{basr} instruction instead,
14491 which does not have this limitation.
14497 When @option{-m31} is specified, generate code compliant to the
14498 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
14499 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
14500 particular to generate 64-bit instructions. For the @samp{s390}
14501 targets, the default is @option{-m31}, while the @samp{s390x}
14502 targets default to @option{-m64}.
14508 When @option{-mzarch} is specified, generate code using the
14509 instructions available on z/Architecture.
14510 When @option{-mesa} is specified, generate code using the
14511 instructions available on ESA/390. Note that @option{-mesa} is
14512 not possible with @option{-m64}.
14513 When generating code compliant to the GNU/Linux for S/390 ABI,
14514 the default is @option{-mesa}. When generating code compliant
14515 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
14521 Generate (or do not generate) code using the @code{mvcle} instruction
14522 to perform block moves. When @option{-mno-mvcle} is specified,
14523 use a @code{mvc} loop instead. This is the default unless optimizing for
14530 Print (or do not print) additional debug information when compiling.
14531 The default is to not print debug information.
14533 @item -march=@var{cpu-type}
14535 Generate code that will run on @var{cpu-type}, which is the name of a system
14536 representing a certain processor type. Possible values for
14537 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
14538 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
14539 When generating code using the instructions available on z/Architecture,
14540 the default is @option{-march=z900}. Otherwise, the default is
14541 @option{-march=g5}.
14543 @item -mtune=@var{cpu-type}
14545 Tune to @var{cpu-type} everything applicable about the generated code,
14546 except for the ABI and the set of available instructions.
14547 The list of @var{cpu-type} values is the same as for @option{-march}.
14548 The default is the value used for @option{-march}.
14551 @itemx -mno-tpf-trace
14552 @opindex mtpf-trace
14553 @opindex mno-tpf-trace
14554 Generate code that adds (does not add) in TPF OS specific branches to trace
14555 routines in the operating system. This option is off by default, even
14556 when compiling for the TPF OS@.
14559 @itemx -mno-fused-madd
14560 @opindex mfused-madd
14561 @opindex mno-fused-madd
14562 Generate code that uses (does not use) the floating point multiply and
14563 accumulate instructions. These instructions are generated by default if
14564 hardware floating point is used.
14566 @item -mwarn-framesize=@var{framesize}
14567 @opindex mwarn-framesize
14568 Emit a warning if the current function exceeds the given frame size. Because
14569 this is a compile time check it doesn't need to be a real problem when the program
14570 runs. It is intended to identify functions which most probably cause
14571 a stack overflow. It is useful to be used in an environment with limited stack
14572 size e.g.@: the linux kernel.
14574 @item -mwarn-dynamicstack
14575 @opindex mwarn-dynamicstack
14576 Emit a warning if the function calls alloca or uses dynamically
14577 sized arrays. This is generally a bad idea with a limited stack size.
14579 @item -mstack-guard=@var{stack-guard}
14580 @itemx -mstack-size=@var{stack-size}
14581 @opindex mstack-guard
14582 @opindex mstack-size
14583 If these options are provided the s390 back end emits additional instructions in
14584 the function prologue which trigger a trap if the stack size is @var{stack-guard}
14585 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
14586 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
14587 the frame size of the compiled function is chosen.
14588 These options are intended to be used to help debugging stack overflow problems.
14589 The additionally emitted code causes only little overhead and hence can also be
14590 used in production like systems without greater performance degradation. The given
14591 values have to be exact powers of 2 and @var{stack-size} has to be greater than
14592 @var{stack-guard} without exceeding 64k.
14593 In order to be efficient the extra code makes the assumption that the stack starts
14594 at an address aligned to the value given by @var{stack-size}.
14595 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
14598 @node Score Options
14599 @subsection Score Options
14600 @cindex Score Options
14602 These options are defined for Score implementations:
14607 Compile code for big endian mode. This is the default.
14611 Compile code for little endian mode.
14615 Disable generate bcnz instruction.
14619 Enable generate unaligned load and store instruction.
14623 Enable the use of multiply-accumulate instructions. Disabled by default.
14627 Specify the SCORE5 as the target architecture.
14631 Specify the SCORE5U of the target architecture.
14635 Specify the SCORE7 as the target architecture. This is the default.
14639 Specify the SCORE7D as the target architecture.
14643 @subsection SH Options
14645 These @samp{-m} options are defined for the SH implementations:
14650 Generate code for the SH1.
14654 Generate code for the SH2.
14657 Generate code for the SH2e.
14661 Generate code for the SH3.
14665 Generate code for the SH3e.
14669 Generate code for the SH4 without a floating-point unit.
14671 @item -m4-single-only
14672 @opindex m4-single-only
14673 Generate code for the SH4 with a floating-point unit that only
14674 supports single-precision arithmetic.
14678 Generate code for the SH4 assuming the floating-point unit is in
14679 single-precision mode by default.
14683 Generate code for the SH4.
14687 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
14688 floating-point unit is not used.
14690 @item -m4a-single-only
14691 @opindex m4a-single-only
14692 Generate code for the SH4a, in such a way that no double-precision
14693 floating point operations are used.
14696 @opindex m4a-single
14697 Generate code for the SH4a assuming the floating-point unit is in
14698 single-precision mode by default.
14702 Generate code for the SH4a.
14706 Same as @option{-m4a-nofpu}, except that it implicitly passes
14707 @option{-dsp} to the assembler. GCC doesn't generate any DSP
14708 instructions at the moment.
14712 Compile code for the processor in big endian mode.
14716 Compile code for the processor in little endian mode.
14720 Align doubles at 64-bit boundaries. Note that this changes the calling
14721 conventions, and thus some functions from the standard C library will
14722 not work unless you recompile it first with @option{-mdalign}.
14726 Shorten some address references at link time, when possible; uses the
14727 linker option @option{-relax}.
14731 Use 32-bit offsets in @code{switch} tables. The default is to use
14736 Enable the use of bit manipulation instructions on SH2A.
14740 Enable the use of the instruction @code{fmovd}.
14744 Comply with the calling conventions defined by Renesas.
14748 Comply with the calling conventions defined by Renesas.
14752 Comply with the calling conventions defined for GCC before the Renesas
14753 conventions were available. This option is the default for all
14754 targets of the SH toolchain except for @samp{sh-symbianelf}.
14757 @opindex mnomacsave
14758 Mark the @code{MAC} register as call-clobbered, even if
14759 @option{-mhitachi} is given.
14763 Increase IEEE-compliance of floating-point code.
14764 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14765 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14766 comparisons of NANs / infinities incurs extra overhead in every
14767 floating point comparison, therefore the default is set to
14768 @option{-ffinite-math-only}.
14770 @item -minline-ic_invalidate
14771 @opindex minline-ic_invalidate
14772 Inline code to invalidate instruction cache entries after setting up
14773 nested function trampolines.
14774 This option has no effect if -musermode is in effect and the selected
14775 code generation option (e.g. -m4) does not allow the use of the icbi
14777 If the selected code generation option does not allow the use of the icbi
14778 instruction, and -musermode is not in effect, the inlined code will
14779 manipulate the instruction cache address array directly with an associative
14780 write. This not only requires privileged mode, but it will also
14781 fail if the cache line had been mapped via the TLB and has become unmapped.
14785 Dump instruction size and location in the assembly code.
14788 @opindex mpadstruct
14789 This option is deprecated. It pads structures to multiple of 4 bytes,
14790 which is incompatible with the SH ABI@.
14794 Optimize for space instead of speed. Implied by @option{-Os}.
14797 @opindex mprefergot
14798 When generating position-independent code, emit function calls using
14799 the Global Offset Table instead of the Procedure Linkage Table.
14803 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14804 if the inlined code would not work in user mode.
14805 This is the default when the target is @code{sh-*-linux*}.
14807 @item -multcost=@var{number}
14808 @opindex multcost=@var{number}
14809 Set the cost to assume for a multiply insn.
14811 @item -mdiv=@var{strategy}
14812 @opindex mdiv=@var{strategy}
14813 Set the division strategy to use for SHmedia code. @var{strategy} must be
14814 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14815 inv:call2, inv:fp .
14816 "fp" performs the operation in floating point. This has a very high latency,
14817 but needs only a few instructions, so it might be a good choice if
14818 your code has enough easily exploitable ILP to allow the compiler to
14819 schedule the floating point instructions together with other instructions.
14820 Division by zero causes a floating point exception.
14821 "inv" uses integer operations to calculate the inverse of the divisor,
14822 and then multiplies the dividend with the inverse. This strategy allows
14823 cse and hoisting of the inverse calculation. Division by zero calculates
14824 an unspecified result, but does not trap.
14825 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14826 have been found, or if the entire operation has been hoisted to the same
14827 place, the last stages of the inverse calculation are intertwined with the
14828 final multiply to reduce the overall latency, at the expense of using a few
14829 more instructions, and thus offering fewer scheduling opportunities with
14831 "call" calls a library function that usually implements the inv:minlat
14833 This gives high code density for m5-*media-nofpu compilations.
14834 "call2" uses a different entry point of the same library function, where it
14835 assumes that a pointer to a lookup table has already been set up, which
14836 exposes the pointer load to cse / code hoisting optimizations.
14837 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14838 code generation, but if the code stays unoptimized, revert to the "call",
14839 "call2", or "fp" strategies, respectively. Note that the
14840 potentially-trapping side effect of division by zero is carried by a
14841 separate instruction, so it is possible that all the integer instructions
14842 are hoisted out, but the marker for the side effect stays where it is.
14843 A recombination to fp operations or a call is not possible in that case.
14844 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14845 that the inverse calculation was nor separated from the multiply, they speed
14846 up division where the dividend fits into 20 bits (plus sign where applicable),
14847 by inserting a test to skip a number of operations in this case; this test
14848 slows down the case of larger dividends. inv20u assumes the case of a such
14849 a small dividend to be unlikely, and inv20l assumes it to be likely.
14851 @item -mdivsi3_libfunc=@var{name}
14852 @opindex mdivsi3_libfunc=@var{name}
14853 Set the name of the library function used for 32 bit signed division to
14854 @var{name}. This only affect the name used in the call and inv:call
14855 division strategies, and the compiler will still expect the same
14856 sets of input/output/clobbered registers as if this option was not present.
14858 @item -mfixed-range=@var{register-range}
14859 @opindex mfixed-range
14860 Generate code treating the given register range as fixed registers.
14861 A fixed register is one that the register allocator can not use. This is
14862 useful when compiling kernel code. A register range is specified as
14863 two registers separated by a dash. Multiple register ranges can be
14864 specified separated by a comma.
14866 @item -madjust-unroll
14867 @opindex madjust-unroll
14868 Throttle unrolling to avoid thrashing target registers.
14869 This option only has an effect if the gcc code base supports the
14870 TARGET_ADJUST_UNROLL_MAX target hook.
14872 @item -mindexed-addressing
14873 @opindex mindexed-addressing
14874 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14875 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14876 semantics for the indexed addressing mode. The architecture allows the
14877 implementation of processors with 64 bit MMU, which the OS could use to
14878 get 32 bit addressing, but since no current hardware implementation supports
14879 this or any other way to make the indexed addressing mode safe to use in
14880 the 32 bit ABI, the default is -mno-indexed-addressing.
14882 @item -mgettrcost=@var{number}
14883 @opindex mgettrcost=@var{number}
14884 Set the cost assumed for the gettr instruction to @var{number}.
14885 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14889 Assume pt* instructions won't trap. This will generally generate better
14890 scheduled code, but is unsafe on current hardware. The current architecture
14891 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14892 This has the unintentional effect of making it unsafe to schedule ptabs /
14893 ptrel before a branch, or hoist it out of a loop. For example,
14894 __do_global_ctors, a part of libgcc that runs constructors at program
14895 startup, calls functions in a list which is delimited by @minus{}1. With the
14896 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14897 That means that all the constructors will be run a bit quicker, but when
14898 the loop comes to the end of the list, the program crashes because ptabs
14899 loads @minus{}1 into a target register. Since this option is unsafe for any
14900 hardware implementing the current architecture specification, the default
14901 is -mno-pt-fixed. Unless the user specifies a specific cost with
14902 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14903 this deters register allocation using target registers for storing
14906 @item -minvalid-symbols
14907 @opindex minvalid-symbols
14908 Assume symbols might be invalid. Ordinary function symbols generated by
14909 the compiler will always be valid to load with movi/shori/ptabs or
14910 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14911 to generate symbols that will cause ptabs / ptrel to trap.
14912 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14913 It will then prevent cross-basic-block cse, hoisting and most scheduling
14914 of symbol loads. The default is @option{-mno-invalid-symbols}.
14917 @node SPARC Options
14918 @subsection SPARC Options
14919 @cindex SPARC options
14921 These @samp{-m} options are supported on the SPARC:
14924 @item -mno-app-regs
14926 @opindex mno-app-regs
14928 Specify @option{-mapp-regs} to generate output using the global registers
14929 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14932 To be fully SVR4 ABI compliant at the cost of some performance loss,
14933 specify @option{-mno-app-regs}. You should compile libraries and system
14934 software with this option.
14937 @itemx -mhard-float
14939 @opindex mhard-float
14940 Generate output containing floating point instructions. This is the
14944 @itemx -msoft-float
14946 @opindex msoft-float
14947 Generate output containing library calls for floating point.
14948 @strong{Warning:} the requisite libraries are not available for all SPARC
14949 targets. Normally the facilities of the machine's usual C compiler are
14950 used, but this cannot be done directly in cross-compilation. You must make
14951 your own arrangements to provide suitable library functions for
14952 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14953 @samp{sparclite-*-*} do provide software floating point support.
14955 @option{-msoft-float} changes the calling convention in the output file;
14956 therefore, it is only useful if you compile @emph{all} of a program with
14957 this option. In particular, you need to compile @file{libgcc.a}, the
14958 library that comes with GCC, with @option{-msoft-float} in order for
14961 @item -mhard-quad-float
14962 @opindex mhard-quad-float
14963 Generate output containing quad-word (long double) floating point
14966 @item -msoft-quad-float
14967 @opindex msoft-quad-float
14968 Generate output containing library calls for quad-word (long double)
14969 floating point instructions. The functions called are those specified
14970 in the SPARC ABI@. This is the default.
14972 As of this writing, there are no SPARC implementations that have hardware
14973 support for the quad-word floating point instructions. They all invoke
14974 a trap handler for one of these instructions, and then the trap handler
14975 emulates the effect of the instruction. Because of the trap handler overhead,
14976 this is much slower than calling the ABI library routines. Thus the
14977 @option{-msoft-quad-float} option is the default.
14979 @item -mno-unaligned-doubles
14980 @itemx -munaligned-doubles
14981 @opindex mno-unaligned-doubles
14982 @opindex munaligned-doubles
14983 Assume that doubles have 8 byte alignment. This is the default.
14985 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14986 alignment only if they are contained in another type, or if they have an
14987 absolute address. Otherwise, it assumes they have 4 byte alignment.
14988 Specifying this option avoids some rare compatibility problems with code
14989 generated by other compilers. It is not the default because it results
14990 in a performance loss, especially for floating point code.
14992 @item -mno-faster-structs
14993 @itemx -mfaster-structs
14994 @opindex mno-faster-structs
14995 @opindex mfaster-structs
14996 With @option{-mfaster-structs}, the compiler assumes that structures
14997 should have 8 byte alignment. This enables the use of pairs of
14998 @code{ldd} and @code{std} instructions for copies in structure
14999 assignment, in place of twice as many @code{ld} and @code{st} pairs.
15000 However, the use of this changed alignment directly violates the SPARC
15001 ABI@. Thus, it's intended only for use on targets where the developer
15002 acknowledges that their resulting code will not be directly in line with
15003 the rules of the ABI@.
15005 @item -mimpure-text
15006 @opindex mimpure-text
15007 @option{-mimpure-text}, used in addition to @option{-shared}, tells
15008 the compiler to not pass @option{-z text} to the linker when linking a
15009 shared object. Using this option, you can link position-dependent
15010 code into a shared object.
15012 @option{-mimpure-text} suppresses the ``relocations remain against
15013 allocatable but non-writable sections'' linker error message.
15014 However, the necessary relocations will trigger copy-on-write, and the
15015 shared object is not actually shared across processes. Instead of
15016 using @option{-mimpure-text}, you should compile all source code with
15017 @option{-fpic} or @option{-fPIC}.
15019 This option is only available on SunOS and Solaris.
15021 @item -mcpu=@var{cpu_type}
15023 Set the instruction set, register set, and instruction scheduling parameters
15024 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
15025 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
15026 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
15027 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
15028 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
15030 Default instruction scheduling parameters are used for values that select
15031 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
15032 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
15034 Here is a list of each supported architecture and their supported
15039 v8: supersparc, hypersparc
15040 sparclite: f930, f934, sparclite86x
15042 v9: ultrasparc, ultrasparc3, niagara, niagara2
15045 By default (unless configured otherwise), GCC generates code for the V7
15046 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
15047 additionally optimizes it for the Cypress CY7C602 chip, as used in the
15048 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
15049 SPARCStation 1, 2, IPX etc.
15051 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
15052 architecture. The only difference from V7 code is that the compiler emits
15053 the integer multiply and integer divide instructions which exist in SPARC-V8
15054 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
15055 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
15058 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
15059 the SPARC architecture. This adds the integer multiply, integer divide step
15060 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
15061 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
15062 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
15063 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
15064 MB86934 chip, which is the more recent SPARClite with FPU@.
15066 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
15067 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
15068 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
15069 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
15070 optimizes it for the TEMIC SPARClet chip.
15072 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
15073 architecture. This adds 64-bit integer and floating-point move instructions,
15074 3 additional floating-point condition code registers and conditional move
15075 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
15076 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
15077 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
15078 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
15079 @option{-mcpu=niagara}, the compiler additionally optimizes it for
15080 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
15081 additionally optimizes it for Sun UltraSPARC T2 chips.
15083 @item -mtune=@var{cpu_type}
15085 Set the instruction scheduling parameters for machine type
15086 @var{cpu_type}, but do not set the instruction set or register set that the
15087 option @option{-mcpu=@var{cpu_type}} would.
15089 The same values for @option{-mcpu=@var{cpu_type}} can be used for
15090 @option{-mtune=@var{cpu_type}}, but the only useful values are those
15091 that select a particular cpu implementation. Those are @samp{cypress},
15092 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
15093 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
15094 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
15099 @opindex mno-v8plus
15100 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
15101 difference from the V8 ABI is that the global and out registers are
15102 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
15103 mode for all SPARC-V9 processors.
15109 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
15110 Visual Instruction Set extensions. The default is @option{-mno-vis}.
15113 These @samp{-m} options are supported in addition to the above
15114 on SPARC-V9 processors in 64-bit environments:
15117 @item -mlittle-endian
15118 @opindex mlittle-endian
15119 Generate code for a processor running in little-endian mode. It is only
15120 available for a few configurations and most notably not on Solaris and Linux.
15126 Generate code for a 32-bit or 64-bit environment.
15127 The 32-bit environment sets int, long and pointer to 32 bits.
15128 The 64-bit environment sets int to 32 bits and long and pointer
15131 @item -mcmodel=medlow
15132 @opindex mcmodel=medlow
15133 Generate code for the Medium/Low code model: 64-bit addresses, programs
15134 must be linked in the low 32 bits of memory. Programs can be statically
15135 or dynamically linked.
15137 @item -mcmodel=medmid
15138 @opindex mcmodel=medmid
15139 Generate code for the Medium/Middle code model: 64-bit addresses, programs
15140 must be linked in the low 44 bits of memory, the text and data segments must
15141 be less than 2GB in size and the data segment must be located within 2GB of
15144 @item -mcmodel=medany
15145 @opindex mcmodel=medany
15146 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
15147 may be linked anywhere in memory, the text and data segments must be less
15148 than 2GB in size and the data segment must be located within 2GB of the
15151 @item -mcmodel=embmedany
15152 @opindex mcmodel=embmedany
15153 Generate code for the Medium/Anywhere code model for embedded systems:
15154 64-bit addresses, the text and data segments must be less than 2GB in
15155 size, both starting anywhere in memory (determined at link time). The
15156 global register %g4 points to the base of the data segment. Programs
15157 are statically linked and PIC is not supported.
15160 @itemx -mno-stack-bias
15161 @opindex mstack-bias
15162 @opindex mno-stack-bias
15163 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
15164 frame pointer if present, are offset by @minus{}2047 which must be added back
15165 when making stack frame references. This is the default in 64-bit mode.
15166 Otherwise, assume no such offset is present.
15169 These switches are supported in addition to the above on Solaris:
15174 Add support for multithreading using the Solaris threads library. This
15175 option sets flags for both the preprocessor and linker. This option does
15176 not affect the thread safety of object code produced by the compiler or
15177 that of libraries supplied with it.
15181 Add support for multithreading using the POSIX threads library. This
15182 option sets flags for both the preprocessor and linker. This option does
15183 not affect the thread safety of object code produced by the compiler or
15184 that of libraries supplied with it.
15188 This is a synonym for @option{-pthreads}.
15192 @subsection SPU Options
15193 @cindex SPU options
15195 These @samp{-m} options are supported on the SPU:
15199 @itemx -merror-reloc
15200 @opindex mwarn-reloc
15201 @opindex merror-reloc
15203 The loader for SPU does not handle dynamic relocations. By default, GCC
15204 will give an error when it generates code that requires a dynamic
15205 relocation. @option{-mno-error-reloc} disables the error,
15206 @option{-mwarn-reloc} will generate a warning instead.
15209 @itemx -munsafe-dma
15211 @opindex munsafe-dma
15213 Instructions which initiate or test completion of DMA must not be
15214 reordered with respect to loads and stores of the memory which is being
15215 accessed. Users typically address this problem using the volatile
15216 keyword, but that can lead to inefficient code in places where the
15217 memory is known to not change. Rather than mark the memory as volatile
15218 we treat the DMA instructions as potentially effecting all memory. With
15219 @option{-munsafe-dma} users must use the volatile keyword to protect
15222 @item -mbranch-hints
15223 @opindex mbranch-hints
15225 By default, GCC will generate a branch hint instruction to avoid
15226 pipeline stalls for always taken or probably taken branches. A hint
15227 will not be generated closer than 8 instructions away from its branch.
15228 There is little reason to disable them, except for debugging purposes,
15229 or to make an object a little bit smaller.
15233 @opindex msmall-mem
15234 @opindex mlarge-mem
15236 By default, GCC generates code assuming that addresses are never larger
15237 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
15238 a full 32 bit address.
15243 By default, GCC links against startup code that assumes the SPU-style
15244 main function interface (which has an unconventional parameter list).
15245 With @option{-mstdmain}, GCC will link your program against startup
15246 code that assumes a C99-style interface to @code{main}, including a
15247 local copy of @code{argv} strings.
15249 @item -mfixed-range=@var{register-range}
15250 @opindex mfixed-range
15251 Generate code treating the given register range as fixed registers.
15252 A fixed register is one that the register allocator can not use. This is
15253 useful when compiling kernel code. A register range is specified as
15254 two registers separated by a dash. Multiple register ranges can be
15255 specified separated by a comma.
15258 @itemx -mdual-nops=@var{n}
15259 @opindex mdual-nops
15260 By default, GCC will insert nops to increase dual issue when it expects
15261 it to increase performance. @var{n} can be a value from 0 to 10. A
15262 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
15263 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
15265 @item -mhint-max-nops=@var{n}
15266 @opindex mhint-max-nops
15267 Maximum number of nops to insert for a branch hint. A branch hint must
15268 be at least 8 instructions away from the branch it is effecting. GCC
15269 will insert up to @var{n} nops to enforce this, otherwise it will not
15270 generate the branch hint.
15272 @item -mhint-max-distance=@var{n}
15273 @opindex mhint-max-distance
15274 The encoding of the branch hint instruction limits the hint to be within
15275 256 instructions of the branch it is effecting. By default, GCC makes
15276 sure it is within 125.
15279 @opindex msafe-hints
15280 Work around a hardware bug which causes the SPU to stall indefinitely.
15281 By default, GCC will insert the @code{hbrp} instruction to make sure
15282 this stall won't happen.
15286 @node System V Options
15287 @subsection Options for System V
15289 These additional options are available on System V Release 4 for
15290 compatibility with other compilers on those systems:
15295 Create a shared object.
15296 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
15300 Identify the versions of each tool used by the compiler, in a
15301 @code{.ident} assembler directive in the output.
15305 Refrain from adding @code{.ident} directives to the output file (this is
15308 @item -YP,@var{dirs}
15310 Search the directories @var{dirs}, and no others, for libraries
15311 specified with @option{-l}.
15313 @item -Ym,@var{dir}
15315 Look in the directory @var{dir} to find the M4 preprocessor.
15316 The assembler uses this option.
15317 @c This is supposed to go with a -Yd for predefined M4 macro files, but
15318 @c the generic assembler that comes with Solaris takes just -Ym.
15322 @subsection V850 Options
15323 @cindex V850 Options
15325 These @samp{-m} options are defined for V850 implementations:
15329 @itemx -mno-long-calls
15330 @opindex mlong-calls
15331 @opindex mno-long-calls
15332 Treat all calls as being far away (near). If calls are assumed to be
15333 far away, the compiler will always load the functions address up into a
15334 register, and call indirect through the pointer.
15340 Do not optimize (do optimize) basic blocks that use the same index
15341 pointer 4 or more times to copy pointer into the @code{ep} register, and
15342 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
15343 option is on by default if you optimize.
15345 @item -mno-prolog-function
15346 @itemx -mprolog-function
15347 @opindex mno-prolog-function
15348 @opindex mprolog-function
15349 Do not use (do use) external functions to save and restore registers
15350 at the prologue and epilogue of a function. The external functions
15351 are slower, but use less code space if more than one function saves
15352 the same number of registers. The @option{-mprolog-function} option
15353 is on by default if you optimize.
15357 Try to make the code as small as possible. At present, this just turns
15358 on the @option{-mep} and @option{-mprolog-function} options.
15360 @item -mtda=@var{n}
15362 Put static or global variables whose size is @var{n} bytes or less into
15363 the tiny data area that register @code{ep} points to. The tiny data
15364 area can hold up to 256 bytes in total (128 bytes for byte references).
15366 @item -msda=@var{n}
15368 Put static or global variables whose size is @var{n} bytes or less into
15369 the small data area that register @code{gp} points to. The small data
15370 area can hold up to 64 kilobytes.
15372 @item -mzda=@var{n}
15374 Put static or global variables whose size is @var{n} bytes or less into
15375 the first 32 kilobytes of memory.
15379 Specify that the target processor is the V850.
15382 @opindex mbig-switch
15383 Generate code suitable for big switch tables. Use this option only if
15384 the assembler/linker complain about out of range branches within a switch
15389 This option will cause r2 and r5 to be used in the code generated by
15390 the compiler. This setting is the default.
15392 @item -mno-app-regs
15393 @opindex mno-app-regs
15394 This option will cause r2 and r5 to be treated as fixed registers.
15398 Specify that the target processor is the V850E1. The preprocessor
15399 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
15400 this option is used.
15404 Specify that the target processor is the V850E@. The preprocessor
15405 constant @samp{__v850e__} will be defined if this option is used.
15407 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
15408 are defined then a default target processor will be chosen and the
15409 relevant @samp{__v850*__} preprocessor constant will be defined.
15411 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
15412 defined, regardless of which processor variant is the target.
15414 @item -mdisable-callt
15415 @opindex mdisable-callt
15416 This option will suppress generation of the CALLT instruction for the
15417 v850e and v850e1 flavors of the v850 architecture. The default is
15418 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
15423 @subsection VAX Options
15424 @cindex VAX options
15426 These @samp{-m} options are defined for the VAX:
15431 Do not output certain jump instructions (@code{aobleq} and so on)
15432 that the Unix assembler for the VAX cannot handle across long
15437 Do output those jump instructions, on the assumption that you
15438 will assemble with the GNU assembler.
15442 Output code for g-format floating point numbers instead of d-format.
15445 @node VxWorks Options
15446 @subsection VxWorks Options
15447 @cindex VxWorks Options
15449 The options in this section are defined for all VxWorks targets.
15450 Options specific to the target hardware are listed with the other
15451 options for that target.
15456 GCC can generate code for both VxWorks kernels and real time processes
15457 (RTPs). This option switches from the former to the latter. It also
15458 defines the preprocessor macro @code{__RTP__}.
15461 @opindex non-static
15462 Link an RTP executable against shared libraries rather than static
15463 libraries. The options @option{-static} and @option{-shared} can
15464 also be used for RTPs (@pxref{Link Options}); @option{-static}
15471 These options are passed down to the linker. They are defined for
15472 compatibility with Diab.
15475 @opindex Xbind-lazy
15476 Enable lazy binding of function calls. This option is equivalent to
15477 @option{-Wl,-z,now} and is defined for compatibility with Diab.
15481 Disable lazy binding of function calls. This option is the default and
15482 is defined for compatibility with Diab.
15485 @node x86-64 Options
15486 @subsection x86-64 Options
15487 @cindex x86-64 options
15489 These are listed under @xref{i386 and x86-64 Options}.
15491 @node i386 and x86-64 Windows Options
15492 @subsection i386 and x86-64 Windows Options
15493 @cindex i386 and x86-64 Windows Options
15495 These additional options are available for Windows targets:
15500 This option is available for Cygwin and MinGW targets. It
15501 specifies that a console application is to be generated, by
15502 instructing the linker to set the PE header subsystem type
15503 required for console applications.
15504 This is the default behaviour for Cygwin and MinGW targets.
15508 This option is available for Cygwin targets. It specifies that
15509 the Cygwin internal interface is to be used for predefined
15510 preprocessor macros, C runtime libraries and related linker
15511 paths and options. For Cygwin targets this is the default behaviour.
15512 This option is deprecated and will be removed in a future release.
15515 @opindex mno-cygwin
15516 This option is available for Cygwin targets. It specifies that
15517 the MinGW internal interface is to be used instead of Cygwin's, by
15518 setting MinGW-related predefined macros and linker paths and default
15520 This option is deprecated and will be removed in a future release.
15524 This option is available for Cygwin and MinGW targets. It
15525 specifies that a DLL - a dynamic link library - is to be
15526 generated, enabling the selection of the required runtime
15527 startup object and entry point.
15529 @item -mnop-fun-dllimport
15530 @opindex mnop-fun-dllimport
15531 This option is available for Cygwin and MinGW targets. It
15532 specifies that the dllimport attribute should be ignored.
15536 This option is available for MinGW targets. It specifies
15537 that MinGW-specific thread support is to be used.
15541 This option is available for Cygwin and MinGW targets. It
15542 specifies that the typical Windows pre-defined macros are to
15543 be set in the pre-processor, but does not influence the choice
15544 of runtime library/startup code.
15548 This option is available for Cygwin and MinGW targets. It
15549 specifies that a GUI application is to be generated by
15550 instructing the linker to set the PE header subsystem type
15554 See also under @ref{i386 and x86-64 Options} for standard options.
15556 @node Xstormy16 Options
15557 @subsection Xstormy16 Options
15558 @cindex Xstormy16 Options
15560 These options are defined for Xstormy16:
15565 Choose startup files and linker script suitable for the simulator.
15568 @node Xtensa Options
15569 @subsection Xtensa Options
15570 @cindex Xtensa Options
15572 These options are supported for Xtensa targets:
15576 @itemx -mno-const16
15578 @opindex mno-const16
15579 Enable or disable use of @code{CONST16} instructions for loading
15580 constant values. The @code{CONST16} instruction is currently not a
15581 standard option from Tensilica. When enabled, @code{CONST16}
15582 instructions are always used in place of the standard @code{L32R}
15583 instructions. The use of @code{CONST16} is enabled by default only if
15584 the @code{L32R} instruction is not available.
15587 @itemx -mno-fused-madd
15588 @opindex mfused-madd
15589 @opindex mno-fused-madd
15590 Enable or disable use of fused multiply/add and multiply/subtract
15591 instructions in the floating-point option. This has no effect if the
15592 floating-point option is not also enabled. Disabling fused multiply/add
15593 and multiply/subtract instructions forces the compiler to use separate
15594 instructions for the multiply and add/subtract operations. This may be
15595 desirable in some cases where strict IEEE 754-compliant results are
15596 required: the fused multiply add/subtract instructions do not round the
15597 intermediate result, thereby producing results with @emph{more} bits of
15598 precision than specified by the IEEE standard. Disabling fused multiply
15599 add/subtract instructions also ensures that the program output is not
15600 sensitive to the compiler's ability to combine multiply and add/subtract
15603 @item -mserialize-volatile
15604 @itemx -mno-serialize-volatile
15605 @opindex mserialize-volatile
15606 @opindex mno-serialize-volatile
15607 When this option is enabled, GCC inserts @code{MEMW} instructions before
15608 @code{volatile} memory references to guarantee sequential consistency.
15609 The default is @option{-mserialize-volatile}. Use
15610 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
15612 @item -mtext-section-literals
15613 @itemx -mno-text-section-literals
15614 @opindex mtext-section-literals
15615 @opindex mno-text-section-literals
15616 Control the treatment of literal pools. The default is
15617 @option{-mno-text-section-literals}, which places literals in a separate
15618 section in the output file. This allows the literal pool to be placed
15619 in a data RAM/ROM, and it also allows the linker to combine literal
15620 pools from separate object files to remove redundant literals and
15621 improve code size. With @option{-mtext-section-literals}, the literals
15622 are interspersed in the text section in order to keep them as close as
15623 possible to their references. This may be necessary for large assembly
15626 @item -mtarget-align
15627 @itemx -mno-target-align
15628 @opindex mtarget-align
15629 @opindex mno-target-align
15630 When this option is enabled, GCC instructs the assembler to
15631 automatically align instructions to reduce branch penalties at the
15632 expense of some code density. The assembler attempts to widen density
15633 instructions to align branch targets and the instructions following call
15634 instructions. If there are not enough preceding safe density
15635 instructions to align a target, no widening will be performed. The
15636 default is @option{-mtarget-align}. These options do not affect the
15637 treatment of auto-aligned instructions like @code{LOOP}, which the
15638 assembler will always align, either by widening density instructions or
15639 by inserting no-op instructions.
15642 @itemx -mno-longcalls
15643 @opindex mlongcalls
15644 @opindex mno-longcalls
15645 When this option is enabled, GCC instructs the assembler to translate
15646 direct calls to indirect calls unless it can determine that the target
15647 of a direct call is in the range allowed by the call instruction. This
15648 translation typically occurs for calls to functions in other source
15649 files. Specifically, the assembler translates a direct @code{CALL}
15650 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
15651 The default is @option{-mno-longcalls}. This option should be used in
15652 programs where the call target can potentially be out of range. This
15653 option is implemented in the assembler, not the compiler, so the
15654 assembly code generated by GCC will still show direct call
15655 instructions---look at the disassembled object code to see the actual
15656 instructions. Note that the assembler will use an indirect call for
15657 every cross-file call, not just those that really will be out of range.
15660 @node zSeries Options
15661 @subsection zSeries Options
15662 @cindex zSeries options
15664 These are listed under @xref{S/390 and zSeries Options}.
15666 @node Code Gen Options
15667 @section Options for Code Generation Conventions
15668 @cindex code generation conventions
15669 @cindex options, code generation
15670 @cindex run-time options
15672 These machine-independent options control the interface conventions
15673 used in code generation.
15675 Most of them have both positive and negative forms; the negative form
15676 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
15677 one of the forms is listed---the one which is not the default. You
15678 can figure out the other form by either removing @samp{no-} or adding
15682 @item -fbounds-check
15683 @opindex fbounds-check
15684 For front-ends that support it, generate additional code to check that
15685 indices used to access arrays are within the declared range. This is
15686 currently only supported by the Java and Fortran front-ends, where
15687 this option defaults to true and false respectively.
15691 This option generates traps for signed overflow on addition, subtraction,
15692 multiplication operations.
15696 This option instructs the compiler to assume that signed arithmetic
15697 overflow of addition, subtraction and multiplication wraps around
15698 using twos-complement representation. This flag enables some optimizations
15699 and disables others. This option is enabled by default for the Java
15700 front-end, as required by the Java language specification.
15703 @opindex fexceptions
15704 Enable exception handling. Generates extra code needed to propagate
15705 exceptions. For some targets, this implies GCC will generate frame
15706 unwind information for all functions, which can produce significant data
15707 size overhead, although it does not affect execution. If you do not
15708 specify this option, GCC will enable it by default for languages like
15709 C++ which normally require exception handling, and disable it for
15710 languages like C that do not normally require it. However, you may need
15711 to enable this option when compiling C code that needs to interoperate
15712 properly with exception handlers written in C++. You may also wish to
15713 disable this option if you are compiling older C++ programs that don't
15714 use exception handling.
15716 @item -fnon-call-exceptions
15717 @opindex fnon-call-exceptions
15718 Generate code that allows trapping instructions to throw exceptions.
15719 Note that this requires platform-specific runtime support that does
15720 not exist everywhere. Moreover, it only allows @emph{trapping}
15721 instructions to throw exceptions, i.e.@: memory references or floating
15722 point instructions. It does not allow exceptions to be thrown from
15723 arbitrary signal handlers such as @code{SIGALRM}.
15725 @item -funwind-tables
15726 @opindex funwind-tables
15727 Similar to @option{-fexceptions}, except that it will just generate any needed
15728 static data, but will not affect the generated code in any other way.
15729 You will normally not enable this option; instead, a language processor
15730 that needs this handling would enable it on your behalf.
15732 @item -fasynchronous-unwind-tables
15733 @opindex fasynchronous-unwind-tables
15734 Generate unwind table in dwarf2 format, if supported by target machine. The
15735 table is exact at each instruction boundary, so it can be used for stack
15736 unwinding from asynchronous events (such as debugger or garbage collector).
15738 @item -fpcc-struct-return
15739 @opindex fpcc-struct-return
15740 Return ``short'' @code{struct} and @code{union} values in memory like
15741 longer ones, rather than in registers. This convention is less
15742 efficient, but it has the advantage of allowing intercallability between
15743 GCC-compiled files and files compiled with other compilers, particularly
15744 the Portable C Compiler (pcc).
15746 The precise convention for returning structures in memory depends
15747 on the target configuration macros.
15749 Short structures and unions are those whose size and alignment match
15750 that of some integer type.
15752 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
15753 switch is not binary compatible with code compiled with the
15754 @option{-freg-struct-return} switch.
15755 Use it to conform to a non-default application binary interface.
15757 @item -freg-struct-return
15758 @opindex freg-struct-return
15759 Return @code{struct} and @code{union} values in registers when possible.
15760 This is more efficient for small structures than
15761 @option{-fpcc-struct-return}.
15763 If you specify neither @option{-fpcc-struct-return} nor
15764 @option{-freg-struct-return}, GCC defaults to whichever convention is
15765 standard for the target. If there is no standard convention, GCC
15766 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
15767 the principal compiler. In those cases, we can choose the standard, and
15768 we chose the more efficient register return alternative.
15770 @strong{Warning:} code compiled with the @option{-freg-struct-return}
15771 switch is not binary compatible with code compiled with the
15772 @option{-fpcc-struct-return} switch.
15773 Use it to conform to a non-default application binary interface.
15775 @item -fshort-enums
15776 @opindex fshort-enums
15777 Allocate to an @code{enum} type only as many bytes as it needs for the
15778 declared range of possible values. Specifically, the @code{enum} type
15779 will be equivalent to the smallest integer type which has enough room.
15781 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
15782 code that is not binary compatible with code generated without that switch.
15783 Use it to conform to a non-default application binary interface.
15785 @item -fshort-double
15786 @opindex fshort-double
15787 Use the same size for @code{double} as for @code{float}.
15789 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
15790 code that is not binary compatible with code generated without that switch.
15791 Use it to conform to a non-default application binary interface.
15793 @item -fshort-wchar
15794 @opindex fshort-wchar
15795 Override the underlying type for @samp{wchar_t} to be @samp{short
15796 unsigned int} instead of the default for the target. This option is
15797 useful for building programs to run under WINE@.
15799 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
15800 code that is not binary compatible with code generated without that switch.
15801 Use it to conform to a non-default application binary interface.
15804 @opindex fno-common
15805 In C code, controls the placement of uninitialized global variables.
15806 Unix C compilers have traditionally permitted multiple definitions of
15807 such variables in different compilation units by placing the variables
15809 This is the behavior specified by @option{-fcommon}, and is the default
15810 for GCC on most targets.
15811 On the other hand, this behavior is not required by ISO C, and on some
15812 targets may carry a speed or code size penalty on variable references.
15813 The @option{-fno-common} option specifies that the compiler should place
15814 uninitialized global variables in the data section of the object file,
15815 rather than generating them as common blocks.
15816 This has the effect that if the same variable is declared
15817 (without @code{extern}) in two different compilations,
15818 you will get a multiple-definition error when you link them.
15819 In this case, you must compile with @option{-fcommon} instead.
15820 Compiling with @option{-fno-common} is useful on targets for which
15821 it provides better performance, or if you wish to verify that the
15822 program will work on other systems which always treat uninitialized
15823 variable declarations this way.
15827 Ignore the @samp{#ident} directive.
15829 @item -finhibit-size-directive
15830 @opindex finhibit-size-directive
15831 Don't output a @code{.size} assembler directive, or anything else that
15832 would cause trouble if the function is split in the middle, and the
15833 two halves are placed at locations far apart in memory. This option is
15834 used when compiling @file{crtstuff.c}; you should not need to use it
15837 @item -fverbose-asm
15838 @opindex fverbose-asm
15839 Put extra commentary information in the generated assembly code to
15840 make it more readable. This option is generally only of use to those
15841 who actually need to read the generated assembly code (perhaps while
15842 debugging the compiler itself).
15844 @option{-fno-verbose-asm}, the default, causes the
15845 extra information to be omitted and is useful when comparing two assembler
15848 @item -frecord-gcc-switches
15849 @opindex frecord-gcc-switches
15850 This switch causes the command line that was used to invoke the
15851 compiler to be recorded into the object file that is being created.
15852 This switch is only implemented on some targets and the exact format
15853 of the recording is target and binary file format dependent, but it
15854 usually takes the form of a section containing ASCII text. This
15855 switch is related to the @option{-fverbose-asm} switch, but that
15856 switch only records information in the assembler output file as
15857 comments, so it never reaches the object file.
15861 @cindex global offset table
15863 Generate position-independent code (PIC) suitable for use in a shared
15864 library, if supported for the target machine. Such code accesses all
15865 constant addresses through a global offset table (GOT)@. The dynamic
15866 loader resolves the GOT entries when the program starts (the dynamic
15867 loader is not part of GCC; it is part of the operating system). If
15868 the GOT size for the linked executable exceeds a machine-specific
15869 maximum size, you get an error message from the linker indicating that
15870 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15871 instead. (These maximums are 8k on the SPARC and 32k
15872 on the m68k and RS/6000. The 386 has no such limit.)
15874 Position-independent code requires special support, and therefore works
15875 only on certain machines. For the 386, GCC supports PIC for System V
15876 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
15877 position-independent.
15879 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15884 If supported for the target machine, emit position-independent code,
15885 suitable for dynamic linking and avoiding any limit on the size of the
15886 global offset table. This option makes a difference on the m68k,
15887 PowerPC and SPARC@.
15889 Position-independent code requires special support, and therefore works
15890 only on certain machines.
15892 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15899 These options are similar to @option{-fpic} and @option{-fPIC}, but
15900 generated position independent code can be only linked into executables.
15901 Usually these options are used when @option{-pie} GCC option will be
15902 used during linking.
15904 @option{-fpie} and @option{-fPIE} both define the macros
15905 @code{__pie__} and @code{__PIE__}. The macros have the value 1
15906 for @option{-fpie} and 2 for @option{-fPIE}.
15908 @item -fno-jump-tables
15909 @opindex fno-jump-tables
15910 Do not use jump tables for switch statements even where it would be
15911 more efficient than other code generation strategies. This option is
15912 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15913 building code which forms part of a dynamic linker and cannot
15914 reference the address of a jump table. On some targets, jump tables
15915 do not require a GOT and this option is not needed.
15917 @item -ffixed-@var{reg}
15919 Treat the register named @var{reg} as a fixed register; generated code
15920 should never refer to it (except perhaps as a stack pointer, frame
15921 pointer or in some other fixed role).
15923 @var{reg} must be the name of a register. The register names accepted
15924 are machine-specific and are defined in the @code{REGISTER_NAMES}
15925 macro in the machine description macro file.
15927 This flag does not have a negative form, because it specifies a
15930 @item -fcall-used-@var{reg}
15931 @opindex fcall-used
15932 Treat the register named @var{reg} as an allocable register that is
15933 clobbered by function calls. It may be allocated for temporaries or
15934 variables that do not live across a call. Functions compiled this way
15935 will not save and restore the register @var{reg}.
15937 It is an error to used this flag with the frame pointer or stack pointer.
15938 Use of this flag for other registers that have fixed pervasive roles in
15939 the machine's execution model will produce disastrous results.
15941 This flag does not have a negative form, because it specifies a
15944 @item -fcall-saved-@var{reg}
15945 @opindex fcall-saved
15946 Treat the register named @var{reg} as an allocable register saved by
15947 functions. It may be allocated even for temporaries or variables that
15948 live across a call. Functions compiled this way will save and restore
15949 the register @var{reg} if they use it.
15951 It is an error to used this flag with the frame pointer or stack pointer.
15952 Use of this flag for other registers that have fixed pervasive roles in
15953 the machine's execution model will produce disastrous results.
15955 A different sort of disaster will result from the use of this flag for
15956 a register in which function values may be returned.
15958 This flag does not have a negative form, because it specifies a
15961 @item -fpack-struct[=@var{n}]
15962 @opindex fpack-struct
15963 Without a value specified, pack all structure members together without
15964 holes. When a value is specified (which must be a small power of two), pack
15965 structure members according to this value, representing the maximum
15966 alignment (that is, objects with default alignment requirements larger than
15967 this will be output potentially unaligned at the next fitting location.
15969 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15970 code that is not binary compatible with code generated without that switch.
15971 Additionally, it makes the code suboptimal.
15972 Use it to conform to a non-default application binary interface.
15974 @item -finstrument-functions
15975 @opindex finstrument-functions
15976 Generate instrumentation calls for entry and exit to functions. Just
15977 after function entry and just before function exit, the following
15978 profiling functions will be called with the address of the current
15979 function and its call site. (On some platforms,
15980 @code{__builtin_return_address} does not work beyond the current
15981 function, so the call site information may not be available to the
15982 profiling functions otherwise.)
15985 void __cyg_profile_func_enter (void *this_fn,
15987 void __cyg_profile_func_exit (void *this_fn,
15991 The first argument is the address of the start of the current function,
15992 which may be looked up exactly in the symbol table.
15994 This instrumentation is also done for functions expanded inline in other
15995 functions. The profiling calls will indicate where, conceptually, the
15996 inline function is entered and exited. This means that addressable
15997 versions of such functions must be available. If all your uses of a
15998 function are expanded inline, this may mean an additional expansion of
15999 code size. If you use @samp{extern inline} in your C code, an
16000 addressable version of such functions must be provided. (This is
16001 normally the case anyways, but if you get lucky and the optimizer always
16002 expands the functions inline, you might have gotten away without
16003 providing static copies.)
16005 A function may be given the attribute @code{no_instrument_function}, in
16006 which case this instrumentation will not be done. This can be used, for
16007 example, for the profiling functions listed above, high-priority
16008 interrupt routines, and any functions from which the profiling functions
16009 cannot safely be called (perhaps signal handlers, if the profiling
16010 routines generate output or allocate memory).
16012 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
16013 @opindex finstrument-functions-exclude-file-list
16015 Set the list of functions that are excluded from instrumentation (see
16016 the description of @code{-finstrument-functions}). If the file that
16017 contains a function definition matches with one of @var{file}, then
16018 that function is not instrumented. The match is done on substrings:
16019 if the @var{file} parameter is a substring of the file name, it is
16020 considered to be a match.
16023 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
16024 will exclude any inline function defined in files whose pathnames
16025 contain @code{/bits/stl} or @code{include/sys}.
16027 If, for some reason, you want to include letter @code{','} in one of
16028 @var{sym}, write @code{'\,'}. For example,
16029 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
16030 (note the single quote surrounding the option).
16032 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
16033 @opindex finstrument-functions-exclude-function-list
16035 This is similar to @code{-finstrument-functions-exclude-file-list},
16036 but this option sets the list of function names to be excluded from
16037 instrumentation. The function name to be matched is its user-visible
16038 name, such as @code{vector<int> blah(const vector<int> &)}, not the
16039 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
16040 match is done on substrings: if the @var{sym} parameter is a substring
16041 of the function name, it is considered to be a match.
16043 @item -fstack-check
16044 @opindex fstack-check
16045 Generate code to verify that you do not go beyond the boundary of the
16046 stack. You should specify this flag if you are running in an
16047 environment with multiple threads, but only rarely need to specify it in
16048 a single-threaded environment since stack overflow is automatically
16049 detected on nearly all systems if there is only one stack.
16051 Note that this switch does not actually cause checking to be done; the
16052 operating system or the language runtime must do that. The switch causes
16053 generation of code to ensure that they see the stack being extended.
16055 You can additionally specify a string parameter: @code{no} means no
16056 checking, @code{generic} means force the use of old-style checking,
16057 @code{specific} means use the best checking method and is equivalent
16058 to bare @option{-fstack-check}.
16060 Old-style checking is a generic mechanism that requires no specific
16061 target support in the compiler but comes with the following drawbacks:
16065 Modified allocation strategy for large objects: they will always be
16066 allocated dynamically if their size exceeds a fixed threshold.
16069 Fixed limit on the size of the static frame of functions: when it is
16070 topped by a particular function, stack checking is not reliable and
16071 a warning is issued by the compiler.
16074 Inefficiency: because of both the modified allocation strategy and the
16075 generic implementation, the performances of the code are hampered.
16078 Note that old-style stack checking is also the fallback method for
16079 @code{specific} if no target support has been added in the compiler.
16081 @item -fstack-limit-register=@var{reg}
16082 @itemx -fstack-limit-symbol=@var{sym}
16083 @itemx -fno-stack-limit
16084 @opindex fstack-limit-register
16085 @opindex fstack-limit-symbol
16086 @opindex fno-stack-limit
16087 Generate code to ensure that the stack does not grow beyond a certain value,
16088 either the value of a register or the address of a symbol. If the stack
16089 would grow beyond the value, a signal is raised. For most targets,
16090 the signal is raised before the stack overruns the boundary, so
16091 it is possible to catch the signal without taking special precautions.
16093 For instance, if the stack starts at absolute address @samp{0x80000000}
16094 and grows downwards, you can use the flags
16095 @option{-fstack-limit-symbol=__stack_limit} and
16096 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
16097 of 128KB@. Note that this may only work with the GNU linker.
16099 @cindex aliasing of parameters
16100 @cindex parameters, aliased
16101 @item -fargument-alias
16102 @itemx -fargument-noalias
16103 @itemx -fargument-noalias-global
16104 @itemx -fargument-noalias-anything
16105 @opindex fargument-alias
16106 @opindex fargument-noalias
16107 @opindex fargument-noalias-global
16108 @opindex fargument-noalias-anything
16109 Specify the possible relationships among parameters and between
16110 parameters and global data.
16112 @option{-fargument-alias} specifies that arguments (parameters) may
16113 alias each other and may alias global storage.@*
16114 @option{-fargument-noalias} specifies that arguments do not alias
16115 each other, but may alias global storage.@*
16116 @option{-fargument-noalias-global} specifies that arguments do not
16117 alias each other and do not alias global storage.
16118 @option{-fargument-noalias-anything} specifies that arguments do not
16119 alias any other storage.
16121 Each language will automatically use whatever option is required by
16122 the language standard. You should not need to use these options yourself.
16124 @item -fleading-underscore
16125 @opindex fleading-underscore
16126 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
16127 change the way C symbols are represented in the object file. One use
16128 is to help link with legacy assembly code.
16130 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
16131 generate code that is not binary compatible with code generated without that
16132 switch. Use it to conform to a non-default application binary interface.
16133 Not all targets provide complete support for this switch.
16135 @item -ftls-model=@var{model}
16136 @opindex ftls-model
16137 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
16138 The @var{model} argument should be one of @code{global-dynamic},
16139 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
16141 The default without @option{-fpic} is @code{initial-exec}; with
16142 @option{-fpic} the default is @code{global-dynamic}.
16144 @item -fvisibility=@var{default|internal|hidden|protected}
16145 @opindex fvisibility
16146 Set the default ELF image symbol visibility to the specified option---all
16147 symbols will be marked with this unless overridden within the code.
16148 Using this feature can very substantially improve linking and
16149 load times of shared object libraries, produce more optimized
16150 code, provide near-perfect API export and prevent symbol clashes.
16151 It is @strong{strongly} recommended that you use this in any shared objects
16154 Despite the nomenclature, @code{default} always means public ie;
16155 available to be linked against from outside the shared object.
16156 @code{protected} and @code{internal} are pretty useless in real-world
16157 usage so the only other commonly used option will be @code{hidden}.
16158 The default if @option{-fvisibility} isn't specified is
16159 @code{default}, i.e., make every
16160 symbol public---this causes the same behavior as previous versions of
16163 A good explanation of the benefits offered by ensuring ELF
16164 symbols have the correct visibility is given by ``How To Write
16165 Shared Libraries'' by Ulrich Drepper (which can be found at
16166 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
16167 solution made possible by this option to marking things hidden when
16168 the default is public is to make the default hidden and mark things
16169 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
16170 and @code{__attribute__ ((visibility("default")))} instead of
16171 @code{__declspec(dllexport)} you get almost identical semantics with
16172 identical syntax. This is a great boon to those working with
16173 cross-platform projects.
16175 For those adding visibility support to existing code, you may find
16176 @samp{#pragma GCC visibility} of use. This works by you enclosing
16177 the declarations you wish to set visibility for with (for example)
16178 @samp{#pragma GCC visibility push(hidden)} and
16179 @samp{#pragma GCC visibility pop}.
16180 Bear in mind that symbol visibility should be viewed @strong{as
16181 part of the API interface contract} and thus all new code should
16182 always specify visibility when it is not the default ie; declarations
16183 only for use within the local DSO should @strong{always} be marked explicitly
16184 as hidden as so to avoid PLT indirection overheads---making this
16185 abundantly clear also aids readability and self-documentation of the code.
16186 Note that due to ISO C++ specification requirements, operator new and
16187 operator delete must always be of default visibility.
16189 Be aware that headers from outside your project, in particular system
16190 headers and headers from any other library you use, may not be
16191 expecting to be compiled with visibility other than the default. You
16192 may need to explicitly say @samp{#pragma GCC visibility push(default)}
16193 before including any such headers.
16195 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
16196 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
16197 no modifications. However, this means that calls to @samp{extern}
16198 functions with no explicit visibility will use the PLT, so it is more
16199 effective to use @samp{__attribute ((visibility))} and/or
16200 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
16201 declarations should be treated as hidden.
16203 Note that @samp{-fvisibility} does affect C++ vague linkage
16204 entities. This means that, for instance, an exception class that will
16205 be thrown between DSOs must be explicitly marked with default
16206 visibility so that the @samp{type_info} nodes will be unified between
16209 An overview of these techniques, their benefits and how to use them
16210 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
16216 @node Environment Variables
16217 @section Environment Variables Affecting GCC
16218 @cindex environment variables
16220 @c man begin ENVIRONMENT
16221 This section describes several environment variables that affect how GCC
16222 operates. Some of them work by specifying directories or prefixes to use
16223 when searching for various kinds of files. Some are used to specify other
16224 aspects of the compilation environment.
16226 Note that you can also specify places to search using options such as
16227 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
16228 take precedence over places specified using environment variables, which
16229 in turn take precedence over those specified by the configuration of GCC@.
16230 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
16231 GNU Compiler Collection (GCC) Internals}.
16236 @c @itemx LC_COLLATE
16238 @c @itemx LC_MONETARY
16239 @c @itemx LC_NUMERIC
16244 @c @findex LC_COLLATE
16245 @findex LC_MESSAGES
16246 @c @findex LC_MONETARY
16247 @c @findex LC_NUMERIC
16251 These environment variables control the way that GCC uses
16252 localization information that allow GCC to work with different
16253 national conventions. GCC inspects the locale categories
16254 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
16255 so. These locale categories can be set to any value supported by your
16256 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
16257 Kingdom encoded in UTF-8.
16259 The @env{LC_CTYPE} environment variable specifies character
16260 classification. GCC uses it to determine the character boundaries in
16261 a string; this is needed for some multibyte encodings that contain quote
16262 and escape characters that would otherwise be interpreted as a string
16265 The @env{LC_MESSAGES} environment variable specifies the language to
16266 use in diagnostic messages.
16268 If the @env{LC_ALL} environment variable is set, it overrides the value
16269 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
16270 and @env{LC_MESSAGES} default to the value of the @env{LANG}
16271 environment variable. If none of these variables are set, GCC
16272 defaults to traditional C English behavior.
16276 If @env{TMPDIR} is set, it specifies the directory to use for temporary
16277 files. GCC uses temporary files to hold the output of one stage of
16278 compilation which is to be used as input to the next stage: for example,
16279 the output of the preprocessor, which is the input to the compiler
16282 @item GCC_EXEC_PREFIX
16283 @findex GCC_EXEC_PREFIX
16284 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
16285 names of the subprograms executed by the compiler. No slash is added
16286 when this prefix is combined with the name of a subprogram, but you can
16287 specify a prefix that ends with a slash if you wish.
16289 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
16290 an appropriate prefix to use based on the pathname it was invoked with.
16292 If GCC cannot find the subprogram using the specified prefix, it
16293 tries looking in the usual places for the subprogram.
16295 The default value of @env{GCC_EXEC_PREFIX} is
16296 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
16297 the installed compiler. In many cases @var{prefix} is the value
16298 of @code{prefix} when you ran the @file{configure} script.
16300 Other prefixes specified with @option{-B} take precedence over this prefix.
16302 This prefix is also used for finding files such as @file{crt0.o} that are
16305 In addition, the prefix is used in an unusual way in finding the
16306 directories to search for header files. For each of the standard
16307 directories whose name normally begins with @samp{/usr/local/lib/gcc}
16308 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
16309 replacing that beginning with the specified prefix to produce an
16310 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
16311 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
16312 These alternate directories are searched first; the standard directories
16313 come next. If a standard directory begins with the configured
16314 @var{prefix} then the value of @var{prefix} is replaced by
16315 @env{GCC_EXEC_PREFIX} when looking for header files.
16317 @item COMPILER_PATH
16318 @findex COMPILER_PATH
16319 The value of @env{COMPILER_PATH} is a colon-separated list of
16320 directories, much like @env{PATH}. GCC tries the directories thus
16321 specified when searching for subprograms, if it can't find the
16322 subprograms using @env{GCC_EXEC_PREFIX}.
16325 @findex LIBRARY_PATH
16326 The value of @env{LIBRARY_PATH} is a colon-separated list of
16327 directories, much like @env{PATH}. When configured as a native compiler,
16328 GCC tries the directories thus specified when searching for special
16329 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
16330 using GCC also uses these directories when searching for ordinary
16331 libraries for the @option{-l} option (but directories specified with
16332 @option{-L} come first).
16336 @cindex locale definition
16337 This variable is used to pass locale information to the compiler. One way in
16338 which this information is used is to determine the character set to be used
16339 when character literals, string literals and comments are parsed in C and C++.
16340 When the compiler is configured to allow multibyte characters,
16341 the following values for @env{LANG} are recognized:
16345 Recognize JIS characters.
16347 Recognize SJIS characters.
16349 Recognize EUCJP characters.
16352 If @env{LANG} is not defined, or if it has some other value, then the
16353 compiler will use mblen and mbtowc as defined by the default locale to
16354 recognize and translate multibyte characters.
16358 Some additional environments variables affect the behavior of the
16361 @include cppenv.texi
16365 @node Precompiled Headers
16366 @section Using Precompiled Headers
16367 @cindex precompiled headers
16368 @cindex speed of compilation
16370 Often large projects have many header files that are included in every
16371 source file. The time the compiler takes to process these header files
16372 over and over again can account for nearly all of the time required to
16373 build the project. To make builds faster, GCC allows users to
16374 `precompile' a header file; then, if builds can use the precompiled
16375 header file they will be much faster.
16377 To create a precompiled header file, simply compile it as you would any
16378 other file, if necessary using the @option{-x} option to make the driver
16379 treat it as a C or C++ header file. You will probably want to use a
16380 tool like @command{make} to keep the precompiled header up-to-date when
16381 the headers it contains change.
16383 A precompiled header file will be searched for when @code{#include} is
16384 seen in the compilation. As it searches for the included file
16385 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
16386 compiler looks for a precompiled header in each directory just before it
16387 looks for the include file in that directory. The name searched for is
16388 the name specified in the @code{#include} with @samp{.gch} appended. If
16389 the precompiled header file can't be used, it is ignored.
16391 For instance, if you have @code{#include "all.h"}, and you have
16392 @file{all.h.gch} in the same directory as @file{all.h}, then the
16393 precompiled header file will be used if possible, and the original
16394 header will be used otherwise.
16396 Alternatively, you might decide to put the precompiled header file in a
16397 directory and use @option{-I} to ensure that directory is searched
16398 before (or instead of) the directory containing the original header.
16399 Then, if you want to check that the precompiled header file is always
16400 used, you can put a file of the same name as the original header in this
16401 directory containing an @code{#error} command.
16403 This also works with @option{-include}. So yet another way to use
16404 precompiled headers, good for projects not designed with precompiled
16405 header files in mind, is to simply take most of the header files used by
16406 a project, include them from another header file, precompile that header
16407 file, and @option{-include} the precompiled header. If the header files
16408 have guards against multiple inclusion, they will be skipped because
16409 they've already been included (in the precompiled header).
16411 If you need to precompile the same header file for different
16412 languages, targets, or compiler options, you can instead make a
16413 @emph{directory} named like @file{all.h.gch}, and put each precompiled
16414 header in the directory, perhaps using @option{-o}. It doesn't matter
16415 what you call the files in the directory, every precompiled header in
16416 the directory will be considered. The first precompiled header
16417 encountered in the directory that is valid for this compilation will
16418 be used; they're searched in no particular order.
16420 There are many other possibilities, limited only by your imagination,
16421 good sense, and the constraints of your build system.
16423 A precompiled header file can be used only when these conditions apply:
16427 Only one precompiled header can be used in a particular compilation.
16430 A precompiled header can't be used once the first C token is seen. You
16431 can have preprocessor directives before a precompiled header; you can
16432 even include a precompiled header from inside another header, so long as
16433 there are no C tokens before the @code{#include}.
16436 The precompiled header file must be produced for the same language as
16437 the current compilation. You can't use a C precompiled header for a C++
16441 The precompiled header file must have been produced by the same compiler
16442 binary as the current compilation is using.
16445 Any macros defined before the precompiled header is included must
16446 either be defined in the same way as when the precompiled header was
16447 generated, or must not affect the precompiled header, which usually
16448 means that they don't appear in the precompiled header at all.
16450 The @option{-D} option is one way to define a macro before a
16451 precompiled header is included; using a @code{#define} can also do it.
16452 There are also some options that define macros implicitly, like
16453 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
16456 @item If debugging information is output when using the precompiled
16457 header, using @option{-g} or similar, the same kind of debugging information
16458 must have been output when building the precompiled header. However,
16459 a precompiled header built using @option{-g} can be used in a compilation
16460 when no debugging information is being output.
16462 @item The same @option{-m} options must generally be used when building
16463 and using the precompiled header. @xref{Submodel Options},
16464 for any cases where this rule is relaxed.
16466 @item Each of the following options must be the same when building and using
16467 the precompiled header:
16469 @gccoptlist{-fexceptions}
16472 Some other command-line options starting with @option{-f},
16473 @option{-p}, or @option{-O} must be defined in the same way as when
16474 the precompiled header was generated. At present, it's not clear
16475 which options are safe to change and which are not; the safest choice
16476 is to use exactly the same options when generating and using the
16477 precompiled header. The following are known to be safe:
16479 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
16480 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
16481 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
16486 For all of these except the last, the compiler will automatically
16487 ignore the precompiled header if the conditions aren't met. If you
16488 find an option combination that doesn't work and doesn't cause the
16489 precompiled header to be ignored, please consider filing a bug report,
16492 If you do use differing options when generating and using the
16493 precompiled header, the actual behavior will be a mixture of the
16494 behavior for the options. For instance, if you use @option{-g} to
16495 generate the precompiled header but not when using it, you may or may
16496 not get debugging information for routines in the precompiled header.
16498 @node Running Protoize
16499 @section Running Protoize
16501 The program @code{protoize} is an optional part of GCC@. You can use
16502 it to add prototypes to a program, thus converting the program to ISO
16503 C in one respect. The companion program @code{unprotoize} does the
16504 reverse: it removes argument types from any prototypes that are found.
16506 When you run these programs, you must specify a set of source files as
16507 command line arguments. The conversion programs start out by compiling
16508 these files to see what functions they define. The information gathered
16509 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
16511 After scanning comes actual conversion. The specified files are all
16512 eligible to be converted; any files they include (whether sources or
16513 just headers) are eligible as well.
16515 But not all the eligible files are converted. By default,
16516 @code{protoize} and @code{unprotoize} convert only source and header
16517 files in the current directory. You can specify additional directories
16518 whose files should be converted with the @option{-d @var{directory}}
16519 option. You can also specify particular files to exclude with the
16520 @option{-x @var{file}} option. A file is converted if it is eligible, its
16521 directory name matches one of the specified directory names, and its
16522 name within the directory has not been excluded.
16524 Basic conversion with @code{protoize} consists of rewriting most
16525 function definitions and function declarations to specify the types of
16526 the arguments. The only ones not rewritten are those for varargs
16529 @code{protoize} optionally inserts prototype declarations at the
16530 beginning of the source file, to make them available for any calls that
16531 precede the function's definition. Or it can insert prototype
16532 declarations with block scope in the blocks where undeclared functions
16535 Basic conversion with @code{unprotoize} consists of rewriting most
16536 function declarations to remove any argument types, and rewriting
16537 function definitions to the old-style pre-ISO form.
16539 Both conversion programs print a warning for any function declaration or
16540 definition that they can't convert. You can suppress these warnings
16543 The output from @code{protoize} or @code{unprotoize} replaces the
16544 original source file. The original file is renamed to a name ending
16545 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
16546 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
16547 for DOS) file already exists, then the source file is simply discarded.
16549 @code{protoize} and @code{unprotoize} both depend on GCC itself to
16550 scan the program and collect information about the functions it uses.
16551 So neither of these programs will work until GCC is installed.
16553 Here is a table of the options you can use with @code{protoize} and
16554 @code{unprotoize}. Each option works with both programs unless
16558 @item -B @var{directory}
16559 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
16560 usual directory (normally @file{/usr/local/lib}). This file contains
16561 prototype information about standard system functions. This option
16562 applies only to @code{protoize}.
16564 @item -c @var{compilation-options}
16565 Use @var{compilation-options} as the options when running @command{gcc} to
16566 produce the @samp{.X} files. The special option @option{-aux-info} is
16567 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
16569 Note that the compilation options must be given as a single argument to
16570 @code{protoize} or @code{unprotoize}. If you want to specify several
16571 @command{gcc} options, you must quote the entire set of compilation options
16572 to make them a single word in the shell.
16574 There are certain @command{gcc} arguments that you cannot use, because they
16575 would produce the wrong kind of output. These include @option{-g},
16576 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
16577 the @var{compilation-options}, they are ignored.
16580 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
16581 systems) instead of @samp{.c}. This is convenient if you are converting
16582 a C program to C++. This option applies only to @code{protoize}.
16585 Add explicit global declarations. This means inserting explicit
16586 declarations at the beginning of each source file for each function
16587 that is called in the file and was not declared. These declarations
16588 precede the first function definition that contains a call to an
16589 undeclared function. This option applies only to @code{protoize}.
16591 @item -i @var{string}
16592 Indent old-style parameter declarations with the string @var{string}.
16593 This option applies only to @code{protoize}.
16595 @code{unprotoize} converts prototyped function definitions to old-style
16596 function definitions, where the arguments are declared between the
16597 argument list and the initial @samp{@{}. By default, @code{unprotoize}
16598 uses five spaces as the indentation. If you want to indent with just
16599 one space instead, use @option{-i " "}.
16602 Keep the @samp{.X} files. Normally, they are deleted after conversion
16606 Add explicit local declarations. @code{protoize} with @option{-l} inserts
16607 a prototype declaration for each function in each block which calls the
16608 function without any declaration. This option applies only to
16612 Make no real changes. This mode just prints information about the conversions
16613 that would have been done without @option{-n}.
16616 Make no @samp{.save} files. The original files are simply deleted.
16617 Use this option with caution.
16619 @item -p @var{program}
16620 Use the program @var{program} as the compiler. Normally, the name
16621 @file{gcc} is used.
16624 Work quietly. Most warnings are suppressed.
16627 Print the version number, just like @option{-v} for @command{gcc}.
16630 If you need special compiler options to compile one of your program's
16631 source files, then you should generate that file's @samp{.X} file
16632 specially, by running @command{gcc} on that source file with the
16633 appropriate options and the option @option{-aux-info}. Then run
16634 @code{protoize} on the entire set of files. @code{protoize} will use
16635 the existing @samp{.X} file because it is newer than the source file.
16639 gcc -Dfoo=bar file1.c -aux-info file1.X
16644 You need to include the special files along with the rest in the
16645 @code{protoize} command, even though their @samp{.X} files already
16646 exist, because otherwise they won't get converted.
16648 @xref{Protoize Caveats}, for more information on how to use
16649 @code{protoize} successfully.